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 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
8 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
9 Alan Hayward <alan.hayward@arm.com>
10 David Sherwood <david.sherwood@arm.com>
12 * match.pd: Cope with polynomial numbers of vector elements.
14 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15 Alan Hayward <alan.hayward@arm.com>
16 David Sherwood <david.sherwood@arm.com>
18 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
19 in a POINTER_PLUS_EXPR.
21 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22 Alan Hayward <alan.hayward@arm.com>
23 David Sherwood <david.sherwood@arm.com>
25 * omp-simd-clone.c (simd_clone_subparts): New function.
26 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
27 (ipa_simd_modify_function_body): Likewise.
29 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30 Alan Hayward <alan.hayward@arm.com>
31 David Sherwood <david.sherwood@arm.com>
33 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
34 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
35 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
36 (expand_vector_condition, vector_element): Likewise.
37 (subparts_gt): New function.
38 (get_compute_type): Use subparts_gt.
39 (count_type_subparts): Delete.
40 (expand_vector_operations_1): Use subparts_gt instead of
43 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
44 Alan Hayward <alan.hayward@arm.com>
45 David Sherwood <david.sherwood@arm.com>
47 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
48 (vect_compile_time_alias): ...this new function. Do the calculation
49 on poly_ints rather than trees.
50 (vect_prune_runtime_alias_test_list): Update call accordingly.
52 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
53 Alan Hayward <alan.hayward@arm.com>
54 David Sherwood <david.sherwood@arm.com>
56 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
58 (vect_schedule_slp_instance): Likewise.
60 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
61 Alan Hayward <alan.hayward@arm.com>
62 David Sherwood <david.sherwood@arm.com>
64 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
65 constant and extern definitions for variable-length vectors.
66 (vect_get_constant_vectors): Note that the number of units
67 is known to be constant.
69 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
70 Alan Hayward <alan.hayward@arm.com>
71 David Sherwood <david.sherwood@arm.com>
73 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
74 of units as polynomial. Choose between WIDE and NARROW based
77 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
78 Alan Hayward <alan.hayward@arm.com>
79 David Sherwood <david.sherwood@arm.com>
81 * tree-vect-stmts.c (simd_clone_subparts): New function.
82 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
84 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
85 Alan Hayward <alan.hayward@arm.com>
86 David Sherwood <david.sherwood@arm.com>
88 * tree-vect-stmts.c (vectorizable_call): Treat the number of
89 vectors as polynomial. Use build_index_vector for
92 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
93 Alan Hayward <alan.hayward@arm.com>
94 David Sherwood <david.sherwood@arm.com>
96 * tree-vect-stmts.c (get_load_store_type): Treat the number of
97 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
98 for variable-length vectors.
99 (vectorizable_mask_load_store): Treat the number of units as
100 polynomial, asserting that it is constant if the condition has
101 already been enforced.
102 (vectorizable_store, vectorizable_load): Likewise.
104 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
105 Alan Hayward <alan.hayward@arm.com>
106 David Sherwood <david.sherwood@arm.com>
108 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
109 of units as polynomial. Punt if we can't tell at compile time
110 which vector contains the final result.
112 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
113 Alan Hayward <alan.hayward@arm.com>
114 David Sherwood <david.sherwood@arm.com>
116 * tree-vect-loop.c (vectorizable_induction): Treat the number
117 of units as polynomial. Punt on SLP inductions. Use an integer
118 VEC_SERIES_EXPR for variable-length integer reductions. Use a
119 cast of such a series for variable-length floating-point
122 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
123 Alan Hayward <alan.hayward@arm.com>
124 David Sherwood <david.sherwood@arm.com>
126 * tree.h (build_index_vector): Declare.
127 * tree.c (build_index_vector): New function.
128 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
129 of units as polynomial, forcibly converting it to a constant if
130 vectorizable_reduction has already enforced the condition.
131 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
132 to create a {1,2,3,...} vector.
133 (vectorizable_reduction): Treat the number of units as polynomial.
134 Choose vectype_in based on the largest scalar element size rather
135 than the smallest number of units. Enforce the restrictions
138 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
139 Alan Hayward <alan.hayward@arm.com>
140 David Sherwood <david.sherwood@arm.com>
142 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
143 number of units as polynomial.
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 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
150 * target.def (autovectorize_vector_sizes): Return the vector sizes
151 by pointer, using vector_sizes rather than a bitmask.
152 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
153 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
154 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
156 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
157 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
158 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
159 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
160 * omp-general.c (omp_max_vf): Likewise.
161 * omp-low.c (omp_clause_aligned_alignment): Likewise.
162 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
163 * tree-vect-loop.c (vect_analyze_loop): Likewise.
164 * tree-vect-slp.c (vect_slp_bb): Likewise.
165 * doc/tm.texi: Regenerate.
166 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
168 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
169 the vector size as a poly_uint64 rather than an unsigned int.
170 (current_vector_size): Change from an unsigned int to a poly_uint64.
171 (get_vectype_for_scalar_type): Update accordingly.
172 * tree.h (build_truth_vector_type): Take the size and number of
173 units as a poly_uint64 rather than an unsigned int.
174 (build_vector_type): Add a temporary overload that takes
175 the number of units as a poly_uint64 rather than an unsigned int.
176 * tree.c (make_vector_type): Likewise.
177 (build_truth_vector_type): Take the number of units as a poly_uint64
178 rather than an unsigned int.
180 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
181 Alan Hayward <alan.hayward@arm.com>
182 David Sherwood <david.sherwood@arm.com>
184 * target.def (get_mask_mode): Take the number of units and length
185 as poly_uint64s rather than unsigned ints.
186 * targhooks.h (default_get_mask_mode): Update accordingly.
187 * targhooks.c (default_get_mask_mode): Likewise.
188 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
189 * doc/tm.texi: Regenerate.
191 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
192 Alan Hayward <alan.hayward@arm.com>
193 David Sherwood <david.sherwood@arm.com>
195 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
196 * omp-general.c (omp_max_vf): Likewise.
197 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
198 (expand_omp_simd): Handle polynomial safelen.
199 * omp-low.c (omplow_simd_context): Add a default constructor.
200 (omplow_simd_context::max_vf): Change from int to poly_uint64.
201 (lower_rec_simd_input_clauses): Update accordingly.
202 (lower_rec_input_clauses): Likewise.
204 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
205 Alan Hayward <alan.hayward@arm.com>
206 David Sherwood <david.sherwood@arm.com>
208 * tree-vectorizer.h (vect_nunits_for_cost): New function.
209 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
210 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
211 (vect_analyze_slp_cost): Likewise.
212 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
213 (vect_model_load_cost): Likewise.
215 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
216 Alan Hayward <alan.hayward@arm.com>
217 David Sherwood <david.sherwood@arm.com>
219 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
220 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
221 from an unsigned int * to a poly_uint64_pod *.
222 (calculate_unrolling_factor): New function.
223 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
225 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
226 Alan Hayward <alan.hayward@arm.com>
227 David Sherwood <david.sherwood@arm.com>
229 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
230 from an unsigned int to a poly_uint64.
231 (_loop_vec_info::slp_unrolling_factor): Likewise.
232 (_loop_vec_info::vectorization_factor): Change from an int
234 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
235 (vect_get_num_vectors): New function.
236 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
237 (vect_get_num_copies): Use vect_get_num_vectors.
238 (vect_analyze_data_ref_dependences): Change max_vf from an int *
239 to an unsigned int *.
240 (vect_analyze_data_refs): Change min_vf from an int * to a
242 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
243 than an unsigned HOST_WIDE_INT.
244 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
245 (vect_analyze_data_ref_dependence): Change max_vf from an int *
246 to an unsigned int *.
247 (vect_analyze_data_ref_dependences): Likewise.
248 (vect_compute_data_ref_alignment): Handle polynomial vf.
249 (vect_enhance_data_refs_alignment): Likewise.
250 (vect_prune_runtime_alias_test_list): Likewise.
251 (vect_shift_permute_load_chain): Likewise.
252 (vect_supportable_dr_alignment): Likewise.
253 (dependence_distance_ge_vf): Take the vectorization factor as a
254 poly_uint64 rather than an unsigned HOST_WIDE_INT.
255 (vect_analyze_data_refs): Change min_vf from an int * to a
257 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
258 vfm1 as a poly_uint64 rather than an int. Make the same change
259 for the returned bound_scalar.
260 (vect_gen_vector_loop_niters): Handle polynomial vf.
261 (vect_do_peeling): Likewise. Update call to
262 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
263 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
265 * tree-vect-loop.c (vect_determine_vectorization_factor)
266 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
267 (vect_get_known_peeling_cost): Likewise.
268 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
269 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
270 (vect_transform_loop): Likewise. Use the lowest possible VF when
271 updating the upper bounds of the loop.
272 (vect_min_worthwhile_factor): Make static. Return an unsigned int
274 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
275 polynomial unroll factors.
276 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
277 (vect_make_slp_decision): Likewise.
278 (vect_supported_load_permutation_p): Likewise, and polynomial
280 (vect_analyze_slp_cost): Handle polynomial vf.
281 (vect_slp_analyze_node_operations): Likewise.
282 (vect_slp_analyze_bb_1): Likewise.
283 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
284 than an unsigned HOST_WIDE_INT.
285 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
286 (vectorizable_load): Handle polynomial vf.
287 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
289 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
291 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
292 Alan Hayward <alan.hayward@arm.com>
293 David Sherwood <david.sherwood@arm.com>
295 * match.pd: Handle bit operations involving three constants
296 and try to fold one pair.
298 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
300 * tree-vect-loop-manip.c: Include gimple-fold.h.
301 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
302 niters_maybe_zero parameters. Handle other cases besides a step of 1.
303 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
304 Add a path that uses a step of VF instead of 1, but disable it
306 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
307 and niters_no_overflow parameters. Update calls to
308 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
309 Create a new SSA name if the latter choses to use a ste other
310 than zero, and return it via niters_vector_mult_vf_var.
311 * tree-vect-loop.c (vect_transform_loop): Update calls to
312 vect_do_peeling, vect_gen_vector_loop_niters and
313 slpeel_make_loop_iterate_ntimes.
314 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
315 (vect_gen_vector_loop_niters): Update declarations after above changes.
317 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
319 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
320 128-bit round to integer instructions.
321 (ceil<mode>2): Likewise.
322 (btrunc<mode>2): Likewise.
323 (round<mode>2): Likewise.
325 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
327 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
328 unaligned VSX load/store on P8/P9.
329 (expand_block_clear): Allow the use of unaligned VSX
332 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
334 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
336 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
337 swap associated with both a load and a store.
339 2018-01-02 Andrew Waterman <andrew@sifive.com>
341 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
342 * config/riscv/riscv.md (clear_cache): Use it.
344 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
346 * web.c: Remove out-of-date comment.
348 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
350 * expr.c (fixup_args_size_notes): Check that any existing
351 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
352 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
353 (emit_single_push_insn): ...here.
355 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
357 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
358 (const_vector_encoded_nelts): New function.
359 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
360 (const_vector_int_elt, const_vector_elt): Declare.
361 * emit-rtl.c (const_vector_int_elt_1): New function.
362 (const_vector_elt): Likewise.
363 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
366 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
368 * expr.c: Include rtx-vector-builder.h.
369 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
370 directly on the tree encoding.
371 (const_vector_from_tree): Likewise.
372 * optabs.c: Include rtx-vector-builder.h.
373 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
374 sequence of "u" values.
375 * vec-perm-indices.c: Include rtx-vector-builder.h.
376 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
377 directly on the vec_perm_indices encoding.
379 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
381 * doc/rtl.texi (const_vector): Describe new encoding scheme.
382 * Makefile.in (OBJS): Add rtx-vector-builder.o.
383 * rtx-vector-builder.h: New file.
384 * rtx-vector-builder.c: Likewise.
385 * rtl.h (rtx_def::u2): Add a const_vector field.
386 (CONST_VECTOR_NPATTERNS): New macro.
387 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
388 (CONST_VECTOR_DUPLICATE_P): Likewise.
389 (CONST_VECTOR_STEPPED_P): Likewise.
390 (CONST_VECTOR_ENCODED_ELT): Likewise.
391 (const_vec_duplicate_p): Check for a duplicated vector encoding.
392 (unwrap_const_vec_duplicate): Likewise.
393 (const_vec_series_p): Check for a non-duplicated vector encoding.
394 Say that the function only returns true for integer vectors.
395 * emit-rtl.c: Include rtx-vector-builder.h.
396 (gen_const_vec_duplicate_1): Delete.
397 (gen_const_vector): Call gen_const_vec_duplicate instead of
398 gen_const_vec_duplicate_1.
399 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
400 (gen_const_vec_duplicate): Use rtx_vector_builder.
401 (gen_const_vec_series): Likewise.
402 (gen_rtx_CONST_VECTOR): Likewise.
403 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
404 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
405 Build a new vector rather than modifying a CONST_VECTOR in-place.
406 (handle_special_swappables): Update call accordingly.
407 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
408 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
409 Build a new vector rather than modifying a CONST_VECTOR in-place.
410 (handle_special_swappables): Update call accordingly.
412 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
414 * simplify-rtx.c (simplify_const_binary_operation): Use
415 CONST_VECTOR_ELT instead of XVECEXP.
417 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
419 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
420 the selector elements to be different from the data elements
421 if the selector is a VECTOR_CST.
422 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
423 ssizetype for the selector.
425 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
427 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
428 before testing each element individually.
429 * tree-vect-generic.c (lower_vec_perm): Likewise.
431 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
433 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
434 * selftest-run-tests.c (selftest::run_tests): Call it.
435 * vector-builder.h (vector_builder::operator ==): New function.
436 (vector_builder::operator !=): Likewise.
437 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
438 (vec_perm_indices::all_from_input_p): New function.
439 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
440 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
441 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
442 instead of reading the VECTOR_CST directly. Detect whether both
443 vector inputs are the same before constructing the vec_perm_indices,
444 and update the number of inputs argument accordingly. Use the
445 utility functions added above. Only construct sel2 if we need to.
447 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
449 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
450 the broadcast of the low byte.
451 (expand_mult_highpart): Use an explicit encoding for the permutes.
452 * optabs-query.c (can_mult_highpart_p): Likewise.
453 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
454 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
455 (vectorizable_bswap): Likewise.
456 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
457 explicit encoding for the power-of-2 permutes.
458 (vect_permute_store_chain): Likewise.
459 (vect_grouped_load_supported): Likewise.
460 (vect_permute_load_chain): Likewise.
462 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
464 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
465 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
466 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
467 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
468 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
469 (vect_gen_perm_mask_any): Likewise.
471 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
473 * int-vector-builder.h: New file.
474 * vec-perm-indices.h: Include int-vector-builder.h.
475 (vec_perm_indices): Redefine as an int_vector_builder.
476 (auto_vec_perm_indices): Delete.
477 (vec_perm_builder): Redefine as a stand-alone class.
478 (vec_perm_indices::vec_perm_indices): New function.
479 (vec_perm_indices::clamp): Likewise.
480 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
481 (vec_perm_indices::new_vector): New function.
482 (vec_perm_indices::new_expanded_vector): Update for new
483 vec_perm_indices class.
484 (vec_perm_indices::rotate_inputs): New function.
485 (vec_perm_indices::all_in_range_p): Operate directly on the
486 encoded form, without computing elided elements.
487 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
488 encoding. Update for new vec_perm_indices class.
489 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
490 the given vec_perm_builder.
491 (expand_vec_perm_var): Update vec_perm_builder constructor.
492 (expand_mult_highpart): Use vec_perm_builder instead of
493 auto_vec_perm_indices.
494 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
495 vec_perm_indices instead of auto_vec_perm_indices. Use a single
496 or double series encoding as appropriate.
497 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
498 vec_perm_indices instead of auto_vec_perm_indices.
499 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
500 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
501 (vect_permute_store_chain): Likewise.
502 (vect_grouped_load_supported): Likewise.
503 (vect_permute_load_chain): Likewise.
504 (vect_shift_permute_load_chain): Likewise.
505 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
506 (vect_transform_slp_perm_load): Likewise.
507 (vect_schedule_slp_instance): Likewise.
508 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
509 (vectorizable_mask_load_store): Likewise.
510 (vectorizable_bswap): Likewise.
511 (vectorizable_store): Likewise.
512 (vectorizable_load): Likewise.
513 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
514 vec_perm_indices instead of auto_vec_perm_indices. Use
515 tree_to_vec_perm_builder to read the vector from a tree.
516 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
517 vec_perm_builder instead of a vec_perm_indices.
518 (have_whole_vector_shift): Use vec_perm_builder and
519 vec_perm_indices instead of auto_vec_perm_indices. Leave the
520 truncation to calc_vec_perm_mask_for_shift.
521 (vect_create_epilog_for_reduction): Likewise.
522 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
523 from auto_vec_perm_indices to vec_perm_indices.
524 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
525 instead of changing individual elements.
526 (aarch64_vectorize_vec_perm_const): Use new_vector to install
527 the vector in d.perm.
528 * config/arm/arm.c (expand_vec_perm_d::perm): Change
529 from auto_vec_perm_indices to vec_perm_indices.
530 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
531 instead of changing individual elements.
532 (arm_vectorize_vec_perm_const): Use new_vector to install
533 the vector in d.perm.
534 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
535 Update vec_perm_builder constructor.
536 (rs6000_expand_interleave): Likewise.
537 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
538 (rs6000_expand_interleave): Likewise.
540 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
542 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
543 to qimode could truncate the indices.
544 * optabs.c (expand_vec_perm_var): Likewise.
546 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
548 * Makefile.in (OBJS): Add vec-perm-indices.o.
549 * vec-perm-indices.h: New file.
550 * vec-perm-indices.c: Likewise.
551 * target.h (vec_perm_indices): Replace with a forward class
553 (auto_vec_perm_indices): Move to vec-perm-indices.h.
554 * optabs.h: Include vec-perm-indices.h.
555 (expand_vec_perm): Delete.
556 (selector_fits_mode_p, expand_vec_perm_var): Declare.
557 (expand_vec_perm_const): Declare.
558 * target.def (vec_perm_const_ok): Replace with...
559 (vec_perm_const): ...this new hook.
560 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
561 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
562 * doc/tm.texi: Regenerate.
563 * optabs.def (vec_perm_const): Delete.
564 * doc/md.texi (vec_perm_const): Likewise.
565 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
566 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
567 expand_vec_perm for constant permutation vectors. Assert that
568 the mode of variable permutation vectors is the integer equivalent
569 of the mode that is being permuted.
570 * optabs-query.h (selector_fits_mode_p): Declare.
571 * optabs-query.c: Include vec-perm-indices.h.
572 (selector_fits_mode_p): New function.
573 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
574 is defined, instead of checking whether the vec_perm_const_optab
575 exists. Use targetm.vectorize.vec_perm_const instead of
576 targetm.vectorize.vec_perm_const_ok. Check whether the indices
577 fit in the vector mode before using a variable permute.
578 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
579 vec_perm_indices instead of an rtx.
580 (expand_vec_perm): Replace with...
581 (expand_vec_perm_const): ...this new function. Take the selector
582 as a vec_perm_indices rather than an rtx. Also take the mode of
583 the selector. Update call to shift_amt_for_vec_perm_mask.
584 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
585 Use vec_perm_indices::new_expanded_vector to expand the original
586 selector into bytes. Check whether the indices fit in the vector
587 mode before using a variable permute.
588 (expand_vec_perm_var): Make global.
589 (expand_mult_highpart): Use expand_vec_perm_const.
590 * fold-const.c: Includes vec-perm-indices.h.
591 * tree-ssa-forwprop.c: Likewise.
592 * tree-vect-data-refs.c: Likewise.
593 * tree-vect-generic.c: Likewise.
594 * tree-vect-loop.c: Likewise.
595 * tree-vect-slp.c: Likewise.
596 * tree-vect-stmts.c: Likewise.
597 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
599 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
600 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
601 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
602 (aarch64_vectorize_vec_perm_const): ...this new function.
603 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
604 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
605 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
606 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
607 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
608 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
609 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
611 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
612 check for NEON modes.
613 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
614 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
615 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
616 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
618 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
619 the old VEC_PERM_CONST conditions.
620 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
621 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
622 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
623 (ia64_vectorize_vec_perm_const_ok): Merge into...
624 (ia64_vectorize_vec_perm_const): ...this new function.
625 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
626 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
627 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
628 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
629 * config/mips/mips.c (mips_expand_vec_perm_const)
630 (mips_vectorize_vec_perm_const_ok): Merge into...
631 (mips_vectorize_vec_perm_const): ...this new function.
632 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
633 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
634 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
635 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
636 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
637 (rs6000_expand_vec_perm_const): Delete.
638 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
640 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
641 (altivec_expand_vec_perm_const_le): Take each operand individually.
642 Operate on constant selectors rather than rtxes.
643 (altivec_expand_vec_perm_const): Likewise. Update call to
644 altivec_expand_vec_perm_const_le.
645 (rs6000_expand_vec_perm_const): Delete.
646 (rs6000_vectorize_vec_perm_const_ok): Delete.
647 (rs6000_vectorize_vec_perm_const): New function.
648 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
649 an element count and rtx array.
650 (rs6000_expand_extract_even): Update call accordingly.
651 (rs6000_expand_interleave): Likewise.
652 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
653 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
654 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
655 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
656 (rs6000_expand_vec_perm_const): Delete.
657 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
658 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
659 (altivec_expand_vec_perm_const_le): Take each operand individually.
660 Operate on constant selectors rather than rtxes.
661 (altivec_expand_vec_perm_const): Likewise. Update call to
662 altivec_expand_vec_perm_const_le.
663 (rs6000_expand_vec_perm_const): Delete.
664 (rs6000_vectorize_vec_perm_const_ok): Delete.
665 (rs6000_vectorize_vec_perm_const): New function. Remove stray
666 reference to the SPE evmerge intructions.
667 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
668 an element count and rtx array.
669 (rs6000_expand_extract_even): Update call accordingly.
670 (rs6000_expand_interleave): Likewise.
671 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
672 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
674 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
676 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
678 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
679 vector mode and that that mode matches the mode of the data
681 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
682 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
683 directly using expand_vec_perm_1 when forcing selectors into
685 (expand_vec_perm_var): New function, split out from expand_vec_perm.
687 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
689 * optabs-query.h (can_vec_perm_p): Delete.
690 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
691 * optabs-query.c (can_vec_perm_p): Split into...
692 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
693 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
694 particular selector is valid.
695 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
696 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
697 (vect_grouped_load_supported): Likewise.
698 (vect_shift_permute_load_chain): Likewise.
699 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
700 (vect_transform_slp_perm_load): Likewise.
701 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
702 (vectorizable_bswap): Likewise.
703 (vect_gen_perm_mask_checked): Likewise.
704 * fold-const.c (fold_ternary_loc): Likewise. Don't take
705 implementations of variable permutation vectors into account
706 when deciding which selector to use.
707 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
708 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
709 with a false third argument.
710 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
711 to test whether the constant selector is valid and can_vec_perm_var_p
712 to test whether a variable selector is valid.
714 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
716 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
717 * optabs-query.c (can_vec_perm_p): Likewise.
718 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
719 instead of vec_perm_indices.
720 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
721 (vect_gen_perm_mask_checked): Likewise,
722 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
723 (vect_gen_perm_mask_checked): Likewise,
725 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
727 * optabs-query.h (qimode_for_vec_perm): Declare.
728 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
729 (qimode_for_vec_perm): ...this new function.
730 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
732 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
734 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
735 does not have a conditional at the top.
737 2018-01-02 Richard Biener <rguenther@suse.de>
739 * ipa-inline.c (big_speedup_p): Fix expression.
741 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
744 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
747 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
751 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
752 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
753 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
754 cond_taken_branch_cost 3->4.
756 2018-01-01 Jakub Jelinek <jakub@redhat.com>
758 PR tree-optimization/83581
759 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
760 TODO_cleanup_cfg if any changes have been made.
763 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
764 convert_modes if target mode has the right side, but different mode
768 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
769 last argument when extracting from CONCAT. If either from_real or
770 from_imag is NULL, use expansion through memory. If result is not
771 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
772 the parts directly to inner mode, if even that fails, use expansion
776 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
777 check for bswap in mode rather than HImode and use that in expand_unop
780 2017-12-31 Jakub Jelinek <jakub@redhat.com>
783 * config/i386/i386.c (ix86_attribute_table): Remove excess
784 initializer for "shared" attribute.
786 2017-12-30 Tom de Vries <tom@codesourcery.com>
789 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
791 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
792 offload_funcs entries that no longer have a corresponding cgraph_node.
793 Mark the remaining ones as DECL_PRESERVE_P.
794 (output_lto): Call prune_offload_funcs.
796 2017-12-30 Jakub Jelinek <jakub@redhat.com>
798 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
799 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
800 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
801 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
802 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
803 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
804 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
805 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
806 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
807 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
808 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
809 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
810 vaesenc_<mode>, vpclmulqdq_<mode>,
811 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
813 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
815 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
816 _Float<N> and _Float<N>X variants for rounding built-in
818 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
819 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
820 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
821 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
822 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
823 * builtins.c (mathfn_built_in_2): Likewise.
824 * internal-fn.def (CEIL): Likewise.
826 (NEARBYINT): Likewise.
830 * convert.c (convert_to_integer_1): Likewise.
831 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
832 (integer_valued_real_call_p): Likewise.
833 * fold-const-call.c (fold_const_call_ss): Likewise.
834 * gencfn-macros.c (print_case_cfn): Change CFN and operator
835 printers to take a const char * suffix instead of a bool.
836 (print_define_operator_list): Likewise.
837 (fltall_suffixes): New list of suffixes, that include the
838 traditional suffixes as well as all of the _Float<N> and
840 (main): For _Float<N> and _Float<N>X functions, emit both
841 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
842 has the _Float<N> and _Float<N>X case names or operators. The
843 <name>_ALL variant has both the traditional and the
844 _Float<N>/_Float<N>X case names or operators.
845 * match.pd (COPYSIGN optimizations): Provide optimizations for
846 _Float<N> and _Float<N>X types where possible.
847 (MIN/MAX optimizations): Likewise.
848 (sqrt optimizations): Likewise.
849 (rounding optimizations): Likewise.
851 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
853 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
854 gen_int_mode rather than GEN_INT.
855 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
857 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
859 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
860 instead of CONSTANT_P.
861 (gen_vec_series): Likewise.
862 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
864 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
866 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
869 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
871 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
872 (aarch64_ld1x2<VDC:mode>): Likewise.
873 (aarch64_simd_ld1<mode>_x2): Likewise.
874 (aarch64_simd_ld1<mode>_x2): Likewise.
875 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
876 (vld1_s8_x2): Likewise.
877 (vld1_u16_x2): Likewise.
878 (vld1_s16_x2): Likewise.
879 (vld1_u32_x2): Likewise.
880 (vld1_s32_x2): Likewise.
881 (vld1_u64_x2): Likewise.
882 (vld1_s64_x2): Likewise.
883 (vld1_f16_x2): Likewise.
884 (vld1_f32_x2): Likewise.
885 (vld1_f64_x2): Likewise.
886 (vld1_p8_x2): Likewise.
887 (vld1_p16_x2): Likewise.
888 (vld1_p64_x2): Likewise.
889 (vld1q_u8_x2): Likewise.
890 (vld1q_s8_x2): Likewise.
891 (vld1q_u16_x2): Likewise.
892 (vld1q_s16_x2): Likewise.
893 (vld1q_u32_x2): Likewise.
894 (vld1q_s32_x2): Likewise.
895 (vld1q_u64_x2): Likewise.
896 (vld1q_s64_x2): Likewise.
897 (vld1q_f16_x2): Likewise.
898 (vld1q_f32_x2): Likewise.
899 (vld1q_f64_x2): Likewise.
900 (vld1q_p8_x2): Likewise.
901 (vld1q_p16_x2): Likewise.
902 (vld1q_p64_x2): Likewise.
904 2017-12-27 Martin Liska <mliska@suse.cz>
906 PR tree-optimization/83552
907 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
908 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
910 2017-12-27 Tom de Vries <tom@codesourcery.com>
912 * config/nvptx/nvptx.c (nvptx_option_override): Disable
913 -gstatement-frontiers.
915 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
917 PR rtl-optimization/83513
918 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
919 before priority comparison.
921 2017-12-25 Jakub Jelinek <jakub@redhat.com>
924 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
925 ix86_isa_flags2 to ix86_isa_flags.
926 * config/i386/i386-c.c (ix86_target_macros_internal): Test
927 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
928 isa_flags rather than isa_flags2.
929 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
930 and -mavx512bitalg from isa2_opts to isa_opts.
931 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
932 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
933 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
935 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
936 OPTION_MASK_ISA_AVX512F ored with another option similarly to
937 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
938 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
939 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
940 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
941 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
942 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
943 builtins from bdesc_args2 to bdesc_args section.
944 (__builtin_ia32_compressstoreuqi512_mask,
945 __builtin_ia32_compressstoreuhi512_mask,
946 __builtin_ia32_compressstoreuqi256_mask,
947 __builtin_ia32_expandloadqi512_mask,
948 __builtin_ia32_expandloadqi512_maskz,
949 __builtin_ia32_expandloadhi512_mask,
950 __builtin_ia32_expandloadhi512_maskz,
951 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
952 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
953 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
954 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
955 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
956 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
957 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
958 __builtin_ia32_vpshldv_v32hi_maskz,
959 __builtin_ia32_vpopcountb_v64qi_mask,
960 __builtin_ia32_vpopcountw_v32hi_mask,
961 __builtin_ia32_vpshufbitqmb512_mask,
962 __builtin_ia32_vpshufbitqmb256_mask): Add
963 " | OPTION_MASK_ISA_AVX512BW".
964 (__builtin_ia32_expandloadqi256_mask,
965 __builtin_ia32_expandloadqi256_maskz,
966 __builtin_ia32_vpopcountb_v32qi_mask): Add
967 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
968 (__builtin_ia32_expandloadhi256_mask,
969 __builtin_ia32_expandloadhi256_maskz,
970 __builtin_ia32_expandloadqi128_mask,
971 __builtin_ia32_expandloadqi128_maskz,
972 __builtin_ia32_expandloadhi128_mask,
973 __builtin_ia32_expandloadhi128_maskz,
974 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
975 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
976 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
977 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
978 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
979 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
980 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
981 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
982 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
983 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
984 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
985 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
986 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
987 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
988 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
989 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
990 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
991 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
992 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
993 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
994 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
995 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
996 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
997 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
998 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
999 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
1000 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
1001 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1002 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1003 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
1004 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
1005 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
1006 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1007 __builtin_ia32_vpopcountw_v8hi_mask): Add
1008 " | OPTION_MASK_ISA_AVX512VL".
1009 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1010 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
1011 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
1012 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
1013 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
1014 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
1015 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1016 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
1017 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
1018 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
1019 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
1020 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
1021 _mm_mask_bitshuffle_epi64_mask): Likewise.
1022 * common/config/i386/i386-common.c
1023 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
1024 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
1025 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1026 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
1027 OPTION_MASK_ISA_AVX512BITALG_UNSET.
1028 (OPTION_MASK_ISA2_AVX512F_UNSET,
1029 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
1030 (ix86_handle_option): For -mno-general-regs-only, clear from
1031 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
1032 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
1033 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
1034 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
1035 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
1036 for moving from ix86_isa_flags2 to ix86_isa_flags.
1038 2017-12-23 Jakub Jelinek <jakub@redhat.com>
1041 * fold-const.c (struct contains_label_data): New type.
1042 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
1043 inside of a SWITCH_BODY seen during the walk.
1044 (contains_label_p): Use walk_tree instead of
1045 walk_tree_without_duplicates, prepare data for contains_label_1 and
1048 2017-12-22 Jakub Jelinek <jakub@redhat.com>
1051 * tree-iterator.c (alloc_stmt_list): Start with cleared
1052 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
1056 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
1057 enable avx512f explicitly in #pragma GCC target.
1058 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
1059 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
1060 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
1061 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
1062 __builtin_ia32_vpdpbusds_v8si_mask,
1063 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
1064 __builtin_ia32_vpdpbusds_v4si_mask,
1065 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
1066 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
1067 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
1068 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
1069 __builtin_ia32_vpdpwssds_v8si_mask,
1070 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
1071 __builtin_ia32_vpdpwssds_v4si_mask,
1072 __builtin_ia32_vpdpwssds_v4si_maskz): Use
1073 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
1074 just OPTION_MASK_ISA_AVX512VNNI.
1076 2017-12-22 Martin Jambor <mjambor@suse.cz>
1079 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
1082 2017-12-22 Julia Koval <julia.koval@intel.com>
1083 Sebastian Peryt <sebastian.peryt@intel.com>
1085 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
1086 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
1087 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
1088 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
1089 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
1090 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
1091 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
1092 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
1093 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
1094 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
1095 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
1096 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
1097 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
1098 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
1099 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
1100 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
1101 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
1102 _mm256_maskz_popcnt_epi64): New intrinsics.
1103 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
1104 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
1105 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
1106 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
1107 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
1108 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
1109 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
1110 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
1111 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
1112 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
1113 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
1114 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
1115 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
1116 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
1117 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1118 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
1119 __builtin_ia32_vpshufbitqmb256_mask,
1120 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
1121 * config/i386/i386-c.c (__AVX512BITALG__): New.
1122 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
1123 (ix86_valid_target_attribute_inner_p): Ditto.
1124 (ix86_expand_args_builtin): Handle new types.
1125 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
1126 * config/i386/i386.opt: Add -mavx512bitalg.
1127 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
1128 avx512bitalgintrin.h.
1129 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
1130 (vpopcount<mode><mask_name>): Add more types.
1131 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
1132 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
1134 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1136 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
1137 Or in OPTION_MASK_ISA_AVX512F_SET.
1138 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1139 OPTION_MASK_ISA_AVX512VNNI_UNSET.
1140 (ix86_handle_option): Adjust for
1141 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
1142 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
1144 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
1145 isa_flag instead of isa_flag2.
1146 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
1147 isa_opts2 to isa_opts.
1148 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
1151 2017-12-22 Mike Stump <mikestump@comcast.net>
1152 Eric Botcazou <ebotcazou@adacore.com>
1154 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
1156 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1159 * config/i386/i386.c (ix86_function_arg_boundary): Return
1160 PARM_BOUNDARY for TYPE_EMPTY_P types.
1163 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
1164 if navail is >= dir.len.
1166 2017-12-21 Steve Ellcey <sellcey@cavium.com>
1168 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
1171 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1173 PR rtl-optimization/80747
1174 PR rtl-optimization/83512
1175 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
1176 succ edge from ENTRY, copy partition from e->dest to the newly
1178 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
1179 ENTRY, use BB_PARTITION of its successor block as current_partition.
1180 Don't copy partition when splitting succ edge from ENTRY.
1182 PR tree-optimization/83523
1183 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
1184 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
1185 (convert_mult_to_fma): Likewise.
1187 PR tree-optimization/83521
1188 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
1189 gimple_build_assign without code on result of
1190 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
1191 a VIEW_CONVERT_EXPR.
1193 2017-12-21 Andrew Pinski <apinski@cavium.com>
1194 Steve Ellcey <sellcey@cavium.com>
1196 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
1197 multi-arch for ilp32.
1199 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
1202 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1203 constraints to operand 2.
1204 (*ashl<mode>3_mask_1): Ditto.
1205 (*<shift_insn><mode>3_mask): Ditto.
1206 (*<shift_insn><mode>3_mask_1): Ditto.
1207 (*<rotate_insn><mode>3_mask): Ditto.
1208 (*<rotate_insn><mode>3_mask_1): Ditto.
1210 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1212 * reorg.c (make_return_insns): Reemit each insn with its own location.
1214 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1216 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1217 generation for cases where splatting a value is not useful.
1218 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1219 across a vec_duplicate and a paradoxical subreg forming a vector
1220 mode to a vec_concat.
1222 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1224 * combine.c (simplify_set): Do not transform subregs to zero_extends
1225 if the destination is not a scalar int mode.
1227 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1230 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1231 shared zero if the input has overflowed.
1233 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1235 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1236 support for Cortex-A55 and Cortex-A75.
1238 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1240 * common/config/arm/arm-common.c (compare_opt_names): Add function
1241 comment. Use strcmp instead of manual loop.
1243 2017-12-21 Martin Liska <mliska@suse.cz>
1245 PR gcov-profile/83509
1246 * gcov-dump.c (dump_gcov_file): Do not read info about
1247 support_unexecuted_blocks for gcda files.
1249 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1251 PR rtl-optimization/82973
1252 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1253 (valid_for_const_vector_p): ... this.
1254 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1255 (valid_for_const_vector_p): ... this. Adjust function comment.
1256 (gen_vec_duplicate): Adjust caller.
1257 * optabs.c (expand_vector_broadcast): Likewise.
1258 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1259 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1261 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1262 instead of GET_CODE == CONST_FIXED.
1263 (simplify_subreg): Use CONST_FIXED_P macro instead of
1264 GET_CODE == CONST_FIXED.
1267 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1268 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1269 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1270 (ix86_option_override_internal): Adjust for
1271 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1272 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1273 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1274 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1275 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1276 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1277 def_builtin2 for CET builtins.
1278 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1279 up formatting in IX86_BUILTIN_RDPID code.
1280 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1281 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1282 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1283 ix86_isa_flags2 to ix86_isa_flags.
1284 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1286 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1287 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1288 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1290 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1291 Or in OPTION_MASK_ISA_AVX512F_SET.
1292 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1293 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1294 (ix86_handle_option): Adjust for
1295 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1296 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
1298 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1299 Alan Hayward <alan.hayward@arm.com>
1300 David Sherwood <david.sherwood@arm.com>
1302 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
1303 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1304 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1305 Track polynomial offsets.
1307 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1308 Alan Hayward <alan.hayward@arm.com>
1309 David Sherwood <david.sherwood@arm.com>
1311 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1312 Treat drb->init as a poly_int. Fail if its misalignment wrt
1313 vector_alignment isn't known.
1315 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1316 Alan Hayward <alan.hayward@arm.com>
1317 David Sherwood <david.sherwood@arm.com>
1319 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1321 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1322 (vect_loop_versioning): Take the loop versioning threshold as a
1324 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1325 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1326 versioning_threshold.
1327 (vect_analyze_loop_2): Compute the loop versioning threshold
1328 whenever loop versioning is needed, and store it in the new
1329 field rather than combining it with the cost model threshold.
1330 (vect_transform_loop): Update call to vect_loop_versioning.
1331 Try to combine the loop versioning and cost thresholds here.
1333 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1334 Alan Hayward <alan.hayward@arm.com>
1335 David Sherwood <david.sherwood@arm.com>
1337 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1338 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1339 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1340 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1341 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1342 (group_compare_offset): Update accordingly.
1343 (split_small_address_groups_p): Likewise.
1344 (record_use): Take addr_offset as a poly_uint64 rather than
1345 an unsigned HOST_WIDE_INT.
1346 (strip_offset): Return the offset as a poly_uint64 rather than
1347 an unsigned HOST_WIDE_INT.
1348 (record_group_use, split_address_groups): Track polynomial offsets.
1349 (add_iv_candidate_for_use): Likewise.
1350 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1351 than a HOST_WIDE_INT.
1352 (strip_offset_1): Return the offset as a poly_int64 rather than
1355 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1356 Alan Hayward <alan.hayward@arm.com>
1357 David Sherwood <david.sherwood@arm.com>
1359 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1360 rather than a HOST_WIDE_INT.
1361 * tree.c (get_binfo_at_offset): Likewise.
1363 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1364 Alan Hayward <alan.hayward@arm.com>
1365 David Sherwood <david.sherwood@arm.com>
1367 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1368 rather than a HOST_WIDE_INT.
1369 * tree-sra.c (build_ref_for_offset): Likewise.
1371 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1372 Alan Hayward <alan.hayward@arm.com>
1373 David Sherwood <david.sherwood@arm.com>
1375 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1377 * tree.c (mem_ref_offset): Likewise.
1378 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1379 * builtins.c (get_object_alignment_2): Likewise.
1380 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1381 * gimple-fold.c (get_base_constructor): Likewise.
1382 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1383 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1385 * ipa-polymorphic-call.c
1386 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1387 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1388 (get_ancestor_addr_info): Likewise.
1389 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1390 * match.pd: Likewise.
1391 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1392 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1393 * tree-eh.c (tree_could_trap_p): Likewise.
1394 * tree-object-size.c (addr_object_size): Likewise.
1395 * tree-ssa-address.c (copy_ref_info): Likewise.
1396 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1397 (indirect_refs_may_alias_p): Likewise.
1398 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1399 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1400 (non_rewritable_mem_ref_base): Likewise.
1401 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1402 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1403 * varasm.c (decode_addr_const): Likewise.
1405 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1406 Alan Hayward <alan.hayward@arm.com>
1407 David Sherwood <david.sherwood@arm.com>
1409 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1410 offsets for MEM_REFs.
1412 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1413 Alan Hayward <alan.hayward@arm.com>
1414 David Sherwood <david.sherwood@arm.com>
1416 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1417 as a poly_uint64 rather than an unsigned int.
1418 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1420 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1421 Alan Hayward <alan.hayward@arm.com>
1422 David Sherwood <david.sherwood@arm.com>
1424 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1426 (decode_addr_const): Update accordingly.
1428 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1429 Alan Hayward <alan.hayward@arm.com>
1430 David Sherwood <david.sherwood@arm.com>
1432 * tree.h (bit_field_size, bit_field_offset): New functions.
1433 * hsa-gen.c (gen_hsa_addr): Use them.
1434 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1435 (simplify_vector_constructor): Likewise.
1436 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1437 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1438 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1439 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1442 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1443 Alan Hayward <alan.hayward@arm.com>
1444 David Sherwood <david.sherwood@arm.com>
1446 * expr.h (emit_group_load, emit_group_load_into_temps)
1447 (emit_group_store): Take the size as a poly_int64 rather than an int.
1448 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1449 (emit_group_load_into_temp, emit_group_store): Likewise.
1451 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1452 Alan Hayward <alan.hayward@arm.com>
1453 David Sherwood <david.sherwood@arm.com>
1455 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1456 an unsigned int to a poly_uint64.
1457 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1458 sizes as poly_uint64s rather than unsigned ints.
1459 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1462 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1463 Alan Hayward <alan.hayward@arm.com>
1464 David Sherwood <david.sherwood@arm.com>
1466 * lra-constraints.c (emit_inc): Change inc_amount from an int
1469 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1470 Alan Hayward <alan.hayward@arm.com>
1471 David Sherwood <david.sherwood@arm.com>
1473 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1475 (add_stack_var, stack_var_cmp, partition_stack_vars)
1476 (dump_stack_var_partition): Update accordingly.
1477 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1478 than a HOST_WIDE_INT.
1479 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1480 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1481 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1482 rather than a HOST_WIDE_INT.
1484 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1485 Alan Hayward <alan.hayward@arm.com>
1486 David Sherwood <david.sherwood@arm.com>
1488 * target.def (return_pops_args): Treat both the input and output
1489 sizes as poly_int64s rather than HOST_WIDE_INTS.
1490 * targhooks.h (default_return_pops_args): Update accordingly.
1491 * targhooks.c (default_return_pops_args): Likewise.
1492 * doc/tm.texi: Regenerate.
1493 * emit-rtl.h (incoming_args): Change pops_args, size and
1494 outgoing_args_size from int to poly_int64_pod.
1495 * function.h (expr_status): Change x_pending_stack_adjust and
1496 x_stack_pointer_delta from int to poly_int64.
1497 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1498 (ARGS_SIZE_RTX): Update accordingly.
1499 * calls.c (highest_outgoing_arg_in_use): Change from int to
1501 (stack_usage_watermark, stored_args_watermark): New variables.
1502 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1503 (emit_call_1): Change the stack_size and rounded_stack_size
1504 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1506 (save_fixed_argument_area): Check stack_usage_watermark.
1507 (initialize_argument_information): Change old_pending_adj from
1508 a HOST_WIDE_INT * to a poly_int64_pod *.
1509 (compute_argument_block_size): Return the size as a poly_int64
1511 (finalize_must_preallocate): Track polynomial argument sizes.
1512 (compute_argument_addresses): Likewise.
1513 (internal_arg_pointer_based_exp): Track polynomial offsets.
1514 (mem_overlaps_already_clobbered_arg_p): Rename to...
1515 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1516 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1517 Check stored_args_used_watermark.
1518 (load_register_parameters): Update accordingly.
1519 (check_sibcall_argument_overlap_1): Likewise.
1520 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1521 args size as a poly_int64 rather than an int. Return a bool
1522 indicating whether the optimization was possible and return
1523 the new adjustment by reference.
1524 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1525 Update stored_args_watermark.
1526 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1527 (expand_call): Likewise. Maintain stack_usage_watermark and
1528 stored_args_watermark. Update calls to
1529 combine_pending_stack_adjustment_and_call.
1530 (emit_library_call_value_1): Handle polynomial argument sizes.
1531 Call stack_region_maybe_used_p and mark_stack_region_used.
1532 Maintain stack_usage_watermark.
1533 (store_one_arg): Likewise. Update call to
1534 mem_overlaps_already_clobbered_arg_p.
1535 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1537 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1538 * config/microblaze/microblaze.c (microblaze_function_prologue):
1540 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1541 TARGET_RETURN_POPS_ARGS interface.
1542 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1544 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1545 * config/i386/i386.c (ix86_return_pops_args): Update for new
1546 TARGET_RETURN_POPS_ARGS interface.
1547 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1548 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1549 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1550 TARGET_RETURN_POPS_ARGS interface.
1551 * config/vax/vax.c (vax_return_pops_args): Likewise.
1552 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1553 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1554 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1555 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1556 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1557 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1558 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1559 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1560 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1561 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1562 and x_stack_pointer_delta from int to poly_int64.
1563 * dojump.c (do_pending_stack_adjust): Update accordingly.
1564 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1565 stack_pointer_deltas.
1566 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1567 (pad_to_arg_alignment): Track polynomial offsets.
1568 (assign_parm_find_stack_rtl): Likewise.
1569 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1570 * toplev.c (output_stack_usage): Update reference to
1571 current_function_pushed_stack_size.
1573 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1574 Alan Hayward <alan.hayward@arm.com>
1575 David Sherwood <david.sherwood@arm.com>
1577 * function.c (in_arg_offset, var_offset, dynamic_offset)
1578 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1579 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1580 rather than a HOST_WIDE_INT.
1581 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1582 (instantiate_virtual_regs_in_insn): Likewise.
1584 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1585 Alan Hayward <alan.hayward@arm.com>
1586 David Sherwood <david.sherwood@arm.com>
1588 * rtl.h (get_args_size, add_args_size_note): New functions.
1589 (find_args_size_adjust): Return a poly_int64 rather than a
1591 (fixup_args_size_notes): Likewise. Make the same change to the
1592 end_args_size parameter.
1593 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1594 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1595 * calls.c (emit_call_1): Likewise.
1596 * explow.c (adjust_stack_1): Likewise.
1597 * cfgcleanup.c (old_insns_match_p): Update use of
1598 find_args_size_adjust.
1599 * combine.c (distribute_notes): Track polynomial arg sizes.
1600 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1601 end_true_args_size, beg_delay_args_size and end_delay_args_size
1602 from HOST_WIDE_INT to poly_int64.
1603 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1604 than a HOST_WIDE_INT.
1605 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1606 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1607 polynomial arg sizes.
1608 * emit-rtl.c (try_split): Use get_args_size.
1609 * recog.c (peep2_attempt): Likewise.
1610 * reload1.c (reload_as_needed): Likewise.
1611 * expr.c (find_args_size_adjust): Return the adjustment as a
1612 poly_int64 rather than a HOST_WIDE_INT.
1613 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1614 to a poly_int64 and change the return type in the same way.
1615 (emit_single_push_insn): Track polynomial arg sizes.
1617 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1618 Alan Hayward <alan.hayward@arm.com>
1619 David Sherwood <david.sherwood@arm.com>
1621 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1622 from HOST_WIDE_INT to poly_int64.
1623 * expr.c (push_block, emit_push_insn): Likewise.
1625 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1626 Alan Hayward <alan.hayward@arm.com>
1627 David Sherwood <david.sherwood@arm.com>
1629 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1631 (get_frame_size): Return the size as a poly_int64 rather than a
1633 (frame_offset_overflow): Take the offset as a poly_int64 rather
1634 than a HOST_WIDE_INT.
1635 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1636 (assign_stack_temp): Likewise for the size.
1637 * function.c (get_frame_size): Return a poly_int64 rather than
1639 (frame_offset_overflow): Take the offset as a poly_int64 rather
1640 than a HOST_WIDE_INT.
1641 (try_fit_stack_local): Take the start, length and size as poly_int64s
1642 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1643 rather than a HOST_WIDE_INT.
1644 (add_frame_space): Take the start and end as poly_int64s rather than
1646 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1647 (assign_stack_temp): Likewise for the size.
1648 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1650 (find_temp_slot_from_address): Handle polynomial offsets.
1651 (combine_temp_slots): Likewise.
1652 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1654 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1655 poly_int64 rather than a HOST_WIDE_INT.
1656 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1657 than a HOST_WIDE_INT.
1658 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1659 polynomial frame offsets.
1660 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1661 as a poly_int64 rather than an int.
1662 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1663 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1664 * config/v850/v850.c (compute_frame_size): Likewise.
1665 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1666 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1667 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1668 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1669 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1670 rather than a HOST_WIDE_INT.
1671 * explow.c (get_dynamic_stack_base): Likewise.
1672 * final.c (final_start_function): Use the constant lower bound
1673 of the frame size for -Wframe-larger-than.
1674 * ira.c (do_reload): Adjust for new get_frame_size return type.
1675 * lra.c (lra): Likewise.
1676 * reload1.c (reload): Likewise.
1677 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1678 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1679 * rtlanal.c (get_initial_register_offset): Return the offset as
1680 a poly_int64 rather than a HOST_WIDE_INT.
1682 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1683 Alan Hayward <alan.hayward@arm.com>
1684 David Sherwood <david.sherwood@arm.com>
1686 * reload1.c (elim_table): Change initial_offset, offset and
1687 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1688 (offsets_at): Change the target array's element type from
1689 HOST_WIDE_INT to poly_int64_pod.
1690 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1691 (elimination_costs_in_insn, update_eliminable_offsets)
1692 (verify_initial_elim_offsets, set_offsets_for_label)
1693 (init_eliminable_invariants): Update after above changes.
1695 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1696 Alan Hayward <alan.hayward@arm.com>
1697 David Sherwood <david.sherwood@arm.com>
1699 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1700 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1701 (decomposition): Change start and end from HOST_WIDE_INT
1703 (decompose, immune_p): Update accordingly.
1704 (find_inc_amount): Return a poly_int64 rather than an int.
1705 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1708 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1709 Alan Hayward <alan.hayward@arm.com>
1710 David Sherwood <david.sherwood@arm.com>
1712 * tree.h (get_inner_reference): Return the bitsize and bitpos
1713 as poly_int64_pods rather than HOST_WIDE_INT.
1714 * fold-const.h (ptr_difference_const): Return the pointer difference
1715 as a poly_int64_pod rather than a HOST_WIDE_INT.
1716 * expr.c (get_inner_reference): Return the bitsize and bitpos
1717 as poly_int64_pods rather than HOST_WIDE_INT.
1718 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1720 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1721 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1722 (optimize_bit_field_compare): Update call to get_inner_reference.
1723 (decode_field_reference): Likewise.
1724 (fold_unary_loc): Track polynomial offsets and sizes.
1725 (split_address_to_core_and_offset): Return the bitpos as a
1726 poly_int64_pod rather than a HOST_WIDE_INT.
1727 (ptr_difference_const): Likewise for the pointer difference.
1728 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1729 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1730 * dbxout.c (dbxout_expand_expr): Likewise.
1731 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1732 (loc_list_from_tree_1, fortran_common): Likewise.
1733 * gimple-laddress.c (pass_laddress::execute): Likewise.
1734 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1735 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1736 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1737 * tree-affine.c (tree_to_aff_combination): Likewise.
1738 (get_inner_reference_aff): Likewise.
1739 * tree-data-ref.c (split_constant_offset_1): Likewise.
1740 (dr_analyze_innermost): Likewise.
1741 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1742 * tree-sra.c (ipa_sra_check_caller): Likewise.
1743 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1744 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1745 (instrument_bool_enum_load, instrument_object_size): Likewise.
1746 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1747 to get_inner_reference.
1748 * hsa-gen.c (gen_hsa_addr): Likewise.
1749 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1750 * tsan.c (instrument_expr): Likewise.
1751 * match.pd: Update call to ptr_difference_const.
1753 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1754 Alan Hayward <alan.hayward@arm.com>
1755 David Sherwood <david.sherwood@arm.com>
1757 * fold-const.c (fold_comparison): Track sizes and offsets as
1758 poly_int64s rather than HOST_WIDE_INTs when folding address
1761 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1762 Alan Hayward <alan.hayward@arm.com>
1763 David Sherwood <david.sherwood@arm.com>
1765 * expr.h (get_bit_range): Return the bitstart and bitend as
1766 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1767 as a poly_int64 rather than a HOST_WIDE_INT.
1768 * expr.c (get_bit_range): Likewise.
1769 (expand_assignment): Update call accordingly.
1770 * fold-const.c (optimize_bit_field_compare): Likewise.
1772 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1773 Alan Hayward <alan.hayward@arm.com>
1774 David Sherwood <david.sherwood@arm.com>
1776 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1777 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1778 as an optional final parameter.
1779 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1780 (aarch64_legitimate_address_p): Likewise.
1781 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1782 instead of an rtx code.
1783 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1784 (aarch64_legitimate_address_hook_p): Likewise.
1785 (aarch64_print_ldpstp_address): Likewise.
1786 (aarch64_print_operand_address): Likewise.
1787 (aarch64_address_cost): Likewise.
1788 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1789 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1790 (aarch64_mem_pair_lanes_operand): Likewise.
1792 2017-12-20 Richard Biener <rguenther@suse.de>
1794 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1795 update_stmt_if_modified.
1797 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1799 PR tree-optimization/83491
1800 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1801 before walking uses. Improve coding style and comments.
1803 2017-12-20 Tom de Vries <tom@codesourcery.com>
1805 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1807 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1810 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1812 (pass_ipa_free_fn_summary): Add small_p private data member,
1813 initialize to false in the ctor.
1814 (pass_ipa_free_fn_summary::clone,
1815 pass_ipa_free_fn_summary::set_pass_param,
1816 pass_ipa_free_fn_summary::gate): New methods.
1817 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1818 | TODO_dump_symtab if small_p.
1819 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1820 entry and add another instance of the pass with false parm after
1822 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1823 ipa_free_fn_summary here.
1825 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1827 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1829 2017-12-20 Martin Sebor <msebor@redhat.com>
1832 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1833 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1834 (expand_builtin_strncmp): Same.
1836 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1839 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1840 allowed debug stmts before labels.
1841 (expand_gimple_basic_block): Likewise.
1842 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1843 * gimple-iterator.h (gsi_after_labels): Likewise.
1844 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1845 rename reused variable, and simplify using gsi_move_before.
1846 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1847 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1848 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1849 (gimple_verify_flow_info, gimple_block_label): Likewise.
1850 (make_blocks): Move debug markers after adjacent labels.
1851 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1852 allowed debug insns outside blocks.
1853 * df-scan.c (df_insn_delete): Likewise.
1854 * lra-constraints.c (update_ebb_live_info): Likewise.
1855 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1856 (vt_initialize, delete_vta_debug_insns): Likewise.
1857 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1859 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1860 Alan Hayward <alan.hayward@arm.com>
1861 David Sherwood <david.sherwood@arm.com>
1863 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1864 (round_up_to_byte_boundary): Likewise.
1865 * expr.h (get_bit_range): Add temporary shim.
1866 * gimple-ssa-store-merging.c (store_operand_info): Change the
1867 bitsize, bitpos, bitregion_start and bitregion_end fields from
1868 unsigned HOST_WIDE_INT to poly_uint64.
1869 (merged_store_group): Likewise load_align_base.
1870 (compatible_load_p, compatible_load_p): Update accordingly.
1871 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1872 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1873 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1874 bitregion_start and bitregion_end as poly_uint64s rather than
1875 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1876 (handled_load): Take the bitsize, bitpos,
1877 bitregion_start and bitregion_end as poly_uint64s rather than
1878 unsigned HOST_WIDE_INTs.
1879 (pass_store_merging::process_store): Update call to
1880 mem_valid_for_store_merging.
1882 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1883 Alan Hayward <alan.hayward@arm.com>
1884 David Sherwood <david.sherwood@arm.com>
1886 * builtins.c (get_object_alignment_2): Track polynomial offsets
1887 and sizes. Update the alignment handling.
1889 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1890 Alan Hayward <alan.hayward@arm.com>
1891 David Sherwood <david.sherwood@arm.com>
1893 * tree.h (get_inner_reference): Add a version that returns the
1894 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1895 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1896 the case in which bitpos is not associated with the first byte.
1898 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1899 Alan Hayward <alan.hayward@arm.com>
1900 David Sherwood <david.sherwood@arm.com>
1902 * tree-affine.h (get_inner_reference_aff): Return the size as a
1904 * tree-affine.c (get_inner_reference_aff): Likewise.
1905 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1906 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1908 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1909 Alan Hayward <alan.hayward@arm.com>
1910 David Sherwood <david.sherwood@arm.com>
1912 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1913 HOST_WIDE_INT rather than a poly_int64.
1915 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1916 Alan Hayward <alan.hayward@arm.com>
1917 David Sherwood <david.sherwood@arm.com>
1919 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1920 HOST_WIDE_INT to poly_int64_pod.
1921 (perform_symbolic_merge): Update accordingly.
1922 (bswap_replace): Likewise.
1924 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1925 Alan Hayward <alan.hayward@arm.com>
1926 David Sherwood <david.sherwood@arm.com>
1928 * tree-affine.h (aff_tree::offset): Change from widest_int
1930 (wide_int_ext_for_comb): Delete.
1931 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1932 constants as poly_widest_int rather than widest_int.
1933 (aff_combination_constant_multiple_p): Return the multiplier
1934 as a poly_widest_int.
1935 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1937 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1938 version static and add an overload for poly_widest_int.
1939 (aff_combination_const, aff_combination_add_cst)
1940 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1941 the constants as poly_widest_int rather than widest_int.
1942 (tree_to_aff_combination): Generalize INTEGER_CST case to
1944 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1945 (aff_combination_add_product, aff_combination_mult): Handle
1947 (aff_combination_constant_multiple_p): Return the multiplier
1948 as a poly_widest_int.
1949 * tree-predcom.c (determine_offset): Return the offset as a
1951 (split_data_refs_to_components, suitable_component_p): Update
1953 (valid_initializer_p): Update call to
1954 aff_combination_constant_multiple_p.
1955 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1956 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1957 as a poly_int64 rather than a HOST_WIDE_INT.
1958 (get_address_cost): Handle polynomial offsets.
1959 (iv_elimination_compare_lt): Likewise.
1960 (rewrite_use_nonlinear_expr): Likewise.
1962 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1963 Alan Hayward <alan.hayward@arm.com>
1964 David Sherwood <david.sherwood@arm.com>
1966 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1967 as a poly_int64_pod rather than a HOST_WIDE_INT.
1968 (get_addr_base_and_unit_offset): Likewise.
1969 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1970 (get_addr_base_and_unit_offset): Likewise.
1971 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1972 to poly_int64 in example.
1973 * fold-const.c (fold_binary_loc): Update call to
1974 get_addr_base_and_unit_offset.
1975 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1976 (maybe_canonicalize_mem_ref_addr): Likewise.
1977 (gimple_fold_stmt_to_constant_1): Likewise.
1978 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1980 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1981 * match.pd: Likewise.
1982 * omp-low.c (lower_omp_target): Likewise.
1983 * tree-sra.c (build_ref_for_offset): Likewise.
1984 (build_debug_ref_for_model): Likewise.
1985 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1986 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1987 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1988 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1989 (constant_pointer_difference): Likewise.
1990 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1991 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1992 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1993 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1994 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1995 (set_ssa_val_to): Likewise.
1996 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1997 (maybe_diag_stxncpy_trunc): Likewise.
1998 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1999 * tree.c (build_simple_mem_ref_loc): Likewise.
2000 (array_at_struct_end_p): Likewise.
2002 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2003 Alan Hayward <alan.hayward@arm.com>
2004 David Sherwood <david.sherwood@arm.com>
2006 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
2007 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
2008 (get_ref_base_and_extent_hwi): Declare.
2009 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
2010 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
2011 (get_ref_base_and_extent_hwi): New function.
2012 * cfgexpand.c (expand_debug_expr): Update call to
2013 get_ref_base_and_extent.
2014 * dwarf2out.c (add_var_loc_to_decl): Likewise.
2015 * gimple-fold.c (get_base_constructor): Return the offset as a
2016 poly_int64_pod rather than a HOST_WIDE_INT.
2017 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
2018 * ipa-polymorphic-call.c
2019 (ipa_polymorphic_call_context::set_by_invariant)
2020 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
2021 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
2022 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
2023 rather than get_ref_base_and_extent.
2024 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
2025 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
2026 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
2028 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
2029 call to get_ref_base_and_extent.
2030 * tree-sra.c (create_access, get_access_for_expr): Likewise.
2031 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
2032 (stmt_kills_ref_p): Likewise.
2033 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2034 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
2036 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
2037 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
2038 when calling native_encode_expr.
2039 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
2040 call to get_ref_base_and_extent.
2041 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
2042 get_ref_base_and_extent.
2043 * var-tracking.c (track_expr_p): Likewise.
2045 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2046 Alan Hayward <alan.hayward@arm.com>
2047 David Sherwood <david.sherwood@arm.com>
2049 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
2050 HOST_WIDE_INT to poly_int64_pod.
2051 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
2052 polynomail parameter offsets.
2054 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2055 Alan Hayward <alan.hayward@arm.com>
2056 David Sherwood <david.sherwood@arm.com>
2058 * gengtype.c (main): Handle poly_int64_pod.
2059 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
2060 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
2061 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
2062 from HOST_WIDE_INT to poly_int64_pod.
2063 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
2064 (copy_cfa): New function.
2065 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
2066 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
2067 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
2068 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
2069 (initial_return_save): Treat offsets as poly_ints.
2070 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
2071 offset is nonconstant.
2072 (reg_save): Take the offset as a poly_int64. Fall back to
2073 DW_CFA_expression for nonconstant offsets.
2074 (queue_reg_save): Take the offset as a poly_int64.
2075 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
2077 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2078 Alan Hayward <alan.hayward@arm.com>
2079 David Sherwood <david.sherwood@arm.com>
2081 * rtl.h (operand_subword, operand_subword_force): Take the offset
2082 as a poly_uint64 an unsigned int.
2083 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
2085 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2086 Alan Hayward <alan.hayward@arm.com>
2087 David Sherwood <david.sherwood@arm.com>
2089 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
2090 'p' format code. Use INT_LIST rather than SUBREG as the example of
2091 a code with an XINT and an XEXP. Remove the implication that
2092 accessing an rtx field using XINT is expected to work.
2093 * rtl.def (SUBREG): Change format from "ei" to "ep".
2094 * rtl.h (rtunion::rt_subreg): New field.
2095 (XCSUBREG): New macro.
2096 (SUBREG_BYTE): Use it.
2097 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
2098 Update constructor accordingly.
2099 (subreg_shape::operator ==): Update accordingly.
2100 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
2101 than an unsigned int.
2102 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
2103 a poly_uint64 rather than an unsigned int.
2104 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
2105 than an unsigned int.
2106 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
2107 (subreg_size_highpart_offset): Return a poly_uint64 rather than
2108 an unsigned int. Take the sizes as poly_uint64s.
2109 (subreg_offset_from_lsb): Return a poly_uint64 rather than
2110 an unsigned int. Take the shift as a poly_uint64 rather than
2112 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
2113 as a poly_uint64 rather than an unsigned int.
2114 (simplify_subreg_regno): Likewise.
2115 (byte_lowpart_offset): Return the memory offset as a poly_int64
2117 (subreg_memory_offset): Likewise. Take the subreg offset as a
2118 poly_uint64 rather than an unsigned int.
2119 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
2120 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
2121 poly_uint64 rather than an unsigned int.
2122 * rtl.c (rtx_format): Describe 'p' in comment.
2123 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
2124 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
2125 offset as a poly_uint64 rather than an unsigned int.
2126 (byte_lowpart_offset): Return the memory offset as a poly_int64
2128 (subreg_memory_offset): Likewise. Take the subreg offset as a
2129 poly_uint64 rather than an unsigned int.
2130 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
2131 mode sizes as poly_uint64s rather than unsigned ints. Return a
2132 poly_uint64 rather than an unsigned int.
2133 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
2134 (copy_insn_1): Handle 'p'.
2135 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
2136 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
2137 an unsigned int. Return the shift in the same way.
2138 (subreg_lsb): Return the shift as a poly_uint64 rather than an
2140 (subreg_size_offset_from_lsb): Take the sizes and shift as
2141 poly_uint64s rather than unsigned ints. Return the offset as
2143 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
2144 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
2146 * rtlhash.c (add_rtx): Handle 'p'.
2147 * genemit.c (gen_exp): Likewise.
2148 * gengenrtl.c (type_from_format, gendef): Likewise.
2149 * gensupport.c (subst_pattern_match, get_alternatives_number)
2150 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
2151 (subst_dup): Likewise.
2152 * gengtype.c (adjust_field_rtx_def): Likewise.
2153 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
2154 (match_pattern_2): Likewise.
2155 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
2156 (rtx_test::subreg_field): New function.
2157 (operator ==, safe_to_hoist_p, transition_parameter_type)
2158 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
2159 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
2160 * genpeep.c (match_rtx): Likewise.
2161 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
2162 (rtx_writer::print_rtx_operand): Handle 'p'.
2163 (print_value): Handle SUBREG.
2164 * read-rtl.c (apply_int_iterator): Likewise.
2165 (rtx_reader::read_rtx_operand): Handle 'p'.
2166 * alias.c (rtx_equal_for_memref_p): Likewise.
2167 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
2168 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
2170 * calls.c (expand_call): Likewise.
2171 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
2172 (make_extraction, gen_lowpart_for_combine): Likewise.
2173 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
2175 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
2176 rather than an unsigned int. Treat subreg offsets as poly_ints.
2177 (exp_equiv_p): Handle 'p'.
2178 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
2179 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
2180 * dse.c (find_shift_sequence): Likewise.
2181 * dwarf2out.c (rtl_for_decl_location): Likewise.
2182 * expmed.c (extract_low_bits): Likewise.
2183 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
2184 (expand_expr_real_2): Likewise.
2185 * final.c (alter_subreg): Likewise.
2186 (leaf_renumber_regs_insn): Handle 'p'.
2187 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
2188 Treat subreg offsets as poly_ints.
2189 * fwprop.c (forward_propagate_and_simplify): Likewise.
2190 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
2191 * ira.c (get_subreg_tracking_sizes): Likewise.
2192 * ira-conflicts.c (go_through_subreg): Likewise.
2193 * ira-lives.c (process_single_reg_class_operands): Likewise.
2194 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
2195 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
2196 as a poly_uint64 rather than an unsigned int.
2197 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
2198 subreg offsets as poly_ints.
2199 * lra-constraints.c (operands_match_p): Handle 'p'.
2200 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
2201 * lra-spills.c (assign_mem_slot): Likewise.
2202 * postreload.c (move2add_valid_value_p): Likewise.
2203 * recog.c (general_operand, indirect_operand): Likewise.
2204 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2205 (copyprop_hardreg_forward_1): Likewise.
2206 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2207 (record_subregs_of_mode): Likewise.
2208 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2209 * reload.c (operands_match_p): Handle 'p'.
2210 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2211 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2212 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2213 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2214 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2215 (simplify_const_poly_int_tests<N>::run): Likewise.
2216 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2217 a poly_uint64 rather than an unsigned int.
2218 * valtrack.c (debug_lowpart_subreg): Likewise.
2219 * var-tracking.c (var_lowpart): Likewise.
2220 (loc_cmp): Handle 'p'.
2222 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2223 Alan Hayward <alan.hayward@arm.com>
2224 David Sherwood <david.sherwood@arm.com>
2226 * ira.c (get_subreg_tracking_sizes): New function.
2227 (init_live_subregs): Take an integer size rather than a register.
2228 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2229 to init_live_subregs.
2231 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2232 Alan Hayward <alan.hayward@arm.com>
2233 David Sherwood <david.sherwood@arm.com>
2235 * expr.c (store_constructor_field): Change bitsize from a
2236 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2237 HOST_WIDE_INT to a poly_int64.
2238 (store_constructor): Change size from a HOST_WIDE_INT to
2240 (store_field): Likewise bitsize and bitpos.
2242 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2243 Alan Hayward <alan.hayward@arm.com>
2244 David Sherwood <david.sherwood@arm.com>
2246 * expmed.h (store_bit_field): Change bitregion_start and
2247 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2248 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2249 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2250 (store_fixed_bit_field, store_split_bit_field): Likewise.
2251 * expr.c (store_constructor_field, store_field): Likewise.
2252 (optimize_bitfield_assignment_op): Likewise. Make the same change
2253 to bitsize and bitpos.
2254 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2255 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2256 same change in the constructor arguments.
2257 (get_best_mode): Change bitregion_start and bitregion_end from
2258 unsigned HOST_WIDE_INT to poly_uint64.
2259 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2260 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2262 (bit_field_mode_iterator::next_mode): Update for new types
2263 of m_bitregion_start and m_bitregion_end.
2264 (get_best_mode): Change bitregion_start and bitregion_end from
2265 unsigned HOST_WIDE_INT to poly_uint64.
2267 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2268 Alan Hayward <alan.hayward@arm.com>
2269 David Sherwood <david.sherwood@arm.com>
2271 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2272 accepts poly_uint64 offsets.
2273 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2274 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2275 * expmed.c (lowpart_bit_field_p): Likewise.
2276 (extract_bit_field_as_subreg): New function, split out from...
2277 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2278 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2279 extractions, check that BITSIZE matches the size of the extracted
2280 value and that BITNUM is an exact multiple of that size.
2281 If all else fails, try forcing the value into memory if
2282 BITNUM is variable, and adjusting the address so that the
2283 offset is constant. Split the part that can only handle constant
2284 bitsize and bitnum out into...
2285 (extract_integral_bit_field): ...this new function.
2286 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2287 rather than unsigned HOST_WIDE_INTs.
2289 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2290 Alan Hayward <alan.hayward@arm.com>
2291 David Sherwood <david.sherwood@arm.com>
2293 * expmed.h (store_bit_field): Take bitsize and bitnum as
2294 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2295 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2296 that returns the byte size.
2297 (store_bit_field_1): Take bitsize and bitnum as
2298 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2299 to simple_mem_bitfield_p. Split the part that can only handle
2300 constant bitsize and bitnum out into...
2301 (store_integral_bit_field): ...this new function.
2302 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2303 than unsigned HOST_WIDE_INTs.
2304 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2306 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2307 Alan Hayward <alan.hayward@arm.com>
2308 David Sherwood <david.sherwood@arm.com>
2310 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2311 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2313 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2314 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2315 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2316 offset as a poly_int64 rather than an int.
2317 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2318 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2319 * lra-constraints.c (equiv_address_substitution): Track offsets
2321 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2322 (curr_insn_transform): Handle the new form of sp_offset.
2323 * lra-eliminations.c (lra_elim_table): Change previous_offset
2324 and offset from HOST_WIDE_INT to poly_int64.
2325 (print_elim_table, update_reg_eliminate): Update accordingly.
2326 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2327 (get_elimination): Update accordingly.
2328 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2329 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2330 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2331 poly_int64 offsets generally.
2332 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2333 (mark_not_eliminable, init_elimination): Update accordingly.
2334 (remove_reg_equal_offset_note): Return a bool and pass the new
2335 offset back by pointer as a poly_int64.
2336 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2337 rather than a HOST_WIDE_INT.
2338 (do_remat): Track offsets poly_int64s.
2339 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2341 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2342 Alan Hayward <alan.hayward@arm.com>
2343 David Sherwood <david.sherwood@arm.com>
2345 * rtl.h (mem_attrs): Add a default constructor. Change size and
2346 offset from HOST_WIDE_INT to poly_int64.
2347 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2348 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2349 (widen_memory_access): Take the sizes and offsets as poly_int64s
2350 rather than HOST_WIDE_INTs.
2351 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2352 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2354 (adjust_offset_for_component_ref): Change the offset from a
2355 HOST_WIDE_INT to a poly_int64.
2356 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2357 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2358 * dce.c (find_call_stack_args): Likewise.
2359 * dse.c (record_store): Likewise.
2360 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2361 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2362 * read-rtl-function.c (test_loading_mem): Likewise.
2363 * rtlanal.c (may_trap_p_1): Likewise.
2364 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2365 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2366 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2367 (mem_attrs::mem_attrs): New function.
2368 (set_mem_attributes_minus_bitpos): Change bitpos from a
2369 HOST_WIDE_INT to poly_int64.
2370 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2371 (clear_mem_offset, clear_mem_size, change_address)
2372 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2373 initialize mem_attrs.
2374 (set_mem_offset, set_mem_size, adjust_address_1)
2375 (adjust_automodify_address_1, offset_address, widen_memory_access):
2376 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2378 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2379 Alan Hayward <alan.hayward@arm.com>
2380 David Sherwood <david.sherwood@arm.com>
2382 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2383 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2384 rather than 0 to represent an unknown size. Assert that the size
2385 is known when the mode isn't BLKmode.
2386 (may_trap_p_1): Use -1 for unknown sizes.
2387 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2389 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2390 Alan Hayward <alan.hayward@arm.com>
2391 David Sherwood <david.sherwood@arm.com>
2393 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2394 to poly_int64. Update commentary for positions_needed.large.
2395 (read_info_type): Change offset and width from HOST_WIDE_INT
2397 (set_usage_bits): Likewise.
2398 (canon_address): Return the offset as a poly_int64 rather than
2399 a HOST_WIDE_INT. Use strip_offset_and_add.
2400 (set_all_positions_unneeded, any_positions_needed_p): Use
2401 positions_needed.large to track stores with non-constant widths.
2402 (all_positions_needed_p): Likewise. Take the offset and width
2403 as poly_int64s rather than ints. Assert that rhs is nonnull.
2404 (record_store): Cope with non-constant offsets and widths.
2405 Nullify the rhs of an earlier store if we can't tell which bytes
2407 (find_shift_sequence): Take the access_size and shift as poly_int64s
2409 (get_stored_val): Take the read_offset and read_width as poly_int64s
2410 rather than HOST_WIDE_INTs.
2411 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2412 non-constant offsets and widths.
2414 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2415 Alan Hayward <alan.hayward@arm.com>
2416 David Sherwood <david.sherwood@arm.com>
2418 * inchash.h (inchash::hash::add_poly_int): New function.
2419 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2420 Use poly_int64 rather than HOST_WIDE_INT.
2421 (ao_ref::max_size_known_p): New function.
2422 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2423 rather than HOST_WIDE_INT.
2424 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2425 to temporaries until its interface is adjusted to match.
2426 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2427 (aliasing_component_refs_p, decl_refs_may_alias_p)
2428 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2429 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2430 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2432 * alias.c (ao_ref_from_mem): Likewise.
2433 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2434 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2435 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2436 (maybe_trim_complex_store, maybe_trim_constructor_store)
2437 (live_bytes_read, dse_classify_store): Likewise.
2438 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2439 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2440 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2441 (vn_reference_lookup_3): Likewise.
2442 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2444 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2445 Alan Hayward <alan.hayward@arm.com>
2446 David Sherwood <david.sherwood@arm.com>
2448 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2449 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2450 instead of ranges_overlap_p.
2452 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2453 Alan Hayward <alan.hayward@arm.com>
2454 David Sherwood <david.sherwood@arm.com>
2456 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2457 sizes as poly_int64s rather than HOST_WIDE_INTs.
2459 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2460 Alan Hayward <alan.hayward@arm.com>
2461 David Sherwood <david.sherwood@arm.com>
2463 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2464 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2465 * gimple-fold.c (fold_ctor_reference): Likewise.
2467 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2468 Alan Hayward <alan.hayward@arm.com>
2469 David Sherwood <david.sherwood@arm.com>
2471 * target.def (dwarf_poly_indeterminate_value): New hook.
2472 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2473 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2474 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2475 * doc/tm.texi: Regenerate.
2476 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2477 offset as a poly_int64.
2478 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2479 offset as a poly_int64.
2480 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2481 Take the offset as a poly_int64.
2482 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2483 (frame_pointer_fb_offset): Change to a poly_int64.
2484 (int_loc_descriptor): Take the offset as a poly_int64. Use
2485 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2486 (based_loc_descr): Take the offset as a poly_int64.
2487 Use strip_offset_and_add to handle (plus X (const)).
2488 Use new_reg_loc_descr instead of an open-coded version of the
2489 previous implementation.
2490 (mem_loc_descriptor): Handle CONST_POLY_INT.
2491 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2492 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2494 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2495 Alan Hayward <alan.hayward@arm.com>
2496 David Sherwood <david.sherwood@arm.com>
2498 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2499 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2500 * inchash.h (inchash::hash::add_poly_hwi): New function.
2501 * gengtype.c (main): Register poly_int64.
2502 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2503 offset as a poly_int.
2504 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2505 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2506 offset as a poly_int64.
2507 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2508 * print-rtl.c (print_poly_int): New function.
2509 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2511 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2513 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2514 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2515 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2516 Enforce track_offset_p for parts of a PARALLEL.
2517 (vt_add_function_parameter): Use const_offset for the final
2518 offset to track. Use get_tracked_reg_offset for the parts
2521 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2522 Alan Hayward <alan.hayward@arm.com>
2523 David Sherwood <david.sherwood@arm.com>
2525 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2526 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2527 * doc/tm.texi: Regenerate.
2528 * hooks.h (hook_bool_uint_uint_true): Delete.
2529 (hook_bool_puint64_puint64_true): Declare.
2530 * hooks.c (hook_bool_uint_uint_true): Delete.
2531 (hook_bool_puint64_puint64_true): New function.
2532 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2533 instead of unsigned ints.
2534 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2535 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2537 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2538 Alan Hayward <alan.hayward@arm.com>
2539 David Sherwood <david.sherwood@arm.com>
2541 * optabs.h (expand_operand): Add an int_value field.
2542 (create_expand_operand): Add an int_value parameter and use it
2543 to initialize the new expand_operand field.
2544 (create_integer_operand): Replace with a declaration of a function
2545 that accepts poly_int64s. Move the implementation to...
2546 * optabs.c (create_integer_operand): ...here.
2547 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2548 the mode preserves the value of int_value, instead of calling
2549 const_int_operand on the rtx. Use gen_int_mode to generate
2552 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2553 Alan Hayward <alan.hayward@arm.com>
2554 David Sherwood <david.sherwood@arm.com>
2556 * dumpfile.h (dump_dec): Declare.
2557 * dumpfile.c (dump_dec): New function.
2558 * pretty-print.h (pp_wide_integer): Turn into a function and
2559 declare a poly_int version.
2560 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2562 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2563 Alan Hayward <alan.hayward@arm.com>
2564 David Sherwood <david.sherwood@arm.com>
2566 * doc/generic.texi (POLY_INT_CST): Document.
2567 * tree.def (POLY_INT_CST): New tree code.
2568 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2569 * tree-core.h (tree_poly_int_cst): New struct.
2570 (tree_node): Add a poly_int_cst field.
2571 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2572 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2573 instead of a wide_int_ref.
2574 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2576 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2577 instead of an unsigned HOST_WIDE_INT.
2578 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2579 (ptrdiff_tree_p): Declare.
2580 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2581 extern inline implementations if the target doesn't use POLY_INT_CST.
2582 (poly_int_tree_p): New function.
2583 (wi::unextended_tree): New class.
2584 (wi::int_traits <unextended_tree>): New override.
2585 (wi::extended_tree): Add a default constructor.
2586 (wi::extended_tree::get_tree): New function.
2587 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2588 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2589 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2590 (wi::tree_to_poly_wide_ref): New typedefs.
2591 (wi::ints_for): Provide overloads for extended_tree and
2593 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2594 (wi::to_wide): New functions.
2595 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2596 * tree.c (poly_int_cst_hasher): New struct.
2597 (poly_int_cst_hash_table): New variable.
2598 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2599 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2601 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2602 (init_ttree): Initialize poly_int_cst_hash_table.
2603 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2604 a poly_int64 instead of a HOST_WIDE_INT.
2605 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2606 instead of an unsigned HOST_WIDE_INT.
2607 (wide_int_to_tree): Rename to...
2608 (wide_int_to_tree_1): ...this.
2609 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2610 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2611 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2612 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2613 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2614 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2616 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2617 (streamer_read_tree_body): Likewise.
2618 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2619 (streamer_write_tree_body): Likewise.
2620 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2621 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2622 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2623 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2624 * gimple-expr.h (is_gimple_constant): Likewise.
2625 * gimplify.c (maybe_with_size_expr): Likewise.
2626 * print-tree.c (print_node): Likewise.
2627 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2628 * tree-pretty-print.c (dump_generic_node): Likewise.
2629 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2630 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2631 * tree-vrp.c (compare_values_warnv): Likewise.
2632 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2633 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2634 (force_expr_to_var_cost): Likewise.
2635 * tree-ssa-loop.c (for_each_index): Likewise.
2636 * fold-const.h (build_invariant_address, size_int_kind): Take a
2637 poly_int64 instead of a HOST_WIDE_INT.
2638 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2639 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2641 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2642 (int_const_binop_2): New function, split out from...
2643 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2644 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2645 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2646 * gimple-ssa-strength-reduction.c (slsr_process_add)
2647 (slsr_process_mul): Check for INTEGER_CSTs before using them
2649 * stor-layout.c (bits_from_bytes): New function.
2650 (bit_from_pos): Use it.
2651 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2652 by BITS_PER_UNIT to get the TYPE_SIZE.
2653 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2654 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2656 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2657 Alan Hayward <alan.hayward@arm.com>
2658 David Sherwood <david.sherwood@arm.com>
2660 * doc/rtl.texi (const_poly_int): Document. Also document the
2661 rtl sharing behavior.
2662 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2663 * rtl.h (const_poly_int_def): New struct.
2664 (rtx_def::u): Add a cpi field.
2665 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2666 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2667 (wi::rtx_to_poly_wide_ref): New typedef
2668 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2669 (poly_int_rtx_p): New functions.
2670 (trunc_int_for_mode): Declare a poly_int64 version.
2671 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2672 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2674 (strip_offset): Declare.
2675 (strip_offset_and_add): New function.
2676 * rtl.def (CONST_POLY_INT): New rtx code.
2677 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2678 (shared_const_p): Use poly_int_rtx_p.
2679 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2681 (gen_int_shift_amount): Likewise.
2682 * emit-rtl.c (const_poly_int_hasher): New class.
2683 (const_poly_int_htab): New variable.
2684 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2685 (const_poly_int_hasher::hash): New function.
2686 (const_poly_int_hasher::equal): Likewise.
2687 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2688 (immed_wide_int_const): Rename to...
2689 (immed_wide_int_const_1): ...this and make static.
2690 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2691 instead of a wide_int_ref.
2692 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2693 (gen_lowpart_common): Handle CONST_POLY_INT.
2694 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2695 * cselib.c (cselib_hash_rtx): Likewise.
2696 * dwarf2out.c (const_ok_for_output_1): Likewise.
2697 * expr.c (convert_modes): Likewise.
2698 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2699 * rtlhash.c (add_rtx): Likewise.
2700 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2701 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2702 Handle existing CONST_POLY_INT rtxes.
2703 * expmed.h (expand_shift): Take a poly_int64 instead of a
2705 * expmed.c (expand_shift): Likewise.
2706 * rtlanal.c (strip_offset): New function.
2707 (commutative_operand_precedence): Give CONST_POLY_INT the same
2708 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2710 * rtl-tests.c (const_poly_int_tests): New struct.
2711 (rtl_tests_c_tests): Use it.
2712 * simplify-rtx.c (simplify_const_unary_operation): Handle
2714 (simplify_const_binary_operation): Likewise.
2715 (simplify_binary_operation_1): Fold additions of symbolic constants
2716 and CONST_POLY_INTs.
2717 (simplify_subreg): Handle extensions and truncations of
2719 (simplify_const_poly_int_tests): New struct.
2720 (simplify_rtx_c_tests): Use it.
2721 * wide-int.h (storage_ref): Add default constructor.
2722 (wide_int_ref_storage): Likewise.
2723 (trailing_wide_ints): Use GTY((user)).
2724 (trailing_wide_ints::operator[]): Add a const version.
2725 (trailing_wide_ints::get_precision): New function.
2726 (trailing_wide_ints::extra_size): Likewise.
2728 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2729 Alan Hayward <alan.hayward@arm.com>
2730 David Sherwood <david.sherwood@arm.com>
2732 * emit-rtl.h (gen_int_shift_amount): Declare.
2733 * emit-rtl.c (gen_int_shift_amount): New function.
2734 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2736 * calls.c (shift_return_value): Likewise.
2737 * cse.c (fold_rtx): Likewise.
2738 * dse.c (find_shift_sequence): Likewise.
2739 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2740 (expand_shift, expand_smod_pow2): Likewise.
2741 * lower-subreg.c (shift_cost): Likewise.
2742 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2743 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2744 (expand_vec_perm_var): Likewise.
2745 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2746 (simplify_binary_operation_1): Likewise.
2747 * combine.c (try_combine, find_split_point, force_int_to_mode)
2748 (simplify_shift_const_1, simplify_shift_const): Likewise.
2749 (change_zero_ext): Likewise. Use simplify_gen_binary.
2751 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2753 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2755 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2757 * doc/invoke.texi (ARM Options): Document accepted extension options
2758 for -march=armv8.3-a.
2760 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2763 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2764 or --with-float=softfp, set the default CPU to arm10e.
2766 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2768 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2769 * config/visium/predicates.md (const_shift_operand): Likewise.
2770 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2771 (visium_legitimize_reload_address): Likewise.
2773 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2776 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2778 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2781 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2783 2017-12-20 Julia Koval <julia.koval@intel.com>
2785 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2786 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2787 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2788 * config.gcc: Include vpclmulqdqintrin.h.
2789 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2790 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2792 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2793 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2794 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2795 * config/i386/i386.c (isa2_opts): Add -mcx16.
2796 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2797 (ix86_option_override_internal): Move mcx16 to flags2.
2798 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2799 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2800 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2801 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2802 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2803 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2804 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2805 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2806 * doc/invoke.texi: Add -mvpclmulqdq.
2808 2017-12-20 Tom de Vries <tom@codesourcery.com>
2811 * config/i386/i386.c (ix86_static_chain): Move
2812 DECL_STATIC_CHAIN test ...
2813 * calls.c (rtx_for_static_chain): ... here. New function.
2814 * calls.h (rtx_for_static_chain): Declare.
2815 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2816 instead of targetm.calls.static_chain.
2817 * df-scan.c (df_get_entry_block_def_set): Same.
2819 2017-12-19 Tom de Vries <tom@codesourcery.com>
2821 PR tree-optimization/83493
2822 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2825 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2827 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2828 inform with hardcoded english plural handling.
2830 2017-12-18 Jeff Law <law@redhat.com>
2832 PR tree-optimization/83477
2833 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2834 a non-virtual PHI, always push a new range.
2836 2017-12-19 Martin Sebor <msebor@redhat.com>
2839 * builtins.c (compute_objsize): Handle non-constant offsets.
2841 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2843 PR tree-optimization/83444
2844 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2845 character load case, if get_stridx on MEM_REF's operand doesn't
2846 look usable, retry with get_addr_stridx.
2848 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2851 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2852 (variable_tracking_main_1): Keep markers even when VTA fails.
2855 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2856 even if there are markers before it.
2857 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2859 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2861 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2862 typical order conditions.
2863 * sel-sched.c (extract_new_fences_from): Likewise.
2864 * config/visium/constraints.md (J, K, L): Likewise.
2865 * config/visium/predicates.md (const_shift_operand): Likewise.
2866 * config/visium/visium.c (visium_legitimize_address,
2867 visium_legitimize_reload_address): Likewise.
2868 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2869 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2870 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2871 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2872 SET_DUMP_DETAIL): Likewise.
2873 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2874 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2875 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2876 avr_set_core_architecture, avr_set_current_function,
2877 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2878 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2879 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2880 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2881 avr_map_decompose, avr_fold_builtin): Likewise.
2882 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2883 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2884 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2885 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2886 * config/m32c/m32c.c (m32c_conditional_register_usage,
2887 m32c_address_cost): Likewise.
2888 * config/m32c/predicates.md (shiftcount_operand,
2889 longshiftcount_operand): Likewise.
2890 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2891 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2892 can_use_cdx_ldstw): Likewise.
2893 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2894 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2896 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2897 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2898 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2899 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2900 * config/vax/vax.c (adjacent_operands_p): Likewise.
2901 * config/ft32/constraints.md (L, b, KA): Likewise.
2902 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2904 * cfgexpand.c (expand_stack_alignment): Likewise.
2905 * gcse.c (insert_expr_in_table): Likewise.
2906 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2907 * cgraphunit.c (cgraph_node::expand): Likewise.
2908 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2909 * emit-rtl.c (add_insn): Likewise.
2910 * input.c (dump_location_info): Likewise.
2911 * passes.c (NEXT_PASS): Likewise.
2912 * read-rtl-function.c (parse_note_insn_name,
2913 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2915 * sched-rgn.c (sched_rgn_init): Likewise.
2916 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2917 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2918 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2920 * explow.c (eliminate_constant_term): Likewise.
2921 * final.c (leaf_renumber_regs_insn): Likewise.
2922 * cfgrtl.c (print_rtl_with_bb): Likewise.
2923 * genhooks.c (emit_init_macros): Likewise.
2924 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2925 * tree-data-ref.c (conflict_fn): Likewise.
2926 * selftest.c (assert_streq): Likewise.
2927 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2928 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2929 fold_binary_loc, multiple_of_p): Likewise.
2930 * reload.c (push_reload, find_equiv_reg): Likewise.
2931 * et-forest.c (et_nca, et_below): Likewise.
2932 * dbxout.c (dbxout_symbol_location): Likewise.
2933 * reorg.c (relax_delay_slots): Likewise.
2934 * dojump.c (do_compare_rtx_and_jump): Likewise.
2935 * gengtype-parse.c (type): Likewise.
2936 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2937 simplify_const_relational_operation): Likewise.
2938 * reload1.c (do_output_reload): Likewise.
2939 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2940 * gengtype.c (type_for_name): Likewise.
2941 * gimple-ssa-sprintf.c (format_directive): Likewise.
2943 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2946 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2947 accessing it. Adjust comment.
2949 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2952 * predict.c (zero_one_minusone): New function.
2953 (apply_return_prediction): Avoid return prediction for functions
2954 returning only -1, 0 and 1 values, unless they only return -1 and 0
2957 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2959 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2960 fall-through warning.
2961 (arc_compute_frame_size): Remove unused variables.
2962 (arc_print_operand): Fix fprintif format.
2963 (arc_can_follow_jump): Clean fall-through warning.
2965 2017-12-19 Marek Polacek <polacek@redhat.com>
2968 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2971 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2973 * config/arc/arc.c (overriderregs): New variable.
2974 (arc_override_options): Track fixed/call saved/call options.
2975 (arc_conditional_register_usage): Check against overrideregs
2976 variable whenever we change register properties.
2978 2017-12-19 Nathan Sidwell <nathan@acm.org>
2980 * opts.c (finish_options): Don't prefix dump_base_name if it
2981 already contains directories.
2983 2017-12-19 Martin Liska <mliska@suse.cz>
2985 PR rtl-optimization/82675
2986 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2987 more element in sbitmap.
2989 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2991 * gimplify.c (gimplify_expr): Use error_operand_p.
2993 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2996 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2999 2017-12-19 Daniel Cederman <cederman@gaisler.com>
3001 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
3002 the jump is to a label.
3004 2017-12-19 Jakub Jelinek <jakub@redhat.com>
3006 PR tree-optimization/83444
3007 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
3012 * ipa-inline.c (flatten_remove_node_hook): New function.
3013 (ipa_inline): Keep only nodes with flatten attribute at the end of
3014 the array in the order from ipa_reverse_postorder, only walk that
3015 portion of array for flattening, if there is more than one such
3016 node, temporarily register a removal hook and ignore removed nodes.
3018 PR tree-optimization/80631
3019 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
3020 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
3021 IFN_REDUC_MAX or IFN_REDUC_MIN.
3023 2017-12-18 Jeff Law <law@redhat.com>
3025 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
3026 record symbolic equivalences from backedges in the CFG.
3029 2017-11-19 Jeff Law <law@redhat.com>
3031 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3032 of degenerates resulting from ignoring an edge.
3034 2017-12-18 Martin Sebor <msebor@redhat.com>
3037 PR tree-optimization/78450
3038 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
3039 (handle_builtin_strlen): Call it.
3041 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
3043 PR rtl-optimization/83424
3044 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
3046 2017-12-18 Marek Polacek <polacek@redhat.com>
3049 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
3050 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
3053 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
3055 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
3056 (umaddsidi4, umaddsidi_split): Likewise.
3058 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
3060 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
3063 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3066 * config/s390/s390.c (s390_option_override): Avoid strncpy.
3068 2017-12-18 Richard Biener <rguenther@suse.de>
3070 PR tree-optimization/81877
3071 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
3072 (outermost_indep_loop): Adjust.
3073 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
3074 (can_sm_ref_p): Adjust.
3076 2017-12-18 Richard Biener <rguenther@suse.de>
3079 * tree.c (array_at_struct_end_p): Return true if the underlying
3080 object has space for at least one element in excess of what
3081 the array domain specifies.
3083 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3085 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
3086 example to avoid overfull hbox.
3087 * doc/invoke.texi (Option Summary): Add missing @gol.
3088 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
3091 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3092 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3094 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
3095 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
3096 (AArch64 Options): Document -mverbose-cost-dump.
3097 (ARM Options): Likewise, plus -mflip-thumb.
3099 2017-12-17 Martin Sebor <msebor@redhat.com>
3102 * gimple-ssa-warn-restrict.c
3103 (builtin_memref::offset_out_of_bounds): Correct the handling of
3106 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3108 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
3110 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
3112 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
3114 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
3116 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
3119 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
3121 * doc/invoke.texi: Fix some typos.
3123 2017-12-16 Martin Sebor <msebor@redhat.com>
3125 PR tree-optimization/78918
3126 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
3127 * builtins.c (check_sizes): Rename...
3128 (check_access): ...to this. Rename function arguments for clarity.
3129 (check_memop_sizes): Adjust names.
3130 (expand_builtin_memchr, expand_builtin_memcpy): Same.
3131 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
3132 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
3133 (check_strncat_sizes, expand_builtin_strncat): Same.
3134 (expand_builtin_strncpy, expand_builtin_memset): Same.
3135 (expand_builtin_bzero, expand_builtin_memcmp): Same.
3136 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
3137 (maybe_emit_sprintf_chk_warning): Same.
3138 (expand_builtin_strcpy): Adjust.
3139 (expand_builtin_stpcpy): Same.
3140 (expand_builtin_with_bounds): Detect out-of-bounds accesses
3141 in pointer-checking forms of memcpy, memmove, and mempcpy.
3142 (gcall_to_tree_minimal, max_object_size): Define new functions.
3143 * builtins.h (max_object_size): Declare.
3144 * calls.c (alloc_max_size): Call max_object_size instead of
3145 hardcoding ssizetype limit.
3146 (get_size_range): Handle new argument.
3147 * calls.h (get_size_range): Add a new argument.
3148 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
3149 * doc/invoke.texi (-Wrestrict): Adjust, add example.
3150 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
3152 (gimple_fold_builtin_memory_chk): Same.
3153 (gimple_fold_builtin_stxcpy_chk): New function.
3154 * gimple-ssa-warn-restrict.c: New source.
3155 * gimple-ssa-warn-restrict.h: New header.
3156 * gimple.c (gimple_build_call_from_tree): Propagate location.
3157 * passes.def (pass_warn_restrict): Add new pass.
3158 * tree-pass.h (make_pass_warn_restrict): Declare.
3159 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
3161 (handle_builtin_strcat): Same.
3162 (strlen_optimize_stmt): Rename...
3163 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
3164 stpncpy, strncpy, and their checking forms.
3166 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
3168 PR rtl-optimization/82849
3169 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
3170 and get_max_loop_iterations_int.
3172 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3173 Alan Hayward <alan.hayward@arm.com>
3174 David Sherwood <david.sherwood@arm.com>
3176 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
3177 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
3178 size as a poly_uint64.
3179 (mode_for_vector, mode_for_int_vector): Take the number of vector
3180 elements as a poly_uint64.
3181 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
3182 size as a poly_uint64.
3183 (mode_for_vector, mode_for_int_vector): Take the number of vector
3184 elements as a poly_uint64.
3186 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3187 Alan Hayward <alan.hayward@arm.com>
3188 David Sherwood <david.sherwood@arm.com>
3190 * machmode.h (MACRO_MODE): New macro.
3191 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
3192 * targhooks.c (default_libcall_value, default_secondary_reload)
3193 (default_memory_move_cost, default_register_move_cost)
3194 (default_class_max_nregs): Likewise.
3196 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3197 Alan Hayward <alan.hayward@arm.com>
3198 David Sherwood <david.sherwood@arm.com>
3200 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
3201 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3203 * genautomata.c (main): Likewise.
3204 * genconditions.c (write_header): Likewise.
3205 * genemit.c (main): Likewise.
3206 * genextract.c (print_header): Likewise.
3207 * genopinit.c (main): Likewise.
3208 * genoutput.c (output_prologue): Likewise.
3209 * genpeep.c (main): Likewise.
3210 * genpreds.c (write_insn_preds_c): Likewise.
3211 * genrecog.c (writer_header): Likewise.
3212 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3213 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3214 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3215 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3216 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3217 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3218 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3219 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3220 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3221 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3222 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3223 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3224 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3225 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3226 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3227 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3228 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3229 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3230 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3231 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3232 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3233 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3234 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3235 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3236 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3237 * config/darwin.c (IN_TARGET_CODE): Likewise.
3238 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3239 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3240 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3241 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3242 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3243 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3244 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3245 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3246 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3247 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3248 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3249 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3250 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3251 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3252 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3253 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3254 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3255 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3256 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3257 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3258 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3259 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3260 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3261 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3262 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3263 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3264 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3265 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3266 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3267 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3268 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3269 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3270 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3271 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3272 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3273 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3274 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3275 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3276 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3277 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3278 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3279 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3280 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3281 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3282 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3283 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3284 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3285 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3286 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3287 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3288 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3289 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3290 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3291 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3292 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3293 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3294 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3295 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3296 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3297 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3298 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3299 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3300 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3301 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3302 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3303 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3304 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3305 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3306 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3307 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3308 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3309 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3310 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3311 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3312 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3313 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3314 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3315 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3316 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3317 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3318 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3319 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3320 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3321 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3322 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3323 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3324 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3325 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3326 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3327 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3328 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3329 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3330 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3331 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3332 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3333 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3334 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3335 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3336 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3337 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3338 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3339 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3340 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3341 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3342 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3343 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3345 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3346 Alan Hayward <alan.hayward@arm.com>
3347 David Sherwood <david.sherwood@arm.com>
3349 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3352 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3353 Alan Hayward <alan.hayward@arm.com>
3354 David Sherwood <david.sherwood@arm.com>
3356 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3357 * doc/md.texi (vec_series@var{m}): Document.
3358 * tree.def (VEC_SERIES_EXPR): New tree code.
3359 * tree.h (build_vec_series): Declare.
3360 * tree.c (build_vec_series): New function.
3361 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3362 * tree-pretty-print.c (dump_generic_node): Likewise.
3363 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3364 * tree-inline.c (estimate_operator_cost): Likewise.
3365 * expr.c (expand_expr_real_2): Likewise.
3366 * optabs-tree.c (optab_for_tree_code): Likewise.
3367 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3368 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3369 * expmed.c (make_tree): Handle VEC_SERIES.
3370 * optabs.def (vec_series_optab): New optab.
3371 * optabs.h (expand_vec_series_expr): Declare.
3372 * optabs.c (expand_vec_series_expr): New function.
3373 * tree-vect-generic.c (expand_vector_operations_1): Check that
3374 the operands also have vector type.
3376 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3377 Alan Hayward <alan.hawyard@arm.com>
3378 David Sherwood <david.sherwood@arm.com>
3380 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3381 (VEC_COND_EXPR): Add missing @tindex.
3382 * doc/md.texi (vec_duplicate@var{m}): Document.
3383 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3384 * tree.c (build_vector_from_val): Add stubbed-out handling of
3385 variable-length vectors, using VEC_DUPLICATE_EXPR.
3386 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3387 * cfgexpand.c (expand_debug_expr): Likewise.
3388 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3389 * tree-inline.c (estimate_operator_cost): Likewise.
3390 * tree-pretty-print.c (dump_generic_node): Likewise.
3391 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3392 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3393 (test_vec_duplicate_folding): New function.
3394 (fold_const_c_tests): Call it.
3395 * optabs.def (vec_duplicate_optab): New optab.
3396 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3397 * optabs.h (expand_vector_broadcast): Declare.
3398 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3399 vec_duplicate_optab.
3400 * expr.c (store_constructor): Try using vec_duplicate_optab for
3402 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3404 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3407 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3408 div/mod latencies a bit.
3410 2017-12-15 Jeff Law <law@redhat.com>
3412 PR tree-optimization/36550
3413 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3414 (mark_threaded_blocks): Rewrite code to avoid block copying when
3415 optimizing for size. Don't pessimize blocks which will be
3416 copied, but all the statements will be dead.
3418 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3420 PR tree-optimization/81165
3421 * tree-ssa-threadupdate.c (uses_in_bb): New.
3422 (estimate_threading_killed_stmts): New.
3423 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3424 * tree-ssa-threadedge.c
3425 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3428 2017-12-15 Jeff Law <law@redhat.com>
3430 PR tree-optimization/83410
3431 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3432 threads when parallelizing loops.
3434 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3436 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3438 * config/alpha/alpha.c (vms_attribute_table): Swap
3439 affects_type_identity and handler fields, adjust comments.
3440 * config/mips/mips.c (mips_attribute_table): Likewise.
3441 * config/visium/visium.c (visium_attribute_table): Likewise.
3442 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3443 * config/microblaze/microblaze.c (microblaze_attribute_table):
3445 * config/spu/spu.c (spu_attribute_table): Likewise.
3446 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3447 * config/arc/arc.c (arc_attribute_table): Likewise.
3448 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3449 * config/v850/v850.c (v850_handle_interrupt_attribute,
3450 v850_handle_data_area_attribute): Formatting fixes.
3451 (v850_attribute_table): Swap affects_type_identity and handler
3452 fields, adjust comments.
3453 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3454 * config/arm/arm.c (arm_attribute_table): Likewise.
3455 * config/avr/avr.c (avr_attribute_table): Likewise.
3456 * config/s390/s390.c (s390_attribute_table): Likewise.
3457 * config/sh/sh.c (sh_attribute_table): Likewise.
3458 * config/i386/i386.c (ix86_handle_cconv_attribute,
3459 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3460 (ix86_attribute_table): Swap affects_type_identity and handler
3461 fields, adjust comments.
3462 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3463 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3464 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3465 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3466 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3467 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3468 * config/rx/rx.c (rx_attribute_table): Likewise.
3469 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3470 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3471 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3472 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3473 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3474 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3475 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3476 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3477 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3478 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3479 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3481 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3483 (register_scoped_attributes, decl_attributes): Formatting fixes.
3485 PR tree-optimization/83269
3486 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3487 subtraction in arg0's type if type is signed and arg0 is unsigned.
3491 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3492 convert? on @0 instead of convert. Check type of @1, not @0.
3493 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3494 convert? on @0 instead of convert. Check type of @1, not @0.
3495 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3496 only optimize if either both @1 and @2 types are narrower
3497 precision, or both are wider or equal precision, and in the former
3498 case only if both have undefined overflow.
3500 2017-12-15 Richard Biener <rguenther@suse.de>
3503 * internal-fn.def (IFN_NOP): Add.
3504 * internal-fn.c (expand_NOP): Do nothing.
3505 * lto-streamer-in.c (input_function): Instead of removing
3506 sanitizer calls replace them with IFN_NOP calls.
3508 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3509 Alan Hayward <alan.hayward@arm.com>
3510 David Sherwood <david.sherwood@arm.com>
3512 * dse.c (store_info, read_info_type): Replace begin and end with
3514 (print_range): New function.
3515 (set_all_positions_unneeded, any_positions_needed_p)
3516 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3518 (record_store): Likewise. Optimize the case in which all positions
3520 (get_stored_val): Replace read_begin and read_end with read_offset
3522 (replace_read): Update call accordingly.
3524 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3526 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3527 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3528 (loop_cand::loop_cand): Initialize above members.
3529 (loop_cand::supported_operations): Delete.
3530 (loop_cand::can_interchange_p): Inline above function.
3531 (loop_cand::classify_simple_reduction): Record number of constant
3532 initialized simple reductions.
3533 (should_interchange_loops): New parameters. Check stmt cost of loops
3535 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3536 Update call to should_interchange_loops.
3537 (should_interchange_loop_nest): Update call to
3538 should_interchange_loops.
3540 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3543 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3544 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3545 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3547 2017-12-15 Julia Koval <julia.koval@intel.com>
3549 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3550 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3552 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3553 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3554 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3556 2017-12-15 Julia Koval <julia.koval@intel.com>
3558 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3559 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3560 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3561 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3562 _mm_aesenc_epi128): New intrinsics.
3564 2017-12-15 Julia Koval <julia.koval@intel.com>
3566 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3567 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3569 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3570 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3571 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3573 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3575 * gimple-ssa-strength-reduction.c (analyze_increments):
3576 Distinguish replacement costs for constant strides from those for
3579 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3581 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3583 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3585 * doc/invoke.texi: Document -Wcast-function-type.
3586 * recog.h (stored_funcptr): Change signature.
3587 * tree-dump.c (dump_node): Avoid warning.
3588 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3590 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3593 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3596 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3598 PR tree-optimization/83312
3599 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3600 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3601 "val" param, and to cope with arbitrary basic blocks.
3602 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3603 handle NULL_TREE for "val", dropping "bb" param.
3604 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3605 drop "bb" param. Handle NULL_TREE for "val".
3606 (find_case_label_for_value): Make "switch_stmt" param const.
3607 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3609 (vrp_prop::check_all_array_refs): Reimplement as...
3610 (check_array_bounds_dom_walker::before_dom_children): ...this new
3611 vfunc. Replace linear search through BB block list, excluding
3612 those with non-executable in-edges via dominator walk.
3614 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3616 * config/arm/arm.opt (mverbose-cost-dump): New option.
3617 * config/arm/arm.c (arm_rtx_costs): Use it.
3619 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3622 * reload1.c (emit_input_reload_insns): Skip debug markers.
3624 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3626 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3630 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3631 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3632 * config/sh/sh.c: Skip debug insns besides notes.
3633 * config/sh/sh.md: Likewise.
3634 * config/sh/sh_treg_combine.cc: Likewise.
3635 * config/sh/sync.md: Likewise.
3637 2017-12-14 Tom de Vries <tom@codesourcery.com>
3639 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3640 Add item for weak_undefined.
3642 2017-12-14 Richard Biener <rguenther@suse.de>
3644 PR tree-optimization/67842
3645 * sese.h (bb_in_region): Remove #if 0'ed code.
3647 2017-12-14 Richard Biener <rguenther@suse.de>
3649 PR tree-optimization/83326
3650 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3651 may_be_zero parameter and handle it by not marking the first
3652 peeled copy as not exiting the loop.
3653 (try_peel_loop): Likewise.
3654 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3655 to handle the case of constant or zero iterations and perform
3656 loop header copying on-the-fly.
3658 2017-12-14 Richard Biener <rguenther@suse.de>
3660 PR tree-optimization/83418
3661 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3662 Instead of asserting we don't get unfolded comparisons deal with them.
3664 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3667 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3668 basic blocks. Assert debug bind insns don't appear outside of bbs,
3669 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3672 PR tree-optimization/83198
3673 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3674 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3675 value if arg is a REAL_CST with incompatible type.
3677 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3678 Bin Cheng <bin.cheng@arm.com>
3681 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3683 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3686 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3688 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3689 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3691 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3692 Alan Hayward <alan.hayward@arm.com>
3693 David Sherwood <david.sherwood@arm.com>
3695 * poly-int.h: New file.
3696 * poly-int-types.h: Likewise.
3697 * coretypes.h: Include them.
3698 (POLY_INT_CONVERSION): Define.
3699 * target.def (estimated_poly_value): New hook.
3700 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3701 * doc/tm.texi: Regenerate.
3702 * doc/poly-int.texi: New file.
3703 * doc/gccint.texi: Include it.
3704 * doc/rtl.texi: Describe restrictions on subreg modes.
3705 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3706 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3707 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3708 * targhooks.h (default_estimated_poly_value): Declare.
3709 * targhooks.c (default_estimated_poly_value): New function.
3710 * target.h (estimated_poly_value): Likewise.
3711 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3712 (wi::unary_traits): Delete.
3713 (wi::binary_traits::signed_shift_result_type): Define for
3714 offset_int << HOST_WIDE_INT, etc.
3715 (generic_wide_int::operator <<=): Define for all types and use
3716 wi::lshift instead of <<.
3717 (wi::hwi_with_prec): Add a default constructor.
3718 (wi::ints_for): New class.
3719 (operator <<): Define for all wide-int types.
3720 (operator /): New function.
3721 (operator %): Likewise.
3722 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3723 (ASSERT_MAYBE_NE_AT): New macros.
3725 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3726 Dominik Vogt <vogt@linux.vnet.ibm.com>
3729 * emit-rtl.c (init_emit): Remove ??? comment.
3730 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3731 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3732 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3733 alignment of 3 virtual registers to BITS_PER_WORD.
3735 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3737 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3739 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3740 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3742 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3745 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3747 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3749 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3750 with the exception of debug begin stmt markers appear before
3754 * final.c (rest_of_handle_final): Call variable_tracking_main only
3755 if !flag_var_tracking.
3757 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3758 Jakub Jelinek <jakub@redhat.com>
3762 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3763 labels debug stmts that can only appear after labels.
3765 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3767 PR rtl-optimization/82398
3768 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3769 EXPR_USEFULNESS in priority comparison.
3771 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3773 PR rtl-optimization/83393
3774 * combine.c (move_deaths): If reg_stat points to a too new insn in
3775 last_death, do not use it: find the proper insn instead.
3777 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3780 * config/rs6000/rs6000-string.c (expand_block_compare,
3781 expand_strn_compare): Fix set_mem_size() calls.
3783 2017-12-12 Jeff Law <law@redhat.com>
3785 PR tree-optimization/83298
3786 PR tree-optimization/83362
3787 PR tree-optimization/83383
3788 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3789 push_value_range a public interface. Add new argument to
3790 record_ranges_from_stmt.
3791 * gimple-ssa-evrp-analyze.c
3792 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3793 Update comments. Handle recording temporary equivalences.
3794 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3795 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3796 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3797 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3798 gimple-ssa-evrp-analyze.h.
3799 (record_temporary_equivalences_from_phis): Add new argument. When
3800 the PHI arg is an SSA_NAME, set the result's range to the range
3802 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3803 from statements too.
3804 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3805 Pass it down to children as needed.
3806 (thread_outgoing_edges): Likewise.
3807 (thread_across_edge): Likewise. Push/pop range state as needed.
3808 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3810 2017-12-12 Julia Koval <julia.koval@intel.com>
3812 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3813 (PTA_CANNONLAKE): Remove PTA_CLWB.
3815 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3818 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3819 (vcondv2div2df): Likewise.
3820 (vconduv2dfv2di): Likewise.
3822 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3824 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3825 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3826 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3827 ATTR_MATHFN_FPROUNDING_ERRNO.
3829 2017-12-12 Richard Biener <rguenther@suse.de>
3831 PR tree-optimization/83385
3832 * graphite-scop-detection.c (get_order, order): Remove.
3833 (bb_to_rpo): New global.
3835 (build_scops): Sort pbbs in RPO order.
3837 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3839 * combine.c (simplify_set): Do not transform subregs to zero_extends
3840 if the destination mode is a vector mode.
3842 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3844 PR tree-optimization/83359
3845 * tree-cfg.h (fold_loop_internal_call): Declare.
3846 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3847 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3848 (find_loop_dist_alias): New function.
3849 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3850 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3853 PR tree-optimization/80631
3854 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3855 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3856 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3857 hardcoding zero as the value if COND_EXPR is never true. For
3858 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3859 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3860 hardcoding MAX_EXPR as the reduction operation.
3861 (is_nonwrapping_integer_induction): Allow negative step.
3862 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3863 vect_create_epilog_for_reduction, if no value is suitable, don't
3864 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3866 2017-12-12 Richard Biener <rguenther@suse.de>
3868 PR tree-optimization/81889
3869 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3870 range info from the non-wrapping IV instead of just the range
3873 2017-12-12 Julia Koval <julia.koval@intel.com>
3875 * config.gcc: Add vaesintrin.h.
3876 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3878 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3879 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3881 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3882 * config/i386/immintrin.h: Include vaesintrin.h.
3883 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3884 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3885 _mm_aesdec_epi128): New intrinsics.
3887 2017-12-12 Julia Koval <julia.koval@intel.com>
3889 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3890 OPTION_MASK_ISA_VAES_UNSET): New.
3891 (ix86_handle_option): Handle -mvaes.
3892 * config/i386/cpuid.h: Define bit_VAES.
3893 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3894 * config/i386/i386-c.c (__VAES__): New.
3895 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3896 (ix86_valid_target_attribute_inner_p): Ditto.
3897 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3898 * config/i386/i386.opt: Add -mvaes.
3899 * doc/invoke.texi: Ditto.
3901 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3903 * debug.h (gcc_debug_hooks): Add inline_entry.
3904 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3905 * debug.c (do_nothing_debug_hooks): Likewise.
3906 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3907 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3908 (dwarf2_lineno_debug_hooks): Likewise.
3910 * common.opt (gstatement-frontiers): New, setting
3911 debug_nonbind_markers_p.
3912 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3913 * toplev.c (process_options): Autodetect value for debug statement
3915 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3916 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3918 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3919 markers. Integrate source bind into debug stmt expand loop.
3920 (pass_expand::execute): Check debug marker limit. Avoid deep
3921 TER and expand debug locations for debug bind insns only.
3922 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3924 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3925 * final.c (reemit_insn_block_notes): Take current block from
3926 nonbind markers. Declare note where it's first set.
3927 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3928 begin stmt markers if enabled.
3929 (notice_source_line): Handle nonbind markers. Fail if their
3930 location is unknown or that of builtins.
3931 (rest_of_handle_final): Convert begin stmt markers to notes if
3932 var-tracking didn't run.
3933 (rest_of_clean_state): Skip begin stmt markers.
3934 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3936 * function.c (allocate_struct_function): Set begin_stmt_markers.
3937 * function.h (struct function): Add debug_marker_count counter
3938 and debug_nonbind_markers flag.
3939 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3940 * gimple-low.c (lower_function_body): Adjust
3941 debug_nonbind_markers.
3942 (lower_stmt): Drop or skip gimple debug stmts.
3943 (lower_try_catch): Skip debug stmts.
3944 * gimple.c (gimple_build_debug_begin_stmt): New.
3945 (gimple_copy): Increment debug_marker_count if copying one.
3946 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3947 * gimplify.c (rexpr_location): New.
3948 (rexpr_has_location): New.
3949 (warn_switch_unreachable_r): Handle gimple debug stmts.
3950 (shortcut_cond_r): Call expr_location.
3952 (find_goto_label): New.
3953 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3955 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3957 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3958 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3959 (LANG_HOOKS_INITIALIZER): ... this.
3960 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3961 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3963 (update_ebb_live_info): Skip debug insn markers.
3964 * lra.c (debug_insn_static_data): Rename to...
3965 (debug_bind_static_data): ... this.
3966 (debug_marker_static_data): New.
3967 (lra_set_insn_recog_data): Select one of the above depending
3969 (lra_update_isn_regno_info): Don't assume debug insns have
3971 (push_insns): Skip debug insns.
3972 * lto-streamer-in.c (input_function): Drop debug stmts
3973 depending on active options. Adjust debug_nonbind_markers.
3974 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3975 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3976 begin stmt marker notes.
3977 (print_insn): Likewise.
3978 * recog.c (extract_insn): Recognize rtl for debug markers.
3979 * rtl.def (DEBUG_MARKER): New.
3980 * tree-inline.c: Include params.h.
3981 (remap_gimple_stmt): Handle nonbind markers.
3982 (maybe_move_debug_stmts_to_successors): Likewise.
3983 (copy_debug_stmt): Likewise.
3984 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3985 markers regardless of no side effects.
3986 (tsi_link_before): Don't update container's side effects when adding
3987 a begin stmt marker.
3988 (tsi_link_after): Likewise.
3989 (expr_first): Skip begin stmt markers.
3990 (expr_last): Likewise.
3991 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3992 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3993 Disregard nonbind markers.
3994 * tree.c (make_node_stat): Don't set side effects for begin stmt
3996 (build1_stat): Likewise.
3997 * tree.def (DEBUG_BEGIN_STMT): New.
3998 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3999 * var-tracking.c (delete_debug_insns): Renamed to...
4000 (delete_vta_debug_insns): ... this.
4001 (reemit_marker_as_note): New.
4002 (vt_initialize): Reemit markers.
4003 (delete_vta_debug_insns): Likewise.
4004 (vt_debug_insns_local): Reemit or delete markers.
4005 (variable_tracking_main_1): Likewise.
4006 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
4007 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
4008 (gimple_debug_nonbind_marker_p): New.
4009 (gimple_build_debug_bind): Adjust.
4010 (gimple_build_debug_begin_stmt): New.
4011 * doc/invoke.texi (max-debug-marker-count): New param.
4012 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
4013 (debug_parameter_ref, debug_marker): New.
4014 (NOTE_INSN_BEGIN_STMT): New.
4015 (DEBUG_INSN): Describe begin stmt markers.
4017 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
4018 without debug insns, we wouldn't, but clean up debug insns
4019 after a control flow insn nevertheless.
4021 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
4022 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
4023 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
4024 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
4025 (rtl_tidy_fallthru_edge): Likewise.
4026 (rtl_verify_fallthru): Likewise.
4027 (rtl_verify_bb_layout): Likewise.
4028 (skip_insns_after_block): Likewise.
4029 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
4030 * dwarf2out.c: Include print-rtl.h.
4031 (dwarf2out_next_real_insn): New.
4032 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
4033 Dump debug binds in asm comments.
4034 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
4035 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
4036 callers to use gsi_start_nondebug_bb instead.
4037 (gsi_after_labels): Skip gimple debug stmts.
4038 (gsi_start_nondebug): New.
4039 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
4040 (proper_loop_form_for_interchange): Adjust.
4041 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
4042 * gimple.h (gimple_seq_last_nondebug_stmt): New.
4043 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
4044 (collect_fallthrough_labels): Likewise.
4045 (should_warn_for_implicit_fallthrough): Likewise.
4046 (warn_implicit_fallthrough_r): Likewise.
4047 (expand_FALLTHROUGH_r): Likewise.
4048 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
4049 (graphite_copy_stmts_from_block): Skip nonbind markers.
4050 * haifa-sched.c (sched_extend_bb): Skip debug insns.
4051 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
4052 * jump.c (clean_barriers): Skip debug insns.
4053 * omp-expand.c (expand_parallel_call): Skip debug insns.
4054 (expand_task_call): Likewise.
4055 (remove_exit_barrier): Likewise.
4056 (expand_omp_taskreg): Likewise.
4057 (expand_omp_for_init_counts): Likewise.
4058 (expand_omp_for_generic): Likewise.
4059 (expand_omp_for_static_nochunk): Likewise.
4060 (expand_omp_for_static_chunk): Likewise.
4061 (expand_omp_simd): Likewise.
4062 (expand_omp_taskloop_for_outer): Likewise.
4063 (expand_omp_taskloop_for_inner): Likewise.
4064 (expand_oacc_for): Likewise.
4065 (expand_omp_sections): Likewise.
4066 (expand_omp_single): Likewise.
4067 (expand_omp_synch): Likewise.
4068 (expand_omp_atomic_load): Likewise.
4069 (expand_omp_atomic_store): Likewise.
4070 (expand_omp_atomic_fetch_op): Likewise.
4071 (expand_omp_atomic_pipeline): Likewise.
4072 (expand_omp_atomic_mutex): Likewise.
4073 (expand_omp_target): Likewise.
4074 (grid_expand_omp_for_loop): Likewise.
4075 (grid_expand_target_grid_body): Likewise.
4076 (build_omp_regions_1): Likewise.
4077 * omp-low.c (check_combined_parallel): Skip debug stmts.
4078 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
4079 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
4080 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
4082 * tree-cfg.c (make_blobs_1): Skip debug stmts.
4083 (make_edges): Likewise.
4084 (cleanup_dead_labels): Likewise.
4085 (gimple_can_merge_blocks_p): Likewise.
4086 (stmt_starts_bb_p): Likewise.
4087 (gimple_block_label): Likewise.
4088 (gimple_redirect_edge_and_branch): Likewise.
4089 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
4091 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
4093 * tree-pretty-print (print_declaration): Omit initializer in slim
4095 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
4097 (eliminate_unnecessary_stmts): Stabilize block removal order.
4098 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
4099 * var-tracking.c (get_first_insn): New.
4100 (vt_emit_notes): Call it.
4101 (vt_initialize): Walk any insns before the first BB.
4102 (delete_debug_insns): Likewise.
4104 * gimple.h (enum gimple_debug_subcode): Add
4105 GIMPLE_DEBUG_BEGIN_STMT.
4106 (gimple_debug_begin_stmt_p): New.
4107 (gimple_debug_nonbind_marker_p): New.
4108 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
4109 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
4110 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
4111 * insn-notes.def (BEGIN_STMT): New.
4112 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
4113 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
4114 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
4115 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
4116 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
4117 (INSN_DEBUG_MARKER_KIND): New.
4118 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
4119 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
4120 (INSN_VAR_LOCATION_PTR): New.
4121 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
4123 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
4124 deps only if debug bind insns are enabled.
4125 (pass_expand::execute): Avoid deep TER and expand
4126 debug locations for debug bind insns only.
4127 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
4128 debug stmts special handling down to debug bind stmts.
4129 * combine.c (try_combine): Narrow debug insns special handling
4130 down to debug bind insns.
4131 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
4132 Narrow debug insns preexisting special handling down to debug
4134 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
4135 handling down to debug bind insns.
4136 * function.c (instantiate_virtual_regs): Skip debug markers,
4137 adjust handling of debug binds.
4138 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
4139 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
4140 * haifa-sched.c (schedule_insn): Narrow special handling of debug
4141 insns to debug bind insns.
4142 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
4143 special handling of debug stmts to debug bind stmts.
4144 * ipa-split.c (split_function): Likewise.
4145 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
4146 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
4148 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
4149 * regrename.c (build_def_use): Likewise.
4150 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
4151 (pass_cprop_hardreg): Narrow special casing of debug insns to
4153 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
4154 * reload1.c (reload): Likewise.
4155 * sese.c (sese_insert_phis_for_liveouts): Narrow special
4156 casing of debug stmts to debug bind stmts.
4157 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
4158 * ssa-iterators.h (num_imm_uses): Likewise.
4159 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
4160 debug stmts to debug bind stmts.
4161 * tree-inline.c (tree_function_versioning): Narrow special casing
4162 of debug stmts to debug bind stmts.
4163 * tree-loop-distribution.c (generate_loops_for_partition):
4164 Narrow special casing of debug stmts to debug bind stmts.
4165 * tree-sra.c (analyze_access_subtree): Narrow special casing
4166 of debug stmts to debug bind stmts.
4167 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
4168 stmts to debug bind stmts.
4169 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
4170 casing of debug stmts to debug bind stmts.
4171 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
4172 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
4173 casing of debug stmts to debug bind stmts.
4174 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4176 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
4177 debug stmts to debug bind stmts.
4178 (gimple_replace_ssa_lhs): Likewise.
4179 (insert_debug_temp_for_var_def): Likewise.
4180 (insert_debug_temps_for_defs): Likewise.
4181 (reset_debug_uses): Likewise.
4182 * tree-ssanames.c (release_ssa_name_fn): Likewise.
4183 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
4184 (adjust_debug_stmts): Likewise.
4185 (adjust_phi_and_debug_stmts): Likewise.
4186 (vect_do_peeling): Likewise.
4187 * tree-vect-loop.c (vect_transform_loop): Likewise.
4188 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
4189 * var-tracking.c (adjust_mems): Narrow special casing of debug
4190 insns to debug bind insns.
4191 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
4192 (compute_bb_dataflow, vt_find_locations): Likewise.
4193 (vt_expand_loc, emit_notes_for_changes): Likewise.
4194 (vt_init_cfa_base): Likewise.
4195 (vt_emit_notes): Likewise.
4196 (vt_initialize): Likewise.
4197 (vt_finalize): Likewise.
4199 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
4200 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
4201 (next_nonnote_nondebug_insn_bb): New.
4202 (prev_nonnote_nondebug_insn_bb): New.
4203 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4204 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4205 (prev_nonnote_nondebug_insn_bb): Declare.
4206 (next_nonnote_nondebug_insn_bb): Declare.
4207 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4208 * cfgrtl.c (get_last_bb_insn): Likewise.
4209 * lra.c (push_insns): Likewise.
4211 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4214 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4216 * selftest.c (selftest::test_runner::test_runner): New ctor.
4217 (selftest::test_runner::~test_runner): New dtor.
4218 * selftest.h (class selftest::test_runner): New class.
4220 2017-12-11 Carl Love <cel@us.ibm.com>
4222 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4223 vec_extract_fp32_from_shortl]): Add #defines.
4224 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4225 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4226 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4227 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4228 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4229 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4230 * doc/extend.texi: Add documentation for the added builtins.
4232 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4234 PR rtl-optimization/80693
4235 PR rtl-optimization/81019
4236 PR rtl-optimization/81020
4237 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4238 are not mentioned in i3. Place the REG_UNUSED note on i2,
4239 possibly modified to REG_DEAD, if it did not originate in i3.
4241 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4243 * recog.c (store_data_bypass_p_1): New function.
4244 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4245 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4247 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4249 PR rtl-optimization/83361
4250 * ifcvt.c (if_convert): Call fixup_partitions.
4252 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4254 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4255 early folding of splat_u{8,16,32}.
4257 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4259 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4260 output_operand_lossage first argument with capital letter.
4261 (aarch64_override_options): Don't start error and sorry first argument
4262 with capital letter.
4264 2017-12-11 Andi Kleen <ak@linux.intel.com>
4266 PR gcov-profile/83355
4267 * auto-profile.c (string_table::get_index_by_decl): Don't
4268 recurse when abstract origin points to itself.
4270 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4272 PR tree-optimization/83320
4273 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4274 (prune_datarefs_not_in_loop): Ditto.
4276 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4278 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4280 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4282 PR tree-optimization/83337
4283 * gimple-loop-interchange.cc (compute_access_stride): Handle
4284 bitfield DRs properly.
4286 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4288 PR tree-optimization/83338
4289 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4290 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4291 vector of unsigned integers to vector of signed integers.
4293 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4295 PR rtl-optimization/83317
4296 * lra-constraints.c (process_address_1): Add insn code check.
4298 2017-12-08 Michael Matz <matz@suse.de>
4300 Fix PR tree-optimization/83323
4301 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4302 head-controlled loops and loop BBs.
4303 * common.opt (funroll-and-jam): Remove, instead ...
4304 (floop-unroll-and-jam): ... reuse this option.
4305 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4306 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4307 (-floop-unroll-and-jam): ... this option.
4309 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4311 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4312 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4313 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4315 2017-12-08 Richard Biener <rguenther@suse.de>
4318 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4319 handle accesses outside of zero-sized vars.
4321 2017-12-08 Martin Jambor <mjambor@suse.cz>
4323 PR tree-optimization/83141
4324 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4325 test for MEM_REFs implicitely changing types with padding. Remove
4327 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4329 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4331 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4333 * config/avr/avr.c (avr_attribute_table): Likewise.
4334 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4335 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4336 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4337 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4338 of struct attribute_spec.
4339 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4341 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4344 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4346 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4349 * profile-count.c (profile_count::from_gcov_type): Move from
4350 profile-count.h; handle overflow.
4351 * profile-count.h (profile_count::from_gcov_type): Move offline.
4353 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4355 PR rtl-optimization/83304
4356 * combine.c (move_deaths): If we do not know where a register died,
4359 2017-12-08 Richard Biener <rguenther@suse.de>
4361 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4362 Provide -fopt-info-loop feedback when we interchange in a nest.
4364 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4366 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4367 for armv6 ARM CPU IDs.
4369 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4371 * common/config/arm/arm-common.c: Include <algorithm>.
4372 (INCLUDE_VECTOR): Define.
4373 (compare_opt_names): New function.
4374 (arm_rewrite_selected_arch): Only strip out extensions that can be
4375 expressed through -mfpu. Sort the remaining extensions
4378 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4380 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4381 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4382 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4383 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4385 2017-06-08 Tristan Gingold <gindold@adacore.com>
4388 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4389 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4391 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4393 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4394 of struct attribute_spec.
4396 2017-12-08 Julia Koval <julia.koval@intel.com>
4398 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4399 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4400 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4401 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4402 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4403 _mm_maskz_dpwssds_epi32): Ditto.
4405 2017-12-08 Richard Biener <rguenther@suse.de>
4407 PR tree-optimization/81303
4408 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4409 conditions try to create a comparison vector type matching
4410 the data vector type.
4411 (vectorizable_condition): Adjust.
4412 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4413 Leave invariant conditions alone in case we can vectorize those.
4415 2017-12-08 Julia Koval <julia.koval@intel.com>
4417 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4418 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4419 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4420 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4421 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4423 2017-12-08 Julia Koval <julia.koval@intel.com>
4425 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4426 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4427 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4428 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4429 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4430 _mm_maskz_dpbusds_epi32): New intrinsics.
4432 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4434 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4436 (ld<bh>io_signed): Likewise.
4437 (st<bhw>io): Likewise.
4438 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4439 SMALL_INT12 constant integer operand.
4441 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4443 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4445 (nios2_large_constant_allowed): ...this. Adjust uses.
4446 (nios2_plus_symbolic_constant_p): Rename to...
4447 (nios2_plus_large_constant_p): ...this. Adjust uses.
4448 (nios2_legitimate_address_p): Correct CONST_INT handling.
4449 (nios2_symbolic_memory_operand_p): Rename to...
4450 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4451 (nios2_large_constant_p): Check for large constant integers too.
4452 (nios2_split_large_constant): Handle constant integers.
4453 (nios2_split_symbolic_memory_operand): Rename to...
4454 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4455 (nios2_legitimize_constant_address): Handle constant integers.
4456 (r0rel_constant_p): Handle small constant integers.
4457 (nios2_print_operand_address): Handle r0-relative integer addresses.
4458 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4459 * config/nios2/nios2.md: Adjust for renamed functions.
4461 2017-12-07 Andrew Waterman <andrew@sifive.com>
4463 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4464 (TARGET_HAVE_SRODATA_SECTION): New define.
4465 (riscv_select_section): New function.
4467 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4468 Alexander Monakov <amonakov@ispras.ru>
4469 Jakub Jelinek <jakub@redhat.com>
4472 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4474 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4476 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4477 a bool success value. Don't call output_operand_lossage here.
4478 (aarch64_print_ldpstp_address): Return a bool success value.
4479 (aarch64_print_operand_address): Call output_addr_const if
4480 aarch64_print_address_internal fails.
4481 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4482 'y'; call output_operand_lossage instead. Call output_operand_lossage
4483 if aarch64_print_ldpstp_address fails.
4485 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4487 * tree-vector-builder.h
4488 (tree_vector_builder::binary_encoded_nelts): Declare.
4489 * tree-vector-builder.c
4490 (tree_vector_builder::binary_encoded_nelts): New function.
4491 * fold-const.c (negate_expr_p): Likewise.
4492 (operand_equal_p, fold_checksum_tree): Likewise.
4493 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4494 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4495 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4496 (uniform_vector_p): Likewise.
4497 * varasm.c (const_hash_1, compare_constant): Likewise.
4498 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4499 (valid_lattice_transition): Operate directly on the VECTOR_CST
4501 * ipa-icf.c: Include tree-vector-builder.h.
4502 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4503 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4505 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4507 * tree.c (build_vector): Delete.
4508 * tree.h (build_vector): Make static and move into the self-testing
4511 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4513 * vector-builder.h (vector_builder::derived): New const overload.
4514 (vector_builder::elt): New function.
4515 * tree-vector-builder.h (tree_vector_builder::type): New function.
4516 (tree_vector_builder::apply_step): Declare.
4517 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4519 * gimple-fold.h (tree_vector_builder): Declare.
4520 (gimple_build_vector): Take a tree_vector_builder instead of a
4521 type and vector of elements.
4522 * gimple-fold.c (gimple_build_vector): Likewise.
4523 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4525 (get_initial_defs_for_reduction): Likewise.
4526 (vectorizable_induction): Likewise.
4528 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4530 * tree-vector-builder.h
4531 (tree_vector_builder::new_binary_operation): Declare.
4532 * tree-vector-builder.c
4533 (tree_vector_builder::new_binary_operation): New function.
4534 * fold-const.c (fold_relational_const): Use it.
4535 (const_binop): Likewise. Check that both input vectors have
4536 the same number of elements, thus excluding things like WIDEN_SUM.
4537 Check whether it is possible to operate directly on the encodings
4540 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4542 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4543 new_unary_operation, operating only on the encoded elements.
4544 (const_unop): Likewise.
4545 (exact_inverse): Likewise.
4546 (distributes_over_addition_p): New function.
4547 (const_binop): Use tree_vector_builder and new_unary_operation
4548 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4549 on the encoded elements unless the encoding is strided and the
4550 operation does not distribute over addition.
4551 (fold_convert_const): Use tree_vector_builder and
4552 new_unary_operation. Operate only on the encoded elements
4553 for truncating integer conversions, or for non-stepped encodings.
4555 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4557 * config/sparc/sparc.c: Include tree-vector-builder.h.
4558 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4559 * expmed.c: Include tree-vector-builder.h.
4560 (make_tree): Use tree_vector_builder instead of build_vector.
4561 * fold-const.c: Include tree-vector-builder.h.
4562 (const_binop): Use tree_vector_builder instead of build_vector.
4563 (const_unop): Likewise.
4564 (native_interpret_vector): Likewise.
4565 (fold_vec_perm): Likewise.
4566 (fold_ternary_loc): Likewise.
4567 * gimple-fold.c: Include tree-vector-builder.h.
4568 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4570 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4571 (simplify_vector_constructor): Use tree_vector_builder instead
4573 * tree-vect-generic.c: Include tree-vector-builder.h.
4574 (add_rshift): Use tree_vector_builder instead of build_vector.
4575 (expand_vector_divmod): Likewise.
4576 (optimize_vector_constructor): Likewise.
4577 * tree-vect-loop.c: Include tree-vector-builder.h.
4578 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4579 of build_vector. Explicitly use a stepped encoding for
4581 * tree-vect-slp.c: Include tree-vector-builder.h.
4582 (vect_get_constant_vectors): Use tree_vector_builder instead
4584 (vect_transform_slp_perm_load): Likewise.
4585 (vect_schedule_slp_instance): Likewise.
4586 * tree-vect-stmts.c: Include tree-vector-builder.h.
4587 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4588 (vect_gen_perm_mask_any): Likewise.
4589 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4590 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4592 (build_vector_from_val): Likewise. Explicitly use a duplicate
4595 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4597 * doc/generic.texi (VECTOR_CST): Describe new representation of
4599 * vector-builder.h: New file.
4600 * tree-vector-builder.h: Likewise.
4601 * tree-vector-builder.c: Likewise.
4602 * Makefile.in (OBJS): Add tree-vector-builder.o.
4603 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4604 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4605 (tree_vector): Change the number of elements to
4606 vector_cst_encoded_nelts.
4607 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4608 (VECTOR_CST_ELTS): Delete.
4609 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4610 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4611 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4612 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4613 (VECTOR_CST_ENCODED_ELT): Likewise.
4614 (vector_cst_encoded_nelts): New function.
4615 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4616 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4617 (vector_cst_int_elt, vector_cst_elt): Declare.
4618 * tree.c: Include tree-vector-builder.h.
4619 (tree_code_size): Abort if passed VECTOR_CST.
4620 (tree_size): Update for new VECTOR_CST layout.
4621 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4622 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4623 (build_vector): Use tree_vector_builder.
4624 (vector_cst_int_elt, vector_cst_elt): New functions.
4625 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4626 encoded elements and then create the vector in the canonical form.
4627 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4628 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4629 (tree_c_tests): Call test_vector_cst_patterns.
4630 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4632 (hash_tree): Likewise.
4633 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4634 (streamer_write_tree_header): Likewise.
4635 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4636 (streamer_alloc_tree): Likewise. Update call to make_vector.
4637 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4639 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4641 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4642 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4643 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4645 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4646 Richard Biener <rguenther@suse.de>
4648 PR tree-optimization/81303
4649 * Makefile.in (gimple-loop-interchange.o): New object file.
4650 * common.opt (floop-interchange): Reuse the option from graphite.
4651 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4652 -floop-interchange and mention it for -O3.
4653 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4654 * gimple-loop-interchange.cc: New file.
4655 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4656 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4657 * passes.def (pass_linterchange): New pass.
4658 * timevar.def (TV_LINTERCHANGE): New time var.
4659 * tree-pass.h (make_pass_linterchange): New declaration.
4660 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4661 interchange. Record IV before/after increment in new parameters.
4662 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4663 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4665 (check_reduction_path): ...New function here.
4666 * tree-vectorizer.h (check_reduction_path): New declaration.
4668 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4671 PR rtl-optimization/80818
4672 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4673 always early clobbered.
4674 * lra-lives.c (process_bb_lives): Check input hard regs for early
4675 clobbered non-operand hard reg.
4677 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4680 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4681 types_compatible_p, just that TYPE_MODE is the same.
4683 2017-12-07 Martin Sebor <msebor@redhat.com>
4686 * attribs.c (empty_attribute_table): Initialize new member of
4687 struct attribute_spec.
4688 (decl_attributes): Add argument. Handle mutually exclusive
4689 combinations of attributes.
4690 (selftests::test_attribute_exclusions): New function.
4691 (selftests::attribute_c_tests): Ditto.
4692 * attribs.h (decl_attributes): Add default argument.
4693 * selftest.h (attribute_c_tests): Declare.
4694 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4695 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4697 * doc/extend.texi (Common Function Attributes): Update const and pure.
4698 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4699 of struct attribute_spec.
4700 * config/arc/arc.c (arc_attribute_table): Same.
4701 * config/arm/arm.c (arm_attribute_table): Same.
4702 * config/avr/avr.c ( avr_attribute_table): Same.
4703 * config/bfin/bfin.c (bfin_attribute_table): Same.
4704 * config/cr16/cr16.c (cr16_attribute_table): Same.
4705 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4706 * config/h8300/h8300.c (h8300_attribute_table): Same.
4707 * config/i386/i386.c (ix86_attribute_table): Same.
4708 * config/ia64/ia64.c (ia64_attribute_table): Same.
4709 * config/m32c/m32c.c (m32c_attribute_table): Same.
4710 * config/m32r/m32r.c (m32r_attribute_table): Same.
4711 * config/m68k/m68k.c (m68k_attribute_table): Same.
4712 * config/mcore/mcore.c (mcore_attribute_table): Same.
4713 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4714 * config/mips/mips.c (mips_attribute_table): Same.
4715 * config/msp430/msp430.c (msp430_attribute_table): Same.
4716 * config/nds32/nds32.c (nds32_attribute_table): Same.
4717 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4718 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4719 * config/rl78/rl78.c (rl78__attribute_table): Same.
4720 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4721 * onfig/rx/rx.c (rx_attribute_table): Same.
4722 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4723 * config/sh/sh.c (sh_attribute_table): Same.
4724 * config/sparc/sparc.c (sparc_attribute_table): Same.
4725 * config/spu/spu.c (spu_attribute_table): Same.
4726 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4727 * config/v850/v850.c (v850_attribute_table): Same.
4728 * config/visium/visium.c (visium_attribute_table): Same.
4730 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4733 * config/arm/arm.c (INCLUDE_STRING): Define.
4734 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4735 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4738 2017-12-07 Michael Matz <matz@suse.de>
4740 Add unroll and jam pass
4742 * gimple-loop-jam.c: New file.
4743 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4744 * common.opt (funroll-and-jam): New option.
4745 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4746 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4747 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4748 * passes.def: Add pass_loop_jam.
4749 * timevar.def (TV_LOOP_JAM): Add.
4750 * tree-pass.h (make_pass_loop_jam): Declare.
4751 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4752 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4753 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4754 to flow_loop_tree_node_add.
4755 (duplicate_subloops, copy_loops_to): Append to sibling list.
4756 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4757 * doc/invoke.texi (-funroll-and-jam): Document new option.
4758 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4760 2017-12-07 Richard Biener <rguenther@suse.de>
4762 PR tree-optimization/83296
4763 PR tree-optimization/67769
4764 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4765 flow sensitive info in an unrelated BB.
4766 (value_replacement): Use reset_flow_sensitive_info.
4767 (minmax_replacement): Reset flow sensitive info on the def
4768 we move. Do not reset flow sensitive info in the whole BB
4769 we move the stmt to.
4770 (abs_replacement): Likewise.
4772 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4775 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4776 rs6000_cpu to the given -mcpu=, or to the default processor.
4778 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4780 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4781 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4782 instead of rs6000_cpu_attr.
4783 (rs6000_adjust_cost): Ditto.
4784 (is_microcoded_insn): Ditto.
4785 (rs6000_adjust_priority): Ditto.
4786 (rs6000_issue_rate): Ditto.
4787 (rs6000_use_sched_lookahead): Ditto.
4788 (rs6000_use_sched_lookahead_guard): Ditto.
4789 (rs6000_sched_reorder): Ditto.
4790 (force_new_group): Ditto.
4791 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4792 (group_ending_nop): Ditto.
4794 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4796 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4797 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4798 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4799 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4801 (rs6000_reassociation_width): Ditto.
4802 (rs6000_emit_epilogue): Ditto.
4803 (rs6000_adjust_cost): Ditto.
4804 (is_microcoded_insn): Ditto.
4805 (is_cracked_insn): Ditto.
4806 (rs6000_adjust_priority): Ditto.
4807 (rs6000_sched_reorder): Ditto.
4808 (rs6000_sched_reorder2): Ditto.
4809 (insn_must_be_first_in_group): Ditto.
4810 (insn_must_be_last_in_group): Ditto.
4811 (rs6000_register_move_cost): Ditto.
4812 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4815 2017-12-07 Julia Koval <julia.koval@intel.com>
4817 * config.gcc: Add vaesintrin.h.
4818 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4820 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4821 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4823 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4824 * config/i386/immintrin.h: Include vaesintrin.h.
4825 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4826 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4827 _mm_aesdec_epi128): New intrinsics.
4829 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4831 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4832 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4833 c-family/c-spellcheck.cc.
4834 (best_macro_match::best_macro_match): Likewise.
4836 (struct edit_distance_traits<cpp_hashnode *>): Move to
4837 c-family/c-spellcheck.h.
4838 (class best_macro_match): Likewise.
4840 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4842 PR tree-optimization/83293
4843 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4844 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4845 might insert into empty bb.
4848 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4849 simplify for plus with :c added, and pointer_plus without that.
4850 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4851 with undefined overflow and the conversion is not widening,
4852 perform negation in utype and only convert to type afterwards.
4853 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4854 simplify for plus with :c added, and pointer_plus without that.
4855 If type is integral with undefined overflow and the conversion is
4856 not widening, perform minus in utype and only convert to type
4857 afterwards. Move the last pointer_diff_expr simplify into the
4860 2017-12-06 Martin Sebor <msebor@redhat.com>
4862 PR tree-optimization/82646
4863 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4864 strncpy, not maxlen.
4866 2017-12-06 Martin Sebor <msebor@redhat.com>
4868 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4871 PR tree-optimization/83075
4872 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4873 strncat/strncpy don't change length of source string.
4875 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4878 2017-11-29 Martin Aberg <maberg@gaisler.com>
4880 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4881 to prevent b2bst errata sequence.
4882 (sqrtdf2_fix): Likewise.
4884 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4886 PR tree-optimization/81945
4887 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4888 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4889 to dest_cfun has orig_loop_num set, either remap it to the new
4890 loop number if the loop got moved too, or clear it.
4892 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4894 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4896 (thunderx2t99_nothing): New insn reservation.
4897 (thunderx2t99_mrs): New insn reservation.
4898 (thunderx2t99_multiple): New insn reservation.
4899 (thunderx2t99_alu_basi): Add bfx to reservation.
4900 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4902 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4905 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4907 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4909 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4910 tree-ssa-pre.c::remove_dead_inserted_code.
4911 * tree-ssa-dce.h: New file.
4912 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4913 (remove_dead_inserted_code): Move and rename to function
4914 tree-ssa-dce.c::simple_dce_from_worklist.
4915 (pass_pre::execute): Update use.
4917 2017-12-05 Richard Biener <rguenther@suse.de>
4919 PR tree-optimization/83277
4920 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4921 to code-gen liveout vars.
4923 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4925 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4926 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4927 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4930 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4932 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4934 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4935 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4938 2017-12-05 Richard Biener <rguenther@suse.de>
4940 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4941 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4942 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4943 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4945 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4947 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4949 2017-12-05 Martin Liska <mliska@suse.cz>
4950 Jakub Jelinek <jakub@redhat.com>
4952 * doc/invoke.texi: Document the options.
4953 * flag-types.h (enum sanitize_code): Add
4954 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4955 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4956 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4957 * opts.c: Define new sanitizer options.
4958 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4959 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4961 2017-12-05 Julia Koval <julia.koval@intel.com>
4963 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4964 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4965 (ix86_handle_option): Handle -mavx512vnni.
4966 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4967 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4968 * config/i386/i386-c (__AVX512VNNI__): New.
4969 * config/i386/i386.c (ix86_target_string): Handle new option.
4970 (ix86_valid_target_attribute_inner_p): Handle new option.
4971 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4972 * config/i386/i386.opt (mavx512vnni): New option.
4974 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4977 * athlon.md: Disable for generic.
4978 * haswell.md: Enable for generic.
4979 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4980 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4982 (ix86_adjust_cost): Move generic to haswell path.
4984 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4986 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4987 instead of MEM_P in a couple more places. Fix formatting issues.
4989 2017-12-04 Jim Wilson <jimw@sifive.com>
4991 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4992 instead of GP_REG_LAST-1.
4993 (riscv_adjust_libcall_cfi_prologue): Likewise.
4994 (riscv_adjust_libcall_cri_epilogue): Likewise.
4995 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4998 2017-12-04 Luis Machado <luis.machado@linaro.org>
5000 * ipa-pure-const.c (check_decl): Add missing newline.
5001 (state_from_flags): Likewise.
5003 2017-12-04 Jeff Law <law@redhat.com>
5005 PR tree-optimizatin/78496
5006 * gimple-ssa-evrp-analyze.h
5007 (evrp_range_analyzer::get_vr_values): Simplify.
5008 * gimple-ssa-evrp-analyze.c: Corresponding changes.
5009 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
5010 and gimple-ssa-evrp-analyze.h.
5011 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
5012 (simplify_stmt_for_jump_threading): Copy a blob of code from
5013 tree-vrp.c to use ranges to simplify statements.
5014 (dom_opt_dom_walker::before_dom_children): Call
5015 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
5016 (dom_opt_dom_walker::after_dom_children): Similarly for
5017 evrp_range_analyzer::leave.
5018 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
5021 * gimple-ssa-evrp-analyze.c
5022 (evrp_range_analyzer::extract_range_from_stmt): Always use
5023 vr_values::update_value_range so preexisting range info is
5024 medged with new range info, even if the new range is VR_VARYING.
5026 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5028 * combine.c: Adjust comment.
5029 (use_crosses_set_p): Delete.
5030 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
5031 (try_combine): Ditto.
5033 2017-12-04 Richard Biener <rguenther@suse.de>
5035 PR tree-optimization/83255
5036 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
5037 Re-add zero-iteration check.
5039 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5041 PR rtl-optimization/83245
5042 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
5043 hard registers as earlyclobber, also if not in an asm.
5045 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5049 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5052 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5053 rs6000_cpu based on cpu_index, not tune_index.
5055 2017-12-04 Richard Biener <rguenther@suse.de>
5057 PR tree-optimization/83238
5058 * graphite-scop-detection.c (scop_detection::merge_sese): Make
5059 code match comment, rejecting invalid SESE regions.
5061 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
5063 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
5064 require base operand is a REG_POINTER prior to reload on targets
5065 with non-equivalent space registers.
5067 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5069 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
5070 (update_specialized_profile): Fix updating of counts.
5071 (perhaps_add_new_callers): Likewise.
5073 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5076 * x86-tune.def: Remove obsolette FIXMEs.
5077 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
5078 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
5079 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
5081 (X86_TUNE_PAD_RETURNS): Disable for generic.
5083 2017-12-02 Jakub Jelinek <jakub@redhat.com>
5085 PR tree-optimization/83170
5086 PR tree-optimization/83241
5087 * gimple-ssa-store-merging.c
5088 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
5089 gimple_vuse (ins_stmt) in case it has changed.
5090 (imm_store_chain_info::output_merged_store): Likewise.
5092 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
5096 * tree-cfg.c (pass_warn_function_return::execute): Handle
5097 __builtin_ubsan_handle_missing_return like __builtin_unreachable
5098 with BUILTINS_LOCATION.
5102 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
5103 is BLKmode for vector field with vector raw mode, use TYPE_MODE
5104 instead of DECL_MODE.
5106 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
5107 last argument to rtx pointer.
5108 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
5109 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
5110 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
5111 depending on the chosen ISAs.
5112 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
5113 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
5114 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
5116 * config/i386/sse.md (mov<mode>_internal): Likewise.
5117 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
5119 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5121 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
5123 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5125 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
5126 parameters from prototype.
5127 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
5128 parameters. Don't print a comment.
5129 (emit_fusion_gpr_load): Adjust.
5130 (emit_fusion_load_store): Adjust.
5131 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
5132 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
5133 comment on the second line.
5135 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5138 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5139 rs6000_cpu based on cpu_index, not tune_index.
5141 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5143 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
5144 which_alternative instead of which_alternative + 1.
5145 (output_asm_insn): Print an extra tab if the template is short.
5147 2017-12-01 Jim Wilson <jimw@sifive.com>
5149 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
5151 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
5152 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
5154 * doc/tm.texi: Regenerate.
5156 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
5159 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
5160 whether we can allocate pseudos before trying to fix an address.
5161 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
5162 memory address is indexed or indirect.
5163 (floatuns_<mode>si2_hw2): Likewise.
5165 2017-12-01 Jason Merrill <jason@redhat.com>
5167 * Makefile.in (TAGS): Add c-family/*.cc.
5169 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
5171 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
5172 (call_value_insn): Likewise.
5173 (sibcall_insn): Likewise.
5174 (sibcall_value_insn): Likewise.
5175 (movsi_aarch64): Likewise.
5176 (movdi_aarch64): Likewise.
5177 (add_losym_): Likewise.
5178 (ldr_got_small_): Likewise.
5179 (ldr_got_small_sidi): Likewise.
5180 (ldr_got_small_28k_): Likewise.
5181 (ldr_got_small_28k_sidi): Likewise.
5182 * config/aarch64/aarch64.c (aarch64_print_address_internal):
5183 Move output_addr_const to symbolic case. Add error check.
5185 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5187 * config/s390/predicates.md (plus16_Q_operand): New predicate.
5188 * config/s390/s390.md: Disable MVC merging peephole if it would
5189 disable operand forwarding.
5190 (new peephole2): Split MVCs if it would turn them into up to 2
5193 2017-12-01 Richard Biener <rguenther@suse.de>
5195 PR tree-optimization/83232
5196 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
5197 detection of same access. Instead of breaking the group here
5198 do not consider the duplicate. Add comment explaining real fix.
5200 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
5202 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5204 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5206 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5208 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5210 * function.h (struct function): Remove cilk_frame_decl,
5211 is_cilk_function and calls_cilk_spawn fields.
5212 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5214 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5215 cilk_elemental field.
5216 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5217 * target.def: Adjust comment.
5218 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5219 Don't test cilk_elemental.
5221 PR tree-optimization/83233
5222 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5223 bswap_stat name for the struct.
5226 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5227 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5228 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5229 SWITCH_BREAK_LABEL_P set on the label.
5230 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5231 added for default case if it was missing and not all cases covered.
5232 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5233 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5235 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5236 to avoid -Wimplicit-fallthrough warning.
5237 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5238 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5240 PR tree-optimization/83221
5241 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5243 (init_reassoc): Formatting fix.
5246 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5247 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5251 * tree-cfg.c: Include asan.h.
5252 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5254 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5256 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5258 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5260 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5261 vpcmpeqd instruction.
5263 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5265 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5266 registers type for 128bit mode.
5268 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5270 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5272 * gengtype-state.c (read_state_pair): Likewise.
5273 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5274 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5275 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5276 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5277 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5278 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5279 * builtins.c (fold_builtin_3): Likewise.
5280 * graphite-scop-detection.c
5281 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5282 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5285 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5288 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5289 parameter. Return true if ignore_decl_rtl_set_p is true and other
5290 conditions are satisfied.
5291 * asan.h (asan_protect_global): Add new parameter.
5292 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5293 to asan_protect_global calls.
5295 2017-11-30 Jim Wilson <jimw@sifive.com>
5297 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5298 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5299 -mstrict-align, add info on default value. Delete redundant lines for
5300 -mabi. Add missing -mexplicit-relocs docs.
5302 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5304 * config/arc/arc.md (trap): New pattern.
5306 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5308 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5309 instruction to end into a delay slot.
5310 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5311 can be placed into a delay slot against reg_note.
5313 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5315 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5316 labels number of usages.
5318 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5320 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5322 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5324 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5327 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5328 multiplication by power of 2 constant into two shifts + comparison.
5330 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5333 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5335 2017-11-30 Richard Biener <rguenther@suse.de>
5337 PR tree-optimization/83202
5338 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5339 allow_peel argument and guard peeling.
5340 (canonicalize_loop_induction_variables): Likewise.
5341 (canonicalize_induction_variables): Pass false.
5342 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5343 peeling from cunrolli.
5345 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5347 * combine.c (try_combine): Print a message to dump file whenever
5348 I0, I1, or I2 cannot be combined into I3.
5350 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5352 PR rtl-optimization/83156
5353 PR rtl-optimization/82621
5354 * combine.c (try_combine): Don't split an I2 if one of the dests is
5355 set again before I3. Allow unused dests.
5357 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5359 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5361 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5363 PR rtl-optimization/80818
5364 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5365 recursively. Use insn code for clobber.
5366 (lra_set_insn_recog_data): Pass the new arg to
5367 collect_non_operand_hard_regs.
5368 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5370 (lra_update_insn_regno_info): Pass insn to
5371 add_regs_to_insn_regno_info.
5373 2017-11-29 Jim Wilson <jimw@sifive.com>
5374 Andrew Waterman <andrew@sifive.com>
5376 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5377 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5378 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5379 use SINGLE_SHIFT_COST.
5380 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5381 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5383 2017-11-29 Julia Koval <julia.koval@intel.com>
5385 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5386 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5387 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5388 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5389 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5390 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5391 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5392 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5393 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5394 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5395 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5396 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5397 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5398 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5399 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5400 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5401 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5402 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5403 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5404 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5405 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5406 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5407 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5408 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5409 __builtin_ia32_vpshldv_v2di_mask,
5410 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5411 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5412 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5414 2017-11-29 Julia Koval <julia.koval@intel.com>
5416 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5417 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5418 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5419 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5420 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5421 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5422 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5423 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5424 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5425 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5426 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5427 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5428 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5429 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5430 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5431 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5432 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5433 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5434 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5435 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5436 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5438 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5440 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5441 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5444 2017-11-29 Martin Aberg <maberg@gaisler.com>
5446 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5447 to prevent b2bst errata sequence.
5448 (sqrtdf2_fix): Likewise.
5450 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5452 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5453 (div_sqrt_insn_p): New function.
5454 (sparc_do_work_around_errata): Insert NOP instructions to
5455 prevent sequences that could trigger the TN-0013 errata for
5456 certain LEON3 processors.
5457 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5458 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5459 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5460 (in_branch_delay): Prevent div and sqrt in delay slot if
5462 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5464 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5466 * config/sparc/sparc.c (atomic_insn_p): New function.
5467 (sparc_do_work_around_errata): Insert NOP instructions to
5468 prevent sequences that could trigger the TN-0010 errata for
5470 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5471 instruction referable in atomic_insns_p.
5473 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5475 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5476 (atomic_compare_and_swap_leon3_1): Likewise.
5479 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5481 * config/sparc/sparc.c (fpop_insn_p): New function.
5482 (sparc_do_work_around_errata): Insert NOP instructions to
5483 prevent sequences that could trigger the TN-0012 errata for
5485 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5486 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5487 (in_branch_annul_delay): Prevent floating-point instructions
5488 in delay slot of annulled integer branch.
5490 2017-11-29 Richard Biener <rguenther@suse.de>
5492 PR tree-optimization/83202
5493 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5495 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5496 nodes vectorized to the same stmts multiple times.
5497 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5498 (vect_analyze_slp_instance): Adjust.
5499 (scalar_stmts_to_slp_tree_map_t): New typedef.
5500 (vect_schedule_slp_instance): Add a map recording the SLP node
5501 representing the vectorized stmts for a set of scalar stmts.
5502 Avoid code-generating redundancies.
5503 (vect_schedule_slp): Allocate map and pass it down.
5505 2017-11-29 Nathan Sidwell <nathan@acm.org>
5508 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5511 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5513 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5514 cases for printing LDP/STP memory addresses.
5515 (aarch64_print_address_internal): Renamed from
5516 aarch64_print_operand_address, added parameter, add Pmode check.
5517 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5518 (aarch64_print_operand_address): Indirect to
5519 aarch64_print_address_internal.
5520 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5521 'y' operand output specifier.
5523 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5526 * tree.c (build_simple_mem_ref_loc): Handle
5527 get_addr_base_and_unit_offset returning a MEM_REF.
5530 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5533 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5536 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5538 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5540 (layout::move_to_column): Likewise.
5542 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5545 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5546 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5548 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5549 Martin Jambor <mjambor@suse.cz>
5552 * tree.h (expr_type_first_operand_type_p): Declare
5553 * tree.c (expr_type_first_operand_type_p): New function.
5554 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5555 (ipa_value_from_jfunc): Adjust declaration.
5556 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5557 Use it as result type for arithmetics, unless it is NULL in which case
5558 be more conservative.
5559 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5560 ipa_get_jf_pass_through_result.
5561 (propagate_vals_across_pass_through): Likewise.
5562 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5563 is to propagate_vals_across_pass_through.
5564 (propagate_constants_across_call): Pass PARM_TYPE to
5565 propagate_scalar_across_jump_function.
5566 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5567 ipa_value_from_jfunc.
5568 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5569 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5570 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5571 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5572 target_type, pass it to ipa_value_from_jfunc.
5573 (update_indirect_edges_after_inlining): Pass parameter type to
5574 try_make_edge_direct_simple_call.
5576 2017-11-28 Jeff Law <law@redhat.com>
5578 * gimple-ssa-evrp-analyze.c
5579 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5580 refine ranges if scev_initialized_p returns true.
5581 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5583 2017-11-28 Julia Koval <julia.koval@intel.com>
5585 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5586 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5587 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5588 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5589 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5590 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5591 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5592 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5593 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5594 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5595 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5596 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5597 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5598 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5599 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5600 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5601 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5602 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5603 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5604 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5605 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5606 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5608 2017-11-28 Julia Koval <julia.koval@intel.com>
5610 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5611 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5612 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5613 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5614 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5615 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5616 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5617 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5618 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5619 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5620 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5621 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5622 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5623 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5624 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5625 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5626 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5627 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5628 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5629 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5630 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5631 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5632 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5633 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5634 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5635 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5636 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5637 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5638 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5639 __builtin_ia32_vpshld_v2di_mask): New builtins.
5640 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5641 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5643 2017-11-28 Richard Biener <rguenther@suse.de>
5645 PR tree-optimization/80776
5646 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5648 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5650 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5651 If the incoming edge is an effective fallthru because the other
5652 edge only reaches a __builtin_unreachable () then record ranges
5653 derived from the controlling condition in SSA info.
5654 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5655 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5657 2017-11-28 Olivier Hainque <hainque@adacore.com>
5659 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5660 on mingw build hosts.
5662 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5665 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5667 2017-11-28 Richard Biener <rguenther@suse.de>
5670 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5671 copies generated from memcpy use a character array as reference
5674 2017-11-28 Julia Koval <julia.koval@intel.com>
5675 Sebastian Peryt <sebastian.peryt@intel.com>
5677 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5678 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5679 cilk-common.o, cilk.h, cilk-common.c): Remove.
5681 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5682 * builtins.c (is_builtin_name): Remove cilkplus condition.
5683 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5684 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5685 cilk-builtins.def, cilkplus.def): Remove.
5686 * cif-code.def (CILK_SPAWN): Remove.
5687 * cilk-builtins.def: Delete.
5688 * cilk-common.c: Ditto.
5690 * cilkplus.def: Ditto.
5691 * config/darwin.h (fcilkplus): Delete.
5692 * cppbuiltin.c: Ditto.
5693 * doc/extend.texi: Remove cilkplus doc.
5694 * doc/generic.texi: Ditto.
5695 * doc/invoke.texi: Ditto.
5696 * doc/passes.texi: Ditto.
5697 * gcc.c (fcilkplus): Remove.
5698 * gengtype.c (cilk.h): Remove.
5699 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5701 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5703 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5705 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5706 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5707 cilkplus conditions.
5708 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5709 inline_read_section): Ditto.
5710 * ipa-inline-analysis.c (cilk.h): Remove.
5711 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5712 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5713 append_linker_options): Remove condition for fcilkplus.
5714 * lto/lto-lang.c (cilk.h): Remove.
5715 (lto_init): Remove condition for fcilkplus.
5716 * omp-expand.c (expand_cilk_for_call): Delete.
5717 (expand_omp_taskreg, expand_omp_for_static_chunk,
5718 expand_omp_for): Remove cilkplus
5720 (expand_cilk_for): Delete.
5721 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5722 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5723 execute_lower_omp, diagnose_sb_0): Ditto.
5724 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5725 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5726 * tree-nested.c: Ditto.
5727 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5728 (dump_generic_node): Ditto.
5729 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5730 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5732 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5734 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5736 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5737 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5738 for vector mode and !TARGET_SIMD.
5740 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5742 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5744 * tree.h (SWITCH_LABELS): Remove.
5745 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5746 assert SWITCH_BODY is non-NULL.
5747 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5749 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5751 PR tree-optimization/80788
5752 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5753 has TREE_OVERFLOW set, call drop_tree_overflow.
5755 2017-11-28 Richard Biener <rguenther@suse.de>
5757 PR tree-optimization/83158
5758 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5760 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5763 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5764 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5766 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5768 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5769 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5771 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5774 * varasm.c (compare_constant): Take the mode of the constants into
5775 account when comparing floating point constants.
5777 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5779 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5780 from explicit instantiation of debug_helper.
5781 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5783 2017-11-27 Richard Biener <rguenther@suse.de>
5785 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5788 2017-11-27 Richard Biener <rguenther@suse.de>
5790 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5791 (type_hash_canon): Also clear int_cst_hash_table entry for
5793 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5795 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5797 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5798 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5799 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5801 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5803 * hash-map.h (gt_cleare_cache): Avoid UB.
5805 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5807 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5808 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5809 (decide_unroll_runtime_iterations): Likewise.
5810 (decide_unroll_stupid): Likewise.
5812 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5815 * config/i386/i386.md: Add a loop with incssp.
5817 2017-11-27 Martin Jambor <mjambor@suse.cz>
5819 PR tree-optimization/81248
5820 * tree-sra.c (splice_param_accesses): Remove size check.
5821 (decide_one_param_reduction): Fix size check.
5822 * gimple-pretty-print.c (dump_profile): Silence warning.
5823 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5825 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5828 * dbxout.c (lastlineno): New variable.
5829 (dbx_debug_hooks): Use dbxout_switch_text_section as
5830 switch_text_section debug hook.
5831 (dbxout_function_end): Switch to current_function_section
5832 rather than function_section. If crtl->has_bb_partition,
5833 output just one N_FUN, depending on in_cold_section_p.
5834 (dbxout_source_line): Remember last lineno in lastlineno.
5835 (dbxout_switch_text_section): New function.
5836 (dbxout_function_decl): Adjust dbxout_block caller.
5837 (dbx_block_with_cold_children): New function.
5838 (dbxout_block): Return true if any LBRAC/RBRAC have been
5839 emitted. Use dbx_block_with_cold_children at depth == 0
5840 in second partition. Add PARENT_BLOCKNUM argument, pass
5841 it optionally adjusted to children. Output LBRAC/RBRAC
5842 around recursive call only if the block is in the current
5843 partition, if not and anything was output, emit empty
5845 * final.c (final_scan_insn): Compute cold_function_name
5846 before calling switch_text_section debug hook. Call
5847 that hook even if dwarf2out_do_frame if not emitting
5851 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5852 TREE_READONLY decls.
5854 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5856 PR rtl-optimization/82488
5857 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5859 2017-11-26 Julia Koval <julia.koval@intel.com>
5861 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5863 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5866 2017-11-26 Julia Koval <julia.koval@intel.com>
5868 * config/i386/driver-i386.c (host_detect_local_cpu):
5869 Detect skylake-avx512.
5871 2017-11-26 Julia Koval <julia.koval@intel.com>
5873 * config.gcc: Add -march=cannonlake.
5874 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5875 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5876 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5877 (PTA_CANNONLAKE): New.
5878 (processor_target_table): Add cannonlake.
5879 (ix86_option_override_internal): Ditto.
5880 (fold_builtin_cpu): Ditto.
5881 (get_builtin_code_for_version): Handle cannonlake.
5882 (M_INTEL_COREI7_CANNONLAKE): New.
5883 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5884 * doc/invoke.texi: Add -march=cannonlake.
5886 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5888 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5889 (try_init_one_plugin): Alternative implementation for MinGW.
5890 * Makefile.in (plugin_implib): New.
5891 (gengtype-lex.c): Fix broken AIX workaround.
5892 * configure: Regenerate.
5893 * doc/plugins.texi: Document support for MinGW.
5895 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5897 PR rtl-optimization/81553
5898 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5899 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5900 is a shift where C1 has different mode than the whole shift, use C1's
5901 mode for MULT rather than the shift's mode.
5904 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5905 builtins not enabled in the currently selected ISA.
5907 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5909 PR tree-optimization/71026
5910 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5911 (insert_reciprocals): Change to insert reciprocals before a division
5912 by a square and to insert the square of a reciprocal.
5913 (execute_cse_reciprocals_1): Change to consider division by a square.
5914 (register_division_in): Add importance parameter.
5916 2017-11-24 Richard Biener <rguenther@suse.de>
5918 PR tree-optimization/82402
5919 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5920 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5922 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5924 * match.pd (0-ptr): New transformation.
5926 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5929 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5930 when skipping checking.
5932 2017-11-24 Richard Biener <rguenther@suse.de>
5934 PR tree-optimization/83128
5935 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5936 (vn_reference_lookup_3): Likewise.
5938 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5941 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5942 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5945 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5946 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5947 check for the latter with gimple_call_builtin_p. Do not handle
5948 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5950 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5952 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5955 2017-11-23 Julia Koval <julia.koval@intel.com>
5957 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5958 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5959 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5960 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5961 _mm512_maskz_expandloadu_epi16): New intrinsics.
5962 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5963 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5964 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5965 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5966 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5967 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5968 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5969 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5970 _mm256_maskz_expandloadu_epi8): New intrinsics.
5971 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5972 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5973 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5974 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5975 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5976 * config/i386/sse.md (VI248_VLBW): New iterator.
5977 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5979 2017-11-23 Julia Koval <julia.koval@intel.com>
5981 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5982 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5983 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5984 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5985 _mm512_mask_compressstoreu_epi16): New.
5986 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5987 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5988 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5989 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5990 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5991 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5992 _mm256_mask_compressstoreu_epi8): New.
5993 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5994 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5995 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5996 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5997 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5998 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5999 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
6000 __builtin_ia32_compresshi128_mask,
6001 __builtin_ia32_compressstoreuqi512_mask,
6002 __builtin_ia32_compressstoreuhi512_mask,
6003 __builtin_ia32_compressstoreuqi256_mask,
6004 __builtin_ia32_compressstoreuqi128_mask,
6005 __builtin_ia32_compressstoreuhi256_mask,
6006 __builtin_ia32_compressstoreuhi128_mask): New builtins.
6007 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
6009 (ix86_expand_special_args_builtin): Handle new types.
6010 (s4fma_expand): Handle new builtin array.
6011 * config/i386/immintrin.h: Include new headers.
6012 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
6013 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
6015 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6017 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
6018 fixes. Declare temp and g variables at the top in order to avoid
6019 {} in most of the cases.
6021 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
6023 * match.pd (ptr-0): New transformation.
6025 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
6027 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
6028 (struct addr_mode_cost_table): New.
6029 (struct tune_params): Add field addr_mode_costs.
6030 * config/arm/arm.c (generic_addr_mode_costs): New.
6031 (arm_slowmul_tune): Initialise addr_mode_costs field.
6032 (arm_fastmul_tune): Likewise.
6033 (arm_strongarm_tune): Likewise.
6034 (arm_xscale_tune): Likewise.
6035 (arm_9e_tune): Likewise.
6036 (arm_marvell_pj4_tune): Likewise.
6037 (arm_v6t2_tune): Likewise.
6038 (arm_cortex_tune): Likewise.
6039 (arm_cortex_a8_tune): Likewise.
6040 (arm_cortex_a7_tune): Likewise.
6041 (arm_cortex_a15_tune): Likewise.
6042 (arm_cortex_a35_tune): Likewise.
6043 (arm_cortex_a53_tune): Likewise.
6044 (arm_cortex_a57_tune): Likewise.
6045 (arm_exynosm1_tune): Likewise.
6046 (arm_xgene1_tune): Likewise.
6047 (arm_cortex_a5_tune): Likewise.
6048 (arm_cortex_a9_tune): Likewise.
6049 (arm_cortex_a12_tune): Likewise.
6050 (arm_cortex_a73_tune): Likewise.
6051 (arm_v7m_tune): Likewise.
6052 (arm_cortex_m7_tune): Likewise.
6053 (arm_v6m_tune): Likewise.
6054 (arm_fa726te_tune): Likewise.
6055 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
6057 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
6059 * config/arm/arm.c (arm_mem_costs): New function.
6060 (arm_rtx_costs_internal): Use arm_mem_costs.
6062 2017-11-23 Mark Wielaard <mark@klomp.org>
6064 * dwarf2out.c (init_sections_and_labels): Use generation to create
6065 unique ranges_section_label and ranges_base_label. Return generation.
6066 (output_rnglists): Add generation argument. Use generation to create
6067 unique ranges labels.
6068 (dwarf2out_finish): Get generation from init_sections_and_labels
6069 and pass generation to output_rnglists.
6071 2017-11-23 Mike Stump <mikestump@comcast.net>
6072 Eric Botcazou <ebotcazou@adacore.com>
6074 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
6075 * cfgloop.h (struct loop): Add unroll field.
6076 * function.h (struct function): Add has_unroll bitfield.
6077 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
6078 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
6079 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
6080 (pass_rtl_unroll_loops::gate): Likewise.
6081 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
6082 for which loop->unroll==1.
6083 (decide_unroll_constant_iterations): Use note for consistency and
6084 take loop->unroll into account. Return early if loop->unroll is set.
6085 Fix thinko in existing test.
6086 (decide_unroll_runtime_iterations): Use note for consistency and
6087 take loop->unroll into account.
6088 (decide_unroll_stupid): Likewise.
6089 * lto-streamer-in.c (input_cfg): Read loop->unroll.
6090 * lto-streamer-out.c (output_cfg): Write loop->unroll.
6091 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
6093 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
6094 (print_loop): Print loop->unroll if set.
6095 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
6096 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
6097 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
6099 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
6100 loop->unroll is set and smaller than the trip count. Otherwise bypass
6101 entirely the heuristics if loop->unroll is set. Remove dead note.
6102 Fix off-by-one bug in other note.
6103 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
6104 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
6106 (tree_unroll_loops_completely): Make static.
6107 (pass_complete_unroll::execute): Use correct type for variable.
6108 (pass_complete_unrolli::execute): Fix formatting.
6109 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
6111 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
6113 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
6114 enable when TARGET_PREFER_AVX128 is set.
6116 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6118 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
6120 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
6121 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
6123 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6125 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
6126 frequencies. Fix estimation of aggregate parameters.
6128 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6130 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
6133 2017-11-23 Tom de Vries <tom@codesourcery.com>
6135 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
6137 2017-11-23 Tom de Vries <tom@codesourcery.com>
6139 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
6142 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
6145 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
6146 sibcall_value_pcrel_fdpic): Use local variable instead of
6148 (calli_tbr_rel): Add missing operand 2.
6149 (call_valuei_tbr_rel): Add missing operand 3.
6151 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6154 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
6155 bitpos/bitsize covering the whole destination, use store_expr only if
6156 the complex mode is the same. Otherwise, use expand_normal and if
6157 it returns CONCAT, subreg each part separately instead of trying to
6158 subreg the whole result.
6160 2017-11-23 Richard Biener <rguenther@suse.de>
6162 PR tree-optimization/23094
6163 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
6164 come here from walking over backedges in the first iteration.
6165 (vn_reference_lookup_3): Skip clobbers that store the same value.
6167 2017-11-23 Richard Biener <rguenther@suse.de>
6169 PR tree-optimization/81403
6170 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
6171 a block we need a leader relative to.
6172 (phi_translate_1): For nary processing require a leader from
6173 get_representative_for given we run expression simplification
6174 using match-and-simplify. Remove previous fix.
6176 2017-11-22 Jeff Law <law@redhat.com>
6178 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
6179 Use new method allocate_value_range rather than accessing the
6180 vrp_value_range_pool data member directly.
6181 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
6182 to use extract_range_from_stmt method to avoid need for
6183 extract_range_from_assignment method.
6184 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
6185 method rather than setting values_propgated data member directly.
6186 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
6187 and values propagated data members and extract_range_from_assignment
6188 method. Reorder private data members to conform to standards.
6189 Add new methods set_lattice_propagation_complete and
6190 allocate_value_range.
6192 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
6194 PR rtl-optimization/83030
6195 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
6196 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
6197 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
6198 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
6199 CROSSING_JUMP_P flag is already set before setting it.
6200 * cfgrtl.c (fixup_partition_crossing): Likewise.
6201 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6204 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6206 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6207 Handle the case where both arguments are using gen_const_vec_series.
6209 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6212 * pretty-print.c (pp_format): Move quoting implementation to
6213 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6214 to pass address of "quote" local.
6215 (pp_begin_quote): New function.
6216 (pp_end_quote): New function.
6217 * pretty-print.h (printer_fn): Convert penultimate param from bool
6219 (pp_begin_quote): New decl.
6220 (pp_end_quote): New decl.
6221 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6222 param from bool to bool *.
6223 * tree-diagnostic.h (default_tree_printer): Likewise.
6225 2017-11-22 Jeff Law <law@redhat.com>
6227 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6228 Initialize vr_values.
6229 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6230 vr_values via vr_values class instance rather than delegators.
6231 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6232 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6233 (evrp_range_analyzer::push_value_range): Likewise.
6234 (evrp_range_analyzer::pop_value_range): Likewise.
6235 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6236 most delegators. Those remaining are exposed as public interfaces.
6237 Make vr_values a pointer and private.
6238 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6240 (evrp_range_analyzer::get_vr_value): New method.
6241 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6242 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6243 (class evrp_dom_walker): Attach evrp_folder class, initialize
6244 it in the ctor. Remove temporary delegators.
6245 (evrp_dom_walker::before_dom_children): Call methods in attached
6246 evrp_range_analyzer class via class instance pointer. Use
6247 free value_range_constant_singleton to remove need for
6248 op_with_constant_singleton_value delegator method. Do not
6249 create a vrp_prop class instance for every call! Narrow
6250 scope of a couple variables.
6251 (evrp_dom_walker::cleanup): Call methods in attached
6252 evrp_range_analyzer class via class instance pointer.
6253 * vr-values.h (class vr_values): Privatize many methods and
6256 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6258 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6259 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6260 * expr.c (expand_expr_real_2): Likewise.
6261 * fold-const.c (const_unop): Likewise.
6262 * optabs-tree.c (optab_for_tree_code): Likewise.
6263 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6264 * tree-inline.c (estimate_operator_cost): Likewise.
6265 * tree-pretty-print.c (dump_generic_node): Likewise.
6266 (op_code_prio): Likewise.
6267 (op_symbol_code): Likewise.
6268 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6269 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6270 * internal-fn.c (direct_internal_fn_optab): New function.
6271 (direct_internal_fn_array, direct_internal_fn_supported_p
6272 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6273 * fold-const-call.c (fold_const_reduction): New function.
6274 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6276 * tree-vect-loop.c: Include internal-fn.h.
6277 (reduction_code_for_scalar_code): Rename to...
6278 (reduction_fn_for_scalar_code): ...this and return an internal
6280 (vect_model_reduction_cost): Take an internal_fn rather than
6282 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6284 (vectorizable_reduction): Use internal functions rather than tree
6285 codes for the reduction operation. Update calls to the functions
6287 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6288 Use calls to internal functions rather than REDUC tree codes.
6289 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6291 2017-11-22 Olivier Hainque <hainque@adacore.com>
6293 * config/vxworks.c (vxworks_override_options): Pick default
6294 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6295 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6296 DWARF_GNAT_ENCODINGS_DEFAULT.
6297 * config/vxworksae.h: Likewise.
6299 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6301 PR tree-optimization/83104
6302 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6305 2017-11-22 Marek Polacek <polacek@redhat.com>
6306 H.J. Lu <hongjiu.lu@intel.com>
6307 Jason Merrill <jason@redhat.com>
6312 * calls.c (initialize_argument_information): Call
6313 warn_parameter_passing_abi target hook.
6314 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6315 argument onto stack.
6316 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6317 * common.opt: Update -fabi-version description.
6318 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6319 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6321 (ix86_is_empty_record): New function.
6322 (ix86_warn_parameter_passing_abi): New function.
6323 (TARGET_EMPTY_RECORD_P): Redefine.
6324 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6325 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6326 * doc/tm.texi: Regenerated.
6327 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6328 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6329 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6330 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6331 instead of int_size_in_bytes.
6332 * expr.c (copy_blkmode_to_reg): Likewise.
6333 * function.c (aggregate_value_p): Return 0 for empty types.
6334 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6336 (locate_and_pad_parm): Call arg size_in_bytes instead
6338 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6339 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6340 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6342 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6343 (std_gimplify_va_arg_expr): Skip empty records. Call
6344 arg_size_in_bytes instead size_in_bytes.
6345 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6346 * tree-core.h (tree_type_common): Add empty_flag.
6347 (tree_decl_common): Update comments.
6348 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6350 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6351 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6353 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6354 * tree.c (default_is_empty_type): New function.
6355 (default_is_empty_record): New function.
6356 (arg_int_size_in_bytes): New function.
6357 (arg_size_in_bytes): New function.
6358 (get_ultimate_context): New function.
6359 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6360 TRANSLATION_UNIT_WARN_EMPTY_P.
6361 (default_is_empty_record, arg_int_size_in_bytes,
6362 arg_size_in_bytes, get_ultimate_context): Declare.
6364 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6366 * config/arm/arm.c (cmse_clear_registers): New function.
6367 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6368 code by call to cmse_clear_registers.
6369 (cmse_nonsecure_entry_clear_before_return): Likewise.
6371 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6373 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6374 (vdot_s32, vdotq_s32): New.
6375 (vdot_lane_u32, vdotq_lane_u32): New.
6376 (vdot_lane_s32, vdotq_lane_s32): New.
6379 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6382 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6383 for unsigned values with fewer HWIs than the precision.
6384 (test_overflow): New function.
6385 (wide_int_cc_tests): Call it.
6387 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6389 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6390 has the same mode class as Pmode.
6392 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6394 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6395 padding_bits_to_clear_ptr.
6396 (cmse_nonsecure_entry_clear_before_return): Likewise.
6398 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6400 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6401 auto_sbitap instead of integer bitfield to control register needing
6404 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6406 PR tree-optimization/83044
6407 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6408 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6409 that need the upper bound. Subtract offset from
6410 get_addr_base_and_unit_offset only if positive and subtract it
6411 before division by eltsize rather than after it.
6414 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6415 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6416 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6420 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6421 expand_binop with *mul_widen_optab, make sure at least one of the
6422 operands doesn't have VOIDmode.
6425 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6427 PR rtl-optimization/82044
6428 PR tree-optimization/82042
6429 * dse.c (record_store): Check for overflow.
6430 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6431 add_wild_read instead of clear_rhs_from_active_local_stores on
6432 overflow and log it into dump_file.
6434 2017-11-22 Richard Biener <rguenther@suse.de>
6436 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6437 fake edges to exit when looking for a place to insert.
6438 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6440 (insert_into_preds_of_block): Commit edge insertion immediately,
6441 assert that doesn't require new BBs.
6442 (fini_pre): Release expressions.
6443 (pass_pre::execute): Shuffle things around a bit, if the fn
6444 is too large do not compute AVAIL either as this is really the
6447 2017-11-22 Richard Biener <rguenther@suse.de>
6449 PR tree-optimization/83089
6450 * tree-if-conv.c (pass_if_conversion::execute): If anything
6451 changed reset SCEV and free the number of iteration estimates.
6453 2017-11-21 Martin Sebor <msebor@redhat.com>
6455 PR tree-optimization/82945
6456 * calls.h (warn_nonstring_bound): Remove unused function.
6458 2017-11-21 Martin Sebor <msebor@redhat.com>
6460 PR tree-optimization/82945
6461 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6462 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6463 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6465 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6466 * calls.h (get_attr_nonstring_decl): Declare new function.
6467 * doc/extend.texi (attribute nonstring): Update.
6468 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6469 get_attr_nonstring_decl and handle it.
6470 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6471 detection of nul-termination.
6472 (strlen_to_stridx): Change to a pointer.
6473 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6474 (pass_strlen::execute): Same.
6476 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6478 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6479 for the new option -mprefer-vector-width=[none|128|256|512].
6480 * config/i386/i386.c (ix86_target_string): remove old style options
6481 -mprefer-avx256 and make -mprefer-avx128 as alias.
6482 (ix86_option_override_internal): Apply defaults for the
6483 -mprefer-vector-width=[128|256] option.
6484 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6485 Implement macros to work with -mprefer-vector-width=.
6486 * config/i386/i386.opt: Implemented option
6487 -mprefer-vector-width=[none|128|256|512].
6488 * doc/invoke.texi: Documentation for
6489 -mprefer-vector-width=[none|128|256|512].
6491 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6493 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6495 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6497 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6499 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6501 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6503 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6504 * expr.c (expand_expr_real_2): Likewise.
6505 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6506 fold_binary_loc): Likewise.
6507 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6508 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6509 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6510 MINUS_EXPR transformations.
6511 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6512 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6513 * tree-inline.c (estimate_operator_cost): Likewise.
6514 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6515 op_symbol_code): Likewise.
6516 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6517 * vr-values.c (extract_range_from_binary_expr): Likewise.
6518 * varasm.c (initializer_constant_valid_p_1): Likewise.
6519 * tree.def: New tree code POINTER_DIFF_EXPR.
6521 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6523 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6524 integer suffix to movbe mnemonic.
6525 (*bswaphi2_movbe): Ditto.
6526 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6528 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6531 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6532 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6533 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6535 2017-11-21 Martin Liska <mliska@suse.cz>
6537 * tree-inline.c (expand_call_inline): Remove not needed
6540 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6541 Jakub Jelinek <jakub@redhat.com>
6544 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6545 Remove static keyword from f variable.
6547 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6549 PR tree-optimization/83086
6550 * gimple-ssa-store-merging.c
6551 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6552 rather than n.base_addr.
6554 2017-11-21 Martin Liska <mliska@suse.cz>
6556 PR rtl-optimization/82044
6557 PR tree-optimization/82042
6558 * dse.c (check_mem_read_rtx): Check for overflow.
6560 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6562 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6565 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6567 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6568 and its corresponding call arg location note.
6570 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6571 Andrew Burgess <andrew.burgess@embecosm.com>
6573 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6575 (arc_return_slot_offset): Likewise.
6576 (arc_eh_return_address_location): New declaration.
6577 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6578 (MUST_SAVE_REGISTER): Add exception handler case.
6579 (MUST_SAVE_RETURN_ADDR): Likewise.
6580 (arc_frame_pointer_required): Likewise.
6581 (arc_frame_pointer_needed): New function.
6582 (arc_compute_frame_size): Changed.
6583 (arc_expand_prologue): Likewise.
6584 (arc_expand_epilogue): Likewise.
6585 (arc_initial_elimination_offset): Likewise.
6586 (arc_return_slot_offset): Delete.
6587 (arc_eh_return_address_location): New function.
6588 (arc_builtin_setjmp_frame_value): Likewise.
6589 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6590 (EH_RETURN_STACKADJ_RTX): Define.
6591 (EH_RETURN_HANDLER_RTX): Likewise.
6592 * config/arc/arc.md (eh_return): Delete.
6594 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6596 * print-tree.h (debug_vec_tree): Remove prototype.
6597 * gdbinit.in (pvt): Remove macro.
6599 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6601 PR tree-optimization/83047
6602 * gimple-ssa-store-merging.c
6603 (imm_store_chain_info::output_merged_store): If the loads with the
6604 same vuse are in different basic blocks, for load_gsi pick a load
6605 location that is dominated by the other loads.
6608 * config/i386/i386.c (ix86_memmodel_check): Start
6609 -Winvalid-memory-model diagnostics with lowercase letter.
6612 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6613 set in_cold_section_p to the partition containing loc_list->first.
6614 When seeing loc_list->last_before_switch node, update secname and
6615 perform range_across_switch second partition handling only after that.
6618 * run-rtl-passes.c: Include debug.h.
6619 (run_rtl_passes): Call debug_hooks->assembly_start.
6620 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6624 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6625 OPTAB_DIRECT in calls to expand_simple_binop.
6627 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6630 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6632 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6635 * spellcheck.h (best_match::blithely_get_best_candidate): New
6638 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6640 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6641 with lower case letter.
6643 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6645 * config/i386/i386.md (bswaphi2): New expander.
6646 (*bswaphi2_movbe): New insn pattern.
6647 (bswaphi -> rorhi pepehole2): New peephole pattern.
6649 2017-11-20 Jeff Law <law@redhat.com>
6651 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6652 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6653 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6654 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6655 gimple-ssa-evrp-analyze.h.
6657 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6658 set BB_VISITED here.
6659 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6661 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6663 * tree-predcom.c: Add general comment on Store-Store chains.
6664 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6666 (get_chain_last_ref_at): Rename into...
6667 (get_chain_last_write_at): ...this.
6668 (get_chain_last_write_before_load): New function.
6669 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6670 CT_STORE_STORE when write reference is added.
6671 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6672 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6673 (initialize_root_vars_store_elim_1): Ditto.
6674 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6675 definition is created.
6676 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6677 chain by replacing it with dominant stored value.
6679 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6681 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6683 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6685 * vr-values.c (extract_range_from_binary_expr): Use a full range
6688 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6690 * config/arm/arm.md (R4_REGNUM): Define constant.
6691 (nonsecure_call_internal): Remove r4 clobber.
6692 (nonsecure_call_value_internal): Likewise.
6693 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6694 clobber and resequence match_operands.
6695 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6696 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6697 (nonsecure_call_value_reg_thumb2): Likewise.
6699 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6701 PR tree-optimization/78821
6702 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6703 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6704 to the address of the base rather than the base itself.
6705 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6706 (find_bswap_or_nop_finalize): New function.
6707 (find_bswap_or_nop): Use it.
6708 (bswap_replace): Return a tree rather than bool, change first
6709 argument from gimple * to gimple_stmt_iterator, allow inserting
6710 into an empty sequence, allow ins_stmt to be NULL - then emit
6711 all stmts into gsi. Fix up MEM_REF address gimplification.
6712 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6713 (struct store_immediate_info): Add N and INS_STMT non-static
6715 (store_immediate_info::store_immediate_info): Initialize them
6716 from newly added ctor args.
6717 (merged_store_group::apply_stores): Formatting fixes. Sort by
6719 (stmts_may_clobber_ref_p): For stores call also
6720 refs_anti_dependent_p.
6721 (gather_bswap_load_refs): New function.
6722 (imm_store_chain_info::try_coalesce_bswap): New method.
6723 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6724 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6725 (imm_store_chain_info::output_merged_store): Fail if number of
6726 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6727 and NOP_EXPR rhs_code.
6728 (pass_store_merging::process_store): Compute n and ins_stmt, if
6729 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6730 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6732 (pass_store_merging::execute): Calculate dominators.
6734 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6735 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6736 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6737 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6738 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6739 class pass_optimize_bswap, bswap_replace,
6740 pass_optimize_bswap::execute): Moved to ...
6741 * gimple-ssa-store-merging.c: ... this file.
6742 Include optabs-tree.h.
6743 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6744 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6745 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6746 anonymous namespace, remove static keywords.
6747 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6748 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6750 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6753 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6755 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6757 * vec.h (debug_helper): New function.
6758 (DEFINE_DEBUG_VEC): New macro.
6759 * hash-set.h (debug_helper): New function.
6760 (DEFINE_DEBUG_HASH_SET): New macro.
6761 * cfg.c (debug_slim (edge)): New function.
6762 Call DEFINE_DEBUG_VEC for edges.
6763 Call DEFINE_DEBUG_HASH_SET for edges.
6764 * cfghooks.c (debug_slim (basic_block)): New function.
6765 Call DEFINE_DEBUG_VEC for basic blocks.
6766 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6767 * print-tree.c (debug_slim): New function to handle trees.
6768 Call DEFINE_DEBUG_VEC for trees.
6769 Call DEFINE_DEBUG_HASH_SET for trees.
6770 (debug (vec<tree, va_gc>) &): Remove.
6771 (debug (<vec<tree, va_gc>) *): Remove.
6772 * print-rtl.c (debug_slim): New function to handle const_rtx.
6773 Call DEFINE_DEBUG_VEC for rtx_def.
6774 Call DEFINE_DEBUG_VEC for rtx_insn.
6775 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6776 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6777 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6778 (debug (vec<rtx_insn *> *ptr): Remove.
6779 (debug_insn_vector): Remove.
6780 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6782 2017-11-20 Tom de Vries <tom@codesourcery.com>
6784 PR rtl-optimization/82020
6785 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6786 IF_THEN_ELSE condition.
6788 2017-11-19 Jeff Law <law@redhat.com>
6790 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6791 of degenerates resulting from ignoring an edge.
6793 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6796 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6798 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6801 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6802 for uninitialied counts.
6804 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6807 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6809 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6812 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6814 2017-11-19 Tom de Vries <tom@codesourcery.com>
6816 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6819 2017-11-19 Tom de Vries <tom@codesourcery.com>
6821 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6824 2017-11-19 Tom de Vries <tom@codesourcery.com>
6826 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6829 2017-11-19 Tom de Vries <tom@codesourcery.com>
6831 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6833 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6835 2017-11-19 Tom de Vries <tom@codesourcery.com>
6837 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6839 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6840 semicolon after MCORE_EXPORT_NAME call.
6842 2017-11-19 Tom de Vries <tom@codesourcery.com>
6845 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6846 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6848 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6850 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6851 (cgraph_edge::clone): Cleanup updating of profile.
6852 * ipa-cp.c (update_profiling_info): Likewise.
6853 * ipa-inline-transform.c (inline_transform): Likewise.
6854 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6855 * ipa-split.c (execute_split_functions): Do not split when function
6857 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6858 * profile-count.c (profile_count::dump): Add global0.
6859 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6861 (profile_count::to_sreal_scale): Likewise.
6862 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6863 (profile_count::combine_with_ipa_count): New function.
6864 * profile-count.h (profile_guessed_global0adjusted): New.
6865 (profile_count::adjusted_zero): New.
6866 (profile_count::global0adjusted): New.
6867 (profile_count::combine_with_ipa_count): New.
6868 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6869 correct profile of return block of split functions.
6870 (copy_cfg_body): Remove unused profile_count.
6871 (copy_body): Likewise.
6872 (expand_call_inline): Update.
6873 (tree_function_versioning): Update.
6875 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6877 * hash-set.h (hash_set::empty): New.
6878 * tree-ssa-threadbackward.h: Delete.
6879 * tree-ssa-threadbackward.c (class thread_jumps): New.
6880 Move max_threaded_paths into class.
6881 (fsm_find_thread_path): Remove arguments that are now in class.
6882 (profitable_jump_thread_path): Rename to...
6883 (thread_jumps::profitable_jump_thread_path): ...this.
6884 (convert_and_register_jump_thread_path): Rename to...
6885 (thread_jumps::convert_and_register_current_path): ...this.
6886 (check_subpath_and_update_thread_path): Rename to...
6887 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6888 (register_jump_thread_path_if_profitable): Rename to...
6889 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6890 (handle_phi): Rename to...
6891 (thread_jumps::handle_phi): ...this.
6892 (handle_assignment): Rename to...
6893 (thread_jumps::handle_assignment): ...this.
6894 (fsm_find_control_statement_thread_paths): Rename to...
6895 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6896 (find_jump_threads_backwards): Rename to...
6897 (thread_jumps::find_jump_threads_backwards): ...this.
6898 Initialize path local data.
6899 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6900 from within thread_jumps class.
6901 (pass_early_thread_jumps::execute): Same.
6903 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6905 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6907 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6909 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6910 to canonical location.
6912 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6915 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6917 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6919 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6921 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6922 rcrt1.o%s/grcrt1.o%s for -static-pie.
6924 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6926 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6927 ix86_shift_rotate_cost): Break out from ...
6928 (ix86_rtx_costs): ... here.
6929 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6932 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6934 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6935 when entry block was promoted unlikely.
6936 (estimate_bb_frequencies): Increase frequency scale.
6937 * profile-count.h (profile_count): Export precision info.
6939 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6941 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6942 disturbin profile of entry block.
6944 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6946 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6949 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6951 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6952 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6954 2017-11-17 Jeff Law <law@redhat.com>
6956 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6957 from evrp_dom_walker class. Various methods moved into new class.
6958 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6959 (evrp_range_analyzer::enter): New method.
6960 (evrp_range_analyzer::leave): New method.
6961 (evrp_dom_walker): Remove delegators no longer needed by this class.
6962 Replace vr_values data member with evrp_range_analyzer
6964 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6965 method extracted from evrp_dom_walker::before_dom_children.
6966 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6967 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6969 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6970 Add private copy constructor and move assignment operators.
6971 Privatize methods and class data where trivially possible.
6972 (evrp_dom_walker::cleanup): New function, extracted from
6973 execute_early_vrp. Simplify access to class data.
6975 * vr-values.h (get_output_for_vrp): Prototype.
6976 * vr-values.c (get_output_for_vrp): New function extracted from
6977 vrp_visit_assignment_or_call and extract_range_from_stmt.
6978 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6980 2017-11-17 Luis Machado <luis.machado@linaro.org>
6982 * config/aarch64/aarch64.c
6983 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6984 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6985 tune_params::AUTOPREFETCHER_WEAK.
6987 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6990 * config/arm/arm.c (arm_valid_target_attribute_rec):
6991 Parse "arch=" and "+<ext>".
6992 (arm_valid_target_attribute_tree): Re-init global options.
6993 (arm_option_override): Make non-static.
6994 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6995 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6996 (__ARM_FEATURE_CRC32): Support undef.
6997 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6998 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
7000 2017-11-17 David Malcolm <dmalcolm@redhat.com>
7002 * gdbinit.in (break-on-diagnostic): New command.
7004 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7006 * config/i386/i386.c (ix86_expand_epilogue): Change simple
7007 return to indirect jump for EH return if control-flow
7008 protection is enabled. Change explicit 'false' argument in
7009 pro_epilogue_adjust_stack with a value of
7011 * config/i386/i386.md (simple_return_indirect_internal):
7012 Remove SImode restriction to support 64-bit.
7014 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
7016 * combine.c (added_notes_insn): New.
7017 (try_combine): Handle added_notes_insn like added_links_insn.
7018 Rewrite return value code.
7019 (distribute_notes): Set added_notes_insn to the earliest insn we added
7022 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
7024 PR rtl-optimization/82621
7025 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
7026 dest of one of those SETs is unused.
7028 2017-11-17 Richard Biener <rguenther@suse.de>
7031 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
7032 * tree-pretty-print.c (dump_generic_node): Handle
7033 annot_expr_parallel_kind.
7034 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
7035 * gimplify.c (gimple_boolify): Likewise.
7037 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7039 * config.gcc (extra_headers): Add cet.h for x86 targets.
7040 * config/i386/cet.h: New file.
7041 * doc/install.texi: Add --enable-cet/--disable-cet.
7043 2017-11-17 Richard Biener <rguenther@suse.de>
7045 PR tree-optimization/83017
7046 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
7047 (gen_parallel_loop): Properly count iterations.
7048 (parallelize_loops): Handle loop->can_be_parallel independent
7049 of flag_loop_parallelize_all. Make static profitability test match
7051 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
7052 * invoke.texi (parloops-min-per-thread): Document.
7054 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
7056 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
7057 upstreaming review comments.
7059 2017-11-17 Tamar Christina <tamar.christina@arm.com>
7061 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
7062 with fast unaligned access.
7063 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
7065 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7067 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
7068 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
7070 2017-11-17 Richard Biener <rguenther@suse.de>
7072 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
7073 folding of references.
7075 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
7078 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
7079 of replacing call to strncmp with corresponding call to strcmp when
7082 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
7084 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
7085 option prefer-avx256 for skylake-avx512 configuration.
7086 * config/i386/i386.c (ix86_option_override_internal): Ditto.
7087 (get_builtin_code_for_version): Ditto.
7089 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7090 Monk Chiang <sh.chiang04@gmail.com>
7092 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
7093 (FIXED_REGISTERS): Reserve more register numbers.
7094 (CALL_USED_REGISTERS): Likewise.
7095 (REG_ALLOC_ORDER): Likewise.
7096 (REG_CLASS_CONTENTS): Likewise.
7097 (REGISTER_NAMES): Likewise.
7099 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7100 Kito Cheng <kito.cheng@gmail.com>
7102 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
7104 * config/nds32/iterators.md: Add vector mode iterators and attributes.
7106 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
7108 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
7109 parameter list for vec_splats.
7111 2017-11-16 Joseph Myers <joseph@codesourcery.com>
7113 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
7115 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
7117 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
7120 * doc/install.texi: Document incompatibility of Perl >=5.6.26
7121 with the required version of automake 1.11.6.
7123 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
7125 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
7127 (DU_C2_3_power9): Correct reservation combinations.
7128 (FP_DIV_power9, VEC_DIV_power9): New.
7129 (power9-alu): Split out rotate/shift...
7130 (power9-rot): ...to here, correct dispatch resource.
7131 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
7133 (power9-fp): Correct latency.
7134 (power9-sdiv): Add div/sqrt resource.
7135 (power9-ddiv): Correct latency, add div/sqrt resource.
7136 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
7137 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
7139 (power9-qpdiv, power9-qpmul): Adjust resource usage.
7141 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
7143 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
7144 switch statement mapping KF built-ins to TF built-ins if we don't
7145 have the proper ISA 3.0 assembler support.
7147 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7149 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
7150 (lower_emutls_function_body): Do not compute it.
7152 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7154 * ipa-split.c (split_bb_info): Turn time to sreal.
7155 (split_point): Likewise.
7156 (dump_split_point): Likewise.
7157 (fine_split_points): Likewise.
7158 (execute_split_functions): Only zero split_bbs; turn time to sreals.
7160 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7162 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
7165 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7167 * predict.c (combine_predictions_for_bb): Preserve zero predicted
7169 (expensive_function_p): Remove useless assert.
7170 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
7172 2017-11-16 Martin Sebor <msebor@redhat.com>
7174 PR tree-optimization/82588
7175 PR tree-optimization/82583
7176 * tree-vrp.c (check_array_ref): Handle flexible array members,
7177 string literals, and inner indices.
7178 (search_for_addr_array): Add detail to diagnostics.
7180 2017-11-16 Nathan Sidwell <nathan@acm.org>
7184 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
7185 (SET_DECL_ASSEMBLER_NAME): Forward to
7186 overwrite_decl_assembler_name.
7187 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
7188 (overwrite_decl_assembler_name): Declare.
7189 * tree.c (overwrite_decl_assembler_name): New.
7190 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
7191 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
7192 (LANG_HOOKS_INITIALIZER): Add it.
7193 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
7194 * langhooks.c (lhd_set_decl_assembler_name): Use
7195 SET_DECL_ASSEMBLER_NAME.
7196 (lhd_overwrite_decl_assembler_name): Default implementation.
7198 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7199 Jackson Woodruff <jackson.woodruff@arm.com>
7201 PR tree-optimization/71026
7202 * match.pd: Canonicalize constant multiplies in division.
7204 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7206 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7207 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7208 Remove OPT_fomit_frame_pointer entry.
7209 * common/config/arc/arc-common.c: Likewise.
7210 * common/config/arm/arm-common.c: Likewise.
7211 * common/config/avr/avr-common.c: Likewise.
7212 * common/config/c6x/c6x-common.c: Likewise.
7213 * common/config/cr16/cr16-common.c: Likewise.
7214 * common/config/cris/cris-common.c: Likewise.
7215 * common/config/epiphany/epiphany-common.c: Likewise.
7216 * common/config/fr30/fr30-common.c: Likewise.
7217 * common/config/frv/frv-common.c: Likewise.
7218 * common/config/ia64/ia64-common.c: Likewise.
7219 * common/config/iq2000/iq2000-common.c: Likewise.
7220 * common/config/lm32/lm32-common.c: Likewise.
7221 * common/config/m32r/m32r-common.c: Likewise.
7222 * common/config/mcore/mcore-common.c: Likewise.
7223 * common/config/microblaze/microblaze-common.c: Likewise.
7224 * common/config/mips/mips-common.c: Likewise.
7225 * common/config/mmix/mmix-common.c: Likewise.
7226 * common/config/mn10300/mn10300-common.c: Likewise.
7227 * common/config/nios2/nios2-common.c: Likewise.
7228 * common/config/pa/pa-common.c: Likewise.
7229 * common/config/pdp11/pdp11-common.c: Likewise.
7230 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7231 * common/config/riscv/riscv-common.c: Likewise.
7232 * common/config/rs6000/rs6000-common.c: Likewise.
7233 * common/config/rx/rx-common.c: Likewise.
7234 * common/config/s390/s390-common.c: Likewise.
7235 * common/config/sh/sh-common.c: Likewise.
7236 * common/config/sparc/sparc-common.c: Likewise.
7237 * common/config/tilegx/tilegx-common.c: Likewise.
7238 * common/config/tilepro/tilepro-common.c: Likewise.
7239 * common/config/v850/v850-common.c: Likewise.
7240 * common/config/visium/visium-common.c: Likewise.
7241 * common/config/xstormy16/xstormy16-common.c: Likewise.
7242 * common/config/xtensa/xtensa-common.c: Likewise.
7243 * invoke.texi (-fomit-frame-pointer): Update documentation.
7245 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7247 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7249 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7251 * cfg.c (scale_bbs_frequencies_int,
7252 cale_bbs_frequencies_gcov_type): Remove.
7253 * cfg.h (scale_bbs_frequencies_int,
7254 cale_bbs_frequencies_gcov_type): Remove.
7256 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7258 * tree-ssa-loop-manip.c
7259 (scale_dominated_blocks_in_loop): Update to profile counts.
7260 (tree_transform_and_unroll_loop): Likewise.
7262 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7264 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7265 scale_bbs_frequencies_int.
7267 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7269 * final.c (compute_alignments): Use counts rather than frequencies.
7271 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7273 * cfgloopanal.c: Include sreal.h
7274 (average_num_loop_insns): Use counts and sreal for accounting.
7276 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7278 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7281 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7282 Kito Cheng <kito.cheng@gmail.com>
7284 * config/nds32/constraints.md: Provide more constraints.
7285 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7286 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7287 support constraints usage.
7289 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7291 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7293 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7294 Kito Cheng <kito.cheng@gmail.com>
7296 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7297 * config/nds32/nds32.opt: Refine the layout.
7298 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7299 TARGET_EXT_STRING): Support new options.
7300 * config/nds32/nds32.h: Likewise.
7301 * config/nds32/nds32.md: Likewise.
7302 * config/nds32/nds32-predicates.c: Likewise.
7303 * config/nds32/constraints.md: Likewise.
7304 * common/config/nds32/nds32-common.c: Likewise.
7306 2017-11-16 Julia Koval <julia.koval@intel.com>
7309 * config/i386/gfniintrin.h: Add sse check.
7310 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7312 2017-11-16 Julia Koval <julia.koval@intel.com>
7314 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7315 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7316 (ix86_handle_option): Handle -mavx512vbmi2.
7317 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7318 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7319 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7320 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7321 (ix86_valid_target_attribute_inner_p): Ditto.
7322 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7323 * config/i386/i386.opt (mavx512vbmi2): New option.
7324 * doc/invoke.texi: Add new option.
7326 2017-11-16 Julia Koval <julia.koval@intel.com>
7328 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7329 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7330 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7331 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7332 _mm512_gf2p8mul_epi8): New intrinsics.
7333 * config/i386/i386-builtin-types.def
7334 (V64QI_FTYPE_V64QI_V64QI): New type.
7335 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7336 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7337 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7338 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7339 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7340 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7342 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7344 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7345 explicitly as a stream of bytes.
7347 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7349 * config/rs6000/altivec.h (vec_xst_be): New #define.
7350 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7351 and externalize from *altivec_vperm_<mode>_internal.
7352 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7354 (XL_BE_V8HI): Likewise.
7355 (XL_BE_V4SI): Likewise.
7356 (XL_BE_V4SI): Likewise.
7357 (XL_BE_V2DI): Likewise.
7358 (XL_BE_V4SF): Likewise.
7359 (XL_BE_V2DF): Likewise.
7360 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7361 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7362 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7363 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7364 VSX_BUILTIN_VEC_XST_BE.
7365 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7366 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7368 (altivec_init_builtins): Replace conditional calls to def_builtin
7369 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7370 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7371 with unconditional calls. Remove calls to def_builtin for
7372 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7373 __builtin_vec_xst_be.
7374 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7375 to define_expand, and add alternate RTL generation for P8.
7376 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7377 vsx_ld_elemrev_v8hi.
7378 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7379 add alternate RTL generation for P8.
7380 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7381 vsx_ld_elemrev_v16qi.
7382 (vsx_st_elemrev_v8hi): Convert define_insn
7383 to define_expand, and add alternate RTL generation for P8.
7384 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7385 vsx_st_elemrev_v8hi.
7386 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7387 add alternate RTL generation for P8.
7388 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7389 vsx_st_elemrev_v16qi.
7391 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7394 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7395 TARGET_AVX512ER check.
7396 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7397 neither -mzeroupper nor -mno-zeroupper is used and
7398 TARGET_EMIT_VZEROUPPER is set.
7399 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7400 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7402 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7404 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7405 folding of vector compares.
7406 (fold_build_vec_cmp): New helper function.
7407 (fold_compare_helper): New helper function.
7408 (builtin_function_type): Add compare builtins to the list of functions
7409 having unsigned arguments. Cosmetic updates to comment indentation.
7410 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7411 the not+eq combination.
7413 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7415 PR tree-optimization/82726
7416 PR tree-optimization/70754
7417 * tree-predcom.c (order_drefs_by_pos): New function.
7418 (combine_chains): Move code setting has_max_use_after to...
7419 (try_combine_chains): ...here. New parameter. Sort combined chains
7420 according to position information.
7421 (tree_predictive_commoning_loop): Update call to above function.
7422 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7424 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7426 PR tree-optimization/82726
7428 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7430 PR tree-optimization/70754
7431 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7432 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7433 combined stmt before it if not NULL.
7434 (combine_chains): Process refs reversely and compute dominance point
7438 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7440 PR tree-optimization/79663
7441 * tree-predcom.c (combine_chains): Process refs in reverse order
7442 only for ZERO length chains, and add explaining comment.
7444 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7446 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7447 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7448 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7450 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7452 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7454 2017-11-15 Martin Liska <mliska@suse.cz>
7456 * tree-cfg.c (pass_warn_function_return::execute):
7457 Compare warn_return_type for greater than zero.
7459 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7463 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7464 to return true on Xeon and not on Xeon Phi.
7465 (ix86_check_avx256_register): Changed to ...
7466 (ix86_check_avx_upper_register): ... this. Add extra check for
7467 VALID_AVX512F_REG_OR_XI_MODE.
7468 (ix86_avx_u128_mode_needed): Changed
7469 ix86_check_avx256_register to ix86_check_avx_upper_register.
7470 (ix86_check_avx256_stores): Changed to ...
7471 (ix86_check_avx_upper_stores): ... this. Changed
7472 ix86_check_avx256_register to ix86_check_avx_upper_register.
7473 (ix86_avx_u128_mode_after): Changed
7474 avx_reg256_found to avx_upper_reg_found. Changed
7475 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7476 (ix86_avx_u128_mode_entry): Changed
7477 ix86_check_avx256_register to ix86_check_avx_upper_register.
7478 (ix86_avx_u128_mode_exit): Ditto.
7479 * config/i386/i386.h: (host_detect_local_cpu): New define.
7481 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7483 * config/arm/xgene1.md (xgene1): Split into automatons
7484 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7485 (xgene1_f_load): Adjust reservations and/or types.
7486 (xgene1_f_store): Likewise.
7487 (xgene1_load_pair): Likewise.
7488 (xgene1_store_pair): Likewise.
7489 (xgene1_fp_load1): Likewise.
7490 (xgene1_load1): Likewise.
7491 (xgene1_store1): Likewise.
7492 (xgene1_move): Likewise.
7493 (xgene1_alu): Likewise.
7494 (xgene1_simd): Likewise.
7495 (xgene1_bfm): Likewise.
7496 (xgene1_neon_load1): Likewise.
7497 (xgene1_neon_store1): Likewise.
7498 (xgene1_neon_logic): Likewise.
7499 (xgene1_neon_st1): Likewise.
7500 (xgene1_neon_ld1r): Likewise.
7501 (xgene1_alu_cond): Added.
7502 (xgene1_shift_reg): Likwise.
7503 (xgene1_bfx): Likewise.
7504 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7506 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7509 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7511 (can_widen_mult_without_libcall): New function.
7512 (expand_mul_overflow): If only checking unsigned mul overflow,
7513 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7514 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7515 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7516 (expand_DIVMOD): Formatting fix.
7517 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7518 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7519 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7521 PR tree-optimization/82977
7522 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7523 constructed temporary to strlen_to_stridx.put.
7525 2017-11-15 Martin Liska <mliska@suse.cz>
7527 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7528 * configure: Regenerate.
7530 2017-11-15 Martin Liska <mliska@suse.cz>
7533 * config/sh/sh-mem.cc: Use proper probability for
7536 2017-11-14 Jeff Law <law@redhat.com>
7538 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7539 the red zone for stack_clash_protection_final_dynamic_probe targets
7540 when the total dynamic stack size is zero bytes.
7542 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7543 blocks is post order.
7545 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7547 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7548 * final.c (rest_of_clean_state): Set it for the
7549 -fcompare-debug dump.
7550 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7551 class when TDF_COMPARE_DEBUG is set.
7553 * dwarf2out.c (gen_producer_string): Discard
7556 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7559 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7560 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7561 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7562 (__TGMATH_REAL_2_3): Remove macros.
7563 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7564 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7566 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7568 (remquo): Define using __TGMATH_REAL_3.
7570 2017-11-14 Jeff Law <law@redhat.com>
7572 * vr-values.c: New file with contents extracted from tree-vrp.c.
7573 * Makefile.in (OBJS): Add vr-values.o
7574 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7575 (set_value_range, set_and_canonicalize_value_range): Likewise.
7576 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7577 (value_range_constant_singleton, symbolic_range_p): Likewise.
7578 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7579 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7580 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7581 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7582 (find_case_label_range, find_case_label_index): Likewise.
7583 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7584 (range_int_cst_singleton_p, value_inside_range): Likewise.
7585 (get_single_symbol): Likewise.
7586 (switch_update): Move structure definition here.
7587 (to_remove_edges, to_update_switch_stmts): Provide externs.
7588 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7589 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7590 (vrp_val_is_min, set_value_range): Likewise.
7591 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7592 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7593 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7594 (range_is_nonnull, range_int_cst_p): Likewwise.
7595 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7596 (get_single_symbol, operand_less_p): Likewise
7597 (compare_values_warnv, compare_values): Likewise.
7598 (value_inside_range, value_range_constant_singleton): Likewise.
7599 (zero_nonzero_bitgs_from_vr): Likewise.
7600 (extract_range_from_binary_expr_1): Likewise.
7601 (overflow_comparison_p): Likewise.
7602 (to_remove_edges, to_update_switch_stmts): Likewise.
7603 (find_case_label-index, find_case_label_range): Likewise.
7604 (switch_update, set_value_range_to_nonnegative): Remove.
7605 (set_value_range_to_truthvalue): Likewise.
7606 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7607 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7608 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7609 (find_case_label_ranges, test_for_singularity): Likewise.
7610 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7611 (x_vr_values): Move to its remaining use site.
7613 2017-11-10 Jeff Law <law@redhat.com>
7615 * vr-values.h (VR_INITIALIZER): Move #define here.
7616 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7617 * Makefile.in (OBJS): Add tree-evrp.o
7618 * tree-vrp.h (assert_info): Move structure definition here.
7619 (set_value_range_to_varying): Prototype.
7620 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7621 (infer_value_range, register_edge_assert_for): Likewise.
7622 (stmt_interesting_for_vrp): Likewise.
7623 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7624 (set_value_range_to_varying): No longer static.
7625 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7626 (infer_value_range, register_edge_assert_for): Likewise.
7628 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7630 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7631 generate the XXBRD instruction.
7633 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7634 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7635 as being compatible if -mabi=ieeelongdouble.
7636 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7637 to setup float128 built-ins with hardware support.
7638 (BU_FLOAT128_HW_2): Likewise.
7639 (BU_FLOAT128_HW_3): Likewise.
7640 (BU_FLOAT128_HW_VSX_1): Likewise.
7641 (BU_FLOAT128_HW_VSX_2): Likewise.
7642 (scalar_extract_expq): Change float128 built-in functions to
7643 accommodate having both KFmode and TFmode functions. Use the
7644 KFmode variant as the default.
7645 (scalar_extract_sigq): Likewise.
7646 (scalar_test_neg_qp): Likewise.
7647 (scalar_insert_exp_q): Likewise.
7648 (scalar_insert_exp_qp): Likewise.
7649 (scalar_test_data_class_qp): Likewise.
7650 (sqrtf128_round_to_odd): Delete processing the round to odd
7651 built-in functions as special built-in functions, and define them
7652 as float128 built-ins. Use the KFmode variant as the default.
7653 (truncf128_round_to_odd): Likewise.
7654 (addf128_round_to_odd): Likewise.
7655 (subf128_round_to_odd): Likewise.
7656 (mulf128_round_to_odd): Likewise.
7657 (divf128_round_to_odd): Likewise.
7658 (fmaf128_round_to_odd): Likewise.
7659 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7660 support for KFmode and TFmode xststdcqp calls.
7661 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7662 point, switch the built-in handlers for the get/set float128
7663 exponent, get float128 mantissa, float128 test built-ins, and the
7664 float128 round to odd built-in functions. Eliminate creating the
7665 float128 round to odd built-in functions as special built-ins.
7666 (rs6000_init_builtins): Eliminate special creation of the float128
7667 round to odd built-in functions.
7668 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7669 function insns to support both TFmode and KFmode variants.
7670 (xsxsigqp_<mode>): Likewise.
7671 (xsiexpqpf_<mode>): Likewise.
7672 (xsiexpqp_<mode>): Likewise.
7673 (xststdcqp_<mode>): Likewise.
7674 (xststdcnegqp_<mode>): Likewise.
7675 (xststdcqp_<mode>): Likewise.
7677 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7679 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7680 unused path_in_freq_ptr parameter.
7681 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7683 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7685 * ipa-inline.c (edge_badness): Dump sreal frequency.
7686 (compute_inlined_call_time): Match natural implementaiton ...
7687 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7688 forgotten division by CGRAPH_FREQ_BASE.
7690 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7692 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7693 Solaris 11. Update comment.
7694 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7696 * config/sol2.h (STARTFILE_SPEC): Likewise.
7697 * configure: Regenerate.
7699 2017-11-14 Carl Love <cel@us.ibm.com>
7701 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7702 le_ and be_ prefixes to swap* variables. Remove
7703 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7706 2017-11-14 Jason Merrill <jason@redhat.com>
7708 Support GTY((cache)) on hash_map.
7709 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7710 (ggc_cache_remove): Override it instead of ggc_mx.
7711 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7712 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7713 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7714 (simple_cache_map_traits): Override maybe_mx.
7715 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7716 (hash_map): Friend gt_cleare_cache.
7717 (gt_cleare_cache): New.
7718 * tree.h (tree_cache_traits): New hash_map traits class.
7719 (tree_cache_map): New typedef.
7721 2017-11-14 Richard Biener <rguenther@suse.de>
7723 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7724 paramter and handling.
7725 (cleanup_control_flow_bb): Likewise.
7726 (cleanup_control_flow_pre): New helper performing a DFS walk
7727 to call cleanup_control_flow_bb in PRE order.
7728 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7729 via cleanup_control_flow_pre.
7731 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7733 * config/aarch64/aarch64-simd.md
7734 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7735 (*aarch64_simd_bsl<mode>_alt): Likewise.
7736 (aarch64_simd_bsldi_internal): New.
7737 (aarch64_simd_bsldi_alt): Likewise.
7739 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7741 * tracer.c (better_p): Do not compare frequencies.
7742 * reg-stack.c (better_edge): Likewise.
7743 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7746 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7748 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7749 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7750 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7751 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7752 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7753 * predict.c (maybe_hot_frequency_p): Inline to ...
7754 (maybe_hot_count_p): ... here; rewrite to counts.
7755 (counts_to_freqs): Rename to ...
7756 (update_max_bb_count): ... this one.
7757 (expensive_function_p): Use counts.
7758 (estimate_bb_frequencies): Update.
7759 (rebuild_frequencies): Update.
7760 * predict.h (counts_to_freqs): Rename to ...
7761 (update_max_bb_count): ... this one.
7762 * profile.c (compute_branch_probabilities): Add debug info
7763 * tree-inline.c (expand_call_inline): Update debug info.
7764 (optimize_inline_calls): Use update_max_bb_count..
7765 (tree_function_versioning): Use update_max_bb_count..
7766 * value-prof.c (gimple_value_profile_transformations):
7767 Do not use update_max_bb_count.
7769 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7771 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7772 always use frequencies.
7774 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7776 * bb-reorder.c: Remove frequencies from comments.
7777 (better_edge_p): Use profile counts.
7778 (find_traces): Dump profile counts.
7779 (rotate_loop): Use profile counts.
7780 (find_traces_1_round): Likewise.
7781 (connect_better_edge_p): Use counts instead of probabilities for
7783 (copy_bb_p): Drop early check for non-0 frequency.
7784 (sanitize_hot_paths): Update comments.
7786 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7788 * ipa-split.c (struct split_point): Add count.
7789 (consider_split): Do not compute incoming frequency; compute incoming
7790 count and store it to split_point.
7791 (split_function): Set count of the call to split part correctly.
7793 2017-11-13 Carl Love <cel@us.ibm.com>
7795 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7797 2017-11-13 Tom Tromey <tom@tromey.com>
7799 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7801 2017-11-13 Carl Love <cel@us.ibm.com>
7803 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7804 Add support for builtins:
7805 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7806 vector {un,}signed {char,int,short},
7807 vector {un,}signed {char,int,short}) arguments.
7808 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7809 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7810 Add BU_P9V_AV_2 expansions for the builtins.
7811 * config/rs6000/altivec.h (vec_first_match_index,
7812 vec_first_mismatch_index, vec_first_match_or_eos_index,
7813 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7814 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7815 new extern declaration.
7816 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7817 * config/rs6000/vsx.md (first_match_index_<mode>,
7818 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7819 first_mismatch_or_eos_index_<mode>): Add define expand.
7820 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7821 * doc/extend.texi: Update the built-in documenation file for the new
7824 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7826 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7827 into the min/max operations for _Float<N> and _Float<N>X types.
7829 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7832 * dwarf2out.c (do_eh_frame): New static variable.
7833 (dwarf2out_begin_prologue): Set it.
7834 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7836 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7838 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7840 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7842 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7845 * debug.h (dwarf2out_do_eh_frame): Declare.
7846 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7847 (dwarf2out_do_frame): Use it.
7848 (dwarf2out_do_cfi_asm): Likewise.
7849 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7850 (dwarf2out_assembly_start): Likewise.
7851 (dwarf2out_begin_prologue): Fix comment.
7852 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7853 if the target needs either debug or unwind DWARF2 info.
7854 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7855 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7856 -ftrapping-math, -ftrapv and -fwrapv.
7858 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7860 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7861 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7862 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7863 (estimate_edge_size_and_time): Likewise.
7864 (ipa_merge_fn_summary_after_inlining): Likewise.
7865 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7866 (compute_uninlined_call_time): Use sreal_frequency.
7867 (compute_inlined_call_time): Likewise.
7868 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7869 * profile-count.c: Include sreal.h.
7870 (profile_count::to_sreal_scale): New.
7871 * profile-count.h: Forward declare sreal.
7872 (profile_count::to_sreal_scale): Declare.
7874 2017-11-13 Nathan Sidwell <nathan@acm.org>
7876 * diagnostic.c (maybe_line_and_column): New.
7877 (diagnostic_get_location_text): Use it.
7878 (diagnostic_report_current_module): Likewise.
7879 (test_diagnostic_get_location_text): Add tests.
7881 2017-11-13 Luis Machado <luis.machado@linaro.org>
7883 * doc/md.texi (Specifying processor pipeline description): Fix
7884 incorrect latency for the div instruction example.
7886 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7888 PR tree-optimization/78821
7889 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7890 that bit_not_p is the same.
7891 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7892 (split_group): Count precisely bit_not_p bits in each statement.
7893 (invert_op): New function.
7894 (imm_store_chain_info::output_merged_store): Use invert_op to
7895 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7896 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7898 2017-11-13 Martin Liska <mliska@suse.cz>
7900 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7901 (struct source_info): Likewise.
7902 (add_branch_counts): Likewise.
7903 (add_line_counts): Likewise.
7904 (function_summary): Likewise.
7905 (output_intermediate_line): Likewise.
7906 (generate_results): Likewise.
7908 2017-11-13 Martin Liska <mliska@suse.cz>
7910 * gcov.c (struct block_info): Remove typedef for block_t.
7911 (struct line_info): Likewise.
7912 (line_info::has_block): Likewise.
7913 (EXIT_BLOCK): Likewise.
7914 (unblock): Likewise.
7915 (circuit): Likewise.
7916 (get_cycles_count): Likewise.
7917 (process_file): Likewise.
7918 (read_graph_file): Likewise.
7919 (solve_flow_graph): Likewise.
7920 (find_exception_blocks): Likewise.
7921 (add_line_counts): Likewise.
7922 (accumulate_line_info): Likewise.
7923 (output_line_details): Likewise.
7925 2017-11-13 Martin Liska <mliska@suse.cz>
7927 * gcov.c (struct arc_info): Remove typedef for arc_t.
7928 (struct line_info): Likewise.
7929 (add_branch_counts): Likewise.
7930 (output_branch_count): Likewise.
7931 (function_info::~function_info): Likewise.
7932 (circuit): Likewise.
7933 (output_intermediate_line): Likewise.
7934 (read_graph_file): Likewise.
7935 (solve_flow_graph): Likewise.
7936 (find_exception_blocks): Likewise.
7937 (add_line_counts): Likewise.
7938 (accumulate_line_info): Likewise.
7939 (output_line_details): Likewise.
7940 (output_function_details): Likewise.
7942 2017-11-13 Martin Liska <mliska@suse.cz>
7944 * gcov.c (struct function_info): Remove typedef for function_t.
7945 (struct source_info): Likewise.
7946 (source_info::get_functions_at_location): Likewise.
7947 (solve_flow_graph): Likewise.
7948 (find_exception_blocks): Likewise.
7949 (add_line_counts): Likewise.
7950 (output_intermediate_file): Likewise.
7951 (process_file): Likewise.
7952 (generate_results): Likewise.
7953 (release_structures): Likewise.
7954 (read_graph_file): Likewise.
7955 (read_count_file): Likewise.
7956 (accumulate_line_counts): Likewise.
7957 (output_lines): Likewise.
7959 2017-11-13 Martin Liska <mliska@suse.cz>
7961 * gcov.c (function_info::function_info): Remove num_counts
7962 and add vector<gcov_type>.
7963 (function_info::~function_info): Use the vector.
7964 (process_file): Likewise.
7965 (read_graph_file): Likewise.
7966 (read_count_file): Likewise.
7967 (solve_flow_graph): Likewise.
7969 2017-11-13 Martin Liska <mliska@suse.cz>
7971 * gcov.c (function_info::is_artificial): New function.
7972 (process_file): Erase all artificial early.
7973 (generate_results): Skip as all artificial are already
7976 2017-11-13 Martin Liska <mliska@suse.cz>
7978 * gcov.c (read_graph_file): Store to global vector of functions.
7979 (read_count_file): Iterate the vector.
7980 (process_file): Likewise.
7981 (generate_results): Likewise.
7982 (release_structures): Likewise.
7984 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7986 PR tree-optimization/82954
7987 * gimple-ssa-store-merging.c
7988 (imm_store_chain_info::coalesce_immediate_stores): If
7989 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7991 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7993 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7994 Upddate call to ENDIAN_LANE_N.
7995 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7996 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7997 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7998 and use aarch64_endian_lane_rtx.
7999 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
8001 2017-11-12 Tom de Vries <tom@codesourcery.com>
8003 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
8005 2017-11-12 Tom de Vries <tom@codesourcery.com>
8007 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
8009 2017-11-12 Tom de Vries <tom@codesourcery.com>
8011 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
8013 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
8014 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
8015 * defaults.h (ASM_OUTPUT_LABELREF): Same.
8017 2017-11-11 Martin Sebor <msebor@redhat.com>
8020 * doc/extend.texi (attribute nonstring): Remove spurious argument.
8023 * prefic.c (translate_name): Replace strncpy with memcpy to
8024 avoid -Wstringop-truncation.
8026 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8028 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
8030 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8032 * predict.c (maybe_hot_frequency_p): Do not use cfun.
8034 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8036 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
8039 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
8041 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
8042 enable generating XXBRH if the value is in a vector register.
8043 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
8044 value is in a vector register.
8045 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
8046 register to register bswap64's instead of doing the GPR sequence
8047 used on previous machines.
8048 (bswapdi2_xxbrd): New insn.
8049 (bswapdi2_reg): Disallow on ISA 3.0.
8050 (register to register bswap64 splitter): Do not split the insn on
8051 ISA 3.0 systems that use XXBRD.
8053 2017-11-10 Martin Sebor <msebor@redhat.com>
8056 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
8058 (find_subframework_file): Same.
8060 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8062 * auto-profile.c (afdo_indirect_call): Drop frequency.
8063 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
8064 (cgraph_node::create_edge): Drop frequency argument.
8065 (cgraph_node::create_indirect_edge): Drop frequency argument.
8066 (cgraph_edge::make_speculative): Drop frequency arguments.
8067 (cgraph_edge::resolve_speculation): Do not update frequencies
8068 (cgraph_edge::dump_edge_flags): Do not dump frequency.
8069 (cgraph_node::dump): Check consistency in IPA mode.
8070 (cgraph_edge::maybe_hot_p): Use IPA counter.
8071 (cgraph_edge::verify_count_and_frequency): Rename to ...
8072 (cgraph_edge::verify_count): ... this one; drop frequency checking.
8073 (cgraph_node::verify_node): Update.
8074 * cgraph.h (struct cgraph_edge): Drop frequency.
8075 (cgraph_edge::frequency): New function.
8076 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
8078 (cgraph_edge::rebuild_edges): Likewise.
8079 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
8080 (duplicate_thunk_for_node): Do not pass frequency.
8081 (cgraph_node::create_clone): Scale only counts.
8082 (cgraph_node::create_virtual_clone): Do not pass frequency.
8083 (cgraph_node::create_edge_including_clones): Do not pass frequency.
8084 (cgraph_node::create_version_clone): Do not pass frequency.
8085 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
8086 (cgraph_node::expand_thunk): Do not pass frequency.
8087 (cgraph_node::create_wrapper): Do not pass frequency.
8088 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
8090 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
8091 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
8092 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
8093 (gather_caller_stats): Use frequency function.
8094 (ipcp_cloning_candidate_p): Use frequency function.
8095 (ipcp_propagate_stage): Use frequency function.
8096 (get_info_about_necessary_edges): Use frequency function.
8097 (update_profiling_info): Update only IPA profile.
8098 (update_specialized_profile): Use frequency functoin.
8099 (perhaps_add_new_callers): Update only IPA profile.
8100 * ipa-devirt.c (ipa_devirt): Use IPA profile.
8101 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
8102 (dump_ipa_call_summary): Use frequency function.
8103 (estimate_edge_size_and_time): Use frequency function.
8104 (ipa_merge_fn_summary_after_inlining): Use frequency function.
8105 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
8106 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
8107 (update_noncloned_counts): ... ths one; scale counts only.
8108 (clone_inlined_nodes): Do not scale frequency.
8109 (inline_call): Do not pass frequency.
8110 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
8111 (compute_inlined_call_time): Use IPA profile.
8112 (want_inline_small_function_p): Use IPA profile.
8113 (want_inline_self_recursive_call_p): Use IPA profile.
8114 (edge_badness): Use IPA profile.
8115 (lookup_recursive_calls): Use IPA profile.
8116 (recursive_inlining): Do not pass frequency.
8117 (resolve_noninline_speculation): Do not update frequency.
8118 (inline_small_functions): Collect max of IPA profile.
8119 (dump_overall_stats): Dump IPA porfile.
8120 (dump_inline_stats): Dump IPA porfile.
8121 (ipa_inline): Collect IPA stats.
8122 * ipa-inline.h (clone_inlined_nodes): Update prototype.
8123 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
8124 (ipa_propagate_frequency): Use frequency function.
8125 (ipa_profile): Cleanup.
8126 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
8127 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
8128 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
8129 (input_node): Do not stream frequency.
8130 (input_edge): Do not stream frequency.
8131 (merge_profile_summaries): Scale only IPA profiles.
8132 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
8133 * predict.c (drop_profile): Do not recompute frequency.
8134 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
8135 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
8136 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
8137 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
8138 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
8139 * tree-inline.c (copy_bb): Do not scale frequency.
8140 (expand_call_inline): Do not scale frequency.
8141 (tree_function_versioning): Do not scale frequency.
8142 * ubsan.c (ubsan_create_edge): Do not pass frequency.
8144 2017-11-10 Julia Koval <julia.koval@intel.com>
8146 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
8147 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
8148 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
8149 (_mm256_maskz_gf2p8affine_epi64_epi8)
8150 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
8151 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
8152 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
8153 (__builtin_ia32_vgf2p8affineqb_v32qi)
8154 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
8155 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
8157 2017-11-10 Tamar Christina <tamar.christina@arm.com>
8161 (arm_option_override): Refactor.
8162 (arm_option_reconfigure_globals): New.
8163 (arm_options_perform_arch_sanity_checks): New.
8164 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
8166 (arm_options_perform_arch_sanity_checks): Likewise
8168 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
8170 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
8171 (power9-qpmul): New.
8172 * rs6000/rs6000.md ("type" attr): Add qmul.
8173 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
8174 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
8175 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
8177 2017-11-10 Martin Sebor <msebor@redhat.com>
8180 * builtins.c (compute_objsize): Handle arrays that
8181 compute_builtin_object_size likes to fail for. Make extern.
8182 * builtins.h (compute_objsize): Declare.
8183 (check_strncpy_sizes): New function.
8184 (expand_builtin_strncpy): Call check_strncpy_sizes.
8185 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
8186 -Wstringop-truncation.
8187 (gimple_fold_builtin_strncat): Same.
8188 * gimple.c (gimple_build_call_from_tree): Set call location.
8189 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
8190 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
8191 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
8192 (handle_builtin_strlen): Use strlen_to_stridx.
8193 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
8195 Use strlen_to_stridx.
8196 (pass_strlen::execute): Release strlen_to_stridx.
8197 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
8198 (-Wstringop-truncation): Document new option.
8200 2017-11-10 Martin Liska <mliska@suse.cz>
8202 PR gcov-profile/82702
8203 * gcov.c (main): Handle intermediate files in a different
8205 (get_gcov_intermediate_filename): New function.
8206 (output_gcov_file): Remove support of intermediate files.
8207 (generate_results): Allocate intermediate file.
8208 (release_structures): Clean-up properly fn_end.
8209 (output_intermediate_file): Start iterating with line 1.
8211 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8213 PR tree-optimization/82929
8214 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8215 ops_swapped_p non-static data member.
8216 (store_immediate_info::store_immediate_info): Clear it.
8217 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8218 ops set ops_swapped_p.
8219 (count_multiple_uses): Handle ops_swapped_p.
8221 2017-11-10 Martin Liska <mliska@suse.cz>
8223 * coverage.c (coverage_init): Stream information about
8224 support of has_unexecuted_blocks.
8225 * doc/gcov.texi: Document that.
8226 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8227 * gcov.c (read_graph_file): Likewise.
8228 (output_line_beginning): Fix a small issue with
8231 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8233 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8234 reference of trivial component.
8236 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8239 * gimple-ssa-store-merging.c
8240 (pass_store_merging::terminate_all_aliasing_chains): For
8241 gimple_store_p stmts also call refs_output_dependent_p.
8243 PR rtl-optimization/82913
8244 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8247 2017-11-09 Jeff Law <law@redhat.com>
8249 * vr-values.h: New file with vr_values class.
8250 * tree-vrp.c: Include vr-values.h
8251 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8252 data objects into the vr_values class.
8253 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8254 (get_value_range): Make it a member function within vr_values class.
8255 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8256 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8257 (op_with_constant_singleton_value_range): Likewise.
8258 (extract_range_for_var_from_comparison_expr): Likewise.
8259 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8260 (extract_range_from_binary_expr): Likewise.
8261 (extract_range_from_unary_expr): Likewise.
8262 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8263 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8264 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8265 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8266 (compare_name_with_value, compare_names): Likewise.
8267 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8268 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8269 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8270 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8271 (extract_range_from_phi_node): Likewise.
8272 (simplify_truth_ops_using_ranges): Likewise.
8273 (simplify_div_or_mod_using_ranges): Likewise.
8274 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8275 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8276 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8277 (simplify_float_conversion_using_ranges): Likewise.
8278 (simplify_internal_call_using_ranges): Likewise.
8279 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8280 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8281 poitner via x_vr_values for calls into gimple folder.
8282 (vrp_initialize_lattice): Make this the vr_values ctor.
8283 (vrp_free_lattice): Make this the vr_values dtor.
8284 (set_vr_value): New function.
8285 (class vrp_prop): Add vr_values data member. Add various member
8286 functions as well as member functions that delegate to vr_values.
8287 (check_array_ref): Make a member function within vrp_prop class.
8288 (search_for_addr_array, vrp_initialize): Likewise.
8289 (vrp_finalize): Likewise. Revamp to avoid direct access to
8290 vr_value, values_propagated, etc.
8291 (check_array_bounds): Extract vrp_prop class instance pointer from
8292 walk info structure. Use it to call member functions.
8293 (check_all_array_refs): Make a member function within vrp_prop class.
8294 Smuggle class instance pointer via walk info structure.
8295 (x_vr_values): New local static.
8296 (vrp_valueize): Use x_vr_values to get class instance.
8297 (vr_valueize_1): Likewise.
8298 (class vrp_folder): Add vr_values data member. Add various member
8299 functions as well as member functions that delegate to vr_values.
8300 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8301 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8302 class instance from vr_values. Use it to call member functions.
8303 (vrp_dom_walker): Add vr_values data member.
8304 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8305 instance via x_vr_values.
8306 (identify_jump_threads): Accept vr_values as argument. Store
8307 it into the walker structure.
8308 (evrp_dom_walker): Add vr_values class data member. Add various
8310 (evrp_dom_walker::try_find_new_range): Use vr_values data
8311 member to access the memory allocator.
8312 (evrp_dom_walker::before_dom_children): Store vr_values class
8313 instance into the vrp_folder class.
8314 (evrp_dom_walker::push_value_range): Rework to avoid direct
8315 access to num_vr_values and vr_value.
8316 (evrp_dom_walker::pop_value_range): Likewise.
8317 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8318 Use vr_values to get to dump_all_value_ranges member function.
8319 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8320 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8321 Pass vr_values class instance down to identify_jump_threads.
8322 Remove call to vrp_free_lattice.
8323 (debug_all_value_ranges): Remove.
8325 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8326 (vrp_folder): Likewise.
8328 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8329 Get it from the existing bitmap instead.
8330 (vrp_intersect_ranges_1): Likewise.
8332 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8334 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8336 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8337 set bit_not_p to it.
8338 (imm_store_chain_info::coalesce_immediate_stores): Break group
8339 if bit_not_p is different.
8340 (count_multiple_uses, split_group,
8341 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8342 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8343 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8344 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8345 to store_immediate_info ctor.
8347 2017-11-09 Jim Wilson <jimw@sifive.com>
8349 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8350 (scan_prog_file): Likewise.
8352 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8354 * bb-reorder.c (max_entry_frequency): Remove.
8355 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8356 connect_traces, push_to_next_round_p): Remove prototypes.
8357 (find_traces_1_round): Use counts only.
8358 (push_to_next_round_p): Likewise.
8359 (find_traces): Likewise.
8360 (rotate_loop): Likewise.
8361 (find_traces_1_round): Likewise.
8362 (connect_traces): Likewise.
8363 (edge_order): Likewise.
8365 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8367 * config/arm/arm.c (output_return_instruction): Add comments to
8368 indicate requirement for cmse_nonsecure_entry return to account
8369 for the size of clearing instruction output here.
8370 (thumb_exit): Likewise.
8371 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8372 return in hardfloat mode.
8374 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8376 * config/rs6000/rs6000.c (machine_function): Add a bool,
8377 "toc_is_wrapped_separately".
8378 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8379 if it wasn't explicitly set or unset, we are optimizing for speed, and
8380 doing separate shrink-wrapping.
8381 (rs6000_get_separate_components): Enable the TOC component if
8382 saving the TOC register in the prologue.
8383 (rs6000_components_for_bb): Handle the TOC component.
8384 (rs6000_emit_prologue_components): Store the TOC register where needed.
8385 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8386 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8388 2017-11-09 Martin Jambor <mjambor@suse.cz>
8390 * ipa-param-manipulation.c: New file.
8391 * ipa-param-manipulation.h: Likewise.
8392 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8393 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8394 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8395 (ipa_parm_adjustment): Likewise.
8396 (ipa_parm_adjustment_vec): Likewise.
8397 (ipa_get_vector_of_formal_parms): Moved declaration to
8398 ipa-param-manipulation.h.
8399 (ipa_get_vector_of_formal_parm_types): Likewise.
8400 (ipa_modify_formal_parameters): Likewise.
8401 (ipa_modify_call_arguments): Likewise.
8402 (ipa_combine_adjustments): Likewise.
8403 (ipa_dump_param_adjustments): Likewise.
8404 (ipa_modify_expr): Likewise.
8405 (ipa_get_adjustment_candidate): Likewise.
8406 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8407 ipa-param-manipulation.c.
8408 (ipa_get_vector_of_formal_parm_types): Likewise.
8409 (ipa_modify_formal_parameters): Likewise.
8410 (ipa_modify_call_arguments): Likewise.
8411 (ipa_modify_expr): Likewise.
8412 (get_ssa_base_param): Likewise.
8413 (ipa_get_adjustment_candidate): Likewise.
8414 (index_in_adjustments_multiple_times_p): Likewise.
8415 (ipa_combine_adjustments): Likewise.
8416 (ipa_dump_param_adjustments): Likewise.
8417 * tree-sra.c: Also include ipa-param-manipulation.h
8418 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8421 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8422 Alan Hayward <alan.hayward@arm.com>
8423 David Sherwood <david.sherwood@arm.com>
8425 * doc/sourcebuild.texi (vect_masked_store): Document.
8427 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8428 Alan Hayward <alan.hayward@arm.com>
8429 David Sherwood <david.sherwood@arm.com>
8431 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8433 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8434 Alan Hayward <alan.hayward@arm.com>
8435 David Sherwood <david.sherwood@arm.com>
8437 * doc/sourcebuild.texi (vect_variable_length): Document.
8439 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8440 Alan Hayward <alan.hayward@arm.com>
8441 David Sherwood <david.sherwood@arm.com>
8443 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8445 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8446 Alan Hayward <alan.hayward@arm.com>
8447 David Sherwood <david.sherwood@arm.com>
8449 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8451 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8452 Alan Hayward <alan.hayward@arm.com>
8453 David Sherwood <david.sherwood@arm.com>
8455 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8456 previously undocumented selectors.
8457 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8459 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8461 * doc/rtl.texi (const_vector): Say that elements can be
8462 const_wide_ints too.
8463 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8464 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8465 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8466 * optabs.c (expand_vector_broadcast): Likewise.
8468 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8469 Alan Hayward <alan.hayward@arm.com>
8470 David Sherwood <david.sherwood@arm.com>
8472 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8473 scaled index even if the unscaled address was invalid.
8474 Don't increase the complexity of using a scale in that case.
8476 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8477 Alan Hayward <alan.hayward@arm.com>
8478 David Sherwood <david.sherwood@arm.com>
8480 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8481 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8482 UNITS_PER_WORD bytes.
8483 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8484 based on REGMODE_NATURAL_SIZE of the inner mode.
8485 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8486 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8487 * expmed.c (lowpart_bit_field_p): Divide the value up into
8488 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8489 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8490 whether something is likely to occupy more than one register.
8492 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8495 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8496 profile_count::adjust_for_ipa_scaling.
8497 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8498 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8500 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8502 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8504 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8505 (split_group): Add total_orig and total_new arguments, estimate the
8506 number of statements related to the store group without store merging
8507 and with store merging.
8508 (imm_store_chain_info::output_merged_store): Adjust split_group
8509 callers, punt if estimated number of statements with store merging
8510 is not smaller than estimated number of statements without it.
8512 (handled_load): Remove has_single_use checks.
8513 (pass_store_merging::process_store): Likewise.
8515 2017-11-09 Richard Biener <rguenther@suse.de>
8517 PR tree-optimization/82902
8518 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8520 2017-11-09 Martin Liska <mliska@suse.cz>
8523 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8526 2017-11-09 Martin Liska <mliska@suse.cz>
8528 PR tree-optimization/82669
8529 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8531 2017-11-09 Martin Liska <mliska@suse.cz>
8533 PR gcov-profile/48463
8534 * coverage.c (coverage_begin_function): Output also end locus
8535 of a function and information whether the function is
8537 * gcov-dump.c (tag_function): Parse and print the information.
8538 * gcov.c (INCLUDE_MAP): Add include.
8539 (INCLUDE_SET): Likewise.
8540 (struct line_info): Move earlier in the source file because
8541 of vector<line_info> in function_info structure.
8542 (line_info::line_info): Likewise.
8543 (line_info::has_block): Likewise.
8544 (struct source_info): Add new member index.
8545 (source_info::get_functions_at_location): New function.
8546 (function_info::group_line_p): New function.
8547 (output_intermediate_line): New function.
8548 (output_intermediate_file): Use the mentioned function.
8549 (struct function_start): New.
8550 (struct function_start_pair_hash): Likewise.
8551 (process_file): Add code that identifies group functions.
8552 Assign lines either to global or function scope.
8553 (generate_results): Skip artificial functions.
8554 (find_source): Assign index for each source file.
8555 (read_graph_file): Read new flag artificial and end_line.
8556 (add_line_counts): Assign it either to global of function scope.
8557 (accumulate_line_counts): Isolate core of the function to
8558 accumulate_line_info and call it for both function and global
8560 (accumulate_line_info): New function.
8561 (output_line_beginning): Fix GNU coding style.
8562 (print_source_line): New function.
8563 (output_line_details): Likewise.
8564 (output_function_details): Likewise.
8565 (output_lines): Iterate both source (global) scope and function
8567 (struct function_line_start_cmp): New class.
8568 * doc/gcov.texi: Reflect changes in documentation.
8570 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8573 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8574 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8575 and similarly for not instead of neg.
8577 2017-11-08 Andi Kleen <ak@linux.intel.com>
8579 * config/i386/i386.opt: Add -mforce-indirect-call.
8580 * config/i386/predicates.md: Check for flag_force_indirect_call.
8581 * doc/invoke.texi: Document -mforce-indirect-call
8583 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8585 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8587 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8588 config/riscv/riscv.c (predict.h): New include.
8589 (riscv_slow_unaligned_access_p): No longer static.
8590 (riscv_block_move_straight): Add require.
8591 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8593 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8596 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8597 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8598 nonimmediate_operand predicate for operand 1 instead of
8601 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8603 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8605 * config/aarch64/constraints.md (Uml): New constraint.
8606 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8609 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8611 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8612 of two vec_duplicates into a vec_concat.
8614 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8616 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8617 Simplify vec_merge of vec_duplicate and vec_concat.
8618 * config/aarch64/constraints.md (Utq): New constraint.
8619 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8622 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8624 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8625 Simplify vec_merge of vec_duplicate and const_vector.
8626 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8628 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8629 mode iterator. Update predicate on operand 1 to
8630 handle non-const_vec constants. Delete constraints.
8631 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8633 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8635 PR tree-optimization/78821
8636 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8638 (store_operand_info::store_operand_info): Initialize it to false.
8639 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8640 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8641 store in the group, and if chain_info is non-NULL, to ignore altogether
8643 (compatible_load_p): Fail if bit_not_p does not match.
8644 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8645 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8646 (pass_store_merging::process_store): Adjust
8647 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8648 call terminate_all_aliasing_chains newly when adding a store into
8649 a chain with non-NULL chain_info.
8651 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8653 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8655 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8657 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8659 (aarch64_layout_frame): Initialise emit_frame_chain.
8660 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8661 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8663 2017-11-08 Martin Liska <mliska@suse.cz>
8665 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8666 of gimple_call_internal_p.
8668 2017-11-07 Tom de Vries <tom@codesourcery.com>
8670 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8672 2017-11-07 Tom de Vries <tom@codesourcery.com>
8674 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8677 2017-11-08 Martin Liska <mliska@suse.cz>
8680 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8682 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8684 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8686 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8689 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8690 (*cmp<mode>_ccz_1): New insn with $k alternative.
8693 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8694 RTX_COMM_COMPARE as commutative as well.
8695 (ix86_binary_operator_ok): Formatting fix.
8696 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8697 *<code><mode>3<mask_name><round_saeonly_name>,
8698 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8699 *<s>mul<mode>3_highpart<mask_name>,
8700 *vec_widen_umult_even_v16si<mask_name>,
8701 *vec_widen_umult_even_v8si<mask_name>,
8702 *vec_widen_umult_even_v4si<mask_name>,
8703 *vec_widen_smult_even_v16si<mask_name>,
8704 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8705 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8706 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8707 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8708 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8709 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8710 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8711 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8712 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8713 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8714 ix86_binary_operator_ok. Formatting fixes.
8715 (*<plusminus_insn><mode>3<mask_name><round_name>,
8716 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8719 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8721 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8722 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8723 eq and ne if TARGET_ISEL.
8724 (cmp): New code_iterator.
8725 (UNS, UNSU_, UNSIK): New code_attrs.
8726 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8727 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8729 ("eq<mode>3"): ... this.
8730 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8732 ("ne<mode>3"): ... this.
8734 2017-11-07 Julia Koval <julia.koval@intel.com>
8737 * common/config/i386/i386-common.c
8738 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8739 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8740 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8741 * config/i386/i386.opt: Ditto.
8742 * config/i386/i386.c (ix86_target_string): Ditto.
8743 (ix86_option_override_internal): Ditto.
8744 (ix86_init_mpx_builtins): Move MPX to args2.
8745 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8746 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8747 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8748 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8749 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8750 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8751 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8753 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8756 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8757 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8758 (zero-extendsidi peephole2): Remove peephole.
8760 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8763 * stor-layout.c: Include attribs.h.
8764 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8765 explicit lookup of "aligned" attribute.
8767 2017-11-07 Andrew Waterman <andrew@sifive.com>
8769 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8770 (riscv_expand_block_move): Likewise.
8771 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8773 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8774 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8775 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8776 (riscv_adjust_block_mem): Likewise.
8777 (riscv_block_move_loop): Likewise.
8778 (riscv_expand_block_move): Likewise.
8779 * config/riscv/riscv.md (movmemsi): New pattern.
8781 2017-11-07 Michael Clark <michaeljclark@mac.com>
8783 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8784 (MUSL_DYNAMIC_LINKER): Likewise.
8786 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8788 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8789 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8792 2017-11-07 Richard Biener <rguenther@suse.de>
8794 * match.pd: Fix build.
8796 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8797 Jackson Woodruff <jackson.woodruff@arm.com>
8799 PR tree-optimization/71026
8800 * match.pd: Canonicalize negate in division.
8802 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8805 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8807 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8809 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8810 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8812 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8814 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8815 non-scalar integral types.
8816 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8817 (-(A-B), -(~A)): New transformations.
8819 2017-11-07 Tom de Vries <tom@codesourcery.com>
8821 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8822 semicolon after "do {} while (0)".
8823 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8824 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8825 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8826 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8827 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8829 2017-11-07 Tom de Vries <tom@codesourcery.com>
8831 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8832 after "do {} while (0)".
8833 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8834 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8835 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8836 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8837 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8839 2017-11-07 Tom de Vries <tom@codesourcery.com>
8842 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8844 (arm_rtx_costs_internal): Add missing semicolon after
8845 HANDLE_NARROW_SHIFT_ARITH call.
8847 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8849 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8850 disable isel if it was not set explicitly.
8852 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8854 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8855 (add_type_attribute) likewise.
8857 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8859 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8860 of struct ix86_frame.
8861 (ix86_initial_elimination_offset): Likewise.
8862 (ix86_expand_split_stack_prologue): Likewise.
8864 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8866 * tree-vrp.h (enum value_range_type): Update stale comment.
8868 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8869 Alan Hayward <alan.hayward@arm.com>
8870 David Sherwood <david.sherwood@arm.com>
8872 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8873 (aarch64_expand_vec_perm_const): Take the number of units too.
8874 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8875 (aarch64_expand_vec_perm_const): Likewise.
8876 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8877 (vec_perm<mode>): Update accordingly.
8879 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8880 Alan Hayward <alan.hayward@arm.com>
8881 David Sherwood <david.sherwood@arm.com>
8883 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8884 Take the number of units too.
8885 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8886 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8887 but check for a vector mode before rather than after the call.
8888 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8889 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8890 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8891 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8892 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8893 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8894 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8895 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8896 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8897 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8898 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8899 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8900 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8901 (aarch64_sqdmull2_n<mode>): Update accordingly.
8903 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8904 Alan Hayward <alan.hayward@arm.com>
8905 David Sherwood <david.sherwood@arm.com>
8907 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8908 the number of units too.
8909 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8910 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8911 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8912 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8913 (vec_store_lanesxi<mode>): Update accordingly.
8915 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8916 Alan Hayward <alan.hayward@arm.com>
8917 David Sherwood <david.sherwood@arm.com>
8919 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8920 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8921 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8922 of units rather than the mode.
8923 * config/aarch64/iterators.md (nunits): New mode attribute.
8924 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8925 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8926 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8927 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8928 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8929 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8930 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8931 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8932 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8933 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8934 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8935 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8936 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8937 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8938 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8939 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8940 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8941 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8942 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8943 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8944 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8945 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8946 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8947 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8948 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8949 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8950 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8951 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8952 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8953 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8954 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8955 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8956 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8957 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8958 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8959 (aarch64_simd_vec_setv2di): Likewise.
8961 2017-11-06 Carl Love <cel@us.ibm.com>
8963 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8965 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8966 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8968 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8970 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8971 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8972 Add power 8 macro expansions.
8973 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8974 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8975 power 8 instructions. (VSX_XXBR): Add iterator.
8977 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8979 * config/arm/arm.md (predicable_short_it): Change default to "no",
8980 improve documentation, remove uses that are identical to the default.
8981 (enabled_for_depr_it): Rename to enabled_for_short_it.
8982 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8983 Remove default uses.
8984 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8985 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8986 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8987 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8989 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8992 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8993 float128 helper macros, which are no longer used after deleting
8994 the old 'q' built-in functions, and moving the round to odd
8995 built-in functions to being special built-in functions.
8996 (BU_FLOAT128_2): Likewise.
8997 (BU_FLOAT128_1_HW): Likewise.
8998 (BU_FLOAT128_2_HW): Likewise.
8999 (BU_FLOAT128_3_HW): Likewise.
9000 (FABSQ): Delete old 'q' built-in functions.
9001 (COPYSIGNQ): Likewise.
9002 (SQRTF128_ODD): Move round to odd built-in functions to be
9003 special built-in functions, so that we can handle
9004 -mabi=ieeelongdouble.
9005 (TRUNCF128_ODD): Likewise.
9006 (ADDF128_ODD): Likewise.
9007 (SUBF128_ODD): Likewise.
9008 (MULF128_ODD): Likewise.
9009 (DIVF128_ODD): Likewise.
9010 (FMAF128_ODD): Likewise.
9011 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
9012 built-in names to 'f128'.
9013 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
9014 old 'q' built-in functions, as the machine independent code for
9015 'f128' built-in functions handles this.
9016 (rs6000_expand_builtin): Add expansion for float128 round to odd
9017 functions, keying off on -mabi=ieeelongdouble of whether to use
9018 the KFmode or TFmode variant.
9019 (rs6000_init_builtins): Initialize the _Float128 round to odd
9021 * doc/extend.texi (PowerPC Built-in Functions): Document the old
9022 _Float128 'q' built-in functions are now mapped into the new
9023 'f128' built-in functions.
9025 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
9027 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
9028 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
9030 2017-11-06 David Malcolm <dmalcolm@redhat.com>
9033 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
9034 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
9035 * toplev.c: Include "ipa-fnsummary.h".
9036 (toplev::finalize): Call ipa_fnsummary_c_finalize.
9038 2017-11-06 Jakub Jelinek <jakub@redhat.com>
9040 PR tree-optimization/82838
9041 * gimple-ssa-store-merging.c
9042 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
9043 on a separate gimple_seq which is then appended to seq.
9045 2017-11-06 Jeff Law <law@redhat.com>
9048 * config/i386/i386.c (PROBE_INTERVAL): Remove.
9049 (get_probe_interval): New functions.
9050 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
9051 (ix86_adjust_stack_and_probe): Likewise.
9052 (output_adjust_stack_and_probe): Likewise.
9053 (ix86_emit_probe_stack_range): Likewise.
9054 (ix86_expand_prologue): Likewise.
9056 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9058 PR tree-optimization/82816
9059 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
9060 if the modes of the two types are the same.
9061 (convert_plusminus_to_widen): Likewise.
9063 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9065 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
9067 (usadv16qi): New define_expand.
9068 (usadv8hi): New define_expand.
9070 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
9073 * ipa-inline-transform.c (update_noncloned_frequencies): Always
9075 (inline_transform): Likewise.
9076 * predict.c (counts_to_freqs): Remove useless conditional.
9077 * profile-count.h (profile_count::apply_scale): Move sanity check.
9078 * tree-inline.c (copy_bb): Always scale.
9079 (copy_cfg_body): Likewise.
9081 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
9084 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
9086 (*cmp_ite0): Add enabled_for_depr_it attribute.
9087 (*cmp_ite1): Likewise.
9089 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
9091 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
9094 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9096 * tree-vrp.c (vrp_int_const_binop): Return true on success and
9097 return the value by pointer.
9098 (extract_range_from_multiplicative_op_1): Update accordingly.
9099 Return as soon as an operation fails.
9101 2017-11-05 Tom de Vries <tom@codesourcery.com>
9104 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
9105 (DEF_SANITIZER_BUILTIN): ... here.
9106 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
9107 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
9109 2017-11-05 Tom de Vries <tom@codesourcery.com>
9112 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
9114 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
9115 ASM_OUTPUT_BEFORE_CASE_LABEL call.
9116 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
9118 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9119 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9120 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9122 2017-11-05 Tom de Vries <tom@codesourcery.com>
9125 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
9128 2017-11-04 Michael Clark <michaeljclark@mac.com>
9130 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
9131 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
9133 (*addsi3_extended): Likewise.
9134 (*addsi3_extended2): Likewise.
9135 (<optab>si3): Likewise.
9136 (<optab>di3): Likewise.
9137 (<optab><mode>3): Likewise.
9138 (<*optabe>si3_internal): Likewise.
9139 (zero_extendqi<SUPERQI:mode>2): Likewise.
9140 (*add<mode>hi3): Likewise.
9141 (*xor<mode>hi3): Likewise.
9142 (<optab>di3): Likewise.
9143 (*<optab>si3_extend): Likewise.
9144 (*sge<u>_<X:mode><GPR:mode>): Likewise.
9145 (*slt<u>_<X:mode><GPR:mode>): Likewise.
9146 (*sle<u>_<X:mode><GPR:mode>): Likewise.
9148 2017-11-04 Andrew Waterman <andrew@sifive.com>
9150 * config/riscv/riscv.c (riscv_option_override): Conditionally set
9151 TARGET_STRICT_ALIGN based upon -mtune argument.
9153 2017-11-04 Andrew Waterman <andrew@sifive.com>
9155 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
9157 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
9159 * config/i386/i386.c (choose_basereg): Use optional scratch
9160 register and add assertion.
9161 (x86_emit_outlined_ms2sysv_save): Use scratch register when
9162 needed, and don't allocate stack.
9163 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
9164 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
9165 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
9167 2017-11-03 Jeff Law <law@redhat.com>
9169 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
9170 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
9171 to probe at the start of a noreturn function.
9173 2017-11-03 Jakub Jelinek <jakub@redhat.com>
9175 PR tree-optimization/78821
9176 * gimple-ssa-store-merging.c: Update the file comment.
9177 (MAX_STORE_ALIAS_CHECKS): Define.
9178 (struct store_operand_info): New type.
9179 (store_operand_info::store_operand_info): New constructor.
9180 (struct store_immediate_info): Add rhs_code and ops data members.
9181 (store_immediate_info::store_immediate_info): Add rhscode, op0r
9182 and op1r arguments to the ctor, initialize corresponding data members.
9183 (struct merged_store_group): Add load_align_base and load_align
9185 (merged_store_group::merged_store_group): Initialize them.
9186 (merged_store_group::do_merge): Update them.
9187 (merged_store_group::apply_stores): Pick the constant for
9188 encode_tree_to_bitpos from one of the two operands, or skip
9189 encode_tree_to_bitpos if neither operand is a constant.
9190 (class pass_store_merging): Add process_store method decl. Remove
9191 bool argument from terminate_all_aliasing_chains method decl.
9192 (pass_store_merging::terminate_all_aliasing_chains): Remove
9193 var_offset_p argument and corresponding handling.
9194 (stmts_may_clobber_ref_p): New function.
9195 (compatible_load_p): New function.
9196 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
9197 if there is overlap and rhs_code is not INTEGER_CST. For
9198 non-overlapping stores terminate group if rhs is not mergeable.
9199 (get_alias_type_for_stmts): Change first argument from
9200 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
9201 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
9202 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9204 (get_location_for_stmts): Change first argument from
9205 auto_vec<gimple *> & to vec<gimple *> &.
9206 (struct split_store): Remove orig_stmts data member, add orig_stores.
9207 (split_store::split_store): Create orig_stores rather than orig_stmts.
9208 (find_constituent_stmts): Renamed to ...
9209 (find_constituent_stores): ... this. Change second argument from
9210 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9211 to info structures rather than the statements.
9212 (split_group): Rename ALLOW_UNALIGNED argument to
9213 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9214 it. Adjust find_constituent_stores caller.
9215 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9216 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9217 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9218 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9219 (mem_valid_for_store_merging): New function.
9220 (handled_load): New function.
9221 (pass_store_merging::process_store): New method.
9222 (pass_store_merging::execute): Use process_store method. Adjust
9223 terminate_all_aliasing_chains caller.
9225 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9227 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9228 Return true for more constants, symbols and label references.
9229 (aarch64_valid_floating_const): Remove unused function.
9231 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9234 * config/aarch64/aarch64.c (aarch64_layout_frame):
9235 Undo forcing of LR at bottom of frame.
9237 2017-11-03 Jeff Law <law@redhat.com>
9240 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9241 for int_registers_saved.
9243 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9244 extracted from tree-ssa-dom.c.
9245 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9246 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9247 (record_equivalences_from_incoming_edge): Add additional argument
9248 to single_pred_edge_ignoring_loop_edges call.
9249 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9250 (uncprop_dom_walker::before_dom_children): Add additional argument
9251 to single_pred_edge_ignoring_loop_edges call.
9252 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9253 single_pred_edge_ignoring_loop_edges rather than open coding.
9254 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9256 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9258 * match.pd (-(-A)): Rewrite.
9260 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9262 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9263 (rs6000_emit_int_cmove): New declaration.
9264 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9265 (rs6000_emit_sISEL): Delete.
9266 (rs6000_emit_int_cmove): Make non-static.
9267 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9268 instead of rs6000_emit_sISEL.
9270 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9272 * asan.c (create_cond_insert_point): Maintain profile.
9273 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9275 * basic-block.h (struct basic_block_def): Remove frequency.
9276 (EDGE_FREQUENCY): Use to_frequency
9277 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9279 (find_traces): Update to use to_frequency.
9280 (find_traces_1_round): Likewise; use only IPA counts.
9281 (bb_to_key): Likewise.
9282 (connect_traces): Use IPA counts only.
9283 (copy_bb_p): Update to use to_frequency.
9284 (fix_up_crossing_landing_pad): Likewise.
9285 (sanitize_hot_paths): Likewise.
9286 * bt-load.c (basic_block_freq): Likewise.
9287 * cfg.c (init_flow): Set count_max to uninitialized.
9288 (check_bb_profile): Remove frequencies; check counts.
9289 (dump_bb_info): Do not dump frequencies.
9290 (update_bb_profile_for_threading): Update counts only.
9291 (scale_bbs_frequencies_int): Likewise.
9292 (MAX_SAFE_MULTIPLIER): Remove.
9293 (scale_bbs_frequencies_gcov_type): Update counts only.
9294 (scale_bbs_frequencies_profile_count): Update counts only.
9295 (scale_bbs_frequencies): Update counts only.
9296 * cfg.h (struct control_flow_graph): Add count-max.
9297 (update_bb_profile_for_threading): Update prototype.
9298 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9299 (find_many_sub_basic_blocks): Likewise.
9300 * cfgcleanup.c (try_forward_edges): Likewise.
9301 (try_crossjump_to_edge): Likewise.
9302 * cfgexpand.c (expand_gimple_cond): Likewise.
9303 (expand_gimple_tailcall): Likewise.
9304 (construct_init_block): Likewise.
9305 (construct_exit_block): Likewise.
9306 * cfghooks.c (verify_flow_info): Check consistency of counts.
9307 (dump_bb_for_graph): Do not dump frequencies.
9308 (split_block_1): Do not update frequencies.
9309 (split_edge): Do not update frequencies.
9310 (make_forwarder_block): Do not update frequencies.
9311 (duplicate_block): Do not update frequencies.
9312 (account_profile_record): Do not update frequencies.
9313 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9314 for global heuristics.
9315 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9316 (expected_loop_iterations_unbounded): Use counts only.
9317 * cfgloopmanip.c (scale_loop_profile): Simplify.
9318 (create_empty_loop_on_edge): Simplify
9320 (duplicate_loop_to_header_edge): Simplify
9321 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9322 (update_br_prob_note): Take care of removing note when profile
9324 (relink_block_chain): Do not dump frequency.
9325 (rtl_account_profile_record): Use to_frequency.
9326 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9327 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9328 (cgraph_update_edges_for_call_stmt_node): Likewise.
9329 (cgraph_edge::verify_count_and_frequency): Update.
9330 (cgraph_node::verify_node): Temporarily disable frequency verification.
9331 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9332 to_cgraph_frequency.
9333 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9334 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9336 (cgraph_node::expand_thunk): Update profile.
9337 * except.c (dw2_build_landing_pads): Do not update frequency.
9338 * final.c (compute_alignments): Use to_frequency.
9339 (dump_basic_block_info): Do not dump frequency.
9340 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9341 (dump_gimple_bb_header): Do not dump frequency.
9342 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9344 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9345 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9346 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9347 (init_before_recovery): Do not update frequency.
9348 (sched_create_recovery_edges): Do not update frequency.
9349 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9350 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9351 (ipa_cp_c_finalize): Set max_count to uninitialized.
9352 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9353 (param_change_prob): Use counts.
9354 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9356 * ipa-split.c (consider_split): Use to_frequency.
9357 (split_function): Use to_frequency.
9358 * ira-build.c (loop_compare_func): Likewise.
9359 (mark_loops_for_removal): Likewise.
9360 (mark_all_loops_for_removal): Likewise.
9361 * loop-doloop.c (doloop_modify): Do not update frequency.
9362 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9364 * lto-streamer-in.c (input_function): Update count_max.
9365 * omp-expand.c (expand_omp_taskreg): Update count_max.
9366 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9367 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9368 (maybe_hot_count_p): Use ipa counts only.
9369 (maybe_hot_bb_p): Simplify.
9370 (maybe_hot_edge_p): Simplify.
9371 (probably_never_executed): Do not take frequency argument.
9372 (probably_never_executed_bb_p): Do not pass frequency.
9373 (probably_never_executed_edge_p): Likewise.
9374 (combine_predictions_for_bb): Check that profile is nonzero.
9375 (propagate_freq): Do not set frequency.
9376 (drop_profile): Simplify.
9377 (counts_to_freqs): Simplify.
9378 (expensive_function_p): Use to_frequency.
9379 (propagate_unlikely_bbs_forward): Simplify.
9380 (determine_unlikely_bbs): Simplify.
9381 (estimate_bb_frequencies): Add hack to silence graphite issues.
9382 (compute_function_frequency): Use ipa counts.
9383 (pass_profile::execute): Update.
9384 (rebuild_frequencies): Use counts only.
9385 (force_edge_cold): Use counts only.
9386 * profile-count.c (profile_count::dump): Dump new count types.
9387 (profile_count::differs_from_p): Check compatiblity.
9388 (profile_count::to_frequency): New function.
9389 (profile_count::to_cgraph_frequency): New function.
9390 * profile-count.h (struct function): Declare.
9391 (enum profile_quality): Add profile_guessed_local and
9392 profile_guessed_global0.
9393 (class profile_proability): Decrease number of bits to 29;
9394 update from_reg_br_prob_note and to_reg_br_prob_note.
9395 (class profile_count: Update comment; decrease number of bits
9396 to 61. Check compatibility.
9397 (profile_count::compatible_p): New private member function.
9398 (profile_count::ipa_p): New member function.
9399 (profile_count::operator<): Handle global zero correctly.
9400 (profile_count::operator>): Handle global zero correctly.
9401 (profile_count::operator<=): Handle global zero correctly.
9402 (profile_count::operator>=): Handle global zero correctly.
9403 (profile_count::nonzero_p): New member function.
9404 (profile_count::force_nonzero): New member function.
9405 (profile_count::max): New member function.
9406 (profile_count::apply_scale): Handle IPA scalling.
9407 (profile_count::guessed_local): New member function.
9408 (profile_count::global0): New member function.
9409 (profile_count::ipa): New member function.
9410 (profile_count::to_frequency): Declare.
9411 (profile_count::to_cgraph_frequency): Declare.
9412 * profile.c (OVERLAP_BASE): Delete.
9413 (compute_frequency_overlap): Delete.
9414 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9415 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9416 * sched-ebb.c (rank): Use counts only.
9417 * shrink-wrap.c (handle_simple_exit): Use counts only.
9418 (try_shrink_wrapping): Use counts only.
9419 (place_prologue_for_one_component): Use counts only.
9420 * tracer.c (find_best_predecessor): Use to_frequency.
9421 (find_trace): Use to_frequency.
9422 (tail_duplicate): Use to_frequency.
9423 * trans-mem.c (expand_transaction): Do not update frequency.
9424 * tree-call-cdce.c: Do not update frequency.
9425 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9426 (gimple_merge_blocks): Likewise.
9427 (gimple_split_edge): Likewise.
9428 (gimple_duplicate_sese_region): Likewise.
9429 (gimple_duplicate_sese_tail): Likewise.
9430 (move_sese_region_to_fn): Likewise.
9431 (gimple_account_profile_record): Likewise.
9432 (insert_cond_bb): Likewise.
9433 * tree-complex.c (expand_complex_div_wide): Likewise.
9434 * tree-eh.c (lower_resx): Update profile.
9435 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9437 (initialize_cfun): Do not initialize frequencies
9438 (freqs_to_counts): Delete.
9439 (copy_cfg_body): Ignore count parameter.
9440 (copy_body): Update.
9441 (expand_call_inline): Update count_max.
9442 (optimize_inline_calls): Update count_max.
9443 (tree_function_versioning): Update count_max.
9444 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9445 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9447 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9448 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9449 (try_peel_loop): Likewise.
9450 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9452 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9453 (tree_transform_and_unroll_loop): Do not use frequencies
9454 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9455 Use reliable prediction only.
9456 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9457 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9458 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9459 probability scaling.
9460 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9461 not update frequency
9462 (any_remaining_duplicated_blocks): Likewise.
9463 (update_profile): Likewise.
9464 (estimated_freqs_path): Delete.
9465 (freqs_to_counts_path): Delete.
9466 (clear_counts_path): Delete.
9467 (ssa_fix_duplicate_block_edges): Likewise.
9468 (duplicate_thread_path): Likewise.
9469 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9470 * tree-tailcall.c (decrease_profile): Do not update frequency.
9471 (eliminate_tail_call): Likewise.
9472 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9473 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9474 (optimize_mask_stores): Likewise.
9475 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9476 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9477 (ubsan_expand_ptr_ifn): Update profile.
9478 * value-prof.c (gimple_ic): Simplify.
9479 * value-prof.h (gimple_ic): Update prototype.
9480 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9481 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9483 (want_inline_self_recursive_call_p): Likewise.
9484 (resolve_noninline_speculation): Only cummulate defined counts.
9485 (inline_small_functions): Use nonzero_p.
9486 (ipa_inline): Do not access freed node.
9488 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9490 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9491 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9493 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9495 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9496 non-legitimate address.
9498 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9500 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9501 (*lt0_<mode>di, *lt0_<mode>si): New.
9503 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9505 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9506 TARGET_PAIRED_FLOAT.
9508 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9509 Jim Wilson <jim.wilson@linaro.org>
9511 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9512 * config/aarch64/aarch64-tune.md: Regenerated.
9513 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9514 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9516 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9518 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9519 (arc_expand_prologue): Restore blink for millicode.
9520 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9522 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9525 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9526 gen_vec_duplicate after forcing the scalar into a register.
9528 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9530 * combine (try_combine): Print the insns input to try_combine to the
9533 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9536 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9537 Remove second argument from aarch64_process_target_attr call.
9538 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9540 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9542 (aarch64_handle_attr_arch): Remove second argument.
9543 (aarch64_handle_attr_cpu): Ditto.
9544 (aarch64_handle_attr_tune): Ditto.
9545 (aarch64_handle_attr_isa_flags): Ditto.
9546 (aarch64_process_one_target_attr): Ditto.
9547 (aarch64_process_target_attr): Ditto.
9548 (aarch64_option_valid_attribute_p): Remove second argument.
9549 on aarch64_process_target_attr call.
9551 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9553 * diagnostic.c: Include "selftest-diagnostic.h".
9554 (selftest::assert_location_text): New function.
9555 (selftest::test_diagnostic_get_location_text): New function.
9556 (selftest::diagnostic_c_tests): Call it.
9558 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9560 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9561 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9562 (class selftest::test_diagnostic_context): Move to...
9563 * selftest-diagnostic.c: New file.
9564 * selftest-diagnostic.h: New file.
9566 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9568 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9569 offset range for FT32B.
9570 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9571 * config/ft32/ft32.md: Add TARGET_NOPM.
9572 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9573 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9575 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9577 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9579 2017-11-02 Jeff Law <law@redhat.com>
9581 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9582 virtual keyword on FINAL OVERRIDE members.
9584 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9585 virtuals together. Add virtual destructor.
9586 (substitute_and_fold_engine): Similarly.
9588 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9590 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9592 2017-11-02 Richard Biener <rguenther@suse.de>
9594 PR tree-optimization/82795
9595 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9597 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9599 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9601 (gcc_GAS_CHECK_FEATURE): Remove.
9602 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9603 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9604 * configure: Regenerate.
9606 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9608 * config/arc/arc.c (hwloop_optimize): Account for empty
9611 2017-11-02 Richard Biener <rguenther@suse.de>
9614 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9615 Truncate ARRAY_REF index and element size.
9617 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9619 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9621 2017-11-01 Jeff Law <law@redhat.com>
9623 * tree-ssa-ccp.c (ccp_folder): New class derived from
9624 substitute_and_fold_engine.
9625 (ccp_folder::get_value): New member function.
9626 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9627 (ccp_fold_stmt): Remove prototype.
9628 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9629 * tree-ssa-copy.c (copy_folder): New class derived from
9630 substitute_and_fold_engine.
9631 (copy_folder::get_value): Renamed from get_value.
9632 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9633 * tree-vrp.c (vrp_folder): New class derived from
9634 substitute_and_fold_engine.
9635 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9636 (vrp_folder::get_value): New member function.
9637 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9638 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9639 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9640 provide a class interface to folder/substitute routines.
9641 (ssa_prop_fold_stmt_fn): Remove typedef.
9642 (ssa_prop_get_value_fn): Likewise.
9643 (subsitute_and_fold): Remove prototype.
9644 (replace_uses_in): Likewise.
9645 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9646 Renamed from replace_uses_in. Call the virtual member function
9647 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9648 (substitute_and_fold_dom_walker): Remove initialization of
9649 data member entries for calbacks. Add substitute_and_fold_engine
9650 member and initialize it.
9651 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9652 member functions for get_value, replace_phi_args_in c
9653 replace_uses_in, and fold_stmt calls.
9654 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9655 substitute_and_fold. Remove assert. Update ctor call.
9657 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9658 (ssa_prop_visit_phi_fn): Likewise.
9659 (class ssa_propagation_engine): New class to provide an interface
9661 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9663 (ssa_prop_visit_phi): Likewise.
9664 (ssa_propagation_engine::simulate_stmt): Moved into class.
9665 Call visit_phi/visit_stmt from the class rather than via
9666 file scoped static variables.
9667 (ssa_propagation_engine::simulate_block): Moved into class.
9668 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9669 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9670 set file scoped statics for the visit_stmt/visit_phi callbacks.
9671 * tree-complex.c (complex_propagate): New class derived from
9672 ssa_propagation_engine.
9673 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9674 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9675 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9677 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9678 ssa_propagation_engine.
9679 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9680 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9681 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9682 * tree-ssa-copy.c (copy_prop): New class derived from
9683 ssa_propagation_engine.
9684 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9685 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9686 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9687 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9688 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9689 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9690 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9692 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9694 PR rtl-optimization/82778
9695 PR rtl-optimization/82597
9696 * compare-elim.c (struct comparison): Add in_a_setter field.
9697 (find_comparison_dom_walker::before_dom_children): Remove killed
9698 bitmap and df_simulate_find_defs call, instead walk the defs.
9699 Compute last_setter and initialize in_a_setter. Merge definitions
9700 with first initialization for a few variables.
9701 (try_validate_parallel): Use insn_invalid_p instead of
9702 recog_memoized. Return insn rather than just the pattern.
9703 (try_merge_compare): Fix up comment. Don't uselessly test if
9704 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9706 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9709 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9710 Alan Hayward <alan.hayward@arm.com>
9711 David Sherwood <david.sherwood@arm.com>
9713 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9714 aarch64_hard_regno_nregs to get the number of registers
9717 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9718 Alan Hayward <alan.hayward@arm.com>
9719 David Sherwood <david.sherwood@arm.com>
9721 * config/aarch64/constraints.md (Upl): Rename to...
9723 * config/aarch64/aarch64.md
9724 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9727 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9728 Alan Hayward <alan.hayward@arm.com>
9729 David Sherwood <david.sherwood@arm.com>
9731 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9732 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9735 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9736 Alan Hayward <alan.hayward@arm.com>
9737 David Sherwood <david.sherwood@arm.com>
9739 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9740 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9741 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9743 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9744 of a single element.
9745 * config/aarch64/iterators.md: Add a comment above the permute
9746 unspecs to say that they are generated directly by
9747 aarch64_expand_vec_perm_const.
9748 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9750 2017-11-01 Nathan Sidwell <nathan@acm.org>
9752 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9754 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9756 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9757 and medany code models, and describe what they do.
9759 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9761 Revert accidental duplicate:
9763 * combine.c (can_change_dest_mode): Reject changes in
9764 REGMODE_NATURAL_SIZE.
9766 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9768 PR rtl-optimization/64682
9769 PR rtl-optimization/69567
9770 PR rtl-optimization/69737
9771 PR rtl-optimization/82683
9772 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9773 register mentioned in the note, drop the note, unless it came from I3,
9774 in which case it should go to I3 again.
9776 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9778 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9779 and return false if not.
9780 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9782 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9784 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9785 range is known to be empty.
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 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9792 and CONST_VECTOR_NUNITS instead of computing the number of units from
9793 the byte sizes of the vector and element.
9794 (simplify_binary_operation_1): Likewise.
9795 (simplify_const_binary_operation): Likewise.
9796 (simplify_ternary_operation): Likewise.
9798 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9799 Alan Hayward <alan.hayward@arm.com>
9800 David Sherwood <david.sherwood@arm.com>
9802 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9803 (int_mem_offset): ...this new function.
9804 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9805 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9806 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9809 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9810 Alan Hayward <alan.hayward@arm.com>
9811 David Sherwood <david.sherwood@arm.com>
9813 * lower-subreg.c (interesting_mode_p): New function.
9814 (compute_costs, find_decomposable_subregs, decompose_register)
9815 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9816 (resolve_clobber, dump_choices): Use it.
9818 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9819 Alan Hayward <alan.hayward@arm.com>
9820 David Sherwood <david.sherwood@arm.com>
9822 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9824 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9825 Alan Hayward <alan.hayward@arm.com>
9826 David Sherwood <david.sherwood@arm.com>
9828 * alias.c (find_base_value, find_base_term): Only process integer
9829 truncations. Check the precision rather than the size.
9831 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9832 Alan Hayward <alan.hayward@arm.com>
9833 David Sherwood <david.sherwood@arm.com>
9835 * machmode.h (is_narrower_int_mode): New function
9836 * optabs.c (expand_float, expand_fix): Use it.
9837 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9839 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9840 Alan Hayward <alan.hayward@arm.com>
9841 David Sherwood <david.sherwood@arm.com>
9843 * rtl.h (narrower_subreg_mode): New function.
9844 * ira-color.c (update_costs_from_allocno): Use it.
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 * optabs-query.h (convert_optab_p): New function, split out from...
9851 (convert_optab_handler): ...here.
9852 (widening_optab_handler): Delete.
9853 (find_widening_optab_handler): Remove permit_non_widening parameter.
9854 (find_widening_optab_handler_and_mode): Likewise. Provide an
9855 override that operates on mode class wrappers.
9856 * optabs-query.c (widening_optab_handler): Delete.
9857 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9858 parameter. Assert that the two modes are the same class and that
9859 the "from" mode is narrower than the "to" mode. Use
9860 convert_optab_handler instead of widening_optab_handler.
9861 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9862 instead of widening_optab_handler.
9863 * expr.c (expand_expr_real_2): Update calls to
9864 find_widening_optab_handler.
9865 * optabs.c (expand_widen_pattern_expr): Likewise.
9866 (expand_binop_directly): Take the insn_code as a parameter.
9867 (expand_binop): Only call find_widening_optab_handler for
9868 conversion optabs; use optab_handler otherwise. Update calls
9869 to find_widening_optab_handler and expand_binop_directly.
9870 Use convert_optab_handler instead of widening_optab_handler.
9871 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9872 find_widening_optab_handler and use scalar_mode rather than
9874 (convert_plusminus_to_widen): Likewise.
9876 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9877 Alan Hayward <alan.hayward@arm.com>
9878 David Sherwood <david.sherwood@arm.com>
9880 * machmode.h (fixed_size_mode): New class.
9881 * rtl.h (get_pool_mode): Return fixed_size_mode.
9882 * gengtype.c (main): Add fixed_size_mode.
9883 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9884 (get_raw_arg_mode): Likewise.
9885 * doc/tm.texi: Regenerate.
9886 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9887 * targhooks.c (default_get_reg_raw_mode): Likewise.
9888 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9889 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9890 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9891 (msp430_get_raw_result_mode): Likewise.
9892 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9893 * dbxout.c (dbxout_parms): Require fixed-size modes.
9894 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9895 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9896 * omp-low.c (lower_oacc_reductions): Likewise.
9897 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9898 (simplify_subreg): Update accordingly.
9899 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9900 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9901 that aren't fixed-size.
9902 (get_pool_mode): Return a fixed_size_mode.
9903 (output_constant_pool_2): Take a fixed_size_mode.
9905 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9906 Alan Hayward <alan.hayward@arm.com>
9907 David Sherwood <david.sherwood@arm.com>
9909 * doc/rtl.texi (vec_series): Document.
9910 (const): Say that the operand can be a vec_series.
9911 * rtl.def (VEC_SERIES): New rtx code.
9912 * rtl.h (const_vec_series_p_1): Declare.
9913 (const_vec_series_p): New function.
9914 * emit-rtl.h (gen_const_vec_series): Declare.
9915 (gen_vec_series): Likewise.
9916 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9917 (gen_vec_series): Likewise.
9918 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9919 * simplify-rtx.c (simplify_unary_operation): Handle negations
9921 (simplify_binary_operation_series): New function.
9922 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9923 (test_vector_ops_series): New function.
9924 (test_vector_ops): Call it.
9925 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9926 gen_const_vec_series.
9927 (altivec_lvsr): Likewise.
9928 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9930 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9931 Alan Hayward <alan.hayward@arm.com>
9932 David Sherwood <david.sherwood@arm.com>
9934 * doc/rtl.texi (const): Update description of address constants.
9935 Say that vector constants are allowed too.
9936 * common.md (E, F): Use CONSTANT_P instead of checking for
9938 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9939 checking for CONST_VECTOR.
9940 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9942 * expr.c (expand_expr_real_2): Check for vector modes instead
9943 of checking for CONST_VECTOR.
9944 * rtl.h (const_vec_p): New function.
9945 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9946 (unwrap_const_vec_duplicate): Handle them here too.
9948 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9949 David Malcolm <dmalcolm@redhat.com>
9950 Alan Hayward <alan.hayward@arm.com>
9951 David Sherwood <david.sherwood@arm.com>
9953 * rtl.h (vec_duplicate_p): New function.
9954 * selftest-rtl.c (assert_rtx_eq_at): New function.
9955 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9956 (assert_rtx_eq_at): Declare.
9957 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9958 * selftest-run-tests.c (selftest::run_tests): Call it.
9959 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9960 (simplify_unary_operation_1): Recursively handle vector duplicates.
9961 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9963 (simplify_subreg): Handle subregs of vector duplicates.
9964 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9965 (selftest::simplify_rtx_c_tests): New functions.
9967 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9968 Alan Hayward <alan.hayward@arm.com>
9969 David Sherwood <david.sherwood@arm.com>
9971 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9972 (gen_vec_duplicate): Likewise.
9973 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9975 (gen_const_vector): ...here.
9976 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9977 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9978 whose elements are all equal.
9979 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9980 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9981 (simplify_relational_operation): Likewise.
9982 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9984 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9985 (aarch64_expand_vector_init): Likewise.
9986 * config/arm/arm.c (neon_vdup_constant): Likewise.
9987 (neon_expand_vector_init): Likewise.
9988 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9989 (arm_block_set_unaligned_vect): Likewise.
9990 (arm_block_set_aligned_vect): Likewise.
9991 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9992 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9993 (expand_vec_perm_even_odd_pack): Likewise.
9994 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9995 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9996 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9997 gen_const_vec_duplicate.
9998 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9999 * config/mips/mips.c (mips_gen_const_int_vector): Use
10000 gen_const_vec_duplicate.
10001 (mips_expand_vector_init): Use CONST0_RTX.
10002 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
10003 (define_split): Use gen_const_vec_duplicate.
10004 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
10005 (define_split): Use gen_const_vec_duplicate.
10006 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
10007 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
10008 * config/spu/spu.c (spu_const): Likewise.
10010 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
10011 Alan Hayward <alan.hayward@arm.com>
10012 David Sherwood <david.sherwood@arm.com>
10014 * combine.c (can_change_dest_mode): Reject changes in
10015 REGMODE_NATURAL_SIZE.
10017 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
10019 * configure.ac (--enable-libssp): New.
10020 (gcc_cv_libc_provides_ssp): Check for explicit setting before
10021 trying to determine target-specific default. Adjust indentation.
10022 * configure: Regenerated.
10023 * doc/install.texi (Configuration): Expand --disable-libssp
10026 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
10028 config/i386/i386.c (ix86_expand_epilogue): Correct stack
10031 2017-10-31 Martin Jambor <mjambor@suse.cz>
10034 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
10036 2017-10-31 David Malcolm <dmalcolm@redhat.com>
10038 * auto-profile.c (autofdo_source_profile::read): Use
10039 UNKNOWN_LOCATION rather than 0.
10040 * diagnostic-core.h (warning_at_rich_loc): Rename to...
10041 (warning_at): ...this overload.
10042 (warning_at_rich_loc_n): Rename to...
10043 (warning_n): ...this overload.
10044 (error_at_rich_loc): Rename to...
10045 (error_at): ...this overload.
10046 (pedwarn_at_rich_loc): Rename to...
10047 (pedwarn): ...this overload.
10048 (permerror_at_rich_loc): Rename to...
10049 (permerror): ...this overload.
10050 (inform_at_rich_loc): Rename to...
10051 (inform): ...this overload.
10052 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
10053 (diagnostic_n_impl_richloc): Rename to...
10054 (diagnostic_n_impl): ...this rich_location *-based decl.
10055 (inform_at_rich_loc): Rename to...
10056 (inform): ...this, and add an assertion.
10057 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
10058 (warning_at_rich_loc): Rename to...
10059 (warning_at): ...this, and add an assertion.
10060 (warning_at_rich_loc_n): Rename to...
10061 (warning_n): ...this, and add an assertion.
10062 (warning_n): Update location_t-based implementation for removal of
10063 location_t-based diagnostic_n_impl.
10064 (pedwarn_at_rich_loc): Rename to...
10065 (pedwarn): ...this, and add an assertion.
10066 (permerror_at_rich_loc): Rename to...
10067 (permerror): ...this, and add an assertion.
10068 (error_n): Update for removal of location_t-based diagnostic_n_impl.
10069 (error_at_rich_loc): Rename to...
10070 (error_at): ...this, and add an assertion.
10071 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
10072 (driver::do_spec_on_infiles): Likewise.
10073 * substring-locations.c (format_warning_va): Update for renaming
10074 of inform_at_rich_loc.
10076 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
10078 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
10079 _Float<N>X built-in functions so that the variant without the
10080 "__builtin_" prefix is only enabled for the GNU C and Objective C
10081 languages when they are in non-strict ANSI/ISO mode.
10082 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
10083 * target.def (floatn_builtin_p): Add a target hook to control
10084 whether _Float<N> and _Float<N>X built-in functions without the
10085 "__builtin_" prefix are enabled, and return true for C and
10086 Objective C in the default hook. Include langhooks.h in
10088 * targhooks.h (default_floatn_builtin_p): Likewise.
10089 * targhooks.c (default_floatn_builtin_p): Likewise.
10090 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
10091 floatn_builtin_p target hook.
10092 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
10094 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
10095 Eric Botcazou <ebotcazou@adacore.com>
10097 PR rtl-optimization/81803
10098 * lra-constraints.c (curr_insn_transform): Also reload the whole
10099 register for a strict subreg no wider than a word if this is for
10100 a WORD_REGISTER_OPERATIONS target.
10102 2017-10-31 Jason Merrill <jason@redhat.com>
10104 * gdbinit.in: Skip over inlines from timevar.h.
10106 2017-10-31 Martin Liska <mliska@suse.cz>
10108 * doc/gcov.texi: Document new option.
10109 * gcov.c (print_usage): Likewise print it.
10110 (process_args): Support the argument.
10111 (format_count): New function.
10112 (format_gcov): Use the function.
10114 2017-10-31 Martin Liska <mliska@suse.cz>
10116 * gcov.c (struct name_map): do not use typedef.
10117 Define operator== and operator<.
10118 (name_search): Remove.
10119 (name_sort): Remove.
10120 (main): Do not allocate names.
10121 (process_file): Add vertical space.
10122 (generate_results): Use std::find.
10123 (release_structures): Do not release memory.
10124 (find_source): Use std::find.
10126 2017-10-31 Martin Liska <mliska@suse.cz>
10128 * gcov.c (struct line_info): Remove it's typedef.
10129 (line_info::line_info): Add proper ctor.
10130 (line_info::has_block): Do not use a typedef.
10131 (struct source_info): Do not use typedef.
10132 (circuit): Likewise.
10133 (get_cycles_count): Likewise.
10134 (output_intermediate_file): Iterate via vector iterator.
10135 (add_line_counts): Use std::vector methods.
10136 (accumulate_line_counts): Likewise.
10137 (output_lines): Likewise.
10139 2017-10-31 Martin Liska <mliska@suse.cz>
10141 * gcov.c (struct source_info): Remove typedef.
10142 (source_info::source_info): Add proper ctor.
10143 (accumulate_line_counts): Use struct, not it's typedef.
10144 (output_gcov_file): Likewise.
10145 (output_lines): Likewise.
10146 (main): Do not allocate an array.
10147 (output_intermediate_file): Use size of vector container.
10148 (process_file): Resize the vector.
10149 (generate_results): Do not preallocate, use newly added vector
10151 (release_structures): Do not release sources.
10152 (find_source): Use vector methods.
10153 (add_line_counts): Do not use typedef.
10155 2017-10-31 Martin Liska <mliska@suse.cz>
10157 * doc/gcov.texi: Document that.
10158 * gcov.c (add_line_counts): Mark lines with a non-executed
10160 (output_line_beginning): Handle such lines.
10161 (output_lines): Pass new argument.
10162 (output_intermediate_file): Print it in intermediate format.
10164 2017-10-31 Martin Liska <mliska@suse.cz>
10166 * color-macros.h: New file.
10167 * diagnostic-color.c: Factor out color related to macros to
10169 * doc/gcov.texi: Document -k option.
10170 * gcov.c (INCLUDE_STRING): Include string.h.
10171 (print_usage): Add -k option.
10172 (process_args): Parse it.
10173 (pad_count_string): New function.
10174 (output_line_beginning): Likewise.
10175 (DEFAULT_LINE_START): New macro.
10176 (output_lines): Support color output.
10178 2017-10-31 Martin Liska <mliska@suse.cz>
10180 PR gcov-profile/82633
10181 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
10182 their interaction with GCOV infrastructure.
10183 * configure.ac: Add -fkeep-{inline,static}-functions to
10185 * configure: Regenerate.
10187 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
10190 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
10192 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
10195 * config/rs6000/rs6000.md (allocate_stack): Force update interval
10196 into a register if it does not fit into an immediate offset field.
10198 2017-10-31 Olivier Hainque <hainque@adacore.com>
10200 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
10202 2017-10-31 Julia Koval <julia.koval@intel.com>
10204 * config.gcc: Add gfniintrin.h.
10205 * config/i386/gfniintrin.h: New.
10206 * config/i386/i386-builtin-types.def
10207 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10208 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10209 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10210 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10211 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10212 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10213 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10214 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10215 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10216 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10217 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10218 * config/i386/immintrin.h: Include gfniintrin.h.
10219 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10221 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10223 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10225 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10227 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10228 (arm_ashldi3_1bit): Remove pattern.
10229 (ashrdi3): Remove shift by 1 expansion.
10230 (arm_ashrdi3_1bit): Remove pattern.
10231 (lshrdi3): Remove shift by 1 expansion.
10232 (arm_lshrdi3_1bit): Remove pattern.
10233 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10234 cost of ashldi3 by 1.
10235 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10236 (<shift>di3_neon): Likewise.
10238 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10240 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10241 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10242 and (*aarch64_simd_mov<VQ:mode>).
10243 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10244 pattern alternative.
10245 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10246 attributes to match pattern alternative.
10248 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10250 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10251 * config/rs6000/emmintrin.h: New file.
10252 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10254 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10256 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10257 * (movdi_vfp_cortexa8): Remove pattern.
10259 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10261 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10263 (Specific, alpha*-dec-osf5.1): Remove.
10264 (Specific, mips-sgi-irix5): Remove.
10265 (Specific, mips-sgi-irix6): Remove.
10267 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10269 PR middle-end/22141
10270 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10271 arguments to clear_bit_region_be.
10273 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10275 * gimplify.c: Include memmodel.h.
10277 2017-10-30 Martin Jambor <mjambor@suse.cz>
10279 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10280 condition whether loop should be executed at all.
10282 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10284 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10285 gimple folding of vec_madd() intrinsics.
10286 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10287 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10288 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10290 2017-10-30 Richard Biener <rguenther@suse.de>
10292 PR tree-optimization/82762
10294 2017-10-23 Richard Biener <rguenther@suse.de>
10296 PR tree-optimization/82129
10298 2017-08-01 Richard Biener <rguenther@suse.de>
10300 PR tree-optimization/81181
10301 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10302 (compute_antic): ... end of iteration here.
10304 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10306 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10307 -std=iso9899:2017 and -std=gnu17.
10308 * doc/standards.texi (C Language): Document C17 support.
10309 * doc/cpp.texi (Overview): Mention -std=c17.
10310 (Standard Predefined Macros): Document C11 and C17 values of
10311 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10312 * doc/extend.texi (Inline): Do not list individual options for
10313 standards newer than C99.
10314 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10316 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10319 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10321 * asan.c (asan_finish_file): Align asan globals array by shadow
10324 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10326 PR middle-end/22141
10327 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10328 (struct store_immediate_info): Add bitregion_start and bitregion_end
10330 (store_immediate_info::store_immediate_info): Add brs and bre
10331 arguments and initialize bitregion_{start,end} from those.
10332 (struct merged_store_group): Add bitregion_start, bitregion_end,
10333 align_base and mask fields. Drop unnecessary struct keyword from
10334 struct store_immediate_info. Add do_merge method.
10335 (clear_bit_region_be): Use memset instead of loop storing zeros.
10336 (merged_store_group::do_merge): New method.
10337 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10338 stores as long as the surrounding bitregions have no gaps.
10339 (merged_store_group::merge_overlapping): Use do_merge.
10340 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10341 is byte aligned, rather than requiring that start and width are
10342 byte aligned. Drop unnecessary struct keyword from
10343 struct store_immediate_info. Allocate and populate also mask array.
10344 Make start of the arrays relative to bitregion_start rather than
10345 start and size them according to bitregion_{end,start} difference.
10346 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10347 struct store_immediate_info.
10348 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10349 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10350 struct keyword from struct store_immediate_info.
10351 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10352 between stores as long as the surrounding bitregions have no gaps.
10354 (struct split_store): Add orig non-static data member.
10355 (split_store::split_store): Initialize orig to false.
10356 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10357 if there is exactly a single original stmt. Change stmts argument
10358 to pointer from reference, if NULL, don't push anything to it. Add
10359 first argument, use it to optimize skipping over orig stmts that
10360 are known to be before bitpos already. Simplify.
10361 (split_group): Return unsigned int count how many stores are or
10362 would be needed rather than a bool. Add allow_unaligned argument.
10363 Change split_stores argument from reference to pointer, if NULL,
10364 only do a dry run computing how many stores would be produced.
10365 Rewritten algorithm to use both alignment and misalign if
10366 !allow_unaligned and handle bitfield stores with gaps.
10367 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10368 from bitregion_start instead of start. Compute allow_unaligned
10369 here, if true, do 2 split_group dry runs to compute which one
10370 produces fewer stores and prefer aligned if equal. Punt if
10371 new count is bigger or equal than original before emitting any
10372 statements, rather than during that. Remove no longer needed
10373 new_ssa_names tracking. Replace num_stmts with
10374 split_stores.length (). Use 32-bit stack allocated entries
10375 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10376 if possible. Handle bitfields with gaps.
10377 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10378 Compute bitregion_{start,end} for the stores and construct
10379 store_immediate_info with that. Formatting fixes.
10381 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10384 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10385 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10387 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10389 * gimplify.c: Include tm_p.h.
10391 * common.opt (gcoff): Re-add as ignored option.
10392 (gcoff1, gcoff2, gcoff3): Likewise.
10394 * Makefile.in (OBJS): Delete sdbout.o.
10395 (GTFILES): Delete $(srcdir)/sdbout.c.
10396 * debug.h: Delete sdb_debug_hooks.
10397 * final.c: Delete sdbout.h include.
10398 (final_scan_insn): Delete SDB_DEBUG check.
10399 (rest_of_clean_state): Likewise.
10400 * output.h: Delete sdb_begin_function_line.
10401 * sdbout.c: Delete.
10402 * sdbout.h: Delete.
10403 * toplev.c: Delete sdbout.h include.
10404 (process_options): Delete SDB_DEBUG check.
10405 * tree-core.h (tree_type_common): Delete pointer field of
10407 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10408 TYPE_SYMTAB_POINTER.
10409 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10410 (TYPE_SYMTAB_IS_POINTER): Delete.
10411 (TYPE_SYMTAB_IS_DIE): Renumber.
10412 * xcoffout.c: Refer to former sdbout.c file.
10413 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10415 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10416 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10417 (Debugging Options): Delete -gcoff.
10418 (-gstabs): Delete SDB reference.
10420 (-gcoff@var{level}): Delete.
10421 * doc/passes.texi (Debugging information output): Delete SDB and
10422 sdbout.c references.
10423 * doc/tm.texi: Regenerate.
10424 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10425 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10427 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10428 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10429 (SDB_DEBUGGING_INFO): Delete.
10430 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10431 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10432 * target.def (output_source_filename): Delete COFF reference.
10434 * common.opt (gcoff): Delete.
10435 (gxcoff+): Update Negative chain.
10436 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10438 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10439 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10440 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10441 and SDB references.
10442 (expand_function_start): Change sdb reference to past tense.
10443 (expand_function_end): Change sdb reference to past tense.
10444 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10445 * opts.c (debug_type_names): Delete coff entry.
10446 (common_handle_option): Delete OPT_gcoff case.
10447 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10449 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10450 * config/cris/cris.h: Delete SDB reference in comment.
10451 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10452 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10453 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10454 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10456 (ix86_expand_prologue): Likewise.
10457 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10458 * config/ia64/ia64.h: Likewise.
10459 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10460 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10462 * config/mmix/mmix.h: Likewise.
10463 * config/nds32/nds32.c: Likewise.
10464 * config/stormy/storym16.h: Likewise.
10465 * config/visium/visium.h: Likewise.
10466 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10468 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10470 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10471 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10472 FRAME_POINTER_REGNUM point at high end of local var area.
10474 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10476 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10477 Move comment around. Do not reset best_edge for a copiable
10478 destination if the copy would cause a partition change.
10479 (better_edge_p): Remove redundant check.
10481 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10483 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10485 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10487 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10488 for math functions that have _Float<N> and _Float<N>X variants.
10489 (mathfn_built_in_2): Add support for math functions that have
10490 _Float<N> and _Float<N>X variants.
10491 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10492 (expand_builtin_mathfn_ternary): Add support for fma with
10493 _Float<N> and _Float<N>X variants.
10494 (expand_builtin): Likewise.
10495 (fold_builtin_3): Likewise.
10496 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10497 create math function _Float<N> and _Float<N>X variants as external
10499 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10500 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10501 the __builtin_ prefix and if not strict ansi, without the prefix.
10502 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10503 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10504 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10505 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10506 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10507 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10508 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10509 function signatures for fma _Float<N> and _Float<N>X variants.
10510 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10511 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10512 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10513 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10514 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10515 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10516 * gencfn-macros.c (print_case_cfn): Add support for math functions
10517 that have _Float<N> and _Float<N>X variants.
10518 (print_define_operator_list): Likewise.
10519 (fltfn_suffixes): Likewise.
10521 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10522 for math functions that have _Float<N> and _Float<N>X variants.
10523 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10524 and _Float<N>X variants.
10525 (COPYSIGN): Likewise.
10528 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10529 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10531 (integer_valued_read_call_p): Likewise.
10532 * fold-const-call.c (fold_const_call_ss): Likewise.
10533 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10534 _Float<N> and _Float<N>X variants.
10535 (fold_const_call_ssss): Add support for fma _Float<N> and
10536 _Float<N>X variants.
10537 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10538 support for copysign and fma _Float<N> and _Float<N>X variants.
10539 (backprop::process_builtin_call_use): Likewise.
10540 * tree-call-cdce.c (can_test_argument_range); Add support for
10541 sqrt _Float<N> and _Float<N>X variants.
10542 (edom_only_function): Likewise.
10543 (get_no_error_domain): Likewise.
10544 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10545 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10546 copysign _Float<N> and _Float<N>X variants.
10547 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10548 handled by machine independent code.
10549 (FMAF128): Likewise.
10550 * doc/cpp.texi (Common Predefined Macros): Document defining
10551 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10552 fma _Float<N> and _Float<N>X variants.
10554 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10557 * config/i386/i386-modes.def (CCFPU): Remove definition.
10558 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10559 (ix86_cc_modes_compatible): Ditto.
10560 (ix86_expand_carry_flag_compare): Ditto.
10561 (ix86_expand_int_movcc): Ditto.
10562 (ix86_expand_int_addcc): Ditto.
10563 (ix86_reverse_condition): Ditto.
10564 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10565 Return true/false for unordered/ordered fp comparisons.
10566 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10567 (ix86_prepare_fp_compare_args): Update for rename.
10568 (ix86_expand_fp_compare): Update for rename. Generate unordered
10569 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10570 (ix86_expand_sse_compare_and_jump): Ditto.
10571 * config/i386/predicates.md (fcmov_comparison_operator):
10572 Remove CCFPU mode handling.
10573 (ix86_comparison_operator): Ditto.
10574 (ix86_carry_flag_operator): Ditto.
10575 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10576 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10577 (*cmpu<mode>_cc_i387): Ditto.
10578 (FPCMP): Remove mode iterator.
10579 (unord): Remove mode attribute.
10580 (unord_subst): New define_subst transformation
10581 (unord): New define_subst attribute.
10582 (unordered): Ditto.
10583 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10584 (*cmpi<unord>xf_i387): Ditto.
10585 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10586 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10587 using unord_subst transformation.
10588 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10589 (round_saeonly): Also handle CCFP mode.
10590 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10591 Remove UNSPEC_SAHF unspec handling.
10593 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10595 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10597 2017-10-27 Jeff Law <law@redhat.com>
10599 * gimple-ssa-sprintf.c: Include domwalk.h.
10600 (class sprintf_dom_walker): New class, derived from dom_walker.
10601 (sprintf_dom_walker::before_dom_children): New function.
10602 (struct call_info): Moved into sprintf_dom_walker class
10603 (compute_formath_length, handle_gimple_call): Likewise.
10604 (sprintf_length::execute): Call the dominator walker rather
10605 than walking the statements.
10607 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10608 gimple statement locations.
10609 (check_array_bounds): Corresponding changes. Get the statement's
10610 location directly from wi->stmt.
10612 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10615 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10617 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10619 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10620 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10622 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10625 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10626 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10627 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10628 maybe_get_pool_constant.
10629 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10632 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10634 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10635 2.26 caveat. Update gas and gld versions.
10636 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10639 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10641 * cgraph.h (set_malloc_flag): Declare.
10642 * cgraph.c (set_malloc_flag_1): New function.
10643 (set_malloc_flag): Likewise.
10644 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10645 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10647 (read_ipa_call_summary): Add support for reading is_return_callee.
10648 (write_ipa_call_summary): Stream is_return_callee.
10649 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10650 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10651 ipa-prop.h, ipa-fnsummary.h.
10652 (pure_const_names): Change to static.
10653 (malloc_state_e): Define.
10654 (malloc_state_names): Define.
10655 (funct_state_d): Add field malloc_state.
10656 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10657 (check_retval_uses): New function.
10658 (malloc_candidate_p): Likewise.
10659 (analyze_function): Add support for malloc attribute.
10660 (pure_const_write_summary): Stream malloc_state.
10661 (pure_const_read_summary): Add support for reading malloc_state.
10662 (dump_malloc_lattice): New function.
10663 (propagate_malloc): New function.
10664 (warn_function_malloc): New function.
10665 (ipa_pure_const::execute): Call propagate_malloc and
10666 ipa_free_fn_summary.
10667 (pass_local_pure_const::execute): Add support for malloc attribute.
10668 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10669 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10671 2017-10-27 Martin Liska <mliska@suse.cz>
10673 PR gcov-profile/82457
10674 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10675 for fork-like functions to be properly instrumented.
10677 2017-10-27 Richard Biener <rguenther@suse.de>
10679 PR middle-end/81659
10680 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10681 info when we redirected EH.
10683 2017-10-26 Michael Collison <michael.collison@arm.com>
10685 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10687 (<optab>_trunchf<GPI:mode>2: New pattern.
10688 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10689 * config/aarch64/iterators.md (wv): New mode attribute.
10690 (vf, VF): New mode attributes.
10691 (vgp, VGP): New mode attributes.
10692 (s): Update attribute with SImode and DImode prefixes.
10694 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10696 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10697 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10698 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10699 (nios2_option_overide): Initialize it. Don't allow R0-relative
10700 addressing with PIC.
10701 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10702 (nios2_symbolic_constant_p): Likewise.
10703 (nios2_legitimate_address_p): Likewise.
10704 (nios2_r0rel_section_name_p): New.
10705 (nios2_symbol_ref_in_r0rel_data_p): New.
10706 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10707 (r0rel_constant_p): New.
10708 (nios2_print_operand_address): Handle r0rel_constant_p.
10709 (nios2_cdx_narrow_form_p): Likewise.
10710 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10711 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10712 (Nios II Options): Document -mr0rel-sec.
10714 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10716 * config/nios2/nios2.c: Include xregex.h.
10717 (nios2_gprel_sec_regex): New.
10718 (nios2_option_overide): Initialize it. Don't allow GP-relative
10719 addressing with PIC.
10720 (nios2_small_section_name_p): Check for regex match.
10721 * config/nios2/nios2.opt (mgprel-sec=): New option.
10722 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10723 (Nios II Options): Document -mgprel-sec.
10725 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10727 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10729 2017-10-26 Tom de Vries <tom@codesourcery.com>
10731 PR tree-optimization/82707
10732 * gimple.c (gimple_copy): Fix unsharing of
10733 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10735 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10737 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10738 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10739 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10740 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10741 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10742 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10743 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10744 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10745 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10746 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10747 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10748 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10749 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10750 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10751 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10752 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10753 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10755 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10757 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10759 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10760 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10761 warning to rs6000.c. Remove the Undocumented flag, since it has
10763 (-mabi=ibmlongdouble): Likewise.
10764 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10765 already set, set the default format for long double.
10766 (rs6000_debug_reg_global): Print whether long double is IBM or
10768 (rs6000_option_override_internal): Rework setting long double
10769 format. Only warn if the user is changing the long double default
10770 and they did not use -Wno-psabi.
10771 * doc/invoke.texi (PowerPC options): Update the documentation for
10772 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10774 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10775 Alan Hayward <alan.hayward@arm.com>
10776 David Sherwood <david.sherwood@arm.com>
10778 * rtl.h (wider_subreg_mode): New function.
10779 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10780 rather than an unsigned int *.
10781 * ira-color.c (regno_max_ref_width): Replace with...
10782 (regno_max_ref_mode): ...this new variable.
10783 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10784 Use wider_subreg_mode.
10785 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10786 rather than an unsigned int *.
10787 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10788 (process_alt_operands): Likewise.
10789 (invariant_p): Likewise.
10790 * lra-spills.c (assign_mem_slot): Likewise.
10791 (add_pseudo_to_slot): Likewise.
10792 * lra.c (collect_non_operand_hard_regs): Likewise.
10793 (add_regs_to_insn_regno_info): Likewise.
10794 * reload1.c (regno_max_ref_width): Replace with...
10795 (regno_max_ref_mode): ...this new variable.
10796 (reload): Update accordingly. Update call to
10797 ira_sort_regnos_for_alter_reg.
10798 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10799 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10800 (scan_paradoxical_subregs): Likewise.
10802 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10804 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10805 (aarch64_frame): Add emit_frame_chain boolean.
10806 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10807 Move eh_return case to aarch64_layout_frame.
10808 (aarch64_layout_frame): Initialize emit_frame_chain.
10809 (aarch64_expand_prologue): Use emit_frame_chain.
10811 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10813 * config/aarch64/aarch64.c (aarch64_layout_frame):
10814 Ensure LR is always stored at the bottom of the callee-saves.
10815 Remove rarely used frame layout which saves callee-saves at top of
10816 frame, so the store of LR can be used as a valid probe in all cases.
10818 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10820 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10821 Improve unaligned TImode/TFmode base/offset split.
10823 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10824 Alan Hayward <alan.hayward@arm.com>
10825 David Sherwood <david.sherwood@arm.com>
10827 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10828 * combine.c (find_single_use_1): Likewise.
10829 (expand_field_assignment): Likewise.
10830 (move_deaths): Likewise.
10831 * lra-constraints.c (simplify_operand_subreg): Likewise.
10832 (curr_insn_transform): Likewise.
10833 * lra.c (collect_non_operand_hard_regs): Likewise.
10834 (add_regs_to_insn_regno_info): Likewise.
10835 * rtlanal.c (reg_referenced_p): Likewise.
10836 (covers_regno_no_parallel_p): Likewise.
10838 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10840 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10841 Don't print any bits outside the precision of the value.
10842 * wide-int.cc (test_printing): Add some new tests.
10844 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10846 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10847 supports -xbrace_comment option.
10848 * configure: Regenerate.
10849 * config.in: Regenerate.
10850 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10851 (ASM_CPU_SPEC): Use it.
10853 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10855 * target.def (static_rtx_alignment): New hook.
10856 * targhooks.h (default_static_rtx_alignment): Declare.
10857 * targhooks.c (default_static_rtx_alignment): New function.
10858 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10859 * doc/tm.texi: Regenerate.
10860 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10861 instead of targetm.constant_alignment. Remove call to
10862 set_mem_attributes.
10863 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10864 (cris_preferred_mininum_alignment): New function, split out from...
10865 (cris_constant_alignment): ...here.
10866 (cris_static_rtx_alignment): New function.
10867 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10869 (ix86_constant_alignment): ...here.
10870 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10871 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10872 (mmix_static_rtx_alignment): New function.
10873 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10874 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10876 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10879 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10880 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10882 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10884 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10885 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10886 mark as a sign-extending load.
10887 (local_pic_load_u): Define.
10889 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10891 PR middle-end/82062
10892 * fold-const.c (operand_equal_for_comparison_p): Also return true
10893 if ARG0 is a simple variant of ARG1 with narrower precision.
10894 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10896 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10898 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10900 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10901 scatter_static, scatter_per_elt.
10902 * x86-tune-costs.h: Add new cost entries.
10904 2017-10-25 Richard Biener <rguenther@suse.de>
10906 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10907 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10908 class pass_fre): Move to ...
10909 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10910 class pass_fre): ... here and adjust for statistics.
10912 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10915 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10916 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10917 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10918 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10921 2017-10-25 Richard Biener <rguenther@suse.de>
10923 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10924 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10925 eliminate_push_avail, eliminate_insert): Move inside...
10926 (class eliminate_dom_walker): ... this class in preparation
10928 (fini_eliminate): Remove by merging with ...
10929 (eliminate): ... this function. Adjust for class changes.
10930 (pass_pre::execute): Remove fini_eliminate call.
10931 (pass_fre::execute): Likewise.
10933 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10936 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10937 (VPERMI2, VPERMI2I): New mode iterators.
10938 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10939 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10941 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10942 mode iterator. Remove 3 old define_insn patterns.
10943 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10944 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10945 VPERMI2 mode iterator, remove the other two expanders.
10946 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10947 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10948 instructions, remove the other two patterns.
10949 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10950 mode iterator, remove the other two patterns.
10951 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10952 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10953 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10954 and adjust argument order accordingly.
10955 (ix86_expand_vec_perm): Adjust caller.
10956 (expand_vec_perm_1): Likewise.
10957 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10958 (expand_vec_perm_vpermt2_vpshub2): ... this.
10959 (ix86_expand_vec_perm_const_1): Adjust caller.
10960 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10963 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10964 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10965 (vec_shl_<mode>): Remove unused expander.
10966 (avx512bw_<shift_insn><mode>3): New define_insn.
10967 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10968 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10970 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10973 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10976 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10978 PR rtl-optimization/82396
10979 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10980 (autopref_multipass_init): Simplify initialization.
10981 (autopref_rank_data): Simplify sort order.
10982 * gcc/sched-int.h (autopref_multipass_data_): Remove
10983 multi_mem_insn_p, min_offset and max_offset.
10985 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10987 PR middle-end/60580
10988 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10989 Check special value of flag_omit_frame_pointer.
10990 (aarch64_can_eliminate): Likewise.
10991 (aarch64_override_options_after_change_1): Simplify handling of
10992 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10994 2017-10-24 Richard Biener <rguenther@suse.de>
10996 PR tree-optimization/82697
10997 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10998 zero for conditional load and unconditional store.
11000 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
11002 * doc/install.texi: Document bootstrap-cet.
11004 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
11007 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
11008 ENDBR instruction at function entrance if function is only
11011 2017-10-24 Jakub Jelinek <jakub@redhat.com>
11014 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
11015 patterns to better describe from which operation the CF is computed.
11016 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
11017 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
11018 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
11019 is 0, use _0 suffixed expanders instead of emitting a comparison
11022 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
11024 * config/i386/i386.md(*movsf_internal, *movdf_internal):
11025 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
11027 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
11029 PR middle-end/82569
11030 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
11031 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
11032 * loop-iv.c (iv_get_reaching_def): Likewise.
11033 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
11034 variable is promoted and the partition contains undefined values.
11036 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11038 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
11040 (nios2_address_cost): Define.
11041 (nios2_legitimize_address): Recognize (exp + constant) directly.
11042 (TARGET_ADDRESS_COST): Define.
11044 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11046 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
11047 (nios2_symbolic_memory_operand_p): Declare.
11048 (nios2_split_large_constant): Declare.
11049 (nios2_split_symbolic_memory_operand): Declare.
11050 * config/nios2/nios2.c: Adjust includes.
11051 (nios2_symbolic_constant_allowed): New.
11052 (nios2_symbolic_constant_p): New.
11053 (nios2_plus_symbolic_constant_p): New.
11054 (nios2_valid_addr_expr_p): Recognize addresses involving
11055 symbolic constants.
11056 (nios2_legitimate_address_p): Likewise, also LO_SUM.
11057 (nios2_symbolic_memory_operand_p): New.
11058 (nios2_large_constant_p): New.
11059 (nios2_split_large_constant): New.
11060 (nios2_split_plus_large_constant): New.
11061 (nios2_split_symbolic_memory_operand): New.
11062 (nios2_legitimize_address): Code refactoring. Handle addresses
11063 involving symbolic constants.
11064 (nios2_emit_move_sequence): Likewise.
11065 (nios2_print_operand): Improve error output.
11066 (nios2_print_operand_address): Handle LO_SUM.
11067 (nios2_cdx_narrow_form_p): Likewise.
11068 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
11069 operands involving symbolic constants.
11070 (movhi_internal, movsi_internal): Likewise.
11071 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
11072 (extendhisi2, extendqi<mode>2): Likewise.
11074 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11076 * tree-pass.h (PROP_rtl_split_insns): Define.
11077 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
11079 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11081 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
11083 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11086 * target.def (const_not_ok_for_debug_p): Default to
11087 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
11088 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
11089 * targhooks.c (default_const_not_ok_for_debug_p): New function.
11090 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
11091 which targetm.const_not_ok_for_debug_p returned true.
11092 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
11094 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
11096 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
11097 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
11098 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
11100 (ix86_const_not_ok_for_debug_p): New function.
11101 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
11102 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
11104 2017-10-23 David Malcolm <dmalcolm@redhat.com>
11107 * system.h: Conditionally include "unique-ptr.h" if
11108 INCLUDE_UNIQUE_PTR is defined.
11109 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
11110 of defining INCLUDE_UNIQUE_PTR before including "system.h".
11112 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
11114 * config/rl78/rl78.md: New define_expand "subdi3".
11116 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
11119 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
11120 DF_REF_INSN if DF_REF_INSN_INFO is false.
11122 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
11124 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
11125 xmm_move instead of sse_move.
11126 (sse_store_index): New function.
11127 (ix86_register_move_cost): Be more sensible about mismatch stall;
11128 model AVX moves correctly; make difference between sse->integer and
11130 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
11131 moves; make difference between SSE and AVX.
11132 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
11133 and zmm_move. Increase size of sse load and store tables;
11134 add unaligned load and store tables; add ssemmx_to_integer.
11135 * x86-tune-costs.h: Update all entries according to real
11136 move latencies from Agner Fog's manual and chip documentation.
11138 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11141 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
11142 * config/i386/constraints.md (Wf): New constraint.
11143 * config/i386/i386.md (UNSPEC_SBB): New unspec.
11144 (cmp<dwi>_doubleword): Removed.
11145 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
11146 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
11147 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
11148 expand with cmp<dwi>_doubleword. For LTU and GEU use
11149 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
11151 * common.opt (gcolumn-info): Enable by default.
11152 * doc/invoke.texi (gcolumn-info): Document new default.
11154 2017-10-23 Richard Biener <rguenther@suse.de>
11156 PR tree-optimization/82672
11157 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
11158 Fold the stmt if we propagated into it.
11160 2017-10-23 Richard Biener <rguenther@suse.de>
11162 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
11163 (bitmap_remove_expr_from_set): ... this. All callers call this
11164 for non-constant values.
11165 (bitmap_set_subtract): Rename to...
11166 (bitmap_set_subtract_expressions): ... this. Adjust and
11168 (bitmap_set_contains_value): Remove superfluous check.
11169 (bitmap_set_replace_value): Inline into single caller ...
11170 (bitmap_value_replace_in_set): ... here and simplify.
11171 (dependent_clean): Merge into ...
11172 (clean): ... this using an overload. Adjust.
11173 (prune_clobbered_mems): Adjust.
11174 (compute_antic_aux): Likewise.
11175 (compute_partial_antic_aux): Likewise.
11177 2017-10-23 Richard Biener <rguenther@suse.de>
11179 PR tree-optimization/82129
11181 2017-08-01 Richard Biener <rguenther@suse.de>
11183 PR tree-optimization/81181
11184 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11185 (compute_antic): ... end of iteration here.
11187 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11189 * target.def (starting_frame_offset): New hook.
11190 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
11191 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
11192 * doc/tm.texi.in: Regenerate.
11193 * hooks.h (hook_hwi_void_0): Declare.
11194 * hooks.c (hook_hwi_void_0): New function.
11195 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
11196 STARTING_FRAME_OFFSET.
11197 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
11198 * reload1.c (reload): Likewise.
11199 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
11200 instead of STARTING_FRAME_OFFSET.
11201 * function.c (try_fit_stack_local): Likewise.
11202 (assign_stack_local_1): Likewise
11203 (instantiate_virtual_regs): Likewise.
11204 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11205 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11206 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11207 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11208 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11209 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11210 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11211 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11212 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11213 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11214 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11215 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11216 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11217 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11218 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11219 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11220 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11221 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11222 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11223 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11224 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11225 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11226 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11227 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11228 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11229 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11230 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11231 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11232 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11233 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11234 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11235 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11236 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11237 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11238 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11239 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11240 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11241 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11242 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11243 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11244 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11245 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11246 return a HOST_WIDE_INT.
11247 (avr_builtin_setjmp_frame_value): Use it instead of
11248 STARTING_FRAME_OFFSET.
11249 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11250 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11251 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11253 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11254 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11255 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11256 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11257 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11258 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11259 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11260 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11261 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11262 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11263 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11264 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11266 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11267 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11268 * config/mips/mips.c (mips_compute_frame_info): Refer to
11269 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11270 (mips_starting_frame_offset): New function.
11271 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11272 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11273 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11274 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11275 and return a HOST_WIDE_INT.
11276 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11277 (mmix_initial_elimination_offset): Refer to
11278 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11279 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11280 * config/pa/pa.c (pa_starting_frame_offset): New function.
11281 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11282 (pa_expand_prologue): Likewise.
11283 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11284 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11285 !FRAME_GROWS_DOWNWARD handling to...
11286 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11287 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11288 !FRAME_GROWS_DOWNWARD handling to...
11289 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11290 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11291 !FRAME_GROWS_DOWNWARD handling to...
11292 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11293 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11295 (rs6000_starting_frame_offset): New function.
11296 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11297 !FRAME_GROWS_DOWNWARD handling to...
11298 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11299 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11300 !FRAME_GROWS_DOWNWARD handling to...
11301 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11302 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11303 !FRAME_GROWS_DOWNWARD handling to...
11304 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11305 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11306 (rs6000_starting_frame_offset): New function.
11307 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11308 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11309 * config/vax/vax.c (vax_starting_frame_offset): New function.
11310 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11311 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11312 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11313 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11314 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11315 * system.h (STARTING_FRAME_OFFSET): Poison.
11317 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11319 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11320 SCALAR_TYPE_MODE instead of TYPE_MODE.
11322 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11323 Alan Hayward <alan.hayward@arm.com>
11324 David Sherwood <david.sherwood@arm.com>
11326 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11328 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11329 Alan Hayward <alan.hayward@arm.com>
11330 David Sherwood <david.sherwood@arm.com>
11332 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11334 2017-10-23 Richard Biener <rguenther@suse.de>
11336 PR tree-optimization/82129
11337 * tree-ssa-pre.c (bitmap_set_and): Remove.
11338 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11339 canonicalizing expressions in the set to those with lowest
11340 ID rather than taking that from the first edge.
11342 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11344 * combine.c (rtx_equal_for_field_assignment_p): Use
11345 byte_lowpart_offset.
11347 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11348 Alan Hayward <alan.hayward@arm.com>
11349 David Sherwood <david.sherwood@arm.com>
11351 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11352 to a SUBREG_PROMOTED_VAR.
11354 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11355 Alan Hayward <alan.hayward@arm.com>
11356 David Sherwood <david.sherwood@arm.com>
11358 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11359 (expand_debug_source_expr): Likewise.
11360 * combine.c (combine_simplify_rtx): Likewise.
11361 * cse.c (fold_rtx): Likewise.
11362 * optabs.c (expand_float): Likewise.
11363 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11364 (simplify_binary_operation_1): Likewise.
11366 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11367 Alan Hayward <alan.hayward@arm.com>
11368 David Sherwood <david.sherwood@arm.com>
11370 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11371 (record_promoted_value): Likewise.
11372 * expr.c (expand_expr_real_2): Likewise.
11373 * ree.c (update_reg_equal_equiv_notes): Likewise.
11374 (combine_set_extension): Likewise.
11375 * rtlanal.c (low_bitmask_len): Likewise.
11376 * simplify-rtx.c (neg_const_int): Likewise.
11377 (simplify_binary_operation_1): Likewise.
11379 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11380 Alan Hayward <alan.hayward@arm.com>
11381 David Sherwood <david.sherwood@arm.com>
11383 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11384 * regcprop.c (maybe_mode_change): Likewise.
11385 * reload1.c (alter_reg): Likewise.
11387 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11389 * inchash.h (inchash::hash::add_wide_int): New function.
11390 * lto-streamer-out.c (hash_tree): Use it.
11392 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11394 * inchash.h (inchash::hash::add_wide_int): Rename to...
11395 (inchash::hash::add_hwi): ...this.
11396 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11397 (polymorphic_call_target_hasher::hash): Likewise.
11398 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11399 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11400 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11401 * lto-streamer-out.c (hash_tree): Likewise.
11402 * optc-save-gen.awk: Likewise.
11403 * tree.c (add_expr): Likewise.
11405 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11408 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11409 for ordered inequality comparisons even with TARGET_IEEE_FP.
11411 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11414 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11415 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11416 Expand with cmp<dwi>_doubleword.
11418 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11420 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11421 List CET intrinsics.
11422 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11423 specific to -fcf-protection option.
11425 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11427 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11428 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11429 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11430 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11431 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11432 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11433 (extra_objs): Add cet.o for Linux/x86 targets.
11434 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11435 * config/i386/cet.c: New file.
11436 * config/i386/cetintrin.h: Likewise.
11437 * config/i386/t-cet: Likewise.
11438 * config/i386/cpuid.h (bit_SHSTK): New.
11439 (bit_IBT): Likewise.
11440 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11441 pass IBT and SHSTK bits.
11442 * config/i386/i386-builtin-types.def
11443 (VOID_FTYPE_UNSIGNED_PVOID): New.
11444 (VOID_FTYPE_UINT64_PVOID): Likewise.
11445 * config/i386/i386-builtin.def: Add CET intrinsics.
11446 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11447 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11448 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11449 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11451 * config/i386/i386.c (rest_of_insert_endbranch): New.
11452 (pass_data_insert_endbranch): Likewise.
11453 (pass_insert_endbranch): Likewise.
11454 (make_pass_insert_endbranch): Likewise.
11455 (ix86_notrack_prefixed_insn_p): Likewise.
11456 (ix86_target_string): Add -mibt, -mshstk flags.
11457 (ix86_option_override_internal): Add flag_cf_protection
11459 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11460 (ix86_print_operand): Add 'notrack' prefix output.
11461 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11462 (ix86_expand_builtin): Expand CET intrinsics.
11463 (x86_output_mi_thunk): Add 'endbranch' instruction.
11464 * config/i386/i386.h (TARGET_IBT): New.
11465 (TARGET_IBT_P): Likewise.
11466 (TARGET_SHSTK): Likewise.
11467 (TARGET_SHSTK_P): Likewise.
11468 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11469 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11470 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11471 (builtin_setjmp_setup): New pattern.
11472 (builtin_longjmp): Likewise.
11473 (rdssp<mode>): Likewise.
11474 (incssp<mode>): Likewise.
11475 (saveprevssp): Likewise.
11476 (rstorssp): Likewise.
11477 (wrss<mode>): Likewise.
11478 (wruss<mode>): Likewise.
11479 (setssbsy): Likewise.
11480 (clrssbsy): Likewise.
11481 (nop_endbr): Likewise.
11482 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11484 * config/i386/immintrin.h: Include <cetintrin.h>.
11485 * config/i386/linux-common.h
11486 (file_end_indicate_exec_stack_and_cet): New prototype.
11487 (TARGET_ASM_FILE_END): New.
11489 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11491 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11492 latencies instead of having separate table; make difference between
11493 integer and float costs.
11494 * i386.h (processor_costs): Remove scalar_stmt_cost,
11495 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11496 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11498 * x86-tune-costs.h: Remove entries which has been removed in
11499 procesor_costs from all tables; make cond_taken_branch_cost
11500 and cond_not_taken_branch_cost COST_N_INSNS based.
11502 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11504 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11506 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11508 * config/i386/i386.md (isa): Remove fma_avx512f.
11509 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11510 <avx512>_fmadd_<mode>_mask3<round_name>,
11511 <avx512>_fmsub_<mode>_mask<round_name>,
11512 <avx512>_fmsub_<mode>_mask3<round_name>,
11513 <avx512>_fnmadd_<mode>_mask<round_name>,
11514 <avx512>_fnmadd_<mode>_mask3<round_name>,
11515 <avx512>_fnmsub_<mode>_mask<round_name>,
11516 <avx512>_fnmsub_<mode>_mask3<round_name>,
11517 <avx512>_fmaddsub_<mode>_mask<round_name>,
11518 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11519 <avx512>_fmsubadd_<mode>_mask<round_name>,
11520 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11521 (*vec_widen_umult_even_v16si<mask_name>,
11522 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11523 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11525 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11527 * extend.texi: Add 'nocf_check' documentation.
11528 * gimple.texi: Add second parameter to
11529 gimple_build_call_from_tree.
11530 * invoke.texi: Add -fcf-protection documentation.
11531 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11533 2017-10-20 Richard Biener <rguenther@suse.de>
11535 PR tree-optimization/82473
11536 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11537 the largest input type.
11539 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11541 * c-attribs.c (handle_nocf_check_attribute): New function.
11542 (c_common_attribute_table): Add 'nocf_check' handling.
11543 * gimple-parser.c: Add second argument NULL to
11544 gimple_build_call_from_tree.
11545 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11546 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11548 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11549 * common.opt: Add fcf-protection flag.
11550 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11551 * flag-types.h: Add enum cf_protection_level.
11552 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11553 Add 'nocf_check' attribute propagation to gimple call.
11554 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11555 (gimple_build_call_from_tree): Update prototype.
11556 (gimple_call_nocf_check_p): New function.
11557 (gimple_call_set_nocf_check): Likewise.
11558 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11559 * ipa-icf.c: Add nocf_check attribute in statement hash.
11560 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11561 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11562 * toplev.c (process_options): Add flag_cf_protection handling.
11564 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11566 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11568 2017-10-20 Richard Biener <rguenther@suse.de>
11570 PR tree-optimization/82603
11571 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11572 remove false predicated stores.
11574 2017-10-20 Richard Biener <rguenther@suse.de>
11576 * graphite-isl-ast-to-gimple.c
11577 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11578 Remove return value and simplify, dump copied stmt after lhs
11580 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11581 Reduce dump verbosity.
11582 (gsi_insert_earliest): Likewise.
11583 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11584 * graphite.c (print_global_statistics): Adjust dumping.
11585 (print_graphite_scop_statistics): Likewise.
11586 (print_graphite_statistics): Do not dump loops here.
11587 (graphite_transform_loops): But here.
11589 2017-10-20 Nicolas Roche <roche@adacore.com>
11591 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11592 * configure: Regenerate.
11594 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11597 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11598 functions when optimizing replace GIMPLE_RETURN stmts with
11599 calls to __builtin_unreachable ().
11602 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11603 for -fsanitize=thread link of executables.
11604 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11605 link of executables.
11608 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11609 New mode iterators.
11610 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11611 define_insns for logical vector shifts to use VI248_AVX512BW
11612 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11613 condition, useless isa and prefix attributes. Change the first
11614 2 of these define_insns to ...
11615 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11616 define_insn for avx512vl.
11617 (<shift_insn><mode>3): ... and this, new define_insn without
11618 masking for non-avx512vl.
11621 * config/i386/sse.md (*andnot<mode>3,
11622 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11623 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11624 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11625 not applied use empty suffix even for TARGET_AVX512VL.
11626 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11627 is applied, supply evex,evex or evex,evex,evex instead of just
11630 2017-10-20 Julia Koval <julia.koval@intel.com>
11632 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11633 (OPTION_MASK_ISA_GFNI_UNSET): New.
11634 (ix86_handle_option): Handle OPT_mgfni.
11635 * config/i386/cpuid.h (bit_GFNI): New.
11636 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11637 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11638 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11639 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11640 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11641 * config/i386/i386.opt: Add mgfni.
11643 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11645 * config/msp430/msp430.c (msp430_option_override): Disable
11646 -fdelete-null-pointer-checks.
11647 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11649 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11651 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11652 of x87 and SSE instructions.
11654 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11656 * asan.c (create_cond_insert_point): Do not update edge count.
11657 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11658 (afdo_propagate_circuit): Likewise.
11659 (afdo_calculate_branch_prob): Likewise.
11660 (afdo_annotate_cfg): Likewise.
11661 * basic-block.h (struct edge_def): Remove count.
11662 (edge_def::count): New accessor.
11663 * bb-reorder.c (rotate_loop): Update.
11664 (find_traces_1_round): Update.
11665 (connect_traces): Update.
11666 (sanitize_hot_paths): Update.
11667 * cfg.c (unchecked_make_edge): Update.
11668 (make_single_succ_edge): Update.
11669 (check_bb_profile): Update.
11670 (dump_edge_info): Update.
11671 (update_bb_profile_for_threading): Update.
11672 (scale_bbs_frequencies_int): Update.
11673 (scale_bbs_frequencies_gcov_type): Update.
11674 (scale_bbs_frequencies_profile_count): Update.
11675 (scale_bbs_frequencies): Update.
11676 * cfganal.c (connect_infinite_loops_to_exit): Update.
11677 * cfgbuild.c (compute_outgoing_frequencies): Update.
11678 (find_many_sub_basic_blocks): Update.
11679 * cfgcleanup.c (try_forward_edges): Update.
11680 (try_crossjump_to_edge): Update
11681 * cfgexpand.c (expand_gimple_cond): Update
11682 (expand_gimple_tailcall): Update
11683 (construct_exit_block): Update
11684 * cfghooks.c (verify_flow_info): Update
11685 (redirect_edge_succ_nodup): Update
11686 (split_edge): Update
11687 (make_forwarder_block): Update
11688 (duplicate_block): Update
11689 (account_profile_record): Update
11690 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11691 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11692 * cfgloopmanip.c (scale_loop_profile): Update.
11694 (lv_adjust_loop_entry_edge): Update.
11695 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11696 (force_nonfallthru_and_redirect): Update.
11697 (purge_dead_edges): Update.
11698 (rtl_flow_call_edges_add): Update.
11699 * cgraphunit.c (init_lowered_empty_function): Update.
11700 (cgraph_node::expand_thunk): Update.
11701 * gimple-pretty-print.c (dump_probability): Update.
11702 (dump_edge_probability): Update.
11703 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11704 * haifa-sched.c (sched_create_recovery_edges): Update.
11705 * hsa-gen.c (convert_switch_statements): Update.
11706 * ifcvt.c (dead_or_predicable): Update.
11707 * ipa-inline-transform.c (inline_transform): Update.
11708 * ipa-split.c (split_function): Update.
11709 * ipa-utils.c (ipa_merge_profiles): Update.
11710 * loop-doloop.c (add_test): Update.
11711 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11712 * lto-streamer-in.c (input_cfg): Update.
11713 (input_function): Update.
11714 * lto-streamer-out.c (output_cfg): Update.
11715 * modulo-sched.c (sms_schedule): Update.
11716 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11717 * predict.c (maybe_hot_edge_p): Update.
11718 (unlikely_executed_edge_p): Update.
11719 (probably_never_executed_edge_p): Update.
11720 (dump_prediction): Update.
11721 (drop_profile): Update.
11722 (propagate_unlikely_bbs_forward): Update.
11723 (determine_unlikely_bbs): Update.
11724 (force_edge_cold): Update.
11725 * profile.c (compute_branch_probabilities): Update.
11726 * reg-stack.c (better_edge): Update.
11727 * shrink-wrap.c (handle_simple_exit): Update.
11728 * tracer.c (better_p): Update.
11729 * trans-mem.c (expand_transaction): Update.
11730 (split_bb_make_tm_edge): Update.
11731 * tree-call-cdce.c: Update.
11732 * tree-cfg.c (gimple_find_sub_bbs): Update.
11733 (gimple_split_edge): Update.
11734 (gimple_duplicate_sese_region): Update.
11735 (gimple_duplicate_sese_tail): Update.
11736 (gimple_flow_call_edges_add): Update.
11737 (insert_cond_bb): Update.
11738 (execute_fixup_cfg): Update.
11739 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11740 * tree-complex.c (expand_complex_div_wide): Update.
11741 * tree-eh.c (lower_resx): Update.
11742 (unsplit_eh): Update.
11743 (cleanup_empty_eh_move_lp): Update.
11744 * tree-inline.c (copy_edges_for_bb): Update.
11745 (freqs_to_counts): Update.
11746 (copy_cfg_body): Update.
11747 * tree-ssa-dce.c (remove_dead_stmt): Update.
11748 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11749 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11750 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11751 (unloop_loops): Update.
11752 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11753 * tree-ssa-loop-split.c (connect_loops): Update.
11754 (split_loop): Update.
11755 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11756 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11757 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11758 * tree-ssa-reassoc.c (branch_fixup): Update.
11759 * tree-ssa-tail-merge.c (replace_block_by): Update.
11760 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11761 (compute_path_counts): Update.
11762 (update_profile): Update.
11763 (recompute_probabilities): Update.
11764 (update_joiner_offpath_counts): Update.
11765 (estimated_freqs_path): Update.
11766 (freqs_to_counts_path): Update.
11767 (clear_counts_path): Update.
11768 (ssa_fix_duplicate_block_edges): Update.
11769 (duplicate_thread_path): Update.
11770 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11771 (case_bit_test_cmp): Update.
11772 (collect_switch_conv_info): Update.
11773 (gen_inbound_check): Update.
11774 (do_jump_if_equal): Update.
11775 (emit_cmp_and_jump_insns): Update.
11776 * tree-tailcall.c (decrease_profile): Update.
11777 (eliminate_tail_call): Update.
11778 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11779 (vect_do_peeling): Update.
11780 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11781 * ubsan.c (ubsan_expand_null_ifn): Update.
11782 (ubsan_expand_ptr_ifn): Update.
11783 * value-prof.c (gimple_divmod_fixed_value): Update.
11784 (gimple_mod_pow2): Update.
11785 (gimple_mod_subtract): Update.
11786 (gimple_ic): Update.
11787 (gimple_stringop_fixed_value): Update.
11789 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11792 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11794 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11796 PR rtl-optimization/82395
11797 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11798 based on non_spilled_static_chain_regno_p.
11800 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11802 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11803 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11804 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11806 2017-10-19 Martin Sebor <msebor@redhat.com>
11808 PR tree-optimization/82596
11809 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11811 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11813 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11814 (get_mem_refs_of_builtin_call): Likewise.
11815 * builtins.c (expand_builtin_apply): Adjust call to
11816 allocate_dynamic_stack_space.
11817 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11818 the third argument to allocate_dynamic_stack_space, otherwise -1.
11819 (expand_builtin): Deal with all alloca variants.
11820 (is_inexpensive_builtin): Likewise.
11821 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11822 * calls.c (special_function_p): Deal with all alloca variants.
11823 (initialize_argument_information): Adjust call to
11824 allocate_dynamic_stack_space.
11825 (expand_call): Likewise.
11826 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11827 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11828 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11829 use it for the stack usage computation.
11830 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11831 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11832 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11833 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11834 (in_loop_p): Remove first argument and useless check.
11835 (pass_walloca::execute): Remove useless test and adjust call to above.
11836 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11837 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11838 (gimplify_call_expr): Deal with all alloca variants.
11839 * hsa-gen.c (gen_hsa_alloca): Likewise.
11840 (gen_hsa_insns_for_call): Likewise.
11841 * ipa-pure-const.c (special_builtin_state): Likewise.
11842 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11843 * tree-object-size.c (alloc_object_size): Likewise.
11844 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11845 (call_may_clobber_ref_p_1): Likewise.
11846 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11847 (ccp_fold_stmt): Likewise.
11848 (optimize_stack_restore): Likewise.
11849 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11850 (mark_all_reaching_defs_necessary_1): Likewise.
11851 (propagate_necessity): Likewise.
11852 (eliminate_unnecessary_stmts): Likewise.
11853 * tree.c (build_common_builtin_nodes): Build
11854 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11855 (build_alloca_call_expr): New function.
11856 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11857 (CASE_BUILT_IN_ALLOCA): Likewise.
11858 (build_alloca_call_expr): Declare.
11859 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11861 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11864 * dwarf2out.c (new_die_raw): New static inline function.
11865 (new_die): Use it to create the DIE.
11866 (add_AT_external_die_ref): Likewise.
11867 (clone_die): Likewise.
11868 (clone_as_declaration): Likewise.
11869 (dwarf2out_vms_debug_main_pointer): Likewise.
11870 (base_type_die): Likewise. Remove early return for corner cases.
11871 Do not call add_pubtype on the DIE here.
11872 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11873 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11874 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11875 native order exists for base types, attach the DIE manually and call
11876 add_pubtype on it. Do not equate a reverse order DIE to the type.
11878 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11880 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11881 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11883 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11885 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11887 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11888 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11889 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11890 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11891 * opts.c (sanitizer_opts): Add builtin.
11892 * ubsan.c (instrument_builtin): New function.
11893 (pass_ubsan::execute): Call it.
11894 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11895 * doc/invoke.texi: Document -fsanitize=builtin.
11897 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11898 builtins, store max (log2 (align), 0) into uchar field instead of
11899 align into uptr field.
11900 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11901 store uchar 0 field instead of uptr 0 field.
11902 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11903 instead of passing one address of struct with 2 locations pass
11904 two addresses of structs with 1 location each.
11905 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11906 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11907 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11908 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11909 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11910 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11911 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11912 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11914 2017-10-19 Martin Liska <mliska@suse.cz>
11917 * file-find.c (remove_prefix): Remove.
11918 * file-find.h (remove_prefix): Likewise.
11919 * gcc-ar.c: Remove smartness of lookup.
11921 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11923 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11924 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11925 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11927 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11930 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11931 (setcc + and to xor + setcc): New peephole2.
11933 2017-10-19 Tom de Vries <tom@codesourcery.com>
11935 * doc/sourcebuild.texi (Test Directives, Variants of
11936 dg-require-support): Add dg-require-stack-size.
11938 2017-10-19 Martin Liska <mliska@suse.cz>
11941 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11942 that have a large alignment.
11943 (gimplify_target_expr): Likewise.
11945 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11947 PR rtl-optimization/82602
11948 * ira.c (rtx_moveable_p): Return false for volatile asm.
11950 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11953 * config/i386/i386-modes.def (CCGZ): New CC mode.
11954 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11955 * config/i386/predicates.md (ix86_comparison_operator):
11957 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11958 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11959 with double-word subtraction.
11960 (put_condition_code): Handle CCGZmode.
11962 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11964 * wide-int.cc (debug (const wide_int &)): New.
11965 (debug (const wide_int *)): New.
11966 (debug (const widest_int &)): New.
11967 (debug (const widest_int *)): New.
11969 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11971 PR middle-end/82556
11972 * lra-constraints.c (curr_insn_transform): Use non-input operand
11973 instead of output one for matched reload.
11975 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11977 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11978 (tree-ssa-loop-ivopts.h): New header file.
11979 (struct builtin_info): New fields.
11980 (classify_builtin_1): Compute and record base and offset parts for
11981 memset builtin partition by calling strip_offset.
11982 (offset_cmp, fuse_memset_builtins): New functions.
11983 (finalize_partitions): Fuse adjacent memset partitions by calling
11985 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11986 Expose the interface.
11987 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11989 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11991 PR tree-optimization/82574
11992 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11993 that data reference must be executed exactly once per iteration
11994 against the outermost loop in nest.
11995 (classify_partition): Update call to above function.
11997 2017-10-18 Richard Biener <rguenther@suse.de>
11999 PR tree-optimization/82591
12000 * graphite.c (graphite_transform_loops): Move code gen message
12002 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12003 Here. Handle scop_to_isl_ast failing.
12004 (scop_to_isl_ast): Limit the number of ISL operations.
12006 2017-10-18 Richard Biener <rguenther@suse.de>
12008 * graphite-isl-ast-to-gimple.c
12009 (translate_isl_ast_to_gimple::set_rename): Simplify.
12010 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
12011 (graphite_copy_stmts_from_block): ... here.
12012 (copy_bb_and_scalar_dependences): Simplify.
12013 (add_parameters_to_ivs_params): Canonicalize.
12014 (generate_entry_out_of_ssa_copies): Simplify.
12015 * graphite-sese-to-poly.c (extract_affine_name): Simplify
12016 by passing in ISL dimension.
12017 (parameter_index_in_region_1): Rename to ...
12018 (parameter_index_in_region): ... this.
12019 (extract_affine): Adjust assert, pass down parameter index.
12020 (add_param_constraints): Use range-info when available.
12021 (build_scop_context): Adjust.
12022 * sese.c (new_sese_info): Adjust.
12023 (free_sese_info): Likewise.
12024 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
12025 Remove unused typedefs.
12026 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
12028 2017-10-18 Martin Liska <mliska@suse.cz>
12030 * combine.c (simplify_compare_const): Add gcc_fallthrough.
12032 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12034 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
12035 (s390_sched_init): Do not reset s390_sched_state if we entered the
12036 current basic block via a fallthru edge and all others are unlikely.
12038 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12040 * config/s390/s390.c (NUM_SIDES): New variable.
12041 (LONGRUNNING_THRESHOLD): New variable.
12042 (LATENCY_FACTOR): New variable.
12043 (s390_sched_score): Decrease score for long-running instructions on
12045 (s390_sched_variable_issue): Perform bookkeeping for long-running
12048 2017-10-18 Richard Biener <rguenther@suse.de>
12050 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
12051 Simplify with removal of the parameter rename map.
12052 (set_rename): Likewise.
12053 (should_copy_to_new_region): Likewise.
12054 (graphite_copy_stmts_from_block): Likewise.
12055 (copy_bb_and_scalar_dependences): Remove initialization of
12056 unused copied_bb_map.
12057 (copy_def): Remove.
12058 (copy_internal_parameters): Likewise.
12059 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
12060 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
12061 Use INTEGRAL_TYPE_P.
12062 (parameter_index_in_region_1): Rename to ...
12063 (assign_parameter_index_in_region): ... this. Assert we have
12064 a parameter we handle.
12065 (scan_tree_for_params): Adjust.
12066 * sese.h (parameter_rename_map_t): Remove.
12067 (struct sese_info_t): Remove unused parameter_rename_map and
12068 copied_bb_map members.
12069 * sese.c (new_sese_info): Adjust.
12070 (free_sese_info): Likewise.
12072 2017-10-18 Martin Liska <mliska@suse.cz>
12075 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
12076 on an abnormal edge.
12078 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
12080 * doc/invoke.texi (ffunction-sections and fdata-sections):
12083 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12085 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
12086 the use statement can throw internally.
12088 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12090 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
12091 any RTX present on the RHS of a SET.
12092 * compare-elim.c (try_eliminate_compare): Restore comment.
12094 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12096 * langhooks.h (struct lang_hooks): Document that tree_size langhook
12097 may be also called on tcc_type nodes.
12098 * langhooks.c (lhd_tree_size): Likewise.
12100 2017-10-17 David Malcolm <dmalcolm@redhat.com>
12102 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
12103 format_warning_at_substring.
12104 (maybe_warn): Convert source_range * param to a location_t. Pass
12105 UNKNOWN_LOCATION rather than NULL to fmtwarn.
12106 (format_directive): Remove code to extract source_ranges and
12107 source_range * in favor of just a location_t.
12108 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
12110 * substring-locations.c (format_warning_va): Convert
12111 source_range * param to a location_t.
12112 (format_warning_at_substring): Likewise.
12113 * substring-locations.h (format_warning_va): Likewise.
12114 (format_warning_at_substring): Likewise.
12116 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
12118 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
12120 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
12121 and scatter/gather ops.
12123 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
12124 vec_gather_load and vec_scatter_store.
12125 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
12126 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
12127 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
12128 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
12129 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
12130 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
12132 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
12134 * reg-stack.c (compare_for_stack_reg): Add bool argument.
12135 Detect FTST instruction and handle its register pops. Only pop
12136 second operand if can_pop_second_op is true.
12137 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
12138 set can_pop_second_op to false in the compare_for_stack_reg call.
12140 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
12141 output_fp_compare for stack register operands.
12142 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
12143 instructions here. Do not emit stack register pops here. Assert
12144 that FCOMPP pops next to top stack register. Rewrite function.
12146 2017-10-17 Nathan Sidwell <nathan@acm.org>
12148 PR middle-end/82577
12149 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
12150 use DECL_ASSEMBLER_NAME_RAW.
12152 PR middle-end/82546
12153 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
12156 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
12157 Wilco Dijkstra <wilco.dijkstra@arm.com>
12159 * builtins.c (expand_builtin_update_setjmp_buf): Add a
12160 converstion to Pmode from the buf_addr.
12162 2017-10-17 Richard Biener <rguenther@suse.de>
12164 * graphite-dependences.c (scop_get_reads_and_writes): Change
12165 output parameters to references.
12167 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
12169 PR 71026/tree-optimization
12170 * fold-const.c (distribute_real_division): Removed.
12171 (fold_binary_loc): Remove calls to distribute_real_divison.
12173 2017-10-17 Richard Biener <rguenther@suse.de>
12175 * graphite-scop-detection.c
12176 (scop_detection::stmt_has_simple_data_refs_p): Always use
12177 the full nest as region.
12178 (try_generate_gimple_bb): Likewise.
12179 * sese.c (scalar_evolution_in_region): Simplify now that
12180 SCEV can handle instantiation in regions.
12181 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
12182 in the non-loop part of a function if requested.
12184 2017-10-17 Richard Biener <rguenther@suse.de>
12186 PR tree-optimization/82563
12187 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
12189 (graphite_regenerate_ast_isl): Call it.
12190 * graphite-scop-detection.c (build_scops): Remove entry edge split.
12192 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12194 PR tree-optimization/82549
12195 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
12196 Formatting fixes. Instead of calling make_bit_field_ref with negative
12199 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
12201 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
12202 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12203 _mm_maskz_reduce_ss): New.
12204 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12205 __builtin_ia32_reducess_mask): Ditto..
12206 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12207 * config/i386/sse.md (reduces<mode>): Renamed to ...
12208 (reduces<mode><mask_scalar_name>): ... this.
12209 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12211 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12212 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12214 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12216 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12217 * selftest-run-tests.c (selftest::run_tests): Call
12218 selftest::unique_ptr_tests_cc_tests.
12219 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12220 * unique-ptr-tests.cc: New file.
12222 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12225 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12227 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12228 (make_hard_regno_born, make_hard_regno_dead): Update
12229 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12231 2017-10-16 Jeff Law <law@redhat.com>
12233 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12235 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12237 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12239 2017-10-16 Olivier Hainque <hainque@adacore.com>
12241 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12242 with_cpu if we were configured for an e500v2 target cpu name.
12244 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12246 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12247 * doc/invoke.texi: Document +nodsp as a valid extension for
12250 2017-10-16 Martin Liska <mliska@suse.cz>
12252 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12253 (test_set_range): Likewise.
12254 (test_range_functions): Rename to ...
12255 (test_bit_in_range): ... this.
12256 (sbitmap_c_tests): Add new test.
12258 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12260 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12262 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12263 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12265 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12267 * config/aarch64/aarch64-builtins.c
12268 (aarch64_types_quadopu_lane_qualifiers): New.
12269 (TYPES_QUADOPU_LANE): New.
12270 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12271 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12272 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12273 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12274 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12275 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12276 (Vdottype, DOTPROD): New.
12277 (sur): Add SDOT and UDOT.
12279 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12281 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12282 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12283 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12284 Add TARGET_DOTPROD.
12285 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12286 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12287 Enable TARGET_DOTPROD.
12288 (cortex-a75.cortex-a55): Likewise.
12289 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12291 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12293 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12294 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12296 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12298 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12299 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12300 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12301 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12302 * config/arm/types.md (neon_dot, neon_dot_q): New.
12303 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12305 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12307 * config/arm/arm.h (TARGET_DOTPROD): New.
12308 * config/arm/arm.c (arm_arch_dotprod): New.
12309 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12310 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12311 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12312 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12313 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12314 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12315 * doc/invoke.texi (armv8.2-a): Document dotprod
12317 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12319 * i386.c (ix86_vec_cost): New function.
12320 (ix86_rtx_costs): Handle vector operations better.
12321 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12322 * x86-tune-costs.h: Add new costs to all tables.
12324 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12326 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12328 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12329 divsd, sqrtss and sqrtsd
12330 * x86-tune-costs.h: Add new entries to all costs.
12331 (znver1_cost): Fix to match real instruction latencies.
12333 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12334 Michael Collison <michael.collison@arm.com>
12336 * compare-elim.c: Include emit-rtl.h.
12337 (can_merge_compare_into_arith): New function.
12338 (try_validate_parallel): Likewise.
12339 (try_merge_compare): Likewise.
12340 (try_eliminate_compare): Call the above when no previous clobber
12342 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12345 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12347 PR middle-end/62263
12348 PR middle-end/82498
12349 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12350 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12352 PR middle-end/62263
12353 PR middle-end/82498
12354 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12355 to be any operand_equal_p operands. For & (B - 1) require
12356 B to be power of 2. Recognize
12357 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12359 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12362 * optabs.c (expand_memory_blockage): Fix call of
12363 targetm.have_memory_blockage.
12365 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12368 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12369 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12370 objects to extra_objs instead of overwriting it.
12372 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12374 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12375 Use any_fp_register_operand as operand[3] predicate. Simplify
12376 equality test for operands[2] and operands[4] memory location.
12377 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12378 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12379 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12380 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12381 any_fp_register_operand as operand[1] predicate. Simplify
12382 equality test for operands[0] and operands[3] memory location.
12383 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12384 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12385 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12387 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12389 * target-insns.def: Add memory_blockage.
12390 * optabs.c (expand_memory_blockage): New function.
12391 (expand_asm_memory_barrier): Rename ...
12392 (expand_asm_memory_blockage): ... to this.
12393 (expand_mem_thread_fence): Call expand_memory_blockage
12394 instead of expand_asm_memory_barrier.
12395 (expand_mem_singnal_fence): Ditto.
12396 (expand_atomic_load): Ditto.
12397 (expand_atomic_store): Ditto.
12398 * doc/md.texi (Standard Pattern Names For Generation):
12399 Document memory_blockage instruction pattern.
12401 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12403 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12404 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12405 * config/rl78/rl78.md: New define_expand "adddi3".
12407 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12409 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12412 2017-10-13 Jeff Law <law@redhat.com>
12414 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12416 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12419 * internal-fn.c (expand_mul_overflow): If both operands have
12420 the same highpart of -1 or 0 and the topmost bit of lowpart
12421 is different, overflow is if res <= 0 rather than res < 0.
12423 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12425 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12426 TARGET_P9_VECTOR code for unaligned_load case.
12428 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12430 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12432 2017-10-13 Nathan Sidwell <nathan@acm.org>
12434 * tree-core.h (tree_contains_struct): Make bool.
12435 * tree.c (tree_contains_struct): Likewise.
12436 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12437 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12438 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12439 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12440 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12442 2017-10-13 Richard Biener <rguenther@suse.de>
12444 * graphite-isl-ast-to-gimple.c
12445 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12446 parameters and dominance check.
12447 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12448 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12449 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12450 Do not update SSA form here or do intermediate IL verification.
12451 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12452 (graphite_initialize): Remove check on the number of loops in
12453 the function and inline into graphite_transform_loops.
12454 (graphite_finalize): Inline into graphite_transform_loops.
12455 (graphite_transform_loops): Perform SSA update and IL verification
12457 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12459 2017-10-13 Richard Biener <rguenther@suse.de>
12461 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12462 graphite_expression_type_precision): Avoid global constructor
12464 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12465 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12466 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12467 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12468 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12469 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12471 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12474 * config/i386/i386.h (ix86_red_zone_size): New.
12475 * config/i386/i386.md (push peephole2s): Replace
12476 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12478 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12479 Alan Hayward <alan.hayward@arm.com>
12480 David Sherwood <david.sherwood@arm.com>
12482 * combine.c (can_change_dest_mode): Reject changes in
12483 REGMODE_NATURAL_SIZE.
12485 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12486 Alan Hayward <alan.hayward@arm.com>
12487 David Sherwood <david.sherwood@arm.com>
12489 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12490 (expand_debug_source_expr): Likewise.
12491 * combine.c (combine_simplify_rtx): Likewise.
12492 * cse.c (fold_rtx): Likewise.
12493 * fwprop.c (canonicalize_address): Likewise.
12494 * targhooks.c (default_shift_truncation_mask): Likewise.
12496 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12497 Alan Hayward <alan.hayward@arm.com>
12498 David Sherwood <david.sherwood@arm.com>
12500 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12501 (widened_mode): Likewise.
12502 (expand_unop): Likewise.
12503 * ree.c (transform_ifelse): Likewise.
12504 (merge_def_and_ext): Likewise.
12505 (combine_reaching_defs): Likewise.
12506 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12508 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12509 Alan Hayward <alan.hayward@arm.com>
12510 David Sherwood <david.sherwood@arm.com>
12512 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12513 * combine.c (gen_lowpart_for_combine): Likewise.
12514 * dwarf2out.c (rtl_for_decl_location): Likewise.
12515 * final.c (alter_subreg): Likewise.
12516 * rtlhooks.c (gen_lowpart_general): Likewise.
12517 (gen_lowpart_if_possible): Likewise.
12519 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12520 Alan Hayward <alan.hayward@arm.com>
12521 David Sherwood <david.sherwood@arm.com>
12523 * calls.c (expand_call): Use subreg_lowpart_offset.
12524 * cse.c (cse_insn): Likewise.
12525 * regcprop.c (copy_value): Likewise.
12526 (copyprop_hardreg_forward_1): Likewise.
12528 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12531 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12532 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12533 =Q constraints to +Q and into insn condition add check
12534 that operands[0] and operands[1] are equal.
12535 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12536 =Q constraints to +Q and into insn condition add check
12537 that operands[0] is equal to either operands[1] or operands[2].
12540 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12541 instead of handling MINUS_EXPR twice (once for each argument),
12542 canonicalize operand order and handle just once, use rtype where
12543 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12546 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12547 any values of __C while still being pattern recognizable as a simple
12548 rotate instruction.
12550 2017-10-13 Richard Biener <rguenther@suse.de>
12552 PR tree-optimization/82451
12554 2017-10-02 Richard Biener <rguenther@suse.de>
12556 PR tree-optimization/82355
12557 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12558 a mapping for the enclosing loop but avoid generating one for
12559 the loop tree root.
12560 (copy_bb_and_scalar_dependences): Remove premature codegen
12561 error on PHIs in blocks duplicated into multiple places.
12562 * graphite-scop-detection.c
12563 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12564 in the region use it as loop and nest to analyze the DR in.
12565 (try_generate_gimple_bb): Likewise.
12566 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12567 (add_loop_constraints): For blocks in a loop not in the region
12568 create a dimension with a single iteration.
12569 * sese.h (gbb_loop_at_index): Remove assert.
12571 * cfgloop.c (loop_preheader_edge): For the loop tree root
12572 return the single successor of the entry block.
12573 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12574 Reset the SCEV hashtable and niters.
12575 * graphite-scop-detection.c
12576 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12577 assert that we only have POLYNOMIAL_CHREC that vary in loops
12578 contained in the region.
12579 (scop_detection::graphite_can_represent_expr): Adjust.
12580 (scop_detection::stmt_has_simple_data_refs_p): For loops
12581 not in the region set loop to NULL. The nest is now the
12582 entry edge to the region.
12583 (try_generate_gimple_bb): Likewise.
12584 * sese.c (scalar_evolution_in_region): Adjust for
12585 instantiate_scev change.
12586 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12587 Make nest parameter the edge into the region.
12588 (create_data_ref): Likewise.
12589 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12590 entry edge into a region and adjust instantiate_scev calls.
12591 (create_data_ref): Likewise.
12592 (graphite_find_data_references_in_stmt): Likewise.
12593 (find_data_references_in_stmt): Pass the loop preheader edge
12594 from the nest argument.
12595 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12596 parameter the edge into the region.
12597 (instantiate_parameters): Use the loop preheader edge as entry.
12598 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12600 (get_instantiated_value_entry): Make instantiate_below parameter
12601 the edge into the region.
12602 (instantiate_scev_name): Likewise. Adjust dominance checks,
12603 when we cannot use loop-based instantiation instantiate by
12604 walking use-def chains.
12605 (instantiate_scev_poly): Adjust.
12606 (instantiate_scev_binary): Likewise.
12607 (instantiate_scev_convert): Likewise.
12608 (instantiate_scev_not): Likewise.
12609 (instantiate_array_ref): Remove.
12610 (instantiate_scev_3): Likewise.
12611 (instantiate_scev_2): Likewise.
12612 (instantiate_scev_1): Likewise.
12613 (instantiate_scev_r): Do not blindly handle N-operand trees.
12614 Do not instantiate array-refs. Handle all constants and invariants.
12615 (instantiate_scev): Make instantiate_below parameter
12616 the edge into the region.
12617 (resolve_mixers): Use the loop preheader edge for the region
12618 parameter to instantiate_scev_r.
12619 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12621 2017-10-13 Richard Biener <rguenther@suse.de>
12623 PR tree-optimization/82525
12624 * graphite-isl-ast-to-gimple.c
12625 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12627 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12628 Fail code generation when we cannot represent the isl integer.
12629 (binary_op_to_tree): Elide modulo operations that are no-ops
12630 in the type we code generate. Remove now superfluous code
12633 2017-10-13 Richard Biener <rguenther@suse.de>
12635 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12636 (scop_detection::harmful_loop_in_region): Remove premature
12637 IV type restriction.
12638 (scop_detection::graphite_can_represent_scev): We can handle
12639 pointer IVs just fine.
12641 2017-10-13 Alan Modra <amodra@gmail.com>
12643 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12644 "Clobbers and Scratch Registers". Add paragraph on
12645 alternative to clobbers for scratch registers and OpenBLAS
12648 2017-10-13 Alan Modra <amodra@gmail.com>
12650 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12651 example of a memory input for a string of known length. Move
12652 commentary out of table. Add a number of new examples
12653 covering array memory inputs.
12655 2017-10-12 Martin Liska <mliska@suse.cz>
12657 PR tree-optimization/82493
12658 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12659 (test_range_functions): New function.
12660 (sbitmap_c_tests): Likewise.
12661 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12662 * selftest.h (sbitmap_c_tests): New function.
12664 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12666 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12668 * config/rs6000/amo.h: Fix spacing issue.
12670 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12673 * config/i386/i386.md (*ashl<mode>3_mask_1,
12674 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12675 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12678 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12680 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12681 (profile_probability): Set max_probability
12682 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12684 * profile-count.c (profile_probability::differs_from_p): Do not
12685 rely on max_probaiblity == 10000
12687 2017-10-12 Jeff Law <law@redhat.com>
12689 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12692 2017-10-12 Martin Sebor <msebor@redhat.com>
12696 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12697 (handle_alias_pairs): Call it.
12698 * common.opt (-Wattribute-alias): New option.
12699 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12700 * doc/invoke.texi (-Wattribute-alias): Document.
12702 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12705 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12707 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12709 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12710 (make_hard_regno_born, make_hard_regno_dead): Update
12711 bb_killed_pseudos and bb_gen_pseudos.
12713 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12715 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12717 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12719 * config/alpha/alpha.c (alpha_split_conditional_move):
12720 Use std::swap instead of manually swapping.
12721 (alpha_stdarg_optimize_hook): Ditto.
12722 (alpha_canonicalize_comparison): Ditto.
12724 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12726 * tree-loop-distribution.c (struct builtin_info): New struct.
12727 (struct partition): Refactor fields into struct builtin_info.
12728 (partition_free): Free struct builtin_info.
12729 (build_size_arg_loc, build_addr_arg_loc): Delete.
12730 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12731 information from struct builtin_info.
12732 (find_single_drs): New function refactored from classify_partition.
12733 Also moved builtin validity checks to this function.
12734 (compute_access_range, alloc_builtin): New functions.
12735 (classify_builtin_st, classify_builtin_ldst): New functions.
12736 (classify_partition): Refactor code into functions find_single_drs,
12737 classify_builtin_st and classify_builtin_ldst.
12738 (distribute_loop): Don't do runtime alias check when distributing
12740 (find_seed_stmts_for_distribution): New function.
12741 (pass_loop_distribution::execute): Refactor code finding seed
12742 stmts into above function. Support distribution for the innermost
12743 two-level loop nest. Adjust dump information.
12745 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12747 * tree-loop-distribution.c: Adjust the general comment.
12748 (NUM_PARTITION_THRESHOLD): New macro.
12749 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12750 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12751 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12752 in call to build_partition_graph.
12753 (finalize_partitions): New parameter. Make loop distribution more
12754 conservative by fusing more partitions.
12755 (distribute_loop): Don't do runtime alias check in case of loop nest
12757 (find_seed_stmts_for_distribution): New function.
12758 (prepare_perfect_loop_nest): New function.
12759 (pass_loop_distribution::execute): Refactor code finding seed stmts
12760 and loop nest into above functions. Support loop nest distribution.
12761 Adjust dump information accordingly.
12763 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12765 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12766 and set PTYPE_SEQUENTIAL for merged partition.
12768 2017-10-12 Richard Biener <rguenther@suse.de>
12770 PR tree-optimization/69728
12772 2017-09-19 Richard Biener <rguenther@suse.de>
12774 PR tree-optimization/69728
12775 * graphite-sese-to-poly.c (schedule_error): New global.
12776 (add_loop_schedule): Handle empty domain by failing the
12778 (build_original_schedule): Handle schedule_error.
12780 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12781 domain by returning an unchanged schedule.
12783 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12785 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12786 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12788 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12790 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12793 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12796 * expr.c (store_field): Don't optimize away bitsize == 0 store
12797 from CALL_EXPR with addressable return type.
12799 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12801 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12802 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12803 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12804 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12805 TARGET_ISEL instead of TARGET_ISEL<sel>.
12807 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12809 * config/rs6000/rs6000.c
12810 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12812 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12814 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12816 (reg_or_cint_operand): Fix comment.
12817 (reg_or_zero_operand): New predicate.
12818 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12819 * config/rs6000/rs6000.c (output_isel): Delete.
12820 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12821 instead of reg_or_cint_operand. Output instruction directly (not via
12823 (isel_unsigned_<mode>): Ditto.
12824 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12825 gpc_reg_operand. Add an instruction alternative for this. Output
12826 instruction directly.
12827 (*isel_reversed_unsigned_<mode>): Ditto.
12829 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12831 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12832 (TARGET_CANONICALIZE_COMPARISON): Define.
12834 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12837 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12838 Check whether the dest is REG before adding REG_EQUIV note.
12840 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12843 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12845 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12846 (make_hard_regno_born, make_hard_regno_dead): Update
12847 bb_killed_pseudos and bb_gen_pseudos.
12849 2017-10-11 Nathan Sidwell <nathan@acm.org>
12851 * incpath.h (enum incpath_kind): Name enum, prefix values.
12852 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12853 * incpath.c (heads, tails): Use INC_MAX.
12854 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12855 (merge_include_chains, split_quote_chain,
12856 register_include_chains): Update incpath_kind names.
12857 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12858 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12860 (add_framework_path, darwin_register_objc_includes): Likewise.
12861 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12863 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12865 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12866 Do not use float_operator operator predicate.
12867 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12868 * config/i386/predicates.md (float_operator): Remove predicate.
12870 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12872 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12873 (*jccxf_i387): Ditto.
12874 (*jcc<mode>_i387): Ditto.
12875 (*jccu<mode>_i387): Ditto.
12876 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12877 (*jcc_*_i387 splitters): Remove.
12878 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12879 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12880 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12883 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12885 * profile-count.h (slow_safe_scale_64bit): New function.
12886 (safe_scale_64bit): New inline.
12887 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12888 * profile-count.c: Include wide-int.h
12889 (slow_safe_scale_64bit): New.
12891 2017-10-11 Nathan Sidwell <nathan@acm.org>
12893 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12894 HAS_DECL_ASSEMBLER_NAME_P.
12895 * gimple-expr.c (gimple_decl_printable_name: Check
12896 HAS_DECL_ASSEMBLER_NAME_P too.
12897 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12898 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12899 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12900 * passes.c (rest_of_decl_compilation): Check
12901 HAS_DECL_ASSEMBLER_NAME_P too.
12902 * recog.c (verify_changes): Likewise.
12903 * tree-pretty-print.c (dump_decl_name): Likewise.
12904 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12906 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12907 (SET_DECL_ASSEMBLER_NAME): Use it.
12908 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12909 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12910 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12912 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12914 * config.gcc (i386, x86_64): Add extra objects.
12915 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12916 (ix86_min_insn_size): Declare.
12917 (ix86_issue_rate): Declare.
12918 (ix86_adjust_cost): Declare.
12919 (ia32_multipass_dfa_lookahead): Declare.
12920 (ix86_macro_fusion_p): Declare.
12921 (ix86_macro_fusion_pair_p): Declare.
12922 (ix86_bd_has_dispatch): Declare.
12923 (ix86_bd_do_dispatch): Declare.
12924 (ix86_core2i7_init_hooks): Declare.
12925 (ix86_atom_sched_reorder): Declare.
12926 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12927 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12928 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12929 (rip_relative_addr_p): Rename to ...
12930 (ix86_rip_relative_addr_p): ... this one; export.
12931 (memory_address_length): Update.
12932 (ix86_issue_rate): Move to x86-tune-sched.c.
12933 (ix86_flags_dependent): Move to x86-tune-sched.c.
12934 (ix86_agi_dependent): Move to x86-tune-sched.c.
12935 (exact_dependency_1): Move to x86-tune-sched.c.
12936 (exact_store_load_dependency): Move to x86-tune-sched.c.
12937 (ix86_adjust_cost): Move to x86-tune-sched.c.
12938 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12939 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12940 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12941 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12942 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12943 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12944 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12945 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12946 (min_insn_size): Rename to ...
12947 (ix86_min_insn_size): ... this one; export.
12948 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12949 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12950 (core2i7_first_cycle_multipass_backtrack): Move to
12951 x86-tune-sched-core.c.
12952 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12953 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12954 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12955 (ix86_avoid_jump_mispredicts): Update.
12956 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12957 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12958 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12959 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12960 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12961 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12962 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12963 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12964 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12965 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12966 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12967 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12968 (BIG): Move to ix86-tune-sched-bd.c.
12969 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12970 (enum insn_path): Move to ix86-tune-sched-bd.c.
12971 (get_mem_group): Move to ix86-tune-sched-bd.c.
12972 (is_cmp): Move to ix86-tune-sched-bd.c.
12973 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12974 (is_branch): Move to ix86-tune-sched-bd.c.
12975 (is_prefetch): Move to ix86-tune-sched-bd.c.
12976 (init_window): Move to ix86-tune-sched-bd.c.
12977 (allocate_window): Move to ix86-tune-sched-bd.c.
12978 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12979 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12980 (process_end_window): Move to ix86-tune-sched-bd.c.
12981 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12982 (find_constant): Move to ix86-tune-sched-bd.c.
12983 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12984 (has_immediate): Move to ix86-tune-sched-bd.c.
12985 (get_insn_path): Move to ix86-tune-sched-bd.c.
12986 (get_insn_group): Move to ix86-tune-sched-bd.c.
12987 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12988 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12989 (add_insn_window): Move to ix86-tune-sched-bd.c.
12990 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12991 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12992 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12993 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12994 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12995 (do_dispatch): Move to ix86-tune-sched-bd.c.
12996 (has_dispatch): Move to ix86-tune-sched-bd.c.
12997 * config/i386/t-i386: Add new object files.
12998 * config/i386/x86-tune-costs.h: New file.
12999 * config/i386/x86-tune-sched-atom.c: New file.
13000 * config/i386/x86-tune-sched-bd.c: New file.
13001 * config/i386/x86-tune-sched-core.c: New file.
13002 * config/i386/x86-tune-sched.c: New file.
13004 2017-10-11 Liu Hao <lh_mouse@126.com>
13006 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
13007 the generic version below instead.
13008 (should_colorize): Recognize Windows consoles as terminals
13010 * pretty-print.c [__MINGW32__] (write_all): New function.
13011 [__MINGW32__] (find_esc_head): Likewise.
13012 [__MINGW32__] (find_esc_terminator): Likewise.
13013 [__MINGW32__] (eat_esc_sequence): Likewise.
13014 [__MINGW32__] (mingw_ansi_fputs): New function that handles
13016 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
13019 2017-10-11 Richard Biener <rguenther@suse.de>
13021 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
13022 Properly call analyze_scalar_evolution with the loop of the stmt.
13024 2017-10-11 Richard Biener <rguenther@suse.de>
13026 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
13027 * tree-core.h (tree_base): Add chrec_var union member.
13028 * tree.h (CHREC_VAR): Remove.
13029 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
13030 * tree-chrec.h (build_polynomial_chrec): Adjust.
13031 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
13032 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
13034 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
13036 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
13037 * match.pd: ... here.
13038 ((T) X == (T) Y): Relax condition.
13040 2017-10-11 Bin Cheng <bin.cheng@arm.com>
13042 PR tree-optimization/82472
13043 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
13045 (break_alias_scc_partitions): Update postorder number.
13047 2017-10-11 Martin Liska <mliska@suse.cz>
13050 * opts.c (parse_no_sanitize_attribute): Do not use error_value
13052 * opts.h (parse_no_sanitize_attribute): Remove last argument.
13054 2017-10-11 Martin Liska <mliska@suse.cz>
13056 * print-rtl.c (print_insn): Move declaration of idbuf
13057 to same scope as name.
13059 2017-10-11 Martin Liska <mliska@suse.cz>
13064 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13065 volatile arguments.
13067 2017-10-11 Martin Liska <mliska@suse.cz>
13070 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13071 volatile arguments.
13073 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
13075 * config.gcc (default_gnu_indirect_function): Default to yes for
13076 arm*-*-linux* with glibc.
13078 2017-10-11 Richard Biener <rguenther@suse.de>
13080 * tree-scalar-evolution.c (get_scalar_evolution): Handle
13081 default-defs and types we do not want to analyze.
13082 (interpret_loop_phi): Replace unreachable code with an assert.
13083 (compute_scalar_evolution_in_loop): Remove and inline ...
13084 (analyze_scalar_evolution_1): ... here, replacing condition with
13085 what makes the intent clearer. Remove handling of cases
13086 get_scalar_evolution now handles.
13088 2017-10-10 Jim Wilson <wilson@tuliptree.org>
13090 PR rtl-optimization/81434
13091 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
13092 comment for main loop. In sched_group_found if, also add checks for
13093 pass and min_cost_group.
13095 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
13097 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
13098 (rs6000_insn_cost): New function.
13099 * config/rs6000/rs6000.md (cost): New attribute.
13101 2017-10-10 Jakub Jelinek <jakub@redhat.com>
13102 H.J. Lu <hongjiu.lu@intel.com>
13106 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
13107 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
13108 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
13109 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
13110 to OPTION_MASK_ISA_AVX512VL - builtins that have both
13111 OPTION_MASK_ISA_MMX and some other bit set require both
13112 mmx and the ISAs without the mmx bit.
13113 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
13114 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
13115 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
13116 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
13117 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
13118 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
13119 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
13120 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
13121 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
13122 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
13123 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
13124 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
13125 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
13126 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
13127 Add OPTION_MASK_ISA_MMX.
13129 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
13131 * config.gcc (armv7*-*-freebsd*): New target.
13132 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
13134 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
13136 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
13137 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
13140 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
13142 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
13143 a template parameter.
13144 (WIDE_INT_REF_FOR): Update accordingly.
13145 * tree.h (wi::int_traits <const_tree>): Delete.
13146 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
13147 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
13148 (wi::tree_to_wide_ref): New typedef.
13149 (wi::to_wide): New function.
13150 * calls.c (get_size_range): Use wi::to_wide when operating on
13151 trees as wide_ints.
13152 * cgraph.c (cgraph_node::create_thunk): Likewise.
13153 * config/i386/i386.c (ix86_data_alignment): Likewise.
13154 (ix86_local_alignment): Likewise.
13155 * dbxout.c (stabstr_O): Likewise.
13156 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
13157 * expr.c (const_vector_from_tree): Likewise.
13158 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
13159 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
13160 (fold_negate_expr_1, int_const_binop_1, const_binop)
13161 (fold_convert_const_int_from_real, optimize_bit_field_compare)
13162 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
13163 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
13164 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
13165 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
13166 (fold_not_const, round_up_loc): Likewise.
13167 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
13168 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
13169 (alloca_call_type): Likewise.
13170 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
13171 * godump.c (go_output_typedef): Likewise.
13172 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
13173 * internal-fn.c (get_min_precision): Likewise.
13174 * ipa-cp.c (ipcp_store_vr_results): Likewise.
13175 * ipa-polymorphic-call.c
13176 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
13177 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
13178 (ipa_modify_call_arguments): Likewise.
13179 * match.pd: Likewise.
13180 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
13181 * print-tree.c (print_node_brief, print_node): Likewise.
13182 * stmt.c (expand_case): Likewise.
13183 * stor-layout.c (layout_type): Likewise.
13184 * tree-affine.c (tree_to_aff_combination): Likewise.
13185 * tree-cfg.c (group_case_labels_stmt): Likewise.
13186 * tree-data-ref.c (dr_analyze_indices): Likewise.
13187 (prune_runtime_alias_test_list): Likewise.
13188 * tree-dump.c (dequeue_and_dump): Likewise.
13189 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
13190 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
13191 * tree-pretty-print.c (dump_generic_node): Likewise.
13192 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
13193 (simple_iv_with_niters): Likewise.
13194 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
13195 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
13196 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
13197 * tree-ssa-loop-niter.c (split_to_var_and_offset)
13198 (refine_value_range_using_guard, number_of_iterations_ne_max)
13199 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
13200 (get_cst_init_from_scev, record_nonwrapping_iv)
13201 (scev_var_range_cant_overflow): Likewise.
13202 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13203 * tree-ssa-pre.c (compute_avail): Likewise.
13204 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13205 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13206 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13207 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13208 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13209 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13210 (dump_case_nodes, try_switch_expansion): Likewise.
13211 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13212 (vect_do_peeling): Likewise.
13213 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13214 * tree-vect-stmts.c (vectorizable_load): Likewise.
13215 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13216 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13217 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13218 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13219 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13220 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13221 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13222 (evrp_dom_walker::before_dom_children): Likewise.
13223 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13224 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13225 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13226 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13227 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13228 (get_range_pos_neg): Likewise.
13229 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13230 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13231 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13232 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13233 * config/avr/avr.c (avr_fold_builtin): Likewise.
13234 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13235 * config/msp430/msp430.c (msp430_attr): Likewise.
13236 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13237 * config/powerpcspe/powerpcspe-c.c
13238 (altivec_resolve_overloaded_builtin): Likewise.
13239 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13240 (rs6000_expand_ternop_builtin): Likewise.
13241 * config/rs6000/rs6000-c.c
13242 (altivec_resolve_overloaded_builtin): Likewise.
13243 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13244 (rs6000_expand_ternop_builtin): Likewise.
13245 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13247 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13249 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13250 when copying loop nest with only one inner loop.
13252 2017-10-10 Richard Biener <rguenther@suse.de>
13254 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13255 blocks if SCEV is active.
13256 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13258 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13259 (scev_initialize): Assert we are not yet initialized.
13261 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13263 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13264 inner loop's exit stmt by making it always exit the loop, otherwise
13265 we would generate an infinite empty loop.
13267 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13269 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13270 renaming variables in new preheader if it's deleted.
13272 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13274 * tree-loop-distribution.c (struct partition): Remove unused field
13275 loops of the structure.
13276 (partition_alloc, partition_free): Ditto.
13277 (build_rdg_partition_for_vertex): Ditto.
13279 2017-10-09 Jeff Law <law@redhat.com>
13281 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13282 return type to match prototype and documentation.
13284 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13286 * config/rs6000/rs6000.c (processor_costs): Move to ...
13287 * config/rs6000/rs6000.h: ... here.
13288 (rs6000_cost): Declare.
13290 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13292 * except.c (setjmp_fn): New global variable.
13293 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13294 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13295 if DONT_USE_BUILTIN_SETJMP is defined.
13297 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13299 * target.def (insn_cost): New hook.
13300 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13301 * doc/tm.texi: Regenerate.
13302 * rtlanal.c (insn_cost): Use the new hook.
13304 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13306 * combine.c (combine_validate_cost): Compute the new insn_cost,
13307 not just pattern_cost.
13308 (try_combine): Adjust comment.
13310 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13312 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13314 * combine.c (uid_insn_cost): Adjust comment.
13315 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13317 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13318 * dse.c (find_shift_sequence): Ditto.
13319 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13320 (bb_valid_for_noce_process_p): Use pattern_cost.
13321 * rtl.h (insn_rtx_cost): Delete.
13322 (pattern_cost): New prototype.
13323 (insn_cost): New prototype.
13324 * rtlanal.c (insn_rtx_cost): Rename to...
13325 (pattern_cost): ... this.
13328 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13330 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13331 (*jcc<mode>_0_r_i387): Ditto.
13332 (*jccxf_r_i387): Ditto.
13333 (*jcc<mode>_r_i387): Ditto.
13334 (*jccu<mode>_r_i387): Ditto.
13335 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13336 (*jcc): Rename from *jcc_1.
13338 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13340 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13341 deferred rescans after the lvx/stvx recombination pre-pass.
13343 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13345 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13346 memory operation instruction support.
13347 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13348 (rs6000-ibm-aix[789]*): Likewise.
13349 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13350 Document new functions.
13352 2017-10-09 Richard Biener <rguenther@suse.de>
13354 PR tree-optimization/82397
13355 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13356 equality only for semantically equal trees.
13358 2017-10-09 Richard Biener <rguenther@suse.de>
13360 PR tree-optimization/82449
13361 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13362 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13363 allow constant addresses.
13364 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13367 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13369 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13372 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13375 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13378 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13381 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13383 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13386 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13387 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13389 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13391 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13392 (WI_BINARY_PREDICATE_RESULT): Likewise.
13393 (wi::binary_traits::operator_result): New type.
13394 (wi::binary_traits::predicate_result): Likewise.
13395 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13396 (generic_wide_int::operator==, generic_wide_int::operator!=)
13397 (generic_wide_int::operator&, generic_wide_int::and_not)
13398 (generic_wide_int::operator|, generic_wide_int::or_not)
13399 (generic_wide_int::operator^, generic_wide_int::operator+
13400 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13402 (operator~, unary operator-, operator==, operator!=, operator&)
13403 (operator|, operator^, operator+, binary operator-, operator*): New
13405 * expr.c (get_inner_reference): Use wi::bit_and_not.
13406 * fold-const.c (fold_binary_loc): Likewise.
13407 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13408 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13409 (bit_value_binop): Likewise.
13410 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13411 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13412 (extract_range_from_binary_expr_1): Likewise.
13413 (masked_increment): Likewise.
13414 (simplify_bit_ops_using_ranges): Likewise.
13416 2017-10-09 Martin Jambor <mjambor@suse.cz>
13419 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13420 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13421 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13423 (hsa_fixup_mov_insn_type): New function.
13424 (hsa_op_with_type::get_in_type): Use it.
13425 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13426 immediates in an assert.
13427 (hsa_op_with_type::extend_int_to_32bit): New method.
13428 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13429 types. Convert to dest type if necessary.
13430 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13431 (reg_for_gimple_ssa): Pass false as min32int to
13432 hsa_type_for_scalar_tree_type.
13433 (gen_hsa_addr): Fixup type when creating addresable temporary.
13434 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13435 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13436 necessary. Call hsa_fixup_mov_insn_type.
13437 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13438 extend operands and convert to dest type if necessary.
13439 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13440 to dest type if necessary.
13441 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13442 if conversion nt necessary and size matches.
13443 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13444 to dest type if necessary.
13445 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13446 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13448 (gen_hsa_clrsb): Likewise.
13449 (gen_hsa_ffs): Likewise.
13450 (gen_hsa_divmod): Extend operands and convert to dest type if
13452 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13454 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13456 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13457 Remove empty default arguments. Use a brace block as output
13459 (conditional return): Ditto.
13461 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13462 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13464 (group_ending_nop): Ditto.
13465 (doloop_end): Ditto.
13466 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13467 (splitters for those): Ditto.
13469 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13471 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13472 a conditional jump (and the compare for it) so that pc_rtx is the
13474 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13475 for the deleted and renamed ctr<mode>_internal[234] patterns.
13476 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13477 Delete second conditional return pattern.
13478 (ctr<mode>_internal2): Delete this second bdnz pattern.
13479 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13480 (ctr<mode>_internal4): Delete this second bdz pattern.
13482 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13484 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13485 (always_initialized_rtx_for_ssa_name_p): New predicate.
13486 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13487 (finish_out_of_ssa): Free new field of SA.
13488 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13489 * tree-ssa-coalesce.c: Include tree-ssa.h.
13490 (get_parm_default_def_partitions): Remove extern keyword.
13491 (get_undefined_value_partitions): New function.
13492 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13493 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13494 uninitialized bits.
13495 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13497 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13499 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13501 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13503 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13504 for targets that preffer 128bit.
13506 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13508 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13510 2017-10-08 Olivier Hainque <hainque@adacore.com>
13512 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13513 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13515 (thumb_set_return_address): Likewise.
13517 2017-10-08 Olivier Hainque <hainque@adacore.com>
13519 * common/config/arm/arm-common.c (arm_except_unwind_info):
13520 Handle DWARF2_UNWIND_INFO.
13522 2017-10-07 Michael Collison <michael.collison@arm.com>
13524 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13527 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13529 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13530 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13531 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13532 defined, force the creation of a new block for a dispatch label.
13534 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13536 * invoke.texi (Wsuggest-attribute=cold): Document.
13537 * common.opt (Wsuggest-attribute=cold): New
13538 * ipa-pure-const.c (warn_function_cold): New function.
13539 * predict.c (compute_function_frequency): Use it.
13540 * predict.h (warn_function_cold): Declare.
13542 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13544 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13547 2017-10-06 Martin Liska <mliska@suse.cz>
13549 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13550 keyword for member functions.
13551 (struct sanopt_tree_couple): New struct.
13552 (struct sanopt_tree_couple_hash): New function.
13553 (struct sanopt_ctx): Add new hash_map.
13554 (has_dominating_ubsan_ptr_check): New function.
13555 (record_ubsan_ptr_check_stmt): Likewise.
13556 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13557 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13558 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13560 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13563 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13564 aarch64_simd_valid_immediate on CONST_VECTORs.
13565 (aarch64_reg_or_bic_imm): Likewise.
13567 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13569 PR rtl-optimization/82396
13570 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13572 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13574 * graphite-dependences.c (scop_get_reads): Move code to...
13575 (scop_get_must_writes): Move code to...
13576 (scop_get_may_writes): Move code to...
13577 (scop_get_reads_and_writes): ... here.
13578 (scop_get_dependences): Call scop_get_reads_and_writes.
13580 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13582 PR tree-optimization/82434
13583 * fold-const.h (can_native_encode_type_p,
13584 can_native_encode_string_p): Remove.
13585 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13586 don't encode anything, just return what would be otherwise returned.
13587 (native_encode_fixed, native_encode_complex, native_encode_vector):
13589 (native_encode_string): Likewise. Inline by hand
13590 can_native_encode_string_p.
13591 (can_native_encode_type_p): Remove.
13592 (can_native_encode_string_p): Remove.
13593 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13594 STRING_CSTs using can_native_encode_string_p, test all
13595 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13596 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13597 argument from native_encode_expr.
13598 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13599 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13602 2017-10-06 Richard Biener <rguenther@suse.de>
13604 PR tree-optimization/82397
13605 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13606 operand_equal_p but rely on data_ref_compare_tree for detecting
13608 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13609 to match up with dr_group_sort_cmp.
13611 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13614 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13616 * config/s390/s390-builtin-types.def: Regenerate.
13618 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13621 * config/s390/s390-builtin-types.def: Regenerate.
13622 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13623 Change flag from B_VXE to B_VX.
13624 (s390_vec_min_dbl): Remove B_VXE flag.
13626 2017-10-06 Richard Biener <rguenther@suse.de>
13628 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13629 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13630 translate_isl_ast_to_gimple::is_valid_rename,
13631 translate_isl_ast_to_gimple::get_rename,
13632 translate_isl_ast_to_gimple::get_def_bb_for_const,
13633 translate_isl_ast_to_gimple::get_new_name,
13634 translate_isl_ast_to_gimple::collect_all_ssa_names,
13635 translate_isl_ast_to_gimple::copy_loop_phi_args,
13636 translate_isl_ast_to_gimple::collect_all_ssa_names,
13637 translate_isl_ast_to_gimple::copy_loop_phi_args,
13638 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13639 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13640 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13641 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13642 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13643 translate_isl_ast_to_gimple::copy_cond_phi_args,
13644 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13645 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13646 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13647 translate_isl_ast_to_gimple::rename_uses,
13648 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13649 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13650 (set_rename_for_each_def): Likewise.
13651 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13652 here. Handle rewriting SCEV analyzable uses here.
13653 (copy_bb_and_scalar_dependences): Generate code for PHI
13655 (graphite_regenerate_ast_isl): Adjust.
13656 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13657 (add_write, add_read): New functions.
13658 (build_cross_bb_scalars_def): Use it and simplify.
13659 (build_cross_bb_scalars_use): Likewise.
13660 (graphite_find_cross_bb_scalar_vars): Inline into...
13661 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13662 simulating out-of-SSA. Compute liveout and add dependencies.
13663 (build_scops): Force an empty entry block.
13664 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13666 (sese_build_liveouts): Declare.
13667 (sese_trivially_empty_bb_p): Likewise.
13668 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13669 compute liveout and debug_liveout.
13670 (sese_bad_liveouts_use): Remove.
13671 (sese_reset_debug_liveouts_bb): Likewise.
13672 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13673 (sese_build_liveouts): Build liveout and debug_liveout and store
13675 (new_sese_info): Adjust.
13676 (free_sese_info): Likewise.
13677 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13678 do not build liveout here.
13679 (move_sese_in_condition): Adjust region entry.
13680 (scev_analyzable_p): Match up with chrec_apply requirements.
13681 (sese_trivially_empty_bb_p): New.
13682 * tree-into-ssa.c (get_reaching_def): Properly support generating
13683 default-defs for incremental rewrite of anonymous names.
13685 2017-10-06 Richard Biener <rguenther@suse.de>
13687 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13688 precision do not perform modulo reduction.
13690 2017-10-06 Richard Biener <rguenther@suse.de>
13692 PR tree-optimization/82436
13693 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13694 conservatively choose the vectorization factor when checking
13695 whether we can perform the required load permutation.
13696 (vect_transform_slp_perm_load): Assert when we may not fail.
13698 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13700 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13701 message for incompatible -msdata=* and -mcall-* options.
13703 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13705 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13706 rate for post-reload scheduling.
13708 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13710 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13712 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13714 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13715 to improve monte carlo in scimark.
13717 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13719 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13720 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13721 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13722 pentium4_cost, nocona_cost): Set reassociation width to 1.
13723 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13724 width to 2 for fp operations and 1 otherwise.
13725 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13727 (atom_cost): Set reassociation width to 2.
13728 (slm_cost, generic_cost): Set fp reassociation width
13729 to 2 and 1 otherwise.
13730 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13731 (core_cost): Set fp reassociation width to 4 and vector to 2.
13732 (ix86_reassociation_width): Rewrite using cost table; special case
13733 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13734 and TARGET_AVX128_OPTIMAL.
13735 * config/i386/i386.h (processor_costs): Add
13736 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13737 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13738 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13739 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13740 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13741 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13743 2017-10-05 Nathan Sidwell <nathan@acm.org>
13745 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13747 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13749 * config/arm/arm.c (arm_test_fpu_data): New.
13750 (arm_run_selftests): Call arm_test_fpu_data.
13752 2017-10-04 Nathan Sidwell <nathan@acm.org>
13754 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13756 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13758 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13760 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13761 check type for aarch64_simd_valid_immediate.
13762 (aarch64_output_simd_mov_immediate): Update prototype.
13763 (aarch64_simd_valid_immediate): Update prototype.
13764 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13765 support for ORR-immediate.
13766 (and<mode>3): modified pattern to add support for BIC-immediate.
13767 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13768 now checks for valid immediate for BIC and ORR based on new enum
13770 (aarch64_output_simd_mov_immediate): Function now used to output
13771 BIC/ORR imm as well based on new enum argument.
13772 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13774 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13775 (aarch64_reg_or_bic_imm): Likewise.
13777 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13779 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13780 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13782 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13786 PR rtl-optimization/82396
13787 * haifa-sched.c (autopref_multipass_init): Simplify
13789 (autopref_rank_data): Simplify sort order.
13790 * sched-int.h (autopref_multipass_data_): Remove
13791 multi_mem_insn_p, min_offset and max_offset.
13793 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13795 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13797 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13799 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13800 vect_doubleint_cvt.
13802 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13804 * doc/sourcebuild.texi: Document vect_long_mult.
13806 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13808 PR tree-optimization/82413
13809 * fold-const.c (build_range_check): Use widest_int when comparing
13810 the maximum ETYPE value with HIGH.
13812 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13814 PR rtl-optimization/82396
13815 * haifa-sched.c (autopref_multipass_init): Simplify
13817 (autopref_rank_data): Simplify sort order.
13818 * sched-int.h (autopref_multipass_data_): Remove
13819 multi_mem_insn_p, min_offset and max_offset.
13821 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13823 PR tree-optimization/82381
13824 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13825 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13828 PR tree-optimization/82374
13829 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13830 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13831 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13832 current_function_decl to the new decl.
13834 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13836 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13837 helper macro for IEEE float128 hardware built-in functions.
13838 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13840 (TRUNCF128_ODD): Likewise.
13841 (ADDF128_ODD): Likewise.
13842 (SUBF128_ODD): Likewise.
13843 (MULF128_ODD): Likewise.
13844 (DIVF128_ODD): Likewise.
13845 (FMAF128_ODD): Likewise.
13846 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13847 UNSPEC_TRUNC_ROUND_TO_ODD.
13848 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13849 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13850 floating point round to odd instructions.
13851 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13852 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13853 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13854 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13855 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13856 (trunc<mode>sf2_hw): Change the truncate with round to odd
13857 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13858 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13859 to odd hardware instructions.
13860 (sub<mode>3_odd): Likewise.
13861 (mul<mode>3_odd): Likewise.
13862 (div<mode>3_odd): Likewise.
13863 (sqrt<mode>2_odd): Likewise.
13864 (fma<mode>4_odd): Likewise.
13865 (fms<mode>4_odd): Likewise.
13866 (nfma<mode>4_odd): Likewise.
13867 (nfms<mode>4_odd): Likewise.
13868 (trunc<mode>df2_odd): Change the truncate with round to odd
13869 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13871 * doc/extend.texi (PowerPC built-in functions): Update documentation
13872 for existing IEEE float128-bit built-in functions. Add built-in
13873 functions that generate the IEEE 128-bit floating point round to
13876 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13878 PR rtl-optimization/77729
13879 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13880 to (X&(C1&~C2))|C2 transformations.
13882 2017-10-03 Martin Jambor <mjambor@suse.cz>
13884 PR tree-optimization/82363
13885 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13886 mismatch, mark lacc written regardless of racc.
13888 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13890 PR tree-optimization/82381
13891 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13892 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13893 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13894 fallthrough into reassoc_stmt_dominates_stmt_p.
13897 * combine.c (combine_instructions): Don't combine in unreachable
13900 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13903 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13904 function to not use the have_cpu variable. Do not set cpu_index,
13905 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13906 or the default cpu.
13907 (rs6000_valid_attribute_p): Remove duplicate initializations of
13908 old_optimize and func_optimize.
13909 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13910 (rs6000_activate_target_options): Make global.
13911 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13914 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13916 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13917 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13918 Set *poffset to 0 and *psize and *pmax_size to -1 if
13919 *poffset + *psize overflows in HOST_WIDE_INT.
13921 PR tree-optimization/82387
13922 PR tree-optimization/82388
13923 PR tree-optimization/82389
13924 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13925 instead of live_bytes non-NULL.
13927 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13930 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13933 2017-10-02 Richard Biener <rguenther@suse.de>
13935 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13936 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13937 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13939 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13941 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13942 requested precision matches the type's.
13943 * calls.c (alloc_max_size): Calculate the new candidate size as
13944 a widest_int and use wi::to_widest when comparing it with the
13945 current candidate size.
13946 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13947 zero rather than integer_zero_node.
13948 * match.pd: Check for a no-op conversion before using wi::add
13949 rather than after. Use tree_to_uhwi when summing small shift
13950 counts into an unsigned int.
13952 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13953 Alan Hayward <alan.hayward@arm.com>
13954 David Sherwood <david.sherwood@arm.com>
13957 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13958 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13959 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13960 POINTER_AND_FP_REGS.
13962 2017-10-02 Richard Biener <rguenther@suse.de>
13964 PR tree-optimization/82355
13965 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13966 a mapping for the enclosing loop but avoid generating one for
13967 the loop tree root.
13968 (copy_bb_and_scalar_dependences): Remove premature codegen
13969 error on PHIs in blocks duplicated into multiple places.
13970 * graphite-scop-detection.c
13971 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13972 in the region use it as loop and nest to analyze the DR in.
13973 (try_generate_gimple_bb): Likewise.
13974 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13975 (add_loop_constraints): For blocks in a loop not in the region
13976 create a dimension with a single iteration.
13977 * sese.h (gbb_loop_at_index): Remove assert.
13979 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13981 * omp-expand.c (adjust_context_scope): New function.
13982 (expand_parallel_call): Call adjust_context_scope.
13984 2017-10-01 Jeff Law <law@redhat.com>
13986 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13987 dom_opt_dom_walker class with direct access to private members.
13988 Add comments. Call test_for_singularity.
13989 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13990 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13991 m_dummy_cond anymore.
13992 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13994 (pass_dominator:execute): Build the dummy_cond here and pass it
13995 to the dom_opt_dom_walker ctor.
13996 (test_for_singularity): New function.
13998 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13999 Maya Rashish <coypu@sdf.org>
14001 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
14002 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
14003 (alpha*-*-netbsd*) Use nbsd_tm_file.
14004 (arm*-*-netbsdelf*) Likewise.
14005 (i[34567]86-*-netbsdelf*) Likewise.
14006 (x86_64-*-netbsd*) Likewise.
14007 (mips*-*-netbsd*) Likewise.
14008 (powerpc-*-netbsd*) Likewise.
14009 (sh*-*-netbsd*) Likewise.
14010 (sparc-*-netbsdelf*) Likewise.
14011 (sparc64-*-netbsd*) Likewise.
14012 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
14014 (vax-*-netbsdelf*) Likewise.
14015 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
14016 (UINT_FAST8_TYPE) Likewise.
14017 (INT_FAST16_TYPE) Check CHAR_FAST16.
14018 (UINT_FAST16_TYPE) Likewise.
14020 2017-09-30 Jakub Jelinek <jakub@redhat.com>
14023 * config/i386/i386.md
14024 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
14025 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
14026 *divmodsi4_zext_2): New define_insn_and_split.
14027 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
14028 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
14029 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
14030 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
14031 New define_insn_and_split.
14032 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
14033 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
14034 operands[1] having DImode when mode is SImode.
14036 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
14037 always SImode for DIV and MOD in REG_EQUAL notes.
14039 2017-09-29 Yury Gribov <tetra2005@gmail.com>
14041 PR middle-end/82319
14042 * match.pd: Fix handling of NaNs in pattern.
14044 2017-09-29 Jeff Law <law@redhat.com>
14046 * sbitmap.c (bitmap_bit_in_range_p): New function.
14047 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
14048 * tree-ssa-dse.c (live_bytes_read): New function.
14049 (dse_classify_store): Ignore reads of dead bytes.
14051 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
14052 typos and whitespace errors.
14053 * config/i386/predicates.md (address_no_seg_operand): Likewise.
14054 * config/s390/s390.c (s390_emit_prologue): Likewise.
14056 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14059 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
14060 with a symbol for LRA.
14062 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14064 PR rtl-optimization/82338
14065 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
14067 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
14069 * genmodes.c (calc_wider_mode): Suppress qsort macro.
14070 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
14071 (qsort_chk): Declare.
14072 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
14073 (qsort_chk): New function.
14075 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14077 PR tree-optimization/82337
14078 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
14079 phi definition if the PHI result appears in an abnormal PHI.
14080 (find_basis_for_base_expr): Don't record a basis if the LHS of the
14081 basis appears in an abnormal PHI.
14083 2017-09-29 Richard Biener <rguenther@suse.de>
14085 * graphite-isl-ast-to-gimple.c
14086 (translate_isl_ast_to_gimple::set_codegen_error): New function.
14087 (binary_op_to_tree): Use it.
14088 (get_rename_from_scev): Likewise.
14089 (copy_loop_phi_nodes): Likewise.
14090 (copy_bb_and_scalar_dependences): Likewise.
14091 (translate_pending_phi_nodes): Likewise.
14093 2017-09-29 Jakub Jelinek <jakub@redhat.com>
14096 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
14097 for movabsq $(i32 << shift), r64.
14099 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
14101 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
14102 index when encoding %esp as %rsp to avoid 0x67 prefix.
14104 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
14106 * config/i386/i386.md (*movsf_internal, *movdf_internal):
14107 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
14109 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
14111 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
14112 Extensions with more than 16 double VFP registers.
14113 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
14114 to_clear_mask and all code related to it. Replace the remaining
14115 entry by a sbitmap and adapt code accordingly.
14117 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
14119 * brig-builtins.def: Change pure attributes to const.
14121 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14123 * config.gcc (default_gnu_indirect_function): Default to yes for
14124 sparc*-*-linux* with glibc.
14126 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14128 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
14129 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
14130 when creating .init_array and .fini_array sections with priority
14133 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
14136 * config/aarch64/aarch64.c
14137 (aarch64_builtin_support_vector_misalignment): Always return false
14138 when misalignment is unknown.
14140 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
14142 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
14143 this function to return false if the definition used by the swap
14144 instruction is artificial, or if the memory address from which the
14145 constant value is loaded is not represented by a base address held
14146 in a register or if the base address register is a frame or stack
14147 pointer. Additionally, return false if the base address of the
14148 loaded constant is a SYMBOL_REF but is not considered to be a
14150 (replace_swapped_load_constant): New function.
14151 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
14152 loaded constant vector with a load of a swapped constant vector.
14154 2017-09-27 Carl Love <cel@us.ibm.com>
14156 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14157 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14158 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14161 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
14163 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
14164 first, always call autopref_rank_data otherwise.
14166 2017-09-27 Richard Biener <rguenther@suse.de>
14168 * graphite-scop-detection.c (find_scop_parameters): Move
14169 loop bound handling ...
14170 (gather_bbs::before_dom_children): ... here, avoiding the need
14171 to build scop_info->loop_nest.
14172 (record_loop_in_sese): Remove.
14173 * sese.h (sese_info_t::loop_nest): Remove.
14174 * sese.c (new_sese_info): Do not allocate loop_nest.
14175 (free_sese_info): Do not free loop_nest.
14177 2017-09-27 Jakub Jelinek <jakub@redhat.com>
14180 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
14181 lhs from calls if the lhs has addressable type.
14183 2017-09-27 Richard Biener <rguenther@suse.de>
14185 * graphite.h (scop::max_alias_set): New member.
14186 * graphite-scop-detection.c: Remove references to non-existing
14187 --param in comments.
14188 (build_alias_sets): Record the maximum alias set used for drs.
14189 (build_scops): Support zero as unlimited for
14190 --param graphite-max-arrays-per-scop.
14191 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
14192 and inline into ...
14193 (build_poly_sr_1): ... here. Compute alias set based on the
14194 maximum alias set used for drs rather than
14195 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
14197 2017-09-27 Richard Biener <rguenther@suse.de>
14199 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
14200 --param loop-block-tile-size=0 to disable tiling.
14202 2017-09-27 Richard Biener <rguenther@suse.de>
14204 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14205 (graphite-max-nb-scop-params): Document special value zero.
14206 * domwalk.h (dom_walker::STOP): New symbolical constant.
14207 (dom_walker::dom_walker): Add optional parameter for bb to
14209 (dom_walker::~dom_walker): Declare.
14210 (dom_walker::before_dom_children): Document STOP return value.
14211 (dom_walker::m_user_bb_to_rpo): New member.
14212 (dom_walker::m_bb_to_rpo): Likewise.
14213 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14214 mapping here if not provided by the user.
14215 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14216 provided by the user.
14217 (dom_walker::STOP): Define.
14218 (dom_walker::walk): Do not compute bb to RPO mapping here.
14219 Support STOP return value from before_dom_children to stop
14221 * graphite-optimize-isl.c (optimize_isl): If the schedule
14222 is the same still generate code if -fgraphite-identity
14223 or -floop-parallelize-all are given.
14224 * graphite-scop-detection.c: Include cfganal.h.
14225 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14227 (gather_bbs::before_dom_children): Return STOP for BBs
14229 (build_scops): Compute bb to RPO mapping and pass it to
14230 the domwalk. Treat --param graphite-max-nb-scop-params=0
14231 as not limiting the number of params.
14232 * graphite.c (graphite_initialize): Remove limit on the
14233 number of basic-blocks in a function.
14234 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14235 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14236 default value of 10.
14238 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14240 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14241 Adjust code to eliminate needing to do the shift right 32-bits
14242 operation after XSCVDPSPN.
14244 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14246 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14247 ((X / Y) != 0 -> X >= Y): Likewise.
14249 2017-09-26 Carl Love <cel@us.ibm.com>
14251 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14252 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14253 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14254 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14255 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14256 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14257 definitions and overloading.
14258 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14259 statement for P9V_BUILTIN_XST_LEN_R.
14260 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14261 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14262 define_expand and define_insn for the instructions and builtins.
14263 * doc/extend.texi: Update the built-in documentation file for the new
14264 built-in functions.
14265 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14266 define_insn for the instructions
14268 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14271 * gcc/config/netbsd-protos.h: New file.
14272 * gcc/config/netbsd.c: New file.
14273 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14274 * gcc/config/t-netbsd: New file.
14275 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14276 (tmake_file) Add t-netbsd.
14277 (extra_objs) Add netbsd.o.
14279 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14283 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14285 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14287 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14288 sign extension from a vector register to a GPR by doing a 32-bit
14289 direct move and then an EXTSW.
14290 (extendsi<mode>2 splitter): Likewise.
14291 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14292 right or vector extract after doing XSCVDPSPN. Use
14293 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14295 (movdi_from_sf_zero_ext): Likewise.
14296 (reload_gpr_from_vsxsf): Likewise.
14297 (p8_mfvsrd_4_disf): Delete, no longer used.
14298 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14299 then needing to move the SFmode to a GPR to use the XSCVDPSP
14300 instruction instead of FRSP and XSCVDPSPN.
14301 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14302 it is adjacent to the other XSCVSPDP insns.
14303 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14304 SFmode to be in traditional Altivec registers.
14305 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14306 (vsx_xscvspdpn): Likewise.
14307 (vsx_xscvspdpn_scalar): Likewise.
14309 2017-09-26 Martin Jambor <mjambor@suse.cz>
14311 * tree-sra.c (compare_access_positions): Put integral types first,
14312 stabilize sorting of integral types, remove conditions putting
14313 non-full-precision integers last.
14314 (sort_and_splice_var_accesses): Disable scalarization if a
14315 non-integert would be represented by a non-full-precision integer.
14317 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14319 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14320 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14321 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14322 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14323 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14324 conditionals inside the function instead of around it. Call
14325 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14326 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14328 2017-09-26 Richard Biener <rguenther@suse.de>
14330 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14332 (scop_detection::build_scop_breadth): ... this. Removed.
14333 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14334 (scop_detection::harmful_stmt_in_bb): Likewise.
14335 (scop_detection::graphite_can_represent_stmt): Likewise.
14336 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14337 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14338 (scop_detection::can_represent_loop_1): ... this. Removed.
14339 (scop_detection::harmful_loop_in_region): Simplify after inlining
14340 the above and remove more quadraticness.
14341 (build_scops): Adjust.
14342 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14345 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14348 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14349 REGNO (base) == SP_REG if base is a REG.
14351 PR middle-end/35691
14352 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14353 if it is different SSA_NAME.
14354 (optimize_range_tests_cmp_bitwise): New function.
14355 (optimize_range_tests): Call it.
14357 2017-09-26 Richard Biener <rguenther@suse.de>
14359 PR tree-optimization/82321
14360 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14361 for the def being inside the loop.
14363 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14365 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14367 * config/s390/s390-builtins.def: Fix constraint on op4.
14369 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14371 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14372 independent expanders.
14373 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14374 ("vec_ordered", "vec_unordered"): New expanders.
14376 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14378 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14381 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14383 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14384 vec_unpacks_lo_v16qi.
14385 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14387 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14389 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14390 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14391 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14393 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14395 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14397 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14399 ("*vec_slb<mode>"): New insn pattern.
14400 ("vec_shr_<mode>"): New expander.
14401 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14402 and force the shift count operand to V16QImode.
14403 ("vec_srb<mode>"): Set shift count mode to V16QI.
14405 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14407 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14408 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14409 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14411 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14414 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14416 2017-09-26 Richard Biener <rguenther@suse.de>
14418 PR tree-optimization/82320
14419 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14422 2017-09-25 Jeff Law <law@redhat.com>
14424 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14425 prototype for new argument.
14426 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14427 mostly extracted from rs6000_emit_allocate_stack.
14428 (rs6000_emit_probe_stack_range_stack_clash): New function.
14429 (rs6000_emit_allocate_stack): Call
14430 rs6000_emit_probe_stack_range_stack_clash as needed.
14431 (rs6000_emit_probe_stack_range): Add additional argument
14432 to call to gen_probe_stack_range{si,di}.
14433 (output_probe_stack_range): New.
14434 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14435 (output_probe_stack_range_stack_clash): New.
14436 (rs6000_emit_prologue): Emit notes into dump file as requested.
14437 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14438 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14439 Add additional operand and pass it to output_probe_stack_range.
14441 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14443 PR tree-optimization/82163
14444 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14445 (checking_verify_loop_closed_ssa): New parameter.
14446 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14447 (check_loop_closed_ssa_stmt): Delete.
14448 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14449 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14450 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14453 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14455 * brig-builtins.def: Treat HSAIL barrier builtins as
14456 setjmp/longjump style functions.
14458 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14460 * target.def (constant_alignment): New hook.
14461 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14462 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14463 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14464 * doc/tm.texi: Regenerate.
14465 * targhooks.h (default_constant_alignment): Declare.
14466 (constant_alignment_word_strings): Likewise.
14467 * targhooks.c (default_constant_alignment): New function.
14468 (constant_alignment_word_strings): Likewise.
14469 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14470 instead of CONSTANT_ALIGNMENT.
14471 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14472 (force_const_mem): Likewise.
14473 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14474 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14475 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14476 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14477 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14479 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14480 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14481 constant_alignment_word_strings.
14482 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14483 (CONSTANT_ALIGNMENT): Likewise.
14484 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14485 (arm_constant_alignment): New function.
14486 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14487 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14488 constant_alignment_word_strings.
14489 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14490 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14491 constant_alignment_word_strings.
14492 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14493 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14494 (cris_constant_alignment): New function.
14495 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14496 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14497 (epiphany_constant_alignment): New function.
14498 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14499 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14500 constant_alignment_word_strings.
14501 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14502 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14503 constant_alignment_word_strings.
14504 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14505 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14506 constant_alignment_word_strings.
14507 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14508 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14509 * config/i386/i386.c (ix86_constant_alignment): Make static.
14510 Use the same interface as the target hook.
14511 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14512 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14513 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14514 constant_alignment_word_strings.
14515 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14516 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14517 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14518 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14519 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14520 constant_alignment_word_strings.
14521 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14522 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14523 constant_alignment_word_strings.
14524 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14525 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14526 constant_alignment_word_strings.
14527 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14528 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14530 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14531 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14532 * config/mips/mips.c (mips_constant_alignment): New function.
14533 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14534 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14535 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14536 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14537 (mmix_constant_alignment): Make static. Use the same interface
14538 as the target hook.
14539 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14540 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14541 constant_alignment_word_strings.
14542 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14543 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14544 constant_alignment_word_strings.
14545 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14546 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14547 constant_alignment_word_strings.
14548 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14549 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14550 (rs6000_constant_alignment): New function.
14551 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14552 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14553 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14554 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14555 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14556 (rs6000_constant_alignment): New function.
14557 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14558 * config/s390/s390.c (s390_constant_alignment): New function.
14559 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14560 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14561 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14562 constant_alignment_word_strings.
14563 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14564 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14565 (sparc_constant_alignment): New function.
14566 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14567 * config/spu/spu.c (spu_constant_alignment): New function.
14568 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14569 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14570 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14571 constant_alignment_word_strings.
14572 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14573 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14574 constant_alignment_word_strings.
14575 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14576 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14577 constant_alignment_word_strings.
14578 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14579 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14580 (visium_constant_alignment): New function.
14581 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14582 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14583 (xtensa_constant_alignment): New function.
14584 * system.h (CONSTANT_ALIGNMENT): Poison.
14586 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14588 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14589 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14590 (rs6000_builtin_valid_without_lhs): New helper function.
14591 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14592 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14594 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14596 * target.h (vec_perm_indices): Use unsigned short rather than
14598 (auto_vec_perm_indices): Likewise.
14599 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14600 Use unsigned int rather than unsigned char.
14601 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14603 2017-09-25 Richard Biener <rguenther@suse.de>
14605 * cfgloop.h (sort_sibling_loops): Declare.
14606 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14607 (sort_sibling_loops): New function sorting the sibling loop list
14609 * graphite.c (graphite_transform_loops): Sort sibling loops.
14611 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14613 * target.def (vec_perm_const_ok): Change sel parameter to
14615 * optabs-query.c (can_vec_perm_p): Update accordingly.
14616 * doc/tm.texi: Regenerate.
14617 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14618 auto_vec_perm_indices and remove separate nelt field.
14619 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14620 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14621 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14622 (aarch64_expand_vec_perm_const): Update accordingly.
14623 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14624 to vec_perm_indices.
14625 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14626 auto_vec_perm_indices and remove separate nelt field.
14627 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14628 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14629 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14631 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14632 to vec_perm_indices.
14633 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14634 sel to vec_perm_indices.
14635 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14636 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14637 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14639 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14642 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14645 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14646 on the FUNCTION_DECL function context if it has a DIE that is a
14649 2017-09-25 Richard Biener <rguenther@suse.de>
14651 PR tree-optimization/82285
14652 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14655 2017-09-25 Tom de Vries <tom@codesourcery.com>
14659 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14662 2017-09-25 Richard Biener <rguenther@suse.de>
14664 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14665 ISL errors other than isl_error_quota happen. Dump if the
14666 schedule is the same.
14667 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14668 errors instead of aborting inside ISL.
14670 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14673 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14674 of libgcc_eh for m64.
14675 * config/i386/darwin64.h: Likewise.
14677 2017-09-25 Richard Biener <rguenther@suse.de>
14679 PR middle-end/82144
14680 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14681 attribute for incomplete types nor twice for complete ones.
14683 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14686 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14687 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14690 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14693 * config/i386/i386.opt (mprefer-avx256): Use
14694 ix86_target_flags variable.
14695 * config/i386/i386.c (ix86_target_string): Move
14696 -mprefer-avx256 to flag2_opts.
14698 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14700 PR middle-end/35691
14701 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14702 and x != -1 | y != -1 into (x & y) != -1.
14704 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14706 * config.gcc: Add new case statement to set
14707 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14708 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14709 s390x-*-linux* case statements. Added aarch64 to the list of
14710 supported architectures.
14712 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14714 PR tree-optimization/82289
14715 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14716 STMT_VINFO_RELEVANT_P.
14718 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14719 Alan Hayward <alan.hayward@arm.com>
14720 David Sherwood <david.sherwood@arm.com>
14722 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14723 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14724 (extract_range_from_binary_expr_1): Don't call
14725 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14727 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14728 Alan Hayward <alan.hayward@arm.com>
14729 David Sherwood <david.sherwood@arm.com>
14731 * target.def (preferred_vector_alignment): New hook.
14732 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14734 * doc/tm.texi: Regenerate.
14735 * targhooks.h (default_preferred_vector_alignment): Declare.
14736 * targhooks.c (default_preferred_vector_alignment): New function.
14737 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14739 (DR_TARGET_ALIGNMENT): New macro.
14740 (aligned_access_p): Update commentary.
14741 (vect_known_alignment_in_bytes): New function.
14742 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14744 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14745 Calculate the misalignment based on the target alignment rather than
14747 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14748 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14749 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14750 the target alignment, rather than masking the element misalignment
14751 with the number of elements in a vector. Also use the target
14752 alignment when calculating the maximum number of peels.
14753 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14754 instead of TYPE_ALIGN_UNIT.
14755 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14756 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14757 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14758 (vect_create_data_ref_ptr): Likewise.
14759 (vect_setup_realignment): Realign by ANDing with
14760 -DR_TARGET_MISALIGNMENT.
14761 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14762 the number of peels based on DR_TARGET_ALIGNMENT.
14763 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14764 with the guaranteed alignment boundary when deciding whether
14766 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14767 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14768 (ensure_base_align): Remove stmt_info parameter. Get the
14769 target base alignment from DR_TARGET_ALIGNMENT.
14770 (vectorizable_store): Update call accordingly. Interpret
14771 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14773 (vectorizable_load): Likewise.
14775 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14776 Alan Hayward <alan.hayward@arm.com>
14777 David Sherwood <david.sherwood@arm.com>
14779 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14780 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14781 (vect_enhance_data_refs_alignment): Likewise.
14783 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14785 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14786 error. Only quit immediately if parsing is complete.
14787 (BEGIN): Initialize fatal_err and parse_done.
14788 (begin fpu, end fpu): Check number of arguments.
14789 (begin arch, end arch): Likewise.
14790 (begin cpu, end cpu): Likewise.
14791 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14792 (optalias): Likewise.
14794 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14796 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14797 * config/arm/arm-isa.h: Delete. Move definitions to ...
14798 * arm-cpus.in: ... here. Use new feature and fgroup values.
14799 * config/arm/arm.c (arm_option_override): Use lower case for feature
14801 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14802 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14803 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14805 (print_isa_bits_for): New function.
14806 (gen_isa): New function.
14807 (gen_comm_data): Use print_isa_bits_for.
14808 (define feature): New keyword.
14809 (define fgroup): New keyword.
14810 * config/arm/t-arm (TM_H): Remove.
14811 (GTM_H): Add arm-isa.h.
14812 (arm-isa.h): Add rule to generate file.
14813 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14814 case for feature bit names.
14816 2017-09-22 Richard Biener <rguenther@suse.de>
14818 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14820 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14821 print of no dependency loops ...
14822 * graphite.c (graphite_transform_loops): ... here.
14823 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14825 (same_close_phi_node, remove_duplicate_close_phi,
14826 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14827 (canonicalize_loop_closed_ssa): ... here and simplify.
14828 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14829 (optimize_isl): Use dump_printf_loc to tell when we stopped
14830 optimizing because of an ISL timeout.
14832 2017-09-22 Richard Biener <rguenther@suse.de>
14834 PR tree-optimization/82291
14835 * tree-if-conv.c (predicate_mem_writes): Make sure to
14836 remove writes in blocks predicated with false.
14838 2017-09-22 Richard Biener <rguenther@suse.de>
14840 * sese.c: Include cfganal.h.
14841 (if_region_set_false_region): Remove.
14842 (create_if_region_on_edge): Likewise.
14843 (move_sese_in_condition): Re-implement without destroying
14846 2017-09-22 Richard Biener <rguenther@suse.de>
14848 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14849 Verify both BBs contain loop PHI nodes before dispatching to
14850 copy_loop_phi_args.
14851 (graphite_regenerate_ast_isl): Do not recompute dominators,
14852 do not verify three times. Restructure for clarity.
14853 * graphite-scop-detection.c (same_close_phi_node,
14854 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14855 defined_in_loop_p, canonicalize_loop_closed_ssa,
14856 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14857 checking and SSA rewrite, move to ...
14858 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14859 (graphite_initialize): Do not pass in ctx, do not reset the
14860 SCEV cache, compute only dominators.
14861 (graphite_transform_loops): Allocate ISL ctx after
14862 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14863 Maintain post-dominators only around build_scops.
14864 * sese.c (if_region_set_false_region): Make static. Free
14865 and recompute dominators.
14866 (move_sese_in_condition): Assert we don't get called with
14867 post-dominators computed.
14868 * sese.h (if_region_set_false_region): Remove.
14870 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14872 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14873 mode attribute for TARGET_AVX512VL.
14875 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14877 * config/i386/i386.opt (mprefer-avx256): New option.
14878 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14880 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14881 for TARGET_PREFER_AVX256.
14882 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14884 2017-09-21 Jeff Law <law@redhat.com>
14886 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14887 Fix dump output if the only stack space is for pushed registers.
14889 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14891 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14894 2017-09-21 Martin Sebor <msebor@redhat.com>
14897 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14898 code (in C++) or code that triggers warnings.
14900 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14902 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14904 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14906 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14907 * sched-rgn.c: Ditto.
14908 * sel-sched-ir.c: Ditto.
14910 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14912 * toplev.h (set_random_seed): Adjust return type.
14913 * toplev.c (init_local_tick): Move eager initialization of random_seed
14914 to get_random_seed. Adjust comment.
14915 (init_random_seed): Inline to get_random_seed, delete.
14916 (get_random_seed): Initialize random_seed lazily.
14917 (set_random_seed): Do not return previous value.
14918 (print_switch_value): Do not call get_random_seed.
14920 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14922 * cgraph.c (delete_function_version): New, broken out from...
14923 (cgraph_node::delete_function_version): ...here. Rename to
14924 cgraph_node::delete_function_version_by_decl. Update all uses.
14925 (cgraph_node::remove): Call delete_function_version.
14927 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14930 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14931 VAR_DECLs to which addressable non-volatile parameters are mapped
14932 and for id->retvar after the return value assignment. Clear
14933 id->retval and id->retbnd after inlining.
14935 2017-09-21 Richard Biener <rguenther@suse.de>
14937 PR tree-optimization/82276
14938 PR tree-optimization/82244
14939 * tree-vrp.c (build_assert_expr_for): Set
14940 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14942 (remove_range_assertions): Revert earlier change.
14944 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14947 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14949 2017-09-21 Richard Biener <rguenther@suse.de>
14951 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14952 Restore valid IL after code generation errors.
14953 * graphite.c (graphite_transform_loops): Diagnose code
14954 generation issues as MSG_MISSED_OPTIMIZATION and continue
14955 with processing SCOPs.
14957 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14958 Alan Hayward <alan.hayward@arm.com>
14959 David Sherwood <david.sherwood@arm.com>
14961 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14962 rather than choosing between plus_constant and gen_rtx_<CODE>.
14963 * expr.c (emit_push_insn): Likewise.
14964 (expand_expr_real_2): Likewise.
14966 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14967 Alan Hayward <alan.hayward@arm.com>
14968 David Sherwood <david.sherwood@arm.com>
14970 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14972 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14973 Alan Hayward <alan.hayward@arm.com>
14974 David Sherwood <david.sherwood@arm.com>
14976 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14977 calling tree_to_uhwi.
14979 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14980 Alan Hayward <alan.hayward@arm.com>
14981 David Sherwood <david.sherwood@arm.com>
14983 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14984 INTEGER_CST rather than a negative test for ADDR_EXPR.
14986 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14987 Alan Hayward <alan.hayward@arm.com>
14988 David Sherwood <david.sherwood@arm.com>
14990 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14991 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14993 2017-09-21 Richard Biener <rguenther@suse.de>
14995 PR tree-optimization/71351
14996 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14997 graphite_create_new_loop_guard): Remove, fold remaining parts
14999 (translate_isl_ast_node_for): ... here and simplify.
15001 2017-09-21 Jakub Jelinek <jakub@redhat.com>
15004 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
15005 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
15006 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
15007 alternative always use QI mode, for -Os imov (=R,R) alternative
15008 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
15011 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15012 Jeff Law <law@redhat.com>
15014 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
15015 (allocate_stack_space): New function, partially extracted from
15016 s390_emit_prologue.
15017 (s390_emit_prologue): Track offset to most recent stack probe.
15018 Code to allocate space moved into allocate_stack_space.
15019 Dump actions when no stack is allocated.
15020 (s390_prologue_plus_offset): New function.
15021 (s390_emit_stack_probe): Likewise.
15023 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
15025 * common.opt (Wa, Wl, Wp, g, gz=): Add
15027 (gno-column-info): Remove.
15028 (gcolumn-info): Drop RejectNegative.
15029 (gno-): New prefix.
15030 (gno-record-gcc-switches): Remove.
15031 (grecord-gcc-switches): Drop RejectNegative.
15032 (gno-split-dwarf): Remove.
15033 (gsplit-dwarf): Drop RejectNegative.
15034 (gno-strict-dwarf): Remove.
15035 (gstrict-dwarf): Drop RejectNegative.
15036 * config/darwin.opt (gfull, gused): Add RejectNegative.
15037 * dwarf2out.c (gen_producer_string): Drop
15038 gno-record-gcc-switches handler.
15039 * optc-gen.awk: Add g to prefixes with negative forms.
15040 * opts-common.c (remapping_prefix_p): New.
15041 (find_opt): Check it.
15042 (generate_canonical_option): Test g prefix.
15043 (option_map): Add -gno- mapping.
15044 (add_misspelling_candidates): Check remapping_prefix_p.
15046 2017-09-20 Jeff Law <law@redhat.com>
15048 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
15049 thinko in stack clash protection support.
15051 * explow.c (compute_stack_clash_protection_loop_data): Use
15052 CONST_INT_P instead of explicit test. Verify object is a
15053 CONST_INT_P before looking at INTVAL.
15054 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
15055 instead of explicit test.
15057 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
15060 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
15061 address instead of to r1 and r11.
15063 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
15065 * config.gcc: Support "knm".
15066 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
15067 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15069 * config/i386/i386.c (m_KNM): Define.
15070 (processor_target_table): Add "knm".
15072 (ix86_option_override_internal): Add "knm".
15073 (ix86_issue_rate): Add PROCESSOR_KNM.
15074 (ix86_adjust_cost): Ditto.
15075 (ia32_multipass_dfa_lookahead): Ditto.
15076 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
15077 (fold_builtin_cpu): Add M_INTEL_KNM.
15078 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
15079 (processor_type): Add PROCESSOR_KNM.
15080 * config/i386/x86-tune.def: Add m_KNM.
15081 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
15083 2017-09-20 Richard Biener <rguenther@suse.de>
15085 PR tree-optimization/80213
15086 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
15087 are allowed in empty BBs as well.
15088 (canonicalize_loop_closed_ssa): Also look for other complex
15090 (scop_detection::get_sese): Include the loop-closed PHI block
15092 (scop_detection::merge_sese): Remove code adding extra blocks.
15093 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
15094 (build_scops): Assert the final returned scop is invalid.
15096 2017-09-20 Richard Biener <rguenther@suse.de>
15098 PR tree-optimization/82264
15099 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
15101 (vn_phi_lookup): Likewise.
15102 (vn_phi_insert): Likewise.
15104 2017-09-20 Jakub Jelinek <jakub@redhat.com>
15106 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
15107 that fits into uhwi or shwi, add DW_AT_const_value regardless
15108 of early_dwarf without going through RTL, using add_AT_unsigned
15111 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
15112 (DEBUG_LTO_ABBREV_SECTION): Likewise.
15113 (DEBUG_LTO_MACINFO_SECTION): Likewise.
15114 (DEBUG_MACRO_SECTION): Likewise.
15115 (DEBUG_LTO_MACRO_SECTION): Likewise.
15116 (DEBUG_STR_DWO_SECTION): Likewise.
15117 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
15118 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
15119 (DEBUG_LTO_DWO_LINE_SECTION): Define.
15120 (DEBUG_LTO_LINE_STR_SECTION): Define.
15121 (init_sections_and_labels): Initialize debug_line_str_section
15122 variable. Initialize debug_loc_section for -gdwarf-5 to
15123 DEBUG_LOCLISTS_SECTION. Formatting fixes.
15125 2017-09-20 Richard Biener <rguenther@suse.de>
15127 * graphite-sese-to-poly.c (extract_affine): Properly handle
15128 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
15130 2017-09-20 Richard Biener <rguenther@suse.de>
15132 PR tree-optimization/81373
15133 * graphite-scop-detection.c (build_cross_bb_scalars_def):
15134 Force SESE live-out defs to be handled even if they are
15137 2017-09-19 Jeff Law <law@redhat.com>
15139 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
15140 nothing for stack adjustments with REG_STACK_CHECK.
15141 * sched-deps.c (parse_add_or_inc): Reject insns with
15142 REG_STACK_CHECK from dependency breaking.
15143 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
15144 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
15145 * reg-notes.def (STACK_CHECK): New note.
15147 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
15148 (ix86_expand_prologue): Dump stack clash info as needed.
15149 Call ix86_adjust_stack_and_probe_stack_clash as needed.
15151 * function.c (dump_stack_clash_frame_info): New function.
15152 * function.h (dump_stack_clash_frame_info): Prototype.
15153 (enum stack_clash_probes): New enum.
15155 * config/alpha/alpha.c (alpha_expand_prologue): Also check
15156 flag_stack_clash_protection.
15157 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
15158 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
15159 (arm_frame_pointer_required): Likewise.
15160 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
15161 (ia64_expand_prologue): Likewise.
15162 * config/mips/mips.c (mips_expand_prologue): Likewise.
15163 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
15164 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15165 (sparc_flat_expand_prologue): Likewise.
15166 * config/spu/spu.c (spu_expand_prologue): Likewise.
15168 * explow.c: Include "params.h".
15169 (anti_adjust_stack_and_probe_stack_clash): New function.
15170 (get_stack_check_protect): Likewise.
15171 (compute_stack_clash_protection_loop_data): Likewise.
15172 (emit_stack_clash_protection_loop_start): Likewise.
15173 (emit_stack_clash_protection_loop_end): Likewise.
15174 (allocate_dynamic_stack_space): Use get_stack_check_protect.
15175 Use anti_adjust_stack_and_probe_stack_clash.
15176 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
15177 (emit_stack_clash_protection_loop_start): Likewise.
15178 (emit_stack_clash_protection_loop_end): Likewise.
15179 * rtl.h (get_stack_check_protect): Prototype.
15180 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
15181 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
15182 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
15184 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
15186 * doc/tm.texi: Rebuilt.
15187 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
15188 get_stack_check_protect.
15189 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
15190 * config/arm/arm.c (arm_expand_prologue): Likewise.
15191 (arm_frame_pointer_required): Likewise.
15192 * config/i386/i386.c (ix86_expand_prologue): Likewise.
15193 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
15194 * config/mips/mips.c (mips_expand_prologue): Likewise.
15195 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
15196 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
15197 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15198 (sparc_flat_expand_prologue): Likewise.
15200 * common.opt (-fstack-clash-protection): New option.
15201 * flag-types.h (enum stack_check_type): Note difference between
15202 -fstack-check= and -fstack-clash-protection.
15203 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15204 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15205 * toplev.c (process_options): Issue warnings/errors for cases
15206 not handled with -fstack-clash-protection.
15207 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15208 (-fstack-check): Note additional problem with -fstack-check=generic.
15209 Note that -fstack-check is primarily for Ada and refer users
15210 to -fstack-clash-protection for stack-clash-protection.
15211 Document new params for stack clash protection.
15213 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15215 * config/i386/i386.c (ix86_split_long_move): Do not handle
15216 address used for LEA in a special way.
15218 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15220 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15222 2017-09-19 Martin Sebor <msebor@redhat.com>
15225 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15226 of incompatible types.
15228 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15230 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15231 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15232 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15233 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15235 2017-09-19 Richard Biener <rguenther@suse.de>
15237 PR tree-optimization/82244
15238 * tree-vrp.c (remove_range_assertions): Do not propagate
15239 a constant to abnormals but replace the assert with a copy.
15241 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15243 PR rtl-optimization/57878
15244 PR rtl-optimization/68988
15245 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15246 avoidance test involving non_reload_pseudos. Move frequency test
15247 below the general fragmentation avoidance test.
15249 2017-09-19 Richard Biener <rguenther@suse.de>
15251 PR tree-optimization/69728
15252 * graphite-sese-to-poly.c (schedule_error): New global.
15253 (add_loop_schedule): Handle empty domain by failing the
15255 (build_original_schedule): Handle schedule_error.
15257 2017-09-19 Richard Biener <rguenther@suse.de>
15259 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15260 Do not iterate to sibling loops but only to siblings of inner
15263 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15266 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15267 that the registers are different.
15269 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15271 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15272 to processor_model and "amdfam17h" to arch_names_table.
15273 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15275 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15278 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15280 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15281 Alan Hayward <alan.hayward@arm.com>
15282 David Sherwood <david.sherwood@arm.com>
15284 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15286 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15288 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15289 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15290 vect_schedule_slp_instance.
15291 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15292 Update call to vect_slp_analyze_node_operations. Simplify return
15294 (vect_slp_analyze_bb_1): Update call accordingly.
15295 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15296 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15297 (vect_schedule_slp): Update call accordingly.
15299 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15300 Alan Hayward <alan.hayward@arm.com>
15301 David Sherwood <david.sherwood@arm.com>
15303 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15304 with types that aren't in fact scalar.
15306 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15308 * tree-vect-slp.c (vect_record_max_nunits): New function,
15310 (vect_build_slp_tree_1): ...here.
15311 (vect_build_slp_tree_2): Call it for phis too.
15313 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15315 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15316 to vect_get_vec_def_for_operand when getting the mask operand.
15318 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15319 Alan Hayward <alan.hayward@arm.com>
15320 David Sherwood <david.sherwood@arm.com>
15322 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15325 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15326 Alan Hayward <alan.hayward@arm.com>
15327 David Sherwood <david.sherwood@arm.com>
15329 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15330 calculation for vector booleans.
15332 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15333 Alan Hayward <alan.hayward@arm.com>
15334 David Sherwood <david.sherwood@arm.com>
15336 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15338 (vect_transform_stmt): ...here.
15339 (vect_analyze_stmt): Use it instead of calling
15340 vectorizable_live_operation directly.
15342 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15344 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15345 non-SIMT targets in acc vector loops.
15347 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15349 * configure.ac: Add arc and check if assembler supports gdwarf2.
15350 * configure: Regenerate.
15352 2017-09-18 Richard Biener <rguenther@suse.de>
15354 PR tree-optimization/82220
15355 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15356 epilogue niters from the min_profitable_iters compute.
15358 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15361 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15362 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15363 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15365 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15368 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15369 switching to a new text section.
15371 2017-09-18 Richard Biener <rguenther@suse.de>
15373 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15375 (build_alias_set): Reject aliases with no access function.
15377 2017-09-18 Richard Biener <rguenther@suse.de>
15379 PR tree-optimization/79622
15380 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15382 (build_cross_bb_scalars_use): Likewise.
15384 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15386 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15388 2017-09-18 Alan Modra <amodra@gmail.com>
15391 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15392 stack_pointer_rtx for count 0. Update comments. Break up
15393 large rtl expression.
15395 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15397 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15398 Increase to 20 bytes.
15399 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15400 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15401 or avx version of the stub.
15403 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15406 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15407 compute the minimum stack alignment. Also update preferred stack
15408 boundary for leaf functions.
15410 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15412 PR tree-optimization/82228
15413 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15416 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15418 * common/config/nds32/nds32-common.c
15419 (nds32_option_optimization_table): Refine formatting.
15420 (nds32_option_optimization_table): Use -fsched-pressure and
15421 -fomit-frame-pointer for specific optimization level.
15423 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15425 * config/nds32/nds32.c: Refine formatting and comments.
15426 * config/nds32/nds32.h: Likewise.
15427 * config/nds32/nds32.md: Likewise.
15428 * config/nds32/nds32-cost.c: Likewise.
15429 * config/nds32/nds32-isr.c: Likewise.
15430 * config/nds32/nds32-md-auxiliary.c: Likewise.
15431 * config/nds32/nds32-multiple.md: Likewise.
15432 * config/nds32/nds32-predicates.c: Likewise.
15434 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15435 Jakub Jelinek <jakub@redhat.com>
15437 Add support for -std=c++2a.
15438 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15440 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15442 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15445 * doc/extend.texi (Common Function Attributes): Add
15446 references to ARM, AArch64, and S/390 specific attributes.
15447 (Function Specific Option Pragmas): Add AArch64 and S/390
15448 to list of back ends that support the target pragma.
15450 2017-09-15 Nathan Sidwell <nathan@acm.org>
15452 * doc/standards.texi: Fix C++17 description. Update URLs for
15455 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15457 * common.opt (Wcast-align=strict): New warning option.
15458 * doc/invoke.texi: Document -Wcast-align=strict.
15460 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15462 * cgraph.h (cgraph_thunk_info): Add comments.
15463 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15464 assert for VIRTUAL_* arguments stricter.
15466 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15468 PR tree-optimization/71026
15469 * match.pd: Move RDIV patterns from fold-const.c
15470 * fold-const.c (distribute_real_division): Removed.
15471 (fold_binary_loc): Remove calls to distribute_real_divison.
15473 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15475 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15476 c++1z and gnu++1z as deprecated. Change other references to
15477 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15478 Change -Wc++1z-compat to -Wc++17-compat.
15479 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15480 * dwarf2out.c (highest_c_language): Handle C++17.
15481 (gen_compile_unit_die): Likewise.
15483 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15485 PR rtl-optimization/82192
15486 * combine.c (make_extraction): Don't look through non-paradoxical
15487 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15490 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15491 Alan Hayward <alan.hayward@arm.com>
15492 David Sherwood <david.sherwood@arm.com>
15494 * target.def (function_arg_offset): New hook.
15495 * targhooks.h (default_function_arg_offset): Declare.
15496 * targhooks.c (default_function_arg_offset): New function.
15497 * function.c (locate_and_pad_parm): Use
15498 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15499 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15500 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15501 * doc/tm.texi: Regenerate.
15502 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15503 * config/spu/spu.c (spu_function_arg_offset): New function.
15504 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15505 * system.h (FUNCTION_ARG_OFFSET): Poison.
15507 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15508 Alan Hayard <alan.hayward@arm.com>
15509 David Sherwood <david.sherwood@arm.com>
15511 * target.def (truly_noop_truncation): New hook.
15512 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15513 than TRULY_NOOP_TRUNCATION.
15514 * hooks.h (hook_bool_uint_uint_true): Declare.
15515 * hooks.c (hook_bool_uint_uint_true): New function.
15516 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15517 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15518 * doc/tm.texi: Regenerate.
15519 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15520 rather than TRULY_NOOP_TRUNCATION in comments.
15521 (simplify_comparison): Likewise.
15522 (record_truncated_value): Likewise.
15523 * expmed.c (extract_bit_field_1): Likewise.
15524 (extract_split_bit_field): Likewise.
15525 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15526 instead of TRULY_NOOP_TRUNCATION.
15527 * function.c (assign_parm_setup_block): Likewise.
15528 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15529 * rtlhooks.c: Include target.h.
15530 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15531 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15532 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15533 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15534 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15535 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15536 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15537 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15538 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15539 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15540 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15541 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15542 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15543 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15544 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15545 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15546 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15547 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15548 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15549 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15550 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15551 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15552 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15553 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15554 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15555 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15556 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15557 rather than TRULY_NOOP_TRUNCATION in comments.
15558 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15559 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15560 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15561 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15562 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15563 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15564 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15565 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15566 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15567 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15568 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15569 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15570 rather than TRULY_NOOP_TRUNCATION in comments.
15571 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15572 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15573 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15574 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15575 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15576 TRULY_NOOP_TRUNCATION condition.
15577 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15578 (TRULY_NOOP_TRUNCATION): Delete.
15579 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15580 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15581 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15582 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15583 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15584 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15585 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15586 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15587 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15588 rather than TRULY_NOOP_TRUNCATION in comments.
15589 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15590 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15591 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15592 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15593 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15594 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15596 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15599 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15600 (*cmp_ior): Likewise.
15601 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15602 (*ior_scc_scc_cmp): Likewise.
15603 (*and_scc_scc): Likewise.
15604 (*and_scc_scc_cmp): Likewise.
15606 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15607 Alan Hayard <alan.hayward@arm.com>
15608 David Sherwood <david.sherwood@arm.com>
15610 * target.def (can_change_mode_class): New hook.
15611 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15612 (hard_regno_nregs): Likewise.
15613 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15614 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15615 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15616 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15617 (LOAD_EXTEND_OP): Update accordingly.
15618 * doc/tm.texi: Regenerate.
15619 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15620 CANNOT_CHANGE_MODE_CLASS.
15621 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15622 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15623 * combine.c (simplify_set): Update accordingly.
15624 * emit-rtl.c (validate_subreg): Likewise.
15625 * recog.c (general_operand): Likewise.
15626 * regcprop.c (mode_change_ok): Likewise.
15627 * reload1.c (choose_reload_regs): Likewise.
15628 (inherit_piecemeal_p): Likewise.
15629 * rtlanal.c (simplify_subreg_regno): Likewise.
15630 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15631 instead of CANNOT_CHANGE_MODE_CLASS.
15632 (reload_cse_simplify_operands): Likewise.
15633 * reload.c (push_reload): Use targetm.can_change_mode_class
15634 instead of CANNOT_CHANGE_MODE_CLASS.
15635 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15636 REG_CANNOT_CHANGE_MODE_P.
15637 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15638 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15639 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15640 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15641 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15642 (arm_can_change_mode_class): New function.
15643 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15644 than CANNOT_CHANGE_MODE_CLASS in comments.
15645 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15646 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15647 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15648 (ix86_can_change_mode_class): ...this new function, inverting the
15649 sense of the return value.
15650 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15651 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15652 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15653 (ia64_can_change_mode_class): New function.
15654 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15655 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15656 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15657 (m32c_can_change_mode_class): ...this new function, inverting the
15658 sense of the return value.
15659 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15660 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15661 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15662 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15663 (mips_can_change_mode_class): ...this new function, inverting the
15664 sense of the return value.
15665 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15666 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15667 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15668 (msp430_can_change_mode_class): New function.
15669 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15670 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15671 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15672 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15673 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15674 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15675 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15676 (pa_cannot_change_mode_class): Replace with...
15677 (pa_can_change_mode_class): ...this new function, inverting the
15678 sense of the return value.
15679 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15680 than CANNOT_CHANGE_MODE_CLASS in comments.
15681 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15682 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15683 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15684 (pdp11_cannot_change_mode_class): Replace with...
15685 (pdp11_can_change_mode_class): ...this new function, inverting the
15686 sense of the return value.
15687 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15688 * config/powerpcspe/powerpcspe-protos.h
15689 (rs6000_cannot_change_mode_class_ptr): Delete.
15690 * config/powerpcspe/powerpcspe.c
15691 (rs6000_cannot_change_mode_class_ptr): Delete.
15692 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15693 (rs6000_option_override_internal): Assign to
15694 targetm.can_change_mode_class instead of
15695 rs6000_cannot_change_mode_class_ptr.
15696 (rs6000_cannot_change_mode_class): Replace with...
15697 (rs6000_can_change_mode_class): ...this new function, inverting the
15698 sense of the return value.
15699 (rs6000_debug_cannot_change_mode_class): Replace with...
15700 (rs6000_debug_can_change_mode_class): ...this new function.
15701 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15702 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15703 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15704 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15705 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15707 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15708 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15709 (rs6000_option_override_internal): Assign to
15710 targetm.can_change_mode_class instead of
15711 rs6000_cannot_change_mode_class_ptr.
15712 (rs6000_cannot_change_mode_class): Replace with...
15713 (rs6000_can_change_mode_class): ...this new function, inverting the
15714 sense of the return value.
15715 (rs6000_debug_cannot_change_mode_class): Replace with...
15716 (rs6000_debug_can_change_mode_class): ...this new function.
15717 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15718 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15719 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15720 (s390_can_change_mode_class): ...this new function, inverting the
15721 sense of the return value.
15722 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15723 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15724 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15725 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15726 (sh_cannot_change_mode_class): Replace with...
15727 (sh_can_change_mode_class): ...this new function, inverting the
15728 sense of the return value.
15729 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15730 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15731 (sparc_can_change_mode_class): New function.
15732 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15733 * config/spu/spu.c (spu_can_change_mode_class): New function.
15734 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15735 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15736 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15737 (visium_can_change_mode_class): New function.
15738 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15740 2017-09-15 Richard Biener <rguenther@suse.de>
15742 PR tree-optimization/82217
15743 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15744 but not undefined case.
15746 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15749 * postreload.c (reload_cse_simplify_operands): Skip
15750 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15752 2017-09-15 Richard Biener <rguenther@suse.de>
15754 PR tree-optimization/68823
15755 * graphite-scop-detection.c (build_alias_set): If we have a
15756 possible dependence check whether we can handle them by just
15757 looking at the DRs DR_ACCESS_FNs.
15758 (build_scops): If build_alias_set fails, fail the SCOP.
15760 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15762 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15763 to support float128 built-in functions that require the ISA 3.0
15765 (BU_FLOAT128_3_HW): Likewise.
15766 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15767 built-in functions.
15768 (FMAF128): Likewise.
15770 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15771 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15772 floating point instructions.
15773 (rs6000_invalid_builtin): Likewise.
15774 (rs6000_builtin_mask_names): Likewise.
15775 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15776 (RS6000_BTM_FLOAT128_HW): Likewise.
15777 (RS6000_BTM_COMMON): Likewise.
15778 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15780 * doc/extend.texi (RS/6000 built-in functions): Document the
15781 IEEE 128-bit floating point square root and fused multiply-add
15782 built-in functions.
15784 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15786 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15787 reg (r2) isn't in the set of registers defined in the prologue.
15789 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15790 Alan Hayward <alan.hayward@arm.com>
15791 David Sherwood <david.sherwood@arm.com>
15793 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15794 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15795 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15796 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15797 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15799 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15800 max_vectorization_factor.
15801 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15803 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15804 Alan Hayward <alan.hayward@arm.com>
15805 David Sherwood <david.sherwood@arm.com>
15807 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15808 (vect_worthwhile_without_simd_p): Declare.
15809 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15810 (vectorizable_reduction): Use it.
15811 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15812 (vectorizable_operation): Likewise.
15814 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15815 Alan Hayward <alan.hayward@arm.com>
15816 David Sherwood <david.sherwood@arm.com>
15818 * tree-vectorizer.h (vect_get_num_copies): New function.
15819 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15820 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15821 (vectorizable_induction): Likewise.
15822 (vectorizable_live_operation): Likewise.
15823 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15824 (vectorizable_bswap): Likewise.
15825 (vectorizable_call): Likewise.
15826 (vectorizable_conversion): Likewise.
15827 (vectorizable_assignment): Likewise.
15828 (vectorizable_shift): Likewise.
15829 (vectorizable_operation): Likewise.
15830 (vectorizable_store): Likewise.
15831 (vectorizable_load): Likewise.
15832 (vectorizable_condition): Likewise.
15833 (vectorizable_comparison): Likewise.
15834 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15836 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15837 Alan Hayward <alan.hayward@arm.com>
15838 David Sherwood <david.sherwood@arm.com>
15840 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15841 of vect_init_vector.
15843 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15844 Alan Hayward <alan.hayward@arm.com>
15845 David Sherwood <david.sherwood@arm.com>
15847 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15848 an inline wrapper that provides a location.
15849 (gimple_build_vector): Likewise.
15850 * gimple-fold.c (gimple_build_vector_from_val): New function.
15851 (gimple_build_vector): Likewise.
15852 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15853 functions to build the initial value. Always return a gimple value.
15854 (get_initial_defs_for_reduction): Likewise. Only compute
15856 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15857 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15858 (vectorizable_induction): Use gimple_build_vector rather than
15861 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15862 Alan Hayward <alan.hayward@arm.com>
15863 David Sherwood <david.sherwood@arm.com>
15865 * target.h (vec_perm_indices): New typedef.
15866 (auto_vec_perm_indices): Likewise.
15867 * optabs-query.h: Include target.h
15868 (can_vec_perm_p): Take a vec_perm_indices *.
15869 * optabs-query.c (can_vec_perm_p): Likewise.
15870 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15871 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15872 * tree-vect-generic.c (lower_vec_perm): Likewise.
15873 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15874 (vect_grouped_load_supported): Likewise.
15875 (vect_shift_permute_load_chain): Likewise.
15876 (vect_permute_store_chain): Use auto_vec_perm_indices.
15877 (vect_permute_load_chain): Likewise.
15878 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15879 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15880 Update uses of can_vec_perm_p.
15881 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15882 mode with a number of elements. Take a vec_perm_indices *.
15883 (vect_create_epilog_for_reduction): Update accordingly.
15884 Use auto_vec_perm_indices.
15885 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15886 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15887 (vect_transform_slp_perm_load): Likewise.
15888 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15889 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15890 (vect_gen_perm_mask_checked): Likewise.
15891 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15892 (vect_gen_perm_mask_checked): Likewise.
15893 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15894 (vectorizable_store): Likewise.
15895 (vectorizable_load): Likewise.
15896 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15897 (vectorizable_bswap): Likewise.
15899 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15900 Alan Hayward <alan.hayward@arm.com>
15901 David Sherwood <david.sherwood@arm.com>
15903 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15904 * tree.c (build_vector): Likewise.
15905 (build_vector_from_ctor): Update accordingly.
15906 (build_vector_from_val): Likewise.
15907 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15908 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15909 * tree-vect-generic.c (add_rshift): Likewise.
15910 (expand_vector_divmod): Likewise.
15911 (optimize_vector_constructor): Likewise.
15912 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15913 (vect_transform_slp_perm_load): Likewise.
15914 (vect_schedule_slp_instance): Likewise.
15915 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15916 (vectorizable_call): Likewise.
15917 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15918 * expmed.c (make_tree): Likewise.
15919 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15920 a vector passed to build_vector.
15921 (fold_convert_const): Likewise.
15922 (exact_inverse): Likewise.
15923 (fold_ternary_loc): Likewise.
15924 (fold_relational_const): Likewise.
15925 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15926 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15927 (const_unop): Likewise. Store the reduction accumulator in a
15928 variable rather than an array.
15929 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15930 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15931 the new vector, rather than constructing it after the input arrays.
15932 (native_interpret_vector): Use auto_vec<tree> when building
15933 a vector passed to build_vector. Add elements in order.
15934 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15935 auto_vec<tree> when building a vector passed to build_vector.
15936 (vect_create_epilog_for_reduction): Likewise.
15937 (vectorizable_induction): Likewise.
15938 (get_initial_def_for_reduction): Likewise. Fix indentation of
15940 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15942 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15943 passed to build_vector.
15945 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15946 Alan Hayward <alan.hayward@arm.com>
15947 David Sherwood <david.sherwood@arm.com>
15949 * tree-core.h (tree_base::u): Add an "nelts" field.
15950 (tree_vector): Use VECTOR_CST_NELTS as the length.
15951 * tree.c (tree_size): Likewise.
15952 (make_vector): Initialize VECTOR_CST_NELTS.
15953 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15954 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15955 TYPE_VECTOR_SUBPARTS.
15956 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15957 as the number of units, setting it from VECTOR_CST_NELTS.
15958 (const_vector_from_tree): Likewise.
15959 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15960 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15961 (fold_negate_expr_1): Likewise.
15962 (fold_convert_const): Likewise.
15963 (const_binop): Likewise. Differentiate the number of output and
15965 (const_unop): Likewise.
15966 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15967 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15968 in cases that did the opposite.
15970 2017-09-14 Richard Biener <rguenther@suse.de>
15972 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15975 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15977 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15979 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15982 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15983 if and where to split a bb, except for splitting before debug insn
15984 sequences followed by non-label real insn. Delete debug insns
15985 in between basic blocks.
15987 * combine.c (make_compound_operation_int): Formatting fixes.
15989 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15990 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15991 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15992 * config/sol2.h (LINK_EH_SPEC): Likewise.
15993 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15994 * config/s390/linux.h (LINK_SPEC): Likewise.
15995 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15996 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15997 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15998 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15999 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
16000 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16001 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
16002 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
16004 2017-09-13 Jakub Jelinek <jakub@redhat.com>
16006 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
16008 (ENDFILE_LINUX_SPEC): Likewise.
16009 (LINK_EH_SPEC): Likewise.
16010 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
16011 (LINK_OS_LINUX_SPEC32): Likewise.
16012 (LINK_OS_LINUX_SPEC64): Likewise.
16013 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
16014 (LINK_OS_LINUX_SPEC): Likewise.
16016 2017-09-13 Martin Liska <mliska@suse.cz>
16018 PR middle-end/82154
16019 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
16020 CASE_HIGH is NULL_TREE.
16022 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16023 Alan Hayward <alan.hayward@arm.com>
16024 David Sherwood <david.sherwood@arm.com>
16026 * target.def (secondary_memory_needed): New hook.
16027 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
16028 instead of SECONDARY_MEMORY_NEEDED.
16029 (secondary_memory_needed_mode): Likewise.
16030 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
16031 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
16032 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
16033 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
16034 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
16035 * doc/tm.texi: Regenerate.
16036 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
16037 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
16038 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16039 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
16040 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
16041 * config/i386/i386.c (inline_secondary_memory_needed): Put the
16042 mode argument first and change the reg_class arguments to reg_class_t.
16043 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
16044 Make static. Update the call to inline_secondary_memory_needed.
16045 (ix86_register_move_cost): Update the call to
16046 inline_secondary_memory_needed.
16047 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16048 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
16050 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
16051 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16053 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
16054 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
16055 * config/mips/mips.c (mips_secondary_memory_needed): Make static
16056 and match hook interface. Add comment from mips.h.
16057 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16058 * config/mmix/mmix.md (truncdfsf2): Refer to
16059 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16061 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
16062 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
16063 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16064 (pa_secondary_memory_needed): New function.
16065 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
16066 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
16067 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16068 (pdp11_secondary_memory_needed): Make static and match hook interface.
16069 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
16070 * config/powerpcspe/powerpcspe-protos.h
16071 (rs6000_secondary_memory_needed_ptr): Delete.
16072 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
16074 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16075 (rs6000_option_override_internal): Assign to
16076 targetm.secondary_memory_needed rather than
16077 rs6000_secondary_memory_needed_ptr.
16078 (rs6000_secondary_memory_needed): Match hook interface.
16079 (rs6000_debug_secondary_memory_needed): Likewise.
16080 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
16081 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
16082 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
16083 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16084 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
16085 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
16087 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
16088 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16089 (rs6000_option_override_internal): Assign to
16090 targetm.secondary_memory_needed rather than
16091 rs6000_secondary_memory_needed_ptr.
16092 (rs6000_secondary_memory_needed): Match hook interface.
16093 (rs6000_debug_secondary_memory_needed): Likewise.
16094 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
16095 * config/s390/s390.c (s390_secondary_memory_needed): New function.
16096 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16097 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
16098 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16099 (sparc_secondary_memory_needed): New function.
16100 * lra-constraints.c (check_and_process_move): Refer to
16101 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16103 (curr_insn_transform): Likewise.
16104 (process_alt_operands): Use targetm.secondary_memory_needed
16105 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16106 (check_secondary_memory_needed_p): Likewise.
16107 (choose_split_class): Likewise.
16108 * reload.c: Unconditionally include code that was previously
16109 conditional on SECONDARY_MEMORY_NEEDED.
16110 (push_secondary_reload): Use targetm.secondary_memory_needed
16111 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16112 (push_reload): Likewise.
16113 * reload1.c: Unconditionally include code that was previously
16114 conditional on SECONDARY_MEMORY_NEEDED.
16115 (choose_reload_regs): Use targetm.secondary_memory_needed
16116 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16117 (gen_reload): Likewise.
16118 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
16120 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16121 Alan Hayward <alan.hayward@arm.com>
16122 David Sherwood <david.sherwood@arm.com>
16124 * target.def (secondary_memory_needed_mode): New hook:
16125 * targhooks.c (default_secondary_memory_needed_mode): Declare.
16126 * targhooks.h (default_secondary_memory_needed_mode): New function.
16127 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
16128 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
16129 * doc/tm.texi: Regenerate.
16130 * lra-constraints.c (check_and_process_move): Use
16131 targetm.secondary_memory_needed_mode instead of
16132 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
16133 (curr_insn_transform): Likewise.
16134 * reload.c (get_secondary_mem): Likewise.
16135 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16136 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
16138 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16139 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16140 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
16141 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16142 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
16144 * config/powerpcspe/powerpcspe-protos.h
16145 (rs6000_secondary_memory_needed_mode): Delete.
16146 * config/powerpcspe/powerpcspe.c
16147 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16148 (rs6000_secondary_memory_needed_mode): Make static.
16149 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16150 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
16152 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16154 (rs6000_secondary_memory_needed_mode): Make static.
16155 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16156 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
16157 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16158 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16159 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16161 (sparc_secondary_memory_needed_mode): New function.
16162 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
16164 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
16166 * config/aarch64/constraints.md (Umq): New constraint.
16167 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
16169 (mov<mode>): Update condition.
16171 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16173 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
16174 when bitposition is the same.
16176 2017-09-13 Richard Biener <rguenther@suse.de>
16178 * dwarf2out.c (output_die_symbol): Remove.
16179 (output_die): Do not output a DIEs symbol.
16181 2017-09-13 Richard Biener <rguenther@suse.de>
16183 PR middle-end/82128
16184 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
16185 default-def to avoid breaking iterator update with the weird
16186 interaction with cgraph_update_edges_for_call_stmt_node.
16188 2017-09-13 Richard Biener <rguenther@suse.de>
16190 * tree-cfg.c (verify_gimple_assign_binary): Add verification
16191 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
16192 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
16193 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
16195 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
16197 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
16198 Disable pc relative literal load irrespective of
16199 TARGET_FIX_ERR_A53_84341 for default.
16201 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16203 * config/sparc/sparc.c (output_return): Output the source location of
16204 the insn in the delay slot, if any.
16205 (output_sibcall): Likewise.
16207 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16210 * common.opt (-static-pie): New alias.
16211 (shared): Negate static-pie.
16212 (-no-pie): Update help text.
16214 (static-pie): New option.
16215 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16216 -static-pie support.
16217 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16218 (LINK_EH_SPEC): Likewise.
16219 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16220 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16221 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16222 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16223 (init_gcc_specs): Likewise.
16224 (init_spec): Likewise.
16225 (display_help): Update help message for -pie.
16226 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16229 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16231 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16232 (movdi_aarch64): Likewise.
16233 (movti_aarch64): Likewise.
16235 2017-09-12 Simon Wright <simon@pushface.org>
16238 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16239 calculation of the minor version, always output as 0.
16241 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16244 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16245 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16246 on it early, rather than manual conversion late. For
16247 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16248 instead of performing manual conversion.
16250 2017-09-12 Carl Love <cel@us.ibm.com>
16252 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16253 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16255 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16256 VMULOSW): Add definitions.
16257 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16258 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16259 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16260 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16261 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16263 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16265 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16267 (movti_aarch64): Likewise.
16268 (movdf_aarch64): Likewise.
16269 (movtf_aarch64): Likewise.
16270 (load_pairdi): Likewise.
16271 (store_pairdi): Likewise.
16272 (load_pairdf): Likewise.
16273 (store_pairdf): Likewise.
16274 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16275 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16276 (ldr_got_small_<mode>): Likewise.
16277 (ldr_got_small_28k_<mode>): Likewise.
16278 (ldr_got_tiny): Likewise.
16279 * config/aarch64/iterators.md (ldst_sz): New.
16280 (ldpstp_sz): Likewise.
16281 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16283 (thunderx_load): Split load_8 to load_16.
16284 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16286 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16287 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16288 (xgene1_store_pair): Split store_8 to store_16.
16289 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16290 (falkor_st_0_st_sd): Split store_8 to store_16.
16292 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16294 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16295 and store1/2/3/4 to store_4/8/12/16.
16296 * config/aarch64/aarch64.md: Update for rename.
16297 * config/arm/arm.md: Likewise.: Likewise.
16298 * config/arm/arm.c: Likewise.
16299 * config/arm/thumb1.md: Likewise.
16300 * config/arm/thumb2.md: Likewise.
16301 * config/arm/vfp.md: Likewise.
16302 * config/arm/arm-generic.md: Likewise.
16303 * config/arm/arm1020e.md: Likewise.
16304 * config/arm/arm1026ejs.md: Likewise.
16305 * config/arm/arm1136jfs.md: Likewise.
16306 * config/arm/arm926ejs.md: Likewise.
16307 * config/arm/cortex-a15.md: Likewise.
16308 * config/arm/cortex-a17.md: Likewise.
16309 * config/arm/cortex-a5.md: Likewise.
16310 * config/arm/cortex-a53.md: Likewise.
16311 * config/arm/cortex-a57.md: Likewise.
16312 * config/arm/cortex-a7.md: Likewise.
16313 * config/arm/cortex-a8.md: Likewise.
16314 * config/arm/cortex-a9.md: Likewise.
16315 * config/arm/cortex-m4.md: Likewise.
16316 * config/arm/cortex-m7.md: Likewise.
16317 * config/arm/cortex-r4.md: Likewise.
16318 * config/arm/exynos-m1.md: Likewise.
16319 * config/arm/fa526.md: Likewise.
16320 * config/arm/fa606te.md: Likewise.
16321 * config/arm/fa626te.md: Likewise.
16322 * config/arm/fa726te.md: Likewise.
16323 * config/arm/fmp626.md: Likewise.
16324 * config/arm/iwmmxt.md: Likewise.
16325 * config/arm/ldmstm.md: Likewise.
16326 * config/arm/marvell-pj4.md: Likewise.
16327 * config/arm/xgene1.md: Likewise.
16328 * config/aarch64/thunderx.md: Likewise.
16329 * config/aarch64/thunderx2t99.md: Likewise.
16330 * config/aarch64/falkor.md: Likewise.
16332 2017-09-12 Martin Liska <mliska@suse.cz>
16334 * attribs.c (private_lookup_attribute): New function.
16335 * attribs.h (private_lookup_attribute): Declared here.
16336 (lookup_attribute): Called from this place.
16338 2017-09-12 Richard Biener <rguenther@suse.de>
16340 PR tree-optimization/82157
16341 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16342 stmts with side-effects.
16344 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16345 Alan Hayward <alan.hayward@arm.com>
16346 David Sherwood <david.sherwood@arm.com>
16348 * target.def (hard_regno_nregs): New hook.
16349 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16350 * targhooks.h (default_hard_regno_nregs): Declare.
16351 * targhooks.c (default_hard_regno_nregs): New function.
16352 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16353 (TARGET_HARD_REGNO_NREGS): ...this hook.
16354 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16355 (CLASS_MAX_NREGS): Likewise.
16356 * doc/tm.texi: Regenerate.
16357 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16358 instead of HARD_REGNO_NREGS.
16359 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16360 HARD_REGNO_NREGS in the comment.
16361 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16362 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16363 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16364 Return an unsigned int.
16365 (TARGET_HARD_REGNO_NREGS): Redefine.
16366 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16367 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16368 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16369 (arc_hard_regno_nregs): New function.
16370 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16371 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16372 (arm_hard_regno_nregs): New function.
16373 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16374 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16375 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16376 (TARGET_HARD_REGNO_NREGS): Redefine.
16377 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16378 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16379 (HARD_REGNO_NREGS): Delete.
16380 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16381 (cr16_hard_regno_nregs): New function.
16382 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16383 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16384 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16385 (cris_hard_regno_nregs): New function.
16386 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16387 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16388 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16389 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16390 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16391 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16392 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16393 (frv_hard_regno_nregs): Make static. Take and return an
16395 (frv_class_max_nregs): Remove outdated copy of documentation.
16396 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16397 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16398 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16399 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16400 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16401 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16402 (TARGET_HARD_REGNO_NREGS): Redefine.
16403 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16404 (CLASS_MAX_NREGS): Update comment.
16405 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16406 (ia64_hard_regno_nregs): New function.
16407 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16408 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16409 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16410 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16411 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16413 (m32c_hard_regno_nregs): Likewise. Make static.
16414 (TARGET_HARD_REGNO_NREGS): Redefine.
16415 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16416 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16417 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16418 (m68k_hard_regno_nregs): New function.
16419 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16420 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16421 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16422 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16423 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16424 Take and return an unsigned int.
16425 (TARGET_HARD_REGNO_NREGS): Redefine.
16426 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16427 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16428 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16429 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16430 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16431 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16432 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16433 (msp430_hard_regno_nregs): Make static. Take and return an
16435 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16436 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16437 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16438 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16439 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16440 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16441 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16442 (TARGET_HARD_REGNO_NREGS): Redefine.
16443 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16444 (PA_HARD_REGNO_NREGS): ...this.
16445 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16446 (PA_HARD_REGNO_NREGS): ...this.
16447 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16448 (pa_hard_regno_nregs): New function.
16449 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16450 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16451 (pdp11_hard_regno_nregs): New function.
16452 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16453 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16454 (rs6000_hard_regno_nregs_hook): New function.
16455 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16456 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16457 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16458 Take and return an unsigned int. Move earlier in file.
16459 (TARGET_HARD_REGNO_NREGS): Redefine.
16460 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16461 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16462 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16463 (rl78_hard_regno_nregs): Make static. Take and return an
16465 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16466 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16467 (rs6000_hard_regno_nregs_hook): New function.
16468 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16469 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16470 (TARGET_HARD_REGNO_NREGS): Redefine.
16471 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16472 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16473 instead of HARD_REGNO_NREGS.
16474 (s390_hard_regno_nregs): New function.
16475 (s390_hard_regno_mode_ok): Add comment from s390.h.
16476 (TARGET_HARD_REGNO_NREGS): Redefine.
16477 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16478 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16479 (sh_hard_regno_nregs): New function.
16480 (sh_pass_in_reg_p): Use it.
16481 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16482 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16483 (sparc_hard_regno_nregs): New function.
16484 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16485 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16486 (spu_function_arg_advance): Use it, supplying a valid register number.
16487 (TARGET_HARD_REGNO_NREGS): Redefine.
16488 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16489 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16490 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16491 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16492 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16493 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16494 (CLASS_MAX_NREGS): Remove copy of old documentation.
16495 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16496 (visium_hard_regno_nregs): New function.
16497 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16498 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16499 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16500 xtensa_hard_regno_nregs): New function.
16501 * system.h (HARD_REGNO_NREGS): Poison.
16503 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16505 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16506 hard_regno_nregs instead of HARD_REGNO_NREGS.
16507 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16508 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16509 (c6x_expand_epilogue): Likewise.
16510 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16511 (frv_read_iacc_argument): Likewise.
16512 * config/sh/sh.c: Include regs.h.
16513 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16514 (regs_used): Likewise.
16515 (output_stack_adjust): Likewise.
16516 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16517 * expmed.c: Include regs.h.
16518 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16519 * ree.c: Include regs.h.
16520 (combine_reaching_defs): Use hard_regno_nregs instead of
16522 (add_removable_extension): Likewise.
16524 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16526 * regs.h (hard_regno_nregs): Turn into a function.
16527 (end_hard_regno): Update accordingly.
16528 * caller-save.c (setup_save_areas): Likewise.
16529 (save_call_clobbered_regs): Likewise.
16530 (replace_reg_with_saved_mem): Likewise.
16531 (insert_restore): Likewise.
16532 (insert_save): Likewise.
16533 * combine.c (can_change_dest_mode): Likewise.
16534 (move_deaths): Likewise.
16535 (distribute_notes): Likewise.
16536 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16537 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16538 (rs6000_split_multireg_move): Likewise.
16539 (rs6000_register_move_cost): Likewise.
16540 (rs6000_memory_move_cost): Likewise.
16541 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16542 (rs6000_split_multireg_move): Likewise.
16543 (rs6000_register_move_cost): Likewise.
16544 (rs6000_memory_move_cost): Likewise.
16545 * cselib.c (cselib_reset_table): Likewise.
16546 (cselib_lookup_1): Likewise.
16547 * emit-rtl.c (set_mode_and_regno): Likewise.
16548 * function.c (aggregate_value_p): Likewise.
16549 * ira-color.c (setup_profitable_hard_regs): Likewise.
16550 (check_hard_reg_p): Likewise.
16551 (calculate_saved_nregs): Likewise.
16552 (assign_hard_reg): Likewise.
16553 (improve_allocation): Likewise.
16554 (calculate_spill_cost): Likewise.
16555 * ira-emit.c (modify_move_list): Likewise.
16556 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16557 (ira_hard_reg_in_set_p): Likewise.
16558 * ira.c (setup_reg_mode_hard_regset): Likewise.
16559 (clarify_prohibited_class_mode_regs): Likewise.
16560 (check_allocation): Likewise.
16561 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16562 (lra_setup_reg_renumber): Likewise.
16563 (setup_try_hard_regno_pseudos): Likewise.
16564 (spill_for): Likewise.
16565 (assign_hard_regno): Likewise.
16566 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16567 * lra-constraints.c (in_class_p): Likewise.
16568 (lra_constraint_offset): Likewise.
16569 (simplify_operand_subreg): Likewise.
16570 (lra_constraints): Likewise.
16571 (split_reg): Likewise.
16572 (split_if_necessary): Likewise.
16573 (invariant_p): Likewise.
16574 (inherit_in_ebb): Likewise.
16575 * lra-lives.c (process_bb_lives): Likewise.
16576 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16577 (get_hard_regs): Likewise.
16578 (do_remat): Likewise.
16579 * lra-spills.c (assign_spill_hard_regs): Likewise.
16580 * mode-switching.c (create_pre_exit): Likewise.
16581 * postreload.c (reload_combine_recognize_pattern): Likewise.
16582 * recog.c (peep2_find_free_register): Likewise.
16583 * regcprop.c (kill_value_regno): Likewise.
16584 (set_value_regno): Likewise.
16585 (copy_value): Likewise.
16586 (maybe_mode_change): Likewise.
16587 (find_oldest_value_reg): Likewise.
16588 (copyprop_hardreg_forward_1): Likewise.
16589 * regrename.c (check_new_reg_p): Likewise.
16590 (regrename_do_replace): Likewise.
16591 * reload.c (push_reload): Likewise.
16592 (combine_reloads): Likewise.
16593 (find_dummy_reload): Likewise.
16594 (operands_match_p): Likewise.
16595 (find_reloads): Likewise.
16596 (find_equiv_reg): Likewise.
16597 (reload_adjust_reg_for_mode): Likewise.
16598 * reload1.c (count_pseudo): Likewise.
16599 (count_spilled_pseudo): Likewise.
16600 (find_reg): Likewise.
16601 (clear_reload_reg_in_use): Likewise.
16602 (free_for_value_p): Likewise.
16603 (allocate_reload_reg): Likewise.
16604 (choose_reload_regs): Likewise.
16605 (reload_adjust_reg_for_temp): Likewise.
16606 (emit_reload_insns): Likewise.
16607 (delete_output_reload): Likewise.
16608 * rtlanal.c (subreg_get_info): Likewise.
16609 * sched-deps.c (sched_analyze_reg): Likewise.
16610 * sel-sched.c (init_regs_for_mode): Likewise.
16611 (mark_unavailable_hard_regs): Likewise.
16612 (choose_best_reg_1): Likewise.
16613 (verify_target_availability): Likewise.
16614 * valtrack.c (dead_debug_insert_temp): Likewise.
16615 * var-tracking.c (track_loc_p): Likewise.
16616 (emit_note_insn_var_location): Likewise.
16617 * varasm.c (make_decl_rtl): Likewise.
16618 * reginfo.c (choose_hard_reg_mode): Likewise.
16619 (init_reg_modes_target): Refer directly to
16620 this_target_regs->x_hard_regno_nregs.
16622 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16624 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16625 instead of hard_regno_nregs.
16627 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16629 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16630 end_hard_regno instead of hard_regno_nregs.
16631 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16632 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16633 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16634 * ira-color.c (improve_allocation): Likewise.
16635 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16636 * lra-lives.c (mark_regno_live): Likewise.
16637 (mark_regno_dead): Likewise.
16638 * lra-remat.c (operand_to_remat): Likewise.
16639 * lra.c (collect_non_operand_hard_regs): Likewise.
16640 * postreload.c (reload_combine_note_store): Likewise.
16641 (move2add_valid_value_p): Likewise.
16642 * reload.c (regno_clobbered_p): Likewise.
16644 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16646 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16648 * config/v850/v850.c (v850_reorg): Likewise.
16649 * reload.c (refers_to_regno_for_reload_p): Likewise.
16650 (find_equiv_reg): Likewise.
16651 * reload1.c (reload_reg_reaches_end_p): Likewise.
16653 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16655 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16657 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16658 * config/arm/arm.c (output_move_neon): Likewise.
16659 (arm_attr_length_move_neon): Likewise.
16660 (neon_split_vcombine): Likewise.
16661 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16662 (c6x_mark_reg_written): Likewise.
16663 (c6x_dwarf_register_span): Likewise.
16664 * config/i386/i386.c (ix86_save_reg): Likewise.
16665 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16666 (rws_access_reg): Likewise.
16667 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16668 * mode-switching.c (create_pre_exit): Likewise.
16669 * ree.c (combine_reaching_defs): Likewise.
16670 (add_removable_extension): Likewise.
16671 * regcprop.c (find_oldest_value_reg): Likewise.
16672 (copyprop_hardreg_forward_1): Likewise.
16673 * reload.c (reload_inner_reg_of_subreg): Likewise.
16674 (push_reload): Likewise.
16675 (combine_reloads): Likewise.
16676 (find_dummy_reload): Likewise.
16677 (reload_adjust_reg_for_mode): Likewise.
16678 * reload1.c (find_reload_regs): Likewise.
16679 (forget_old_reloads_1): Likewise.
16680 (reload_reg_free_for_value_p): Likewise.
16681 (reload_adjust_reg_for_temp): Likewise.
16682 (emit_reload_insns): Likewise.
16683 (delete_output_reload): Likewise.
16684 * sel-sched.c (choose_best_reg_1): Likewise.
16685 (choose_best_pseudo_reg): Likewise.
16687 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16688 Alan Hayward <alan.hayward@arm.com>
16689 David Sherwood <david.sherwood@arm.com>
16691 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16692 * target.def (slow_unaligned_access): New hook.
16693 * targhooks.h (default_slow_unaligned_access): Declare.
16694 * targhooks.c (default_slow_unaligned_access): New function.
16695 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16696 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16697 * doc/tm.texi: Regenerate.
16698 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16699 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16700 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16702 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16703 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16705 (rs6000_slow_unaligned_access): New function.
16706 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16707 (expand_block_compare): Likewise.
16708 (expand_strn_compare): Likewise.
16709 (rs6000_rtx_costs): Likewise.
16710 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16711 (riscv_slow_unaligned_access): Likewise.
16712 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16713 (riscv_slow_unaligned_access_p): ...this and make static.
16714 (riscv_option_override): Update accordingly.
16715 (riscv_slow_unaligned_access): New function.
16716 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16717 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16718 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16719 (rs6000_slow_unaligned_access): New function.
16720 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16721 (rs6000_rtx_costs): Likewise.
16722 * config/rs6000/rs6000-string.c (expand_block_compare)
16723 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16724 of SLOW_UNALIGNED_ACCESS.
16725 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16726 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16727 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16728 of SLOW_UNALIGNED_ACCESS.
16729 * expmed.c (simple_mem_bitfield_p): Likewise.
16730 * expr.c (alignment_for_piecewise_move): Likewise.
16731 (emit_group_load_1): Likewise.
16732 (emit_group_store): Likewise.
16733 (copy_blkmode_from_reg): Likewise.
16734 (emit_push_insn): Likewise.
16735 (expand_assignment): Likewise.
16736 (store_field): Likewise.
16737 (expand_expr_real_1): Likewise.
16738 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16739 * lra-constraints.c (simplify_operand_subreg): Likewise.
16740 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16741 * gimple-ssa-store-merging.c: Likewise in block comment at start
16743 * tree-ssa-strlen.c: Include target.h.
16744 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16745 of SLOW_UNALIGNED_ACCESS.
16746 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16748 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16750 PR rtl-optimization/82185
16751 * expmed.c (emit_store_flag_int): Only test tem if it has been
16754 2017-09-12 Richard Biener <rguenther@suse.de>
16756 PR middle-end/82149
16757 * match.pd ((FTYPE) N CMP CST): Fix typo.
16759 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16761 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16763 (mips_near_type_p): Add 'short_call' attribute as a synonym
16765 * doc/extend.texi (short_call): Document new function attribute.
16767 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16770 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16771 assertion check that in the condition.
16772 (get_atomic_generic_size): Likewise. Before testing if parameter
16773 has pointer type, if it has array type, call for C++
16774 default_conversion to perform array-to-pointer conversion.
16776 2017-09-12 Richard Biener <rguenther@suse.de>
16778 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16779 for operations we cannot scalarize.
16781 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16783 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16784 vectors heap vectors. Clean up comments.
16785 Make visited_bbs a reference.
16786 (profitable_jump_thread_path): Make GC
16787 vectors heap vectors. Clean up comments.
16789 (convert_and_register_jump_thread_path): Make GC vectors heap
16791 (check_subpath_and_update_thread_path): Same. Clean up comments.
16792 Make visited_bbs a reference.
16793 (handle_phi): Abstract common code to to
16794 register_jump_thread_path_if_profitable.
16795 Rename VAR_BB to DEF_BB.
16797 Make GC vectors heap vectors.
16798 Make visited_bbs a reference.
16799 (handle_assignment): Same.
16800 (register_jump_thread_path_if_profitable): New.
16801 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16803 Make GC vectors heap vectors. Clean up comments.
16804 Make visited_bbs a reference.
16805 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16806 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16809 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16812 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16813 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16815 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16817 Revert r251800 and r251799.
16819 2017-09-11 Martin Jambor <mjambor@suse.cz>
16822 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16823 arguments in advance.
16824 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16825 use it to find predecessor edges.
16826 (naive_outof_ssa): Collect vector of predecessors.
16828 2017-09-08 Jason Merrill <jason@redhat.com>
16830 PR c++/70029 - ICE with ref-qualifier and -flto
16831 * langhooks.h (struct lang_hooks_for_types): Add
16832 copy_lang_qualifiers.
16833 * attribs.c (build_type_attribute_qual_variant): Use it.
16834 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16836 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16837 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16839 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16842 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16843 (*mulsi3_sp64): New instruction.
16844 (mulsi3): New expander.
16846 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16848 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16850 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16852 * sancov.c: Include memmodel.h.
16854 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16857 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16860 2017-09-07 Carl Love <cel@us.ibm.com>
16862 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16863 the sldi instruction.
16865 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16867 * sancov.c: Include tm_p.h.
16869 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16872 * output.h (switch_to_other_text_partition): New declaration.
16873 * varasm.c (switch_to_other_text_partition): New function.
16874 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16875 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16876 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16877 to the other text partition before emitting LCL label and switch back
16878 after emitting the word after it.
16880 2017-09-07 Richard Biener <rguenther@suse.de>
16882 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16883 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16884 critical edges here, after loop init.
16885 (pass_data_pre): Remove PROP_no_crit_edges flags.
16886 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16887 for valueization of call args to avoid leaking VN_TOP.
16888 (visit_use): Assert we do not visit default defs.
16889 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16890 Use error_mark_node to more easily detect leaking VN_TOP.
16891 All default-defs are varying, not VN_TOP. Mark them visited.
16892 (run_scc_vn): Make code match comment.
16894 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16896 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16897 OPTION_MASK_FLOAT128_KEYWORD.
16898 (POWERPC_MASKS): Likewise.
16899 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16900 support for the -mfloat128-type option, and make -mfloat128
16901 default on PowerPC Linux systems. Define or undefine
16902 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16903 Define __float128 to be __ieee128 if IEEE 128-bit support is
16904 enabled, or undefine it.
16905 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16906 Delete defining __FLOAT128_TYPE__.
16907 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16908 -mfloat128-type option and make -mfloat128 default on PowerPC
16910 (TARGET_FLOAT128_TYPE): Likewise.
16911 (-mfloat128-type): Likewise.
16912 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16913 Delete the -mfloat128-type option and make -mfloat128 default on
16914 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16915 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16916 128-bit floating point is enabled. Change tests from using
16917 -mfloat128-type to -mfloat128.
16918 (rs6000_mangle_type): Use the correct mangling for the __float128
16919 type even if normal long double is restricted to 64-bits.
16920 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16922 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16923 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16924 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16926 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16927 documentation for -mfloat128.
16929 2017-09-06 Olivier Hainque <hainque@adacore.com>
16931 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16933 2017-09-06 Wish Wu <wishwu007@gmail.com>
16934 Jakub Jelinek <jakub@redhat.com>
16936 * asan.c (initialize_sanitizer_builtins): Add
16937 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16938 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16939 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16940 BT_FN_VOID_UINT64_PTR variables.
16941 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16942 (BT_FN_VOID_UINT16_UINT16): Likewise.
16943 (BT_FN_VOID_UINT32_UINT32): Likewise.
16944 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16945 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16946 (BT_FN_VOID_UINT64_PTR): Likewise.
16947 * common.opt (flag_sanitize_coverage): New variable.
16948 (fsanitize-coverage=trace-pc): Remove.
16949 (fsanitize-coverage=): Add.
16950 * flag-types.h (enum sanitize_coverage_code): New enum.
16951 * fold-const.c (fold_range_test): Disable non-short-circuit
16952 optimization if flag_sanitize_coverage.
16953 (fold_truth_andor): Likewise.
16954 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16955 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16956 (coverage_sanitizer_opts): New array.
16957 (get_closest_sanitizer_option): Add OPTS argument, handle also
16958 OPT_fsanitize_coverage_.
16959 (parse_sanitizer_options): Adjusted to also handle
16960 OPT_fsanitize_coverage_.
16961 (common_handle_option): Add OPT_fsanitize_coverage_.
16962 * sancov.c (instrument_comparison, instrument_switch): New function.
16963 (sancov_pass): Add trace-cmp support.
16964 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16965 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16966 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16967 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16968 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16969 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16970 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16971 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16972 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16973 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16975 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16977 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16978 error. Only quit immediately if parsing is complete.
16979 (BEGIN): Initialize fatal_err and parse_done.
16980 (begin fpu, end fpu): Check number of arguments.
16981 (begin arch, end arch): Likewise.
16982 (begin cpu, end cpu): Likewise.
16983 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16984 (optalias): Likewise.
16986 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16988 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16989 * config/arm/arm-isa.h: Delete. Move definitions to ...
16990 * arm-cpus.in: ... here. Use new feature and fgroup values.
16991 * config/arm/arm.c (arm_option_override): Use lower case for feature
16993 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16994 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16995 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16997 (print_isa_bits_for): New function.
16998 (gen_isa): New function.
16999 (gen_comm_data): Use print_isa_bits_for.
17000 (define feature): New keyword.
17001 (define fgroup): New keyword.
17002 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
17003 (arm-isa.h): Add rule to generate file.
17004 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
17005 case for feature bit names.
17007 2017-09-06 Richard Biener <rguenther@suse.de>
17009 * tree-ssa-pre.c (NECESSARY): Remove.
17010 (create_expression_by_pieces): Do not touch pass-local flags.
17011 (insert_into_preds_of_block): Likewise.
17012 (do_pre_regular_insertion): Likewise.
17013 (eliminate_insert): Likewise.
17014 (eliminate_dom_walker::before_dom_children): Likewise.
17015 (fini_eliminate): Do not look at inserted_exprs.
17016 (mark_operand_necessary): Remove.
17017 (remove_dead_inserted_code): Replace with simple work-list
17018 algorithm based on inserted_exprs and SSA uses.
17019 (pass_pre::execute): Re-order fini_eliminate and
17020 remove_dead_inserted_code.
17022 2017-09-06 Olivier Hainque <hainque@adacore.com>
17024 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
17025 for VxWorks 7. Adjust surrounding comments.
17027 2017-09-06 Richard Biener <rguenther@suse.de>
17029 * gimple-ssa-strength-reduction.c
17030 (find_candidates_dom_walker::before_dom_children): Also allow
17033 2017-09-06 Richard Biener <rguenther@suse.de>
17035 PR tree-optimization/82108
17036 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
17037 for gap in the non-permutation SLP case.
17039 2017-09-06 Martin Jambor <mjambor@suse.cz>
17041 PR tree-optimization/82078
17042 * tree-sra.c (sort_and_splice_var_accesses): Move call to
17043 add_access_to_work_queue...
17044 (build_accesses_from_assign): ...here.
17045 (propagate_all_subaccesses): Make sure racc is the group
17046 representative, if there is one.
17048 2017-09-06 Jakub Jelinek <jakub@redhat.com>
17050 PR middle-end/82095
17051 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
17054 2017-09-06 Richard Biener <rguenther@suse.de>
17056 * gimple-ssa-strength-reduction.c
17057 (find_candidates_dom_walker::before_doom_children): Use a
17058 type and not a mode check.
17060 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
17063 * config/arm/predicates.md (arm_general_adddi_operand): Create new
17065 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
17067 2017-09-05 Jeff Law <law@redhat.com>
17069 PR tree-optimization/64910
17070 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
17071 cases where we have 3 or more operands.
17073 2017-09-05 Jakub Jelinek <jakub@redhat.com>
17075 PR middle-end/81768
17076 * omp-low.c (lower_omp_for): Recompute tree invariant if
17077 gimple_omp_for_initial/final is ADDR_EXPR.
17079 PR middle-end/81768
17080 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
17081 into gimple val before gimplification fo the COND_EXPR.
17083 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
17085 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
17086 REGION_COPY argument.
17087 (thread_through_all_blocks): Remove unused argument to
17088 duplicate_thread_path.
17090 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17091 Alan Hayward <alan.hayward@arm.com>
17092 David Sherwood <david.sherwood@arm.com>
17094 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
17095 Take a scalar_mode rather than a machine_mode.
17096 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17097 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
17098 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17099 (aarch64_gen_adjusted_ldpstp): Likewise.
17100 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
17102 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17103 Alan Hayward <alan.hayward@arm.com>
17104 David Sherwood <david.sherwood@arm.com>
17106 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
17107 Take a scalar_int_mode instead of a machine_mode.
17108 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17109 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17110 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17111 (aarch64_simd_attr_length_rglist): Delete.
17112 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
17113 a scalar_int_mode instead of a machine_mode.
17114 (aarch64_add_offset): Likewise.
17115 (aarch64_internal_mov_immediate): Likewise
17116 (aarch64_add_constant_internal): Likewise.
17117 (aarch64_add_constant): Likewise.
17118 (aarch64_movw_imm): Likewise.
17119 (aarch64_rtx_arith_op_extract_p): Likewise.
17120 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17121 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17122 Remove assert that the mode isn't a vector.
17123 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17124 (aarch64_expand_mov_immediate): Update calls after above changes.
17125 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
17126 (aarch64_and_bitmask_imm): Check for scalar integer modes.
17127 (aarch64_move_imm): Likewise.
17128 (aarch64_can_const_movi_rtx_p): Likewise.
17129 (aarch64_strip_extend): Likewise.
17130 (aarch64_extr_rtx_p): Likewise.
17131 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
17132 a CONST_INT when the mode parameter is VOIDmode.
17133 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
17135 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17137 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
17138 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17139 (bitwise_type_for_mode): Update accordingly.
17141 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17143 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
17144 * stor-layout.c (mode_for_size_tree): Likewise.
17145 (mode_for_array): Update accordingly.
17146 (layout_decl): Likewise.
17147 (compute_record_mode): Likewise. Only set the mode once.
17149 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17151 * target.def (get_mask_mode): Change return type to opt_mode.
17153 * doc/tm.texi: Regenerate.
17154 * targhooks.h (default_get_mask_mode): Return an opt_mode.
17155 * targhooks.c (default_get_mask_mode): Likewise.
17156 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17157 * optabs-query.c (can_vec_mask_load_store_p): Update use of
17158 targetm.get_mask_mode.
17159 * tree.c (build_truth_vector_type): Likewise.
17161 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17163 * machmode.h (mode_for_vector): Return an opt_mode.
17164 * stor-layout.c (mode_for_vector): Likewise.
17165 (mode_for_int_vector): Update accordingly.
17166 (layout_type): Likewise.
17167 * config/i386/i386.c (emit_memmov): Likewise.
17168 (ix86_expand_set_or_movmem): Likewise.
17169 (ix86_expand_vector_init): Likewise.
17170 (ix86_get_mask_mode): Likewise.
17171 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
17173 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
17174 * expmed.c (extract_bit_field_1): Likewise.
17175 * expr.c (expand_expr_real_2): Likewise.
17176 * optabs-query.c (can_vec_perm_p): Likewise.
17177 (can_vec_mask_load_store_p): Likewise.
17178 * optabs.c (expand_vec_perm): Likewise.
17179 * targhooks.c (default_get_mask_mode): Likewise.
17180 * tree-vect-stmts.c (vectorizable_store): Likewise.
17181 (vectorizable_load): Likewise.
17182 (get_vectype_for_scalar_type_and_size): Likewise.
17184 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17186 * machmode.h (mode_for_int_vector): New function.
17187 * stor-layout.c (mode_for_int_vector): Likewise.
17188 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
17189 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
17190 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17191 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17192 (s390_expand_vcond): Likewise.
17194 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17196 * machmode.h (opt_machine_mode): New type.
17197 (opt_mode<T>): Allow construction from anything that can be
17199 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17200 (mode_for_size): Return an opt_machine_mode.
17201 * stor-layout.c (mode_for_size): Likewise.
17202 (mode_for_size_tree): Update call accordingly.
17203 (bitwise_mode_for_mode): Likewise.
17204 (make_fract_type): Likewise.
17205 (make_accum_type): Likewise.
17206 * caller-save.c (replace_reg_with_saved_mem): Update call
17208 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17209 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17210 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17211 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17212 * expmed.c (extract_bit_field_1): Likewise.
17213 * reload.c (get_secondary_mem): Likewise.
17214 * varasm.c (assemble_integer): Likewise.
17215 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17218 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17220 * machmode.h (decimal_float_mode_for_size): New function.
17221 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17222 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17223 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17224 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17225 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17226 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17228 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17230 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17231 (get_builtin_sync_mode): Likewise.
17232 (expand_ifn_atomic_compare_exchange): Likewise.
17233 (expand_builtin_atomic_clear): Likewise.
17234 (expand_builtin_atomic_test_and_set): Likewise.
17235 (fold_builtin_atomic_always_lock_free): Likewise.
17236 * calls.c (compute_argument_addresses): Likewise.
17237 (emit_library_call_value_1): Likewise.
17238 (store_one_arg): Likewise.
17239 * combine.c (combine_instructions): Likewise.
17240 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17241 * config/arm/arm.c (arm_function_value): Likewise.
17242 (aapcs_allocate_return_reg): Likewise.
17243 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17244 * config/i386/i386.c (construct_container): Likewise.
17245 (ix86_gimplify_va_arg): Likewise.
17246 (ix86_expand_sse_cmp): Likewise.
17247 (emit_memmov): Likewise.
17248 (emit_memset): Likewise.
17249 (expand_small_movmem_or_setmem): Likewise.
17250 (ix86_expand_pextr): Likewise.
17251 (ix86_expand_pinsr): Likewise.
17252 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17253 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17255 * config/mips/mips.c (mips_function_value_1) Likewise.
17256 (mips_block_move_straight): Likewise.
17257 (mips_expand_ins_as_unaligned_store): Likewise.
17258 * config/powerpcspe/powerpcspe.c
17259 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17260 (rs6000_darwin64_record_arg_flush): Likewise.
17261 * config/rs6000/rs6000.c
17262 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17263 (rs6000_darwin64_record_arg_flush): Likewise.
17264 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17265 (sparc_function_value_1): Likewise.
17266 * config/spu/spu.c (adjust_operand): Likewise.
17267 (spu_emit_branch_or_set): Likewise.
17268 (arith_immediate_p): Likewise.
17269 * emit-rtl.c (gen_lowpart_common): Likewise.
17270 * expr.c (expand_expr_real_1): Likewise.
17271 * function.c (assign_parm_setup_block): Likewise.
17272 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17273 * reload1.c (alter_reg): Likewise.
17274 * stor-layout.c (mode_for_vector): Likewise.
17275 (layout_type): Likewise.
17277 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17279 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17280 (spu_convert_move): Likewise.
17281 * lower-subreg.c (resolve_simple_move): Likewise.
17283 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17286 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17287 define_insn to a define_expand.
17288 (altivec_vsum2sws_direct): New define_insn.
17289 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17291 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17293 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17296 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17301 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17302 realign stack if stack alignment needed is less than incoming
17305 2017-09-05 Marek Polacek <polacek@redhat.com>
17308 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17311 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17313 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17315 2017-09-05 Richard Biener <rguenther@suse.de>
17317 PR tree-optimization/82084
17318 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17320 2017-09-05 Richard Biener <rguenther@suse.de>
17322 PR tree-optimization/82102
17323 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17325 2017-09-05 Martin Liska <mliska@suse.cz>
17327 PR tree-optimization/82032
17328 * tree-cfg.c (generate_range_test): New function.
17329 * tree-cfg.h (generate_range_test): Declared here.
17330 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17331 (cleanup_control_expr_graph): Use it.
17332 * tree-switch-conversion.c (try_switch_expansion): Remove
17334 (emit_case_nodes): Use generate_range_test.
17336 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17339 * config/i386/i386.md (*<btsc><mode>_mask): Add
17340 TARGET_USE_BT to insn constraint.
17341 (*btr<mode>_mask): Ditto.
17343 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17345 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17346 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17348 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17351 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17352 TARGET_NEON and TARGET_IWMMXT.
17353 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17354 TARGET_NEON and TARGET_IWMMXT.
17355 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17357 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17359 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17360 (ix86_rewrite_tls_address): Ditto.
17361 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17362 (ix86_rewrite_tls_address_1): Ditto.
17363 (ix86_rewrite_tls_address): Ditto.
17364 * config/i386/predicates.md (tls_address_pattern): New predicate.
17365 * config/i386/i386.md (TLS address splitter): New splitter.
17367 2017-09-04 Richard Biener <rguenther@suse.de>
17369 PR tree-optimization/82084
17370 * fold-const.h (can_native_encode_string_p): Declare.
17371 * fold-const.c (can_native_encode_string_p): Factor out from ...
17372 (native_encode_string): ... here.
17373 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17374 vectorizing stores from constants we later cannot handle.
17376 2017-09-04 Marek Polacek <polacek@redhat.com>
17379 * doc/invoke.texi: Update -Wtautological-compare documentation.
17381 2017-09-04 Jeff Law <law@redhat.com>
17383 PR tree-optimization/64910
17384 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17385 swap the first and last operand if the last is a constant.
17387 2017-09-04 Marek Polacek <polacek@redhat.com>
17390 * convert.c (do_narrow): When sanitizing signed integer overflows,
17391 bail out for signed types.
17392 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17394 2017-09-04 Richard Biener <rguenther@suse.de>
17396 PR tree-optimization/82060
17397 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17398 Move devirtualization after stmt folding and before EH/AB/noreturn
17399 cleanup to get the stmt refs canonicalized. Use a bool instead
17400 of gimple_modified_p since that doesn't work for NOPs. Schedule
17401 NOPs generated by folding for removal.
17403 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17404 Alan Hayward <alan.hayward@arm.com>
17405 David Sherwood <david.sherwood@arm.com>
17407 * coretypes.h (pad_direction): New enum.
17408 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17409 (FUNCTION_ARG_PADDING): Likewise.
17410 * target.def (function_arg_padding): New hook.
17411 * targhooks.h (default_function_arg_padding): Declare.
17412 * targhooks.c (default_function_arg_padding): New function.
17413 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17414 (TARGET_FUNCTION_ARG_PADDING): ...this.
17415 * doc/tm.texi: Regenerate.
17416 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17417 instead of direction.
17418 (compute_argument_addresses): Likewise.
17419 (load_register_parameters): Likewise.
17420 (emit_library_call_value_1): Likewise.
17421 (store_one_arg): Use targetm.calls.function_arg_padding instead
17422 of FUNCTION_ARG_PADDING.
17423 (must_pass_in_stack_var_size_or_pad): Likewise.
17424 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17425 (emit_group_store): Likewise.
17426 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17427 instead of FUNCTION_ARG_PADDING.
17428 (emit_push_insn): Likewise, and propagate enum change throughout
17430 * function.h (direction): Delete.
17431 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17433 * function.c (assign_parm_find_stack_rtl): Likewise.
17434 (assign_parm_setup_block_p): Likewise.
17435 (assign_parm_setup_block): Likewise.
17436 (gimplify_parameters): Likewise.
17437 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17438 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17440 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17441 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17442 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17443 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17444 (aarch64_function_arg_padding): ...this new function.
17445 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17446 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17447 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17448 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17449 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17450 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17451 (arm_pad_arg_upward): Replace with...
17452 (arm_function_arg_padding): ...this new function.
17453 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17455 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17456 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17457 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17458 (ia64_hpux_function_arg_padding): Replace with...
17459 (ia64_function_arg_padding): ...this new function. Use pad_direction
17460 instead of direction. Check for TARGET_HPUX.
17461 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17462 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17463 (iq2000_function_arg_padding): New function.
17464 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17465 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17466 (mips_function_arg_padding): ...this new function.
17467 (mips_pad_reg_upward): Update accordingly.
17468 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17469 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17470 targetm.calls.function_arg_padding.
17471 (FUNCTION_ARG_PADDING): Delete.
17472 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17473 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17474 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17475 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17476 (nios2_block_reg_padding): Return pad_direction instead of direction.
17477 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17478 instead of direction.
17479 (nios2_function_arg_padding): Likewise. Make static.
17480 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17481 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17482 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17483 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17484 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17485 (pa_function_arg_padding): Make static. Return pad_direction instead
17487 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17488 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17489 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17490 instead of direction. Use targetm.calls.function_arg_padding.
17491 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17492 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17493 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17494 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17495 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17497 (function_arg_padding): Rename to...
17498 (rs6000_function_arg_padding): ...this. Make static. Return
17499 pad_direction instead of direction.
17500 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17501 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17502 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17503 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17504 instead of direction. Use targetm.calls.function_arg_padding.
17505 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17506 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17507 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17508 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17509 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17510 (function_arg_padding): Rename to...
17511 (rs6000_function_arg_padding): ...this. Make static. Return
17512 pad_direction instead of direction.
17513 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17514 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17515 * config/s390/s390.c (s390_function_arg_padding): New function.
17516 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17517 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17518 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17519 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17520 (function_arg_padding): Rename to...
17521 (sparc_function_arg_padding): ...this. Make static. Return
17522 pad_direction instead of direction.
17523 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17524 * config/spu/spu.c (spu_function_arg_padding): New function.
17525 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17526 * system.h (FUNCTION_ARG_PADDING): Poison.
17528 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17529 Alan Hayward <alan.hayward@arm.com>
17530 David Sherwood <david.sherwood@arm.com>
17532 * target.def (modes_tieable_p): New hook.
17533 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17534 (TARGET_MODES_TIEABLE_P): ...this.
17535 * doc/tm.texi.in: Regenerate.
17536 * hooks.h (hook_bool_mode_mode_true): Declare.
17537 * hooks.c (hook_bool_mode_mode_true): New function.
17538 * combine.c (subst): Use targetm.modes_tieable_p instead of
17540 * dse.c (find_shift_sequence): Likewise.
17541 * expmed.c (extract_low_bits): Likewise.
17542 * lower-subreg.c: Include target.h.
17543 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17545 * rtlanal.c (rtx_cost): Likewise.
17546 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17547 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17548 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17549 (TARGET_MODES_TIEABLE_P): Redefine.
17550 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17551 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17552 (TARGET_MODES_TIEABLE_P): Redefine.
17553 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17554 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17555 (arc_modes_tieable_p): New function.
17556 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17557 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17558 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17559 (arm_modes_tieable_p): Make static.
17560 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17561 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17562 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17563 (TARGET_MODES_TIEABLE_P): Redefine.
17564 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17565 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17566 (TARGET_MODES_TIEABLE_P): Redefine.
17567 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17568 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17569 (cr16_modes_tieable_p): New function.
17570 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17571 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17572 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17573 (TRULY_NOOP_TRUNCATION): Update comment.
17574 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17575 (TRULY_NOOP_TRUNCATION): Update comment.
17576 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17577 (frv_modes_tieable_p): New function.
17578 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17579 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17580 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17581 (TARGET_MODES_TIEABLE_P): Redefine.
17582 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17583 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17584 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17585 (TARGET_MODES_TIEABLE_P): Redefine.
17586 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17587 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17588 (ia64_modes_tieable_p): New function.
17589 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17590 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17591 (iq2000_modes_tieable_p): New function.
17592 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17593 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17594 (lm32_modes_tieable_p): New function.
17595 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17596 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17597 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17598 (TARGET_MODES_TIEABLE_P): Redefine.
17599 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17600 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17601 (m32r_modes_tieable_p): New function.
17602 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17603 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17604 (m68k_modes_tieable_p): New function.
17605 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17606 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17607 (mcore_modes_tieable_p): New function.
17608 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17609 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17611 (TARGET_MODES_TIEABLE_P): Redefine.
17612 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17613 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17614 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17615 (TARGET_MODES_TIEABLE_P): Redefine.
17616 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17617 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17618 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17619 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17620 (mn10300_modes_tieable_p): ...this and make static.
17621 (TARGET_MODES_TIEABLE_P): Redefine.
17622 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17623 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17624 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17625 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17626 (msp430_modes_tieable_p): Make static.
17627 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17628 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17629 (TARGET_MODES_TIEABLE_P): Redefine.
17630 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17631 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17632 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17633 (TARGET_MODES_TIEABLE_P): Redefine.
17634 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17635 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17636 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17637 (TARGET_MODES_TIEABLE_P): Redefine.
17638 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17639 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17640 (pdp11_modes_tieable_p): New function.
17641 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17642 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17643 (rs6000_modes_tieable_p): New function.
17644 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17645 * config/powerpcspe/powerpcspe.md: Update comment.
17646 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17647 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17648 (TARGET_MODES_TIEABLE_P): Redefine.
17649 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17650 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17651 (rl78_modes_tieable_p): New function.
17652 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17653 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17654 (rs6000_modes_tieable_p): New function.
17655 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17656 * config/rs6000/rs6000.md: Update comment.
17657 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17658 * config/rx/rx.c (rx_modes_tieable_p): New function.
17659 (TARGET_MODES_TIEABLE_P): Redefine.
17660 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17661 * config/s390/s390.c (s390_modes_tieable_p): New function.
17662 (TARGET_MODES_TIEABLE_P): Redefine.
17663 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17664 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17665 (sh_modes_tieable_p): New function.
17666 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17667 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17668 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17669 (sparc_modes_tieable_p): Make static.
17670 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17671 * config/spu/spu.c (spu_modes_tieable_p): New function.
17672 (TARGET_MODES_TIEABLE_P): Redefine.
17673 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17674 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17675 (TARGET_MODES_TIEABLE_P): Redefine.
17676 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17677 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17678 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17679 * config/v850/v850.c (v850_modes_tieable_p): New function.
17680 (TARGET_MODES_TIEABLE_P): Redefine.
17681 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17682 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17683 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17684 (visium_modes_tieable_p): New function.
17685 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17686 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17687 (xtensa_modes_tieable_p): New function.
17688 * system.h (MODES_TIEABLE_P): Poison.
17690 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17691 Alan Hayward <alan.hayward@arm.com>
17692 David Sherwood <david.sherwood@arm.com>
17694 * target.def (hard_regno_mode_ok): New hook.
17695 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17696 (TARGET_HARD_REGNO_MODE_OK): ...this.
17697 * doc/tm.texi.in: Regenerate.
17698 * hooks.h (hook_bool_uint_mode_true): Declare.
17699 * hooks.c (hook_bool_uint_mode_true): New function.
17700 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17701 HARD_REGNO_MODE_OK.
17702 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17703 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17704 instead of HARD_REGNO_MODE_OK.
17705 * caller-save.c: Include target.h.
17706 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17707 HARD_REGNO_MODE_OK.
17708 * combine.c (can_combine_p): Likewise.
17709 (combinable_i3pat): Likewise.
17710 (can_change_dest_mode): Likewise.
17711 * expr.c (init_expr_target): Likewise.
17712 (convert_move): Likewise.
17713 (convert_modes): Likewise.
17714 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17715 (setup_prohibited_mode_move_regs): Likewise.
17716 * ira.h (target_ira): Likewise.
17717 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17718 * lra-constraints.c (process_alt_operands): Likewise.
17719 (split_reg): Likewise.
17720 * recog.c (peep2_find_free_register): Likewise.
17721 * ree.c (combine_reaching_defs): Likewise.
17722 * regcprop.c (maybe_mode_change): Likewise.
17723 * reginfo.c (init_reg_sets_1): Likewise.
17724 (choose_hard_reg_mode): Likewise.
17725 (simplifiable_subregs): Likewise.
17726 * regrename.c (check_new_reg_p): Likewise.
17727 * reload.c (find_valid_class): Likewise.
17728 (find_valid_class_1): Likewise.
17729 (reload_inner_reg_of_subreg): Likewise.
17730 (push_reload): Likewise.
17731 (combine_reloads): Likewise.
17732 (find_dummy_reload): Likewise.
17733 (find_reloads): Likewise.
17734 * reload1.c (find_reg): Likewise.
17735 (set_reload_reg): Likewise.
17736 (allocate_reload_reg): Likewise.
17737 (choose_reload_regs): Likewise.
17738 (reload_adjust_reg_for_temp): Likewise.
17739 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17740 (simplify_subreg_regno): Likewise.
17741 * sel-sched.c (init_regs_for_mode): Likewise.
17742 * varasm.c (make_decl_rtl): Likewise.
17743 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17744 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17745 HARD_REGNO_MODE_OK.
17746 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17747 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17748 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17749 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17750 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17751 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17752 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17753 (arc_mode_class): Delete.
17754 (HARD_REGNO_MODE_OK): Delete.
17755 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17756 (arc_hard_regno_mode_ok): Rename old array to...
17757 (arc_hard_regno_mode_ok_modes): ...this.
17758 (arc_conditional_register_usage): Update accordingly.
17759 (arc_mode_class): Make static.
17760 (arc_hard_regno_mode_ok): New function.
17761 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17762 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17763 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17764 (arm_hard_regno_mode_ok): Make static.
17765 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17766 HARD_REGNO_MODE_OK.
17767 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17768 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17769 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17771 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17772 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17773 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17774 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17775 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17776 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17777 * config/bfin/predicates.md (valid_reg_operand): Use
17778 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17779 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17780 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17781 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17782 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17783 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17784 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17785 (cr16_hard_regno_mode_ok): Make static and return a bool.
17786 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17787 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17788 (cris_hard_regno_mode_ok): New function.
17789 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17790 (epiphany_mode_class): Delete.
17791 (HARD_REGNO_MODE_OK): Delete.
17792 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17793 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17794 (hard_regno_mode_ok): Rename to...
17795 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17796 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17797 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17798 HARD_REGNO_MODE_OK.
17799 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17800 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17801 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17802 (frv_hard_regno_mode_ok): Make static and return a bool.
17803 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17804 HARD_REGNO_MODE_OK.
17805 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17806 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17807 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17808 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17810 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17811 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17812 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17813 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17815 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17816 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17817 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17818 (ia64_hard_regno_mode_ok): New function.
17819 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17820 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17821 (iq2000_hard_regno_mode_ok): New function.
17822 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17823 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17824 (lm32_hard_regno_mode_ok): New function.
17825 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17826 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17827 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17828 instead of HARD_REGNO_MODE_OK.
17829 (m32c_hard_regno_ok): Rename to...
17830 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17831 (m32c_cannot_change_mode_class): Update accordingly.
17832 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17833 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17834 (m32r_mode_class): Delete.
17835 (HARD_REGNO_MODE_OK): Delete.
17836 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17837 (m32r_hard_regno_mode_ok): Rename to...
17838 (m32r_hard_regno_modes): ...this.
17839 (m32r_mode_class): Make static.
17840 (m32r_hard_regno_mode_ok): New function.
17841 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17842 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17843 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17844 (m68k_hard_regno_mode_ok): Make static.
17845 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17846 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17847 (mcore_hard_regno_mode_ok): New function.
17848 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17849 (HARD_REGNO_MODE_OK): Delete.
17850 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17852 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17853 (microblaze_hard_regno_mode_ok): New function.
17854 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17855 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17856 (mips_hard_regno_mode_ok): Delete.
17857 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17858 (mips_hard_regno_mode_ok_p): ...this and make static.
17859 (mips_hard_regno_mode_ok_p): Rename to...
17860 (mips_hard_regno_mode_ok_uncached): ...this.
17861 (mips_hard_regno_mode_ok): New function.
17862 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17863 of HARD_REGNO_MODE_OK.
17864 (mips_option_override): Update after above name changes.
17865 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17866 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17867 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17868 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17869 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17870 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17871 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17872 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17873 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17874 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17875 (msp430_hard_regno_mode_ok): Make static and return a bool.
17876 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17877 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17878 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17880 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17881 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17882 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17883 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17884 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17885 (PA_HARD_REGNO_MODE_OK): ...this
17886 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17887 (PA_HARD_REGNO_MODE_OK): ...this.
17888 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17889 (pa_hard_regno_mode_ok): New function.
17890 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17891 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17892 (pdp11_hard_regno_mode_ok): New function.
17893 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17894 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17896 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17898 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17899 (rs6000_hard_regno_mode_ok): Rename to...
17900 (rs6000_hard_regno_mode_ok_uncached): ...this.
17901 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17902 (rs6000_hard_regno_mode_ok): New function.
17903 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17904 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17905 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17906 (riscv_hard_regno_mode_ok): ...this and make static.
17907 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17908 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17909 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17910 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17911 (rl78_hard_regno_mode_ok): Make static and return bool.
17912 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17913 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17915 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17916 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17917 (rs6000_hard_regno_mode_ok): Rename to...
17918 (rs6000_hard_regno_mode_ok_uncached): ...this.
17919 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17920 (rs6000_hard_regno_mode_ok): New function.
17921 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17922 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17923 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17924 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17925 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17926 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17927 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17928 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17929 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17930 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17931 (sh_hard_regno_mode_ok): Make static.
17932 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17933 instead of HARD_REGNO_MODE_OK.
17934 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17935 (sparc_mode_class): Delete.
17936 (HARD_REGNO_MODE_OK): Delete.
17937 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17938 (hard_regno_mode_classes): Make static.
17939 (sparc_mode_class): Likewise.
17940 (sparc_hard_regno_mode_ok): New function.
17941 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17942 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17943 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17945 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17946 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17947 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17948 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17949 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17950 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17951 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17952 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17953 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17954 (visium_hard_regno_mode_ok): New function.
17955 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17956 instead of HARD_REGNO_MODE_OK.
17957 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17958 (HARD_REGNO_MODE_OK): Delete.
17959 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17960 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17961 (xtensa_option_override): Update accordingly.
17962 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17963 (xtensa_hard_regno_mode_ok): New function.
17964 * system.h (HARD_REGNO_MODE_OK): Poison.
17966 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17967 Alan Hayward <alan.hayward@arm.com>
17968 David Sherwood <david.sherwood@arm.com>
17970 * target.def (hard_regno_call_part_clobbered): New hook.
17971 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17972 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17973 * doc/tm.texi: Regenerate.
17974 * hooks.h (hook_bool_uint_mode_false): Declare.
17975 * hooks.c (hook_bool_uint_mode_false): New function.
17976 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17977 * cselib.c (cselib_process_insn): Use
17978 targetm.hard_regno_call_part_clobbered instead of
17979 HARD_REGNO_CALL_PART_CLOBBERED.
17980 * ira-conflicts.c (ira_build_conflicts): Likewise.
17981 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17982 * lra-constraints.c (need_for_call_save_p): Likewise.
17983 * lra-lives.c: Include target.h.
17984 (check_pseudos_live_through_calls): Use
17985 targetm.hard_regno_call_part_clobbered instead of
17986 HARD_REGNO_CALL_PART_CLOBBERED.
17987 * regcprop.c: Include target.h.
17988 (copyprop_hardreg_forward_1): Use
17989 targetm.hard_regno_call_part_clobbered instead of
17990 HARD_REGNO_CALL_PART_CLOBBERED.
17991 * reginfo.c (choose_hard_reg_mode): Likewise.
17992 * regrename.c (check_new_reg_p): Likewise.
17993 * reload.c (find_equiv_reg): Likewise.
17994 * reload1.c (emit_reload_insns): Likewise.
17995 * sched-deps.c (deps_analyze_insn): Likewise.
17996 * sel-sched.c (init_regs_for_mode): Likewise.
17997 (mark_unavailable_hard_regs): Likewise.
17998 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17999 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18000 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
18002 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18003 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18004 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
18006 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
18008 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18009 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18010 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
18012 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18013 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18014 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
18016 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18017 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
18019 * config/powerpcspe/powerpcspe.c
18020 (rs6000_hard_regno_call_part_clobbered): New function.
18021 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18022 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18023 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
18025 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18026 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18027 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
18029 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18030 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18031 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
18033 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18034 Alan Hayward <alan.hayward@arm.com>
18035 David Sherwood <david.sherwood@arm.com>
18037 * rtl.h (subreg_memory_offset): Declare.
18038 * emit-rtl.c (subreg_memory_offset): New function.
18039 * expmed.c (store_bit_field_1): Use it.
18040 * expr.c (undefined_operand_subword_p): Likewise.
18041 * simplify-rtx.c (simplify_subreg): Likewise.
18043 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
18045 PR rtl-optimization/57448
18048 * optabs.c (expand_atomic_load): Place compiler memory barriers if
18049 using atomic_load pattern.
18050 (expand_atomic_store): Likewise.
18052 2017-09-04 Jakub Jelinek <jakub@redhat.com>
18055 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
18056 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
18057 handling. Use replace_call_with_value with NULL instead of
18058 gsi_replace, unlink_stmt_vdef and release_defs.
18060 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
18063 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
18065 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18068 * rtl.h (emit_library_call_value_1): Declare.
18069 (emit_library_call): Replace declaration with a series of overloads.
18070 Remove the parameter count argument.
18071 (emit_library_call_value): Likewise.
18072 * calls.c (emit_library_call_value_1): Make global. Replace varargs
18073 with an "rtx_mode_t *".
18074 (emit_library_call_value): Delete.
18075 (emit_library_call): Likewise.
18076 * asan.c (asan_emit_stack_protection): Update calls accordingly.
18077 (asan_emit_allocas_unpoison): Likewise.
18078 * builtins.c (expand_builtin_powi): Likewise.
18079 (expand_asan_emit_allocas_unpoison): Likewise.
18080 * cfgexpand.c (expand_main_function): Likewise.
18081 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
18082 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
18083 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
18084 * config/arm/arm.c (arm_trampoline_init): Likewise.
18085 (arm_call_tls_get_addr): Likewise.
18086 (arm_expand_divmod_libfunc): Likewise.
18087 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
18088 (smulsi3_highpart): Likewise.
18089 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
18090 (c6x_expand_compare): Likewise.
18091 (c6x_expand_movmem): Likewise.
18092 * config/frv/frv.c (frv_trampoline_init): Likewise.
18093 * config/i386/i386.c (ix86_trampoline_init): Likewise.
18094 (ix86_expand_divmod_libfunc): Likewise.
18095 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
18096 (ia64_expand_compare): Likewise.
18097 (ia64_profile_hook): Likewise.
18098 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
18099 (nonlocal_goto): Likewise.
18100 (restore_stack_nonlocal): Likewise.
18101 * config/m32r/m32r.c (block_move_call): Likewise.
18102 (m32r_trampoline_init): Likewise.
18103 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
18104 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
18105 (m68k_call_m68k_read_tp): Likewise.
18106 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
18107 (microblaze_expand_divide): Likewise.
18108 * config/mips/mips.h (mips_args): Likewise.
18109 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
18110 (MIPS_ICACHE_SYNC): Likewise.
18111 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
18112 (nios2_trampoline_init): Likewise.
18113 * config/pa/pa.c (hppa_tls_call): Likewise.
18114 (pa_trampoline_init): Likewise.
18115 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
18116 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
18117 (expand_strn_compare): Likewise.
18118 (rs6000_generate_compare): Likewise.
18119 (rs6000_expand_float128_convert): Likewise.
18120 (output_profile_hook): Likewise.
18121 (rs6000_trampoline_init): Likewise.
18122 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
18123 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
18124 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
18125 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
18126 (rs6000_generate_compare): Likewise.
18127 (rs6000_expand_float128_convert): Likewise.
18128 (output_profile_hook): Likewise.
18129 (rs6000_trampoline_init): Likewise.
18130 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
18131 * config/sh/sh.c (sh_trampoline_init): Likewise.
18132 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
18133 (sparc_emit_float_lib_cmp): Likewise.
18134 (sparc32_initialize_trampoline): Likewise.
18135 (sparc64_initialize_trampoline): Likewise.
18136 (sparc_profile_hook): Likewise.
18137 * config/spu/spu.c (ea_load_store): Likewise.
18138 * config/spu/spu.md (floatunssidf2): Likewise.
18139 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
18140 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
18141 * config/visium/visium.c (expand_block_move_4): Likewise.
18142 (expand_block_move_2): Likewise.
18143 (expand_block_move_1): Likewise.
18144 (expand_block_set_4): Likewise.
18145 (expand_block_set_2): Likewise.
18146 (expand_block_set_1): Likewise.
18147 (visium_trampoline_init): Likewise.
18148 (visium_profile_hook): Likewise.
18149 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
18150 (xtensa_setup_frame_addresses): Likewise.
18151 (xtensa_trampoline_init): Likewise.
18152 * except.c (sjlj_emit_function_enter): Likewise.
18153 (sjlj_emit_function_exit): Likewise.
18154 * explow.c (allocate_dynamic_stack_space): Likewise.
18155 (probe_stack_range): Likewise.
18156 * expr.c (convert_mode_scalar): Likewise.
18157 * optabs.c (expand_binop): Likewise.
18158 (expand_twoval_binop_libfunc): Likewise.
18159 (expand_unop): Likewise.
18160 (prepare_cmp_insn): Likewise.
18161 (prepare_float_lib_cmp): Likewise.
18162 (expand_float): Likewise.
18163 (expand_fix): Likewise.
18164 (expand_fixed_convert): Likewise.
18165 (maybe_emit_sync_lock_test_and_set): Likewise.
18166 (expand_atomic_compare_and_swap): Likewise.
18167 (expand_mem_thread_fence): Likewise.
18168 (expand_atomic_fetch_op): Likewise.
18170 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
18172 * doc/generic.texi (OpenACC): Adjust URL.
18173 * doc/invoke.texi (C Dialect Options): Ditto.
18175 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
18177 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
18178 predicate for operand 1. Add (m,<S>) constraint.
18179 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
18180 Prevent memory operand 1 with register operand 2.
18182 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
18184 PR rtl-optimization/82024
18185 * combine.c (try_combine): If the combination result is a PARALLEL,
18186 and we only need to retain the SET in there that would be placed
18187 at I2, check that we can place that at I3 instead, before doing so.
18189 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18192 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
18194 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
18195 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
18198 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
18199 Jeff Law <law@redhat.com>
18201 * varasm.c (bss_initializer_p): Do not put constants into .bss
18202 (categorize_decl_for_section): Handle bss_initializer_p returning
18203 false when DECL_INITIAL is NULL.
18205 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18208 * config/s390/s390.c (s390_can_inline_p): New function.
18210 2017-09-01 Jeff Law <law@redhat.com>
18212 PR tree-optimization/82052
18213 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18214 Always initialize the returned slot after a hash table miss
18215 when INSERT is true.
18217 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18219 * config/s390/s390.md (mem_signal_fence): Remove.
18220 * doc/md.texi (mem_signal_fence): Remove.
18221 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18223 * target-insns.def (mem_signal_fence): Remove.
18225 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18228 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18231 * asan.c (create_odr_indicator): Strip name encoding from assembler
18232 name before appending it after __odr_asan_.
18234 2017-09-01 Martin Liska <mliska@suse.cz>
18236 PR tree-optimization/82059
18237 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18238 frequency only when an edge is redirected.
18240 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18242 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18243 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18244 (arc_conditional_register_usage): Remove ARC600 lp_count
18246 (arc_file_start): Emit Tag_ARC_CPU_variation.
18247 (arc_can_use_doloop_p): New conditions to use ZOLs.
18248 (hwloop_fail): New function.
18249 (hwloop_optimize): Likewise.
18250 (hwloop_pattern_reg): Likewise.
18251 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18252 (arc_reorg_loops): New function, calls reorg_loops.
18253 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18254 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18256 (arc_loop_hazard): Remove function, functionality moved into
18258 (arc_hazard): Remove arc_loop_hazard call.
18259 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18260 into hwloop_optimize.
18261 (arc_label_align): Remove ZOL handling.
18262 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18263 * config/arc/arc.md (doloop_begin): Remove pattern.
18264 (doloop_begin_i): Likewise.
18265 (doloop_end_i): Likewise.
18266 (doloop_fallback): Likewise.
18267 (doloop_fallback_m): Likewise.
18268 (doloop_end): Reimplement expand.
18269 (arc_lp): New pattern for LP instruction.
18270 (loop_end): New pattern.
18271 (loop_fail): Likewise.
18272 (decrement_and_branch_until_zero): Likewise.
18273 * config/arc/arc.opt (mlpc-width): New option.
18274 * doc/invoke.texi (mlpc-width): Document option.
18276 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18278 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18279 (arc_ccfsm_advance): Fix checking for delay slots.
18280 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18282 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18284 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18286 (movhi_insn): Update store instruction constraint which are saving
18287 6-bit short immediates.
18288 (movsi_insn): Consider also short scaled load operations.
18289 (zero_extendhisi2_i): Use Usd constraint instead of T.
18290 (extendhisi2_i): Add q constraint.
18291 (arc_clzsi2): Add type and length attributes.
18292 (arc_ctzsi2): Likewise.
18293 * config/arc/constraints.md (Usc): Update constraint, the
18294 assembler can parse two relocations for a single instruction.
18296 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18298 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18299 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18301 2017-08-31 Olivier Hainque <hainque@adacore.com>
18303 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18304 match as powerpc-wrs-vxworks*.
18306 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18308 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18309 register constraint for by-element operand.
18310 (aarch64_mls_elt_merge<mode>): Likewise.
18312 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18314 * config/arc/arc.c (arc_can_follow_jump): Check for short
18317 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18319 * config.gcc: Use g.opt for arc.
18320 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18321 functionality moved to ...
18322 (legitimate_scaled_address_p): New function, ...here.
18323 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18324 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18325 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18327 (arc_override_options): Handle G option.
18328 (arc_output_pic_addr_const): Correct function definition.
18329 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18330 (arc_decl_anon_ns_mem_p): Delete.
18331 (arc_in_small_data_p): Overhaul this function to take into
18332 consideration the value given via G option.
18333 (arc_rewrite_small_data_1): Renamed and corrected old
18334 arc_rewrite_small_data function.
18335 (arc_rewrite_small_data): New function.
18336 (small_data_pattern): Don't use pic_offset_table_rtx.
18337 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18338 * config/arc/simdext.md (movmisalignv2hi): Use
18339 prepare_move_operands function.
18341 (movmisalign*): Likewise.
18342 * doc/invoke.texi (ARC options): Document -G option.
18344 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18346 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18348 * config/arc/arc.c (arc_print_operand): Output scalled address for
18349 sdata whenever is possible.
18350 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18351 load/stores are available.
18352 (compact_sda_memory_operand): Check for the alignment required by
18353 code density instructions.
18354 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18356 * config/arc/constraints.md (Usd): Update constraint.
18357 (Us0): New constraint.
18358 (Usc): Update constraint.
18360 2017-08-31 Richard Biener <rguenther@suse.de>
18362 PR middle-end/82054
18363 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18364 function only once.
18366 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18368 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18369 Resize type_signature.
18371 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18372 Alan Hayward <alan.hayward@arm.com>
18373 David Sherwood <david.sherwood@arm.com>
18375 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18376 subregs whose inner modes can be stored in GPRs.
18377 (aarch64_classify_index): Likewise.
18379 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18380 Alan Hayward <alan.hayward@arm.com>
18381 David Sherwood <david.sherwood@arm.com>
18383 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18384 (V_INT_EQUIV): ...this.
18385 (v_cmp_result): Rename to...
18386 (v_int_equiv): ...this.
18387 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18388 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18389 (copysign<mode>3): Likewise.
18390 (aarch64_simd_bsl<mode>_internal): Likewise.
18391 (aarch64_simd_bsl<mode>): Likewise.
18392 (vec_cmp<mode><mode>): Likewise.
18393 (vcond<mode><mode>): Likewise.
18394 (vcond<v_cmp_mixed><mode>): Likewise.
18395 (vcondu<mode><v_cmp_mixed>): Likewise.
18396 (aarch64_cm<optab><mode>): Likewise.
18397 (aarch64_cmtst<mode>): Likewise.
18398 (aarch64_fac<optab><mode>): Likewise.
18399 (vec_perm_const<mode>): Likewise.
18400 (vcond_mask_<mode><v_cmp_result>): Rename to...
18401 (vcond_mask_<mode><v_int_equiv>): ...this.
18402 (vec_cmp<mode><v_cmp_result>): Rename to...
18403 (vec_cmp<mode><v_int_equiv>): ...this.
18405 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18406 Alan Hayward <alan.hayward@arm.com>
18407 David Sherwood <david.sherwood@arm.com>
18409 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18411 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18412 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18413 (UNSPEC_LD4_DREG): New unspecs.
18414 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18415 (aarch64_ld2<mode>_dreg_be): Replace with...
18416 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18418 (aarch64_ld3<mode>_dreg_le)
18419 (aarch64_ld3<mode>_dreg_be): Replace with...
18420 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18422 (aarch64_ld4<mode>_dreg_le)
18423 (aarch64_ld4<mode>_dreg_be): Replace with...
18424 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18427 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18429 PR tree-optimization/81987
18430 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18431 insert an initializer in a location not dominated by the stride
18434 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18436 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18437 on the entire header of the finally block in the fallthru case.
18439 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18441 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18443 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18445 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18446 rs6000_emit_move_from_cr and call renamed function.
18447 (rs6000_emit_prologue): Call renamed functions.
18448 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18449 movesi_from_cr, remove volatile CRs.
18451 2017-08-30 Jon Beniston <jon@beniston.com>
18452 Richard Biener <rguenther@suse.de>
18454 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18455 of VECTOR_MODE_P check.
18456 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18457 element vector types.
18459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18461 * df.h (df_read_modify_subreg_p): Remove in favor of...
18462 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18463 const_rtx instead of an rtx.
18464 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18465 * df-problems.c (df_word_lr_mark_ref): Likewise.
18466 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18467 (mark_pseudo_reg_dead): Likewise.
18468 (mark_ref_dead): Likewise.
18469 * reginfo.c (init_subregs_of_mode): Likewise.
18470 * sched-deps.c (sched_analyze_1): Likewise.
18471 * df-scan.c (df_def_record_1): Likewise.
18472 (df_uses_record): Likewise.
18473 (df_read_modify_subreg_p): Remove in favor of...
18474 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18475 const_rtx instead of an rtx.
18477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18478 Alan Hayward <alan.hayward@arm.com>
18479 David Sherwood <david.sherwood@arm.com>
18481 * rtl.h (partial_subreg_p): New function.
18482 * caller-save.c (save_call_clobbered_regs): Use it.
18483 * calls.c (expand_call): Likewise.
18484 * combine.c (combinable_i3pat): Likewise.
18485 (simplify_set): Likewise.
18486 (make_extraction): Likewise.
18487 (make_compound_operation_int): Likewise.
18488 (gen_lowpart_or_truncate): Likewise.
18489 (force_to_mode): Likewise.
18490 (make_field_assignment): Likewise.
18491 (reg_truncated_to_mode): Likewise.
18492 (record_truncated_value): Likewise.
18493 (move_deaths): Likewise.
18494 * cse.c (record_jump_cond): Likewise.
18495 (cse_insn): Likewise.
18496 * cselib.c (cselib_lookup_1): Likewise.
18497 * expmed.c (extract_bit_field_using_extv): Likewise.
18498 * function.c (assign_parm_setup_reg): Likewise.
18499 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18500 * ira-build.c (create_insn_allocnos): Likewise.
18501 * lra-coalesce.c (merge_pseudos): Likewise.
18502 * lra-constraints.c (match_reload): Likewise.
18503 (simplify_operand_subreg): Likewise.
18504 (curr_insn_transform): Likewise.
18505 * lra-lives.c (process_bb_lives): Likewise.
18506 * lra.c (new_insn_reg): Likewise.
18507 (lra_substitute_pseudo): Likewise.
18508 * regcprop.c (mode_change_ok): Likewise.
18509 (maybe_mode_change): Likewise.
18510 (copyprop_hardreg_forward_1): Likewise.
18511 * reload.c (push_reload): Likewise.
18512 (find_reloads): Likewise.
18513 (find_reloads_subreg_address): Likewise.
18514 * reload1.c (alter_reg): Likewise.
18515 (eliminate_regs_1): Likewise.
18516 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18518 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18520 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18521 back to if statements, including unpack.
18523 2017-08-30 Martin Liska <mliska@suse.cz>
18525 PR inline-asm/82001
18526 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18528 (func_checker::compare_asm_inputs_outputs): ... this function.
18529 (func_checker::compare_gimple_asm): Use the function to compare
18530 also ASM constrains.
18531 * ipa-icf-gimple.h: Rename the function.
18533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18534 Alan Hayward <alan.hayward@arm.com>
18535 David Sherwood <david.sherwood@arm.com>
18537 * coretypes.h (complex_mode): New type.
18538 * gdbhooks.py (build_pretty_printer): Handle it.
18539 * machmode.h (complex_mode): New class.
18540 (complex_mode::includes_p): New function.
18541 (is_complex_int_mode): Likewise.
18542 (is_complex_float_mode): Likewise.
18543 * genmodes.c (get_mode_class): Handle complex mode classes.
18544 * function.c (expand_function_end): Use is_complex_int_mode.
18546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18547 Alan Hayward <alan.hayward@arm.com>
18548 David Sherwood <david.sherwood@arm.com>
18550 * coretypes.h (scalar_mode_pod): New typedef.
18551 * gdbhooks.py (build_pretty_printer): Handle it.
18552 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18553 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18554 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18555 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18556 as_a <scalar_mode>.
18558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18559 Alan Hayward <alan.hayward@arm.com>
18560 David Sherwood <david.sherwood@arm.com>
18562 * machmode.h (mode_for_vector): Take a scalar_mode instead
18564 * stor-layout.c (mode_for_vector): Likewise.
18565 * explow.c (promote_mode): Use as_a <scalar_mode>.
18566 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18569 Alan Hayward <alan.hayward@arm.com>
18570 David Sherwood <david.sherwood@arm.com>
18572 * target.def (preferred_simd_mode): Take a scalar_mode
18573 instead of a machine_mode.
18574 * targhooks.h (default_preferred_simd_mode): Likewise.
18575 * targhooks.c (default_preferred_simd_mode): Likewise.
18576 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18577 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18578 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18579 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18580 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18581 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18582 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18583 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18585 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18586 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18587 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18588 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18589 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18590 * doc/tm.texi: Regenerate.
18591 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18595 Alan Hayward <alan.hayward@arm.com>
18596 David Sherwood <david.sherwood@arm.com>
18598 * target.def (scalar_mode_supported_p): Take a scalar_mode
18599 instead of a machine_mode.
18600 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18601 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18602 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18603 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18604 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18605 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18606 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18607 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18608 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18609 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18610 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18611 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18612 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18613 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18615 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18616 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18617 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18618 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18619 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18621 * doc/tm.texi: Regenerate.
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 * coretypes.h (opt_scalar_mode): New typedef.
18628 * gdbhooks.py (build_pretty_printers): Handle it.
18629 * machmode.h (mode_iterator::get_2xwider): Add overload for
18631 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18633 * expr.c (convert_mode_scalar): Likewise.
18634 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18635 * optabs.c (expand_float): Likewise.
18636 (expand_fix): Likewise.
18637 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
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 * optabs.c (expand_float): Explicitly check for scalars before
18644 using a branching expansion.
18645 (expand_fix): Likewise.
18647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18648 Alan Hayward <alan.hayward@arm.com>
18649 David Sherwood <david.sherwood@arm.com>
18651 * expr.c (convert_mode): Split scalar handling out into...
18652 (convert_mode_scalar): ...this new function. Treat the modes
18655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18656 Alan Hayward <alan.hayward@arm.com>
18657 David Sherwood <david.sherwood@arm.com>
18659 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18661 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18664 Alan Hayward <alan.hayward@arm.com>
18665 David Sherwood <david.sherwood@arm.com>
18667 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18668 rather than a machine_mode.
18669 (fixed_from_string): Likewise.
18670 (fixed_convert): Likewise.
18671 (fixed_convert_from_int): Likewise.
18672 (fixed_convert_from_real): Likewise.
18673 (real_convert_from_fixed): Likewise.
18674 * fixed-value.c (fixed_from_double_int): Likewise.
18675 (fixed_from_string): Likewise.
18676 (fixed_convert): Likewise.
18677 (fixed_convert_from_int): Likewise.
18678 (fixed_convert_from_real): Likewise.
18679 (real_convert_from_fixed): Likewise.
18680 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18683 Alan Hayward <alan.hayward@arm.com>
18684 David Sherwood <david.sherwood@arm.com>
18686 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18687 of separate mode class checks. Do not allow vector modes here.
18688 (immed_wide_int_const): Use as_a <scalar_mode>.
18689 * explow.c (trunc_int_for_mode): Likewise.
18690 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18691 (wi::shwi): Likewise.
18692 (wi::min_value): Likewise.
18693 (wi::max_value): Likewise.
18694 * dwarf2out.c (loc_descriptor): Likewise.
18695 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18696 for CONST_WIDE_INT.
18698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18699 Alan Hayward <alan.hayward@arm.com>
18700 David Sherwood <david.sherwood@arm.com>
18702 * tree.h (SCALAR_TYPE_MODE): New macro.
18703 * expr.c (expand_expr_addr_expr_1): Use it.
18704 (expand_expr_real_2): Likewise.
18705 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18706 (fold_convert_const_fixed_from_int): Likewise.
18707 (fold_convert_const_fixed_from_real): Likewise.
18708 (native_encode_fixed): Likewise
18709 (native_encode_complex): Likewise
18710 (native_encode_vector): Likewise.
18711 (native_interpret_fixed): Likewise.
18712 (native_interpret_real): Likewise.
18713 (native_interpret_complex): Likewise.
18714 (native_interpret_vector): Likewise.
18715 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18716 (simd_clone_adjust_argument_types): Likewise.
18717 (simd_clone_init_simd_arrays): Likewise.
18718 (simd_clone_adjust): Likewise.
18719 * stor-layout.c (layout_type): Likewise.
18720 * tree.c (build_minus_one_cst): Likewise.
18721 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18722 * tree-inline.c (estimate_move_cost): Likewise.
18723 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18724 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18725 (vectorizable_reduction): Likewise.
18726 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18727 (vect_recog_mixed_size_cond_pattern): Likewise.
18728 (check_bool_pattern): Likewise.
18729 (adjust_bool_pattern): Likewise.
18730 (search_type_for_mask_1): Likewise.
18731 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18732 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18733 (vectorizable_load): Likewise.
18734 (vectorizable_store): Likewise.
18735 * ubsan.c (ubsan_encode_value): Likewise.
18736 * varasm.c (output_constant): Likewise.
18738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18739 Alan Hayward <alan.hayward@arm.com>
18740 David Sherwood <david.sherwood@arm.com>
18742 * coretypes.h (scalar_mode): New class.
18743 * machmode.h (scalar_mode): Likewise.
18744 (scalar_mode::includes_p): New function.
18745 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18746 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18747 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18748 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18749 * expmed.c (store_bit_field_1): Likewise.
18750 (extract_bit_field_1): Likewise.
18751 * expr.c (write_complex_part): Likewise.
18752 (read_complex_part): Likewise.
18753 (emit_move_complex_push): Likewise.
18754 (expand_expr_real_2): Likewise.
18755 * function.c (assign_parm_setup_reg): Likewise.
18756 (assign_parms_unsplit_complex): Likewise.
18757 * optabs.c (expand_binop): Likewise.
18758 * rtlanal.c (subreg_get_info): Likewise.
18759 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18760 * varasm.c (output_constant_pool_2): Likewise.
18762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18763 Alan Hayward <alan.hayward@arm.com>
18764 David Sherwood <david.sherwood@arm.com>
18766 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18768 (expmed_mult_highpart_optab): Likewise.
18769 (expmed_mult_highpart): Likewise.
18771 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18772 Alan Hayward <alan.hayward@arm.com>
18773 David Sherwood <david.sherwood@arm.com>
18775 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18776 instead of a machine_mode.
18777 (builtin_memset_read_str): Likewise.
18778 * builtins.c (c_readstr): Likewise.
18779 (builtin_memcpy_read_str): Likewise.
18780 (builtin_strncpy_read_str): Likewise.
18781 (builtin_memset_read_str): Likewise.
18782 (builtin_memset_gen_str): Likewise.
18783 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18784 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18785 instead of a machine_mode.
18786 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18788 (make_extraction): Likewise.
18789 (try_widen_shift_mode): Take and return scalar_int_modes instead
18791 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18792 a scalar_int_mode instead of a machine_mode.
18793 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18794 (avr_addr_space_pointer_mode): Likewise.
18795 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18796 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18797 (msp430_unwind_word_mode): Likewise.
18798 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18799 (spu_addr_space_pointer_mode): Likewise.
18800 (spu_addr_space_address_mode): Likewise.
18801 (spu_libgcc_cmp_return_mode): Likewise.
18802 (spu_libgcc_shift_count_mode): Likewise.
18803 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18804 (rl78_addr_space_pointer_mode): Likewise.
18805 (fl78_unwind_word_mode): Likewise.
18806 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18808 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18809 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18810 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18811 (mips_valid_pointer_mode): Likewise.
18812 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18813 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18814 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18816 (ft32_addr_space_address_mode): Likewise.
18817 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18818 scalar_int_mode instead of a machine_mode.
18819 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18821 (m32c_addr_space_address_mode): Likewise.
18822 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18823 (rs6000_eh_return_filter_mode): Likewise.
18824 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18825 (rs6000_eh_return_filter_mode): Likewise.
18826 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18827 (s390_libgcc_shift_count_mode): Likewise.
18828 (s390_unwind_word_mode): Likewise.
18829 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18831 * target.def (mode_rep_extended): Likewise.
18832 (valid_pointer_mode): Likewise.
18833 (addr_space.valid_pointer_mode): Likewise.
18834 (eh_return_filter_mode): Return a scalar_int_mode rather than
18836 (libgcc_cmp_return_mode): Likewise.
18837 (libgcc_shift_count_mode): Likewise.
18838 (unwind_word_mode): Likewise.
18839 (addr_space.pointer_mode): Likewise.
18840 (addr_space.address_mode): Likewise.
18841 * doc/tm.texi: Regenerate.
18842 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18844 (do_jump): Use scalar_int_mode for local variables.
18845 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18846 rather than a machine_mode.
18847 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18848 (scompare_loc_descriptor_wide): Likewise.
18849 (scompare_loc_descriptor_narrow): Likewise.
18850 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18852 * except.c (sjlj_emit_dispatch_table): Likewise.
18853 (expand_builtin_eh_copy_values): Likewise.
18854 * explow.c (convert_memory_address_addr_space_1): Likewise.
18855 Take a scalar_int_mode rather than a machine_mode.
18856 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18857 than a machine_mode.
18858 (memory_address_addr_space): Use scalar_int_mode for local variables.
18859 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18860 rather than a machine_mode.
18861 * expmed.c (mask_rtx): Likewise.
18862 (init_expmed_one_conv): Likewise.
18863 (expand_mult_highpart_adjust): Likewise.
18864 (extract_high_half): Likewise.
18865 (expmed_mult_highpart_optab): Likewise.
18866 (expmed_mult_highpart): Likewise.
18867 (expand_smod_pow2): Likewise.
18868 (expand_sdiv_pow2): Likewise.
18869 (emit_store_flag_int): Likewise.
18870 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18872 (extract_low_bits): Likewise.
18873 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18875 * expr.c (pieces_addr::adjust): Likewise.
18876 (can_store_by_pieces): Likewise.
18877 (store_by_pieces): Likewise.
18878 (clear_by_pieces_1): Likewise.
18879 (expand_expr_addr_expr_1): Likewise.
18880 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18881 (expand_expr_real_1): Likewise.
18882 (try_casesi): Likewise.
18883 * final.c (shorten_branches): Likewise.
18884 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18885 type of "mode" to machine_mode.
18886 * internal-fn.c (expand_arith_overflow_result_store): Take a
18887 scalar_int_mode rather than a machine_mode.
18888 (expand_mul_overflow): Use scalar_int_mode for local variables.
18889 * loop-doloop.c (doloop_modify): Likewise.
18890 (doloop_optimize): Likewise.
18891 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18892 than a machine_mode.
18893 (expand_doubleword_shift_condmove): Likewise.
18894 (expand_doubleword_shift): Likewise.
18895 (expand_doubleword_clz): Likewise.
18896 (expand_doubleword_popcount): Likewise.
18897 (expand_doubleword_parity): Likewise.
18898 (expand_absneg_bit): Use scalar_int_mode for local variables.
18899 (prepare_float_lib_cmp): Likewise.
18900 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18901 rather than a machine_mode.
18902 (convert_memory_address_addr_space): Likewise.
18903 (get_mode_bounds): Likewise.
18904 (get_address_mode): Return a scalar_int_mode rather than a
18906 * rtlanal.c (get_address_mode): Likewise.
18907 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18908 than a machine_mode.
18909 * targhooks.c (default_mode_rep_extended): Likewise.
18910 (default_valid_pointer_mode): Likewise.
18911 (default_addr_space_valid_pointer_mode): Likewise.
18912 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18913 than a machine_mode.
18914 (default_libgcc_cmp_return_mode): Likewise.
18915 (default_libgcc_shift_count_mode): Likewise.
18916 (default_unwind_word_mode): Likewise.
18917 (default_addr_space_pointer_mode): Likewise.
18918 (default_addr_space_address_mode): Likewise.
18919 * targhooks.h (default_eh_return_filter_mode): Likewise.
18920 (default_libgcc_cmp_return_mode): Likewise.
18921 (default_libgcc_shift_count_mode): Likewise.
18922 (default_unwind_word_mode): Likewise.
18923 (default_addr_space_pointer_mode): Likewise.
18924 (default_addr_space_address_mode): Likewise.
18925 (default_mode_rep_extended): Take a scalar_int_mode rather than
18927 (default_valid_pointer_mode): Likewise.
18928 (default_addr_space_valid_pointer_mode): Likewise.
18929 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18931 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18932 rather than a machine_mode.
18933 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18934 for local variables.
18935 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18936 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18937 than a machine_mode.
18939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18940 Alan Hayward <alan.hayward@arm.com>
18941 David Sherwood <david.sherwood@arm.com>
18943 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18944 the mode argument to scalar_int_mode.
18945 (do_jump_by_parts_zero_rtx): Likewise.
18946 (do_jump_by_parts_equality_rtx): Likewise.
18947 (do_jump_by_parts_greater): Take a mode argument.
18948 (do_jump_by_parts_equality): Likewise.
18949 (do_jump_1): Update calls accordingly.
18951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18952 Alan Hayward <alan.hayward@arm.com>
18953 David Sherwood <david.sherwood@arm.com>
18955 * is-a.h (safe_dyn_cast): New function.
18956 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18957 (jump_table_for_label): Likewise.
18958 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18959 instead of an rtx_insn *.
18960 (shorten_branches): Use dyn_cast instead of LABEL_P and
18961 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18962 rtx_jump_table_data::get_data_mode.
18963 (final_scan_insn): Likewise.
18965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18966 Alan Hayward <alan.hayward@arm.com>
18967 David Sherwood <david.sherwood@arm.com>
18969 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18970 trying to combine a full-register integer set with a subreg
18973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18974 Alan Hayward <alan.hayward@arm.com>
18975 David Sherwood <david.sherwood@arm.com>
18977 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18978 that is always either address_mode or pointer_mode.
18980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18981 Alan Hayward <alan.hayward@arm.com>
18982 David Sherwood <david.sherwood@arm.com>
18984 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18985 when the two are known to be equal.
18987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18988 Alan Hayward <alan.hayward@arm.com>
18989 David Sherwood <david.sherwood@arm.com>
18991 * simplify-rtx.c (simplify_const_unary_operation): Use
18992 is_a <scalar_int_mode> instead of checking for a nonzero
18993 precision. Forcibly convert op_mode to a scalar_int_mode
18994 in that case. More clearly differentiate the operand and
18995 result modes and use the former when deciding what the value
18996 of a count-bits operation should be. Use is_int_mode instead
18997 of checking for a MODE_INT. Remove redundant check for whether
18998 this mode has a zero precision.
19000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19001 Alan Hayward <alan.hayward@arm.com>
19002 David Sherwood <david.sherwood@arm.com>
19004 * optabs.c (widen_leading): Change the type of the mode argument
19005 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
19006 (widen_bswap): Likewise.
19007 (expand_parity): Likewise.
19008 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
19009 (expand_ffs): Likewise.
19010 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
19019 (expand_expr_real_1): Update call accordingly.
19021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19022 Alan Hayward <alan.hayward@arm.com>
19023 David Sherwood <david.sherwood@arm.com>
19025 * expmed.c (store_bit_field_using_insv): Add op0_mode and
19026 value_mode arguments. Use scalar_int_mode internally.
19027 (store_bit_field_1): Rename the new integer mode from imode
19028 to op0_mode and use it instead of GET_MODE (op0). Update calls
19029 to store_split_bit_field, store_bit_field_using_insv and
19030 store_fixed_bit_field.
19031 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
19032 Use scalar_int_mode internally. Use a bit count rather than a mode
19033 when calculating the largest bit size for get_best_mode.
19034 Update calls to store_split_bit_field and store_fixed_bit_field_1.
19035 (store_fixed_bit_field_1): Add mode and value_mode arguments.
19036 Remove assertion that OP0 has a scalar integer mode.
19037 (store_split_bit_field): Add op0_mode and value_mode arguments.
19038 Update calls to extract_fixed_bit_field.
19039 (extract_bit_field_using_extv): Add an op0_mode argument.
19040 Use scalar_int_mode internally.
19041 (extract_bit_field_1): Rename the new integer mode from imode to
19042 op0_mode and use it instead of GET_MODE (op0). Update calls to
19043 extract_split_bit_field, extract_bit_field_using_extv and
19044 extract_fixed_bit_field.
19045 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
19046 to extract_split_bit_field and extract_fixed_bit_field_1.
19047 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
19048 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
19049 on the target mode.
19050 (extract_split_bit_field): Add an op0_mode argument. Update call
19051 to extract_fixed_bit_field.
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 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
19058 * explow.c (hard_function_value): Likewise.
19059 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
19060 convert_to_mode call outside the loop.
19061 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
19062 for the mode iterator. Require the mode specified by max_pieces
19064 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
19066 (copy_blkmode_to_reg): Likewise.
19067 (set_storage_via_setmem): Likewise.
19068 * optabs.c (prepare_cmp_insn): Likewise.
19069 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19070 * stor-layout.c (finish_bitfield_representative): Likewise.
19072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19073 Alan Hayward <alan.hayward@arm.com>
19074 David Sherwood <david.sherwood@arm.com>
19076 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
19077 * expr.c (convert_move): Use them.
19078 (convert_modes): Likewise.
19079 (store_expr_with_bounds): Likewise.
19081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19082 Alan Hayward <alan.hayward@arm.com>
19083 David Sherwood <david.sherwood@arm.com>
19085 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
19086 parameter for the mode of "x". Remove the "known_x", "known_mode"
19087 and "known_ret" arguments. Change the type of the mode argument
19088 to scalar_int_mode.
19089 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
19090 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
19091 (reg_num_sign_bit_copies_for_combine): Likewise.
19092 * rtlanal.c (nonzero_bits1): Likewise.
19093 (num_sign_bit_copies1): Likewise.
19094 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
19095 (reg_num_sign_bit_copies_general): Likewise.
19096 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
19097 (reg_nonzero_bits_general): Likewise.
19099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19100 Alan Hayward <alan.hayward@arm.com>
19101 David Sherwood <david.sherwood@arm.com>
19103 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
19104 than in subroutines. Return 1 for non-integer modes.
19105 (cached_num_sign_bit_copies): Change the type of the mode parameter
19106 to scalar_int_mode.
19107 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
19108 classes. Handle CONST_INT_P first and then check whether X also
19109 has a scalar integer mode. Check the same thing for inner registers
19110 of a SUBREG and for values that are being extended or truncated.
19112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19113 Alan Hayward <alan.hayward@arm.com>
19114 David Sherwood <david.sherwood@arm.com>
19116 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
19117 in subroutines. Return the mode mask for non-integer modes.
19118 (cached_nonzero_bits): Change the type of the mode parameter
19119 to scalar_int_mode.
19120 (nonzero_bits1): Likewise. Remove early exit for other mode
19121 classes. Handle CONST_INT_P first and then check whether X
19122 also has a scalar integer mode.
19124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19125 Alan Hayward <alan.hayward@arm.com>
19126 David Sherwood <david.sherwood@arm.com>
19128 * expr.c (widest_int_mode_for_size): Make the comment match the code.
19129 Return a scalar_int_mode and assert that the size is greater than
19131 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
19132 (op_by_pieces_d::op_by_pieces_d): Likewise.
19133 (op_by_pieces_d::run): Likewise.
19134 (can_store_by_pieces): Likewise.
19136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19137 Alan Hayward <alan.hayward@arm.com>
19138 David Sherwood <david.sherwood@arm.com>
19140 * combine.c (extract_left_shift): Add a mode argument and update
19142 (make_compound_operation_int): Change the type of the mode parameter
19143 to scalar_int_mode and update the call to extract_left_shift.
19145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19146 Alan Hayward <alan.hayward@arm.com>
19147 David Sherwood <david.sherwood@arm.com>
19149 * combine.c (simplify_and_const_int): Change the type of the mode
19150 parameter to scalar_int_mode.
19151 (simplify_and_const_int_1): Likewise. Update recursive call.
19153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19154 Alan Hayward <alan.hayward@arm.com>
19155 David Sherwood <david.sherwood@arm.com>
19157 * combine.c (simplify_compare_const): Check that the mode is a
19158 scalar_int_mode (rather than VOIDmode) before testing its
19160 (simplify_comparison): Move COMPARISON_P handling out of the
19161 loop and restrict the latter part of the loop to scalar_int_modes.
19162 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
19163 and when considering SUBREG_REGs. Use is_int_mode instead of
19164 checking GET_MODE_CLASS against MODE_INT.
19166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19167 Alan Hayward <alan.hayward@arm.com>
19168 David Sherwood <david.sherwood@arm.com>
19170 * combine.c (try_widen_shift_mode): Move check for equal modes to...
19171 (simplify_shift_const_1): ...here. Use scalar_int_mode for
19172 shift_unit_mode and for modes involved in scalar shifts.
19174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19175 Alan Hayward <alan.hayward@arm.com>
19176 David Sherwood <david.sherwood@arm.com>
19178 * combine.c (force_int_to_mode): New function, split out from...
19179 (force_to_mode): ...here. Keep xmode up-to-date and use it
19180 instead of GET_MODE (x).
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 * optabs-query.h (extraction_insn::struct_mode): Change type to
19187 opt_scalar_int_mode and update comment.
19188 (extraction_insn::field_mode): Change type to scalar_int_mode.
19189 (extraction_insn::pos_mode): Likewise.
19190 * combine.c (make_extraction): Update accordingly.
19191 * optabs-query.c (get_traditional_extraction_insn): Likewise.
19192 (get_optab_extraction_insn): Likewise.
19193 * recog.c (simplify_while_replacing): Likewise.
19194 * expmed.c (narrow_bit_field_mem): Change the type of the mode
19195 parameter to opt_scalar_int_mode.
19197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19198 Alan Hayward <alan.hayward@arm.com>
19199 David Sherwood <david.sherwood@arm.com>
19201 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
19202 to a scalar_int_mode instead of a machine_mode.
19203 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19204 (get_best_mode): Return a boolean and use a pointer argument to store
19205 the selected mode. Replace the limit mode parameter with a bit limit.
19206 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19207 for the values returned by bit_field_mode_iterator::next_mode.
19208 (store_bit_field): Update call to get_best_mode.
19209 (store_fixed_bit_field): Likewise.
19210 (extract_fixed_bit_field): Likewise.
19211 * expr.c (optimize_bitfield_assignment_op): Likewise.
19212 * fold-const.c (optimize_bit_field_compare): Likewise.
19213 (fold_truth_andor_1): Likewise.
19214 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19215 Update for new type of m_mode.
19216 (get_best_mode): As above.
19218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19219 Alan Hayward <alan.hayward@arm.com>
19220 David Sherwood <david.sherwood@arm.com>
19222 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19223 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19224 (store_bit_field): Check is_a <scalar_int_mode> before calling
19225 strict_volatile_bitfield_p.
19226 (extract_bit_field): Likewise.
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 * target.def (cstore_mode): Return a scalar_int_mode.
19233 * doc/tm.texi: Regenerate.
19234 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19235 * targhooks.h (default_cstore_mode): Likewise.
19236 * targhooks.c (default_cstore_mode): Likewise, using a forced
19238 * expmed.c (emit_cstore): Expect the target of the cstore to be
19241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19242 Alan Hayward <alan.hayward@arm.com>
19243 David Sherwood <david.sherwood@arm.com>
19245 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19247 (niter_desc): Likewise mode.
19248 (iv_analyze): Add a mode parameter.
19250 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19251 and change its type to scalar_int_mode.
19252 * loop-iv.c: Update commentary at head of file.
19253 (iv_constant): Pass the mode paraeter before the rtx it describes
19254 and change its type to scalar_int_mode. Remove VOIDmode handling.
19255 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19256 (iv_extend): Likewise.
19257 (shorten_into_mode): Likewise.
19258 (iv_add): Use scalar_int_mode.
19259 (iv_mult): Likewise.
19260 (iv_shift): Likewise.
19261 (canonicalize_iv_subregs): Likewise.
19262 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19263 it describes and change its mode to scalar_int_mode. Also change
19264 the type of the returned inner_mode to scalar_int_mode.
19265 (get_biv_step): Likewise, turning outer_mode from a pointer
19266 into a direct parameter. Update call to get_biv_step_1.
19267 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19268 iv_constant and get_biv_step.
19269 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19270 and change its type to scalar_int_mode. Don't initialise iv->mode
19271 to VOIDmode and remove later checks for its still being VOIDmode.
19272 Update calls to iv_analyze_op and iv_analyze_expr. Check
19273 is_a <scalar_int_mode> when changing the mode under consideration.
19274 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19275 Update call to iv_analyze_expr.
19276 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19277 inner register is not also a scalar_int_mode. Update call to
19279 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19280 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19281 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19282 separate mode class checks. Update calls to iv_analyze. Remove
19283 fix-up of VOIDmodes after iv_analyze_biv.
19284 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19285 don't have a scalar_int_mode. Update call to biv_p.
19287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19288 Alan Hayward <alan.hayward@arm.com>
19289 David Sherwood <david.sherwood@arm.com>
19291 * cfgexpand.c (convert_debug_memory_address): Use
19292 as_a <scalar_int_mode>.
19293 * combine.c (expand_compound_operation): Likewise.
19294 (make_extraction): Likewise.
19295 (change_zero_ext): Likewise.
19296 (simplify_comparison): Likewise.
19297 * cse.c (cse_insn): Likewise.
19298 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19299 (mem_loc_descriptor): Likewise.
19300 (loc_descriptor): Likewise.
19301 * expmed.c (init_expmed_one_mode): Likewise.
19302 (synth_mult): Likewise.
19303 (emit_store_flag_1): Likewise.
19304 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19305 of a comparison with size.
19306 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19307 (reduce_to_bit_field_precision): Likewise.
19308 * function.c (expand_function_end): Likewise.
19309 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19310 * loop-doloop.c (doloop_modify): Likewise.
19311 * optabs.c (expand_binop): Likewise.
19312 (expand_unop): Likewise.
19313 (expand_copysign_absneg): Likewise.
19314 (prepare_cmp_insn): Likewise.
19315 (maybe_legitimize_operand): Likewise.
19316 * recog.c (const_scalar_int_operand): Likewise.
19317 * rtlanal.c (get_address_mode): Likewise.
19318 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19319 (simplify_cond_clz_ctz): Likewise.
19320 * tree-nested.c (get_nl_goto_field): Likewise.
19321 * tree.c (build_vector_type_for_mode): Likewise.
19322 * var-tracking.c (use_narrower_mode): Likewise.
19324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19325 Alan Hayward <alan.hayward@arm.com>
19326 David Sherwood <david.sherwood@arm.com>
19328 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19329 * builtins.c (expand_builtin_signbit): Use it.
19330 * cfgexpand.c (expand_debug_expr): Likewise.
19331 * dojump.c (do_jump): Likewise.
19332 (do_compare_and_jump): Likewise.
19333 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19334 * expmed.c (make_tree): Likewise.
19335 * expr.c (expand_expr_real_2): Likewise.
19336 (expand_expr_real_1): Likewise.
19337 (try_casesi): Likewise.
19338 * fold-const-call.c (fold_const_call_ss): Likewise.
19339 * fold-const.c (unextend): Likewise.
19340 (extract_muldiv_1): Likewise.
19341 (fold_single_bit_test): Likewise.
19342 (native_encode_int): Likewise.
19343 (native_encode_string): Likewise.
19344 (native_interpret_int): Likewise.
19345 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19346 * internal-fn.c (expand_addsub_overflow): Likewise.
19347 (expand_neg_overflow): Likewise.
19348 (expand_mul_overflow): Likewise.
19349 (expand_arith_overflow): Likewise.
19350 * match.pd: Likewise.
19351 * stor-layout.c (layout_type): Likewise.
19352 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19353 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19354 * tree-ssanames.c (get_range_info): Likewise.
19355 * tree-switch-conversion.c (array_value_type) Likewise.
19356 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19357 (vect_recog_divmod_pattern): Likewise.
19358 (vect_recog_mixed_size_cond_pattern): Likewise.
19359 * tree-vrp.c (extract_range_basic): Likewise.
19360 (simplify_float_conversion_using_ranges): Likewise.
19361 * tree.c (int_fits_type_p): Likewise.
19362 * ubsan.c (instrument_bool_enum_load): Likewise.
19363 * varasm.c (mergeable_string_section): Likewise.
19364 (narrowing_initializer_constant_valid_p): Likewise.
19365 (output_constant): Likewise.
19367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19368 Alan Hayward <alan.hayward@arm.com>
19369 David Sherwood <david.sherwood@arm.com>
19371 * machmode.h (NARROWEST_INT_MODE): New macro.
19372 * expr.c (alignment_for_piecewise_move): Use it instead of
19373 GET_CLASS_NARROWEST_MODE (MODE_INT).
19374 (push_block): Likewise.
19375 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19377 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19380 Alan Hayward <alan.hayward@arm.com>
19381 David Sherwood <david.sherwood@arm.com>
19383 * postreload.c (move2add_valid_value_p): Change the type of the
19384 mode parameter to scalar_int_mode.
19385 (move2add_use_add2_insn): Add a mode parameter and use it instead
19387 (move2add_use_add3_insn): Likewise.
19388 (reload_cse_move2add): Update accordingly.
19390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19391 Alan Hayward <alan.hayward@arm.com>
19392 David Sherwood <david.sherwood@arm.com>
19394 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19396 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19397 * optabs.c (expand_unop): Likewise.
19399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19400 Alan Hayward <alan.hayward@arm.com>
19401 David Sherwood <david.sherwood@arm.com>
19403 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19404 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19405 (popcount_loc_descriptor): Likewise.
19406 (bswap_loc_descriptor): Likewise.
19407 (rotate_loc_descriptor): Likewise.
19408 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19409 calling the functions above.
19411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19412 Alan Hayward <alan.hayward@arm.com>
19413 David Sherwood <david.sherwood@arm.com>
19415 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19417 (try_combine): Likewise.
19418 (simplify_if_then_else): Likewise.
19419 * cse.c (cse_insn): Likewise.
19420 * dwarf2out.c (mem_loc_descriptor): Likewise.
19421 * emit-rtl.c (gen_lowpart_common): Likewise.
19422 * simplify-rtx.c (simplify_truncation): Likewise.
19423 (simplify_binary_operation_1): Likewise.
19424 (simplify_const_relational_operation): Likewise.
19425 (simplify_ternary_operation): Likewise.
19426 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): 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 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19433 * reload.c (push_reload): Likewise.
19434 (find_reloads): Likewise.
19436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19437 Alan Hayward <alan.hayward@arm.com>
19438 David Sherwood <david.sherwood@arm.com>
19440 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19441 (make_compound_operation_int): Likewise.
19442 (change_zero_ext): Likewise.
19443 * expr.c (convert_move): Likewise.
19444 (convert_modes): Likewise.
19445 * fwprop.c (forward_propagate_subreg): Likewise.
19446 * loop-iv.c (get_biv_step_1): Likewise.
19447 * optabs.c (widen_operand): Likewise.
19448 * postreload.c (move2add_valid_value_p): Likewise.
19449 * recog.c (simplify_while_replacing): Likewise.
19450 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19451 (simplify_binary_operation_1): Likewise. Remove redundant
19452 mode equality check.
19454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19455 Alan Hayward <alan.hayward@arm.com>
19456 David Sherwood <david.sherwood@arm.com>
19458 * combine.c (combine_simplify_rtx): Add checks for
19459 is_a <scalar_int_mode>.
19460 (simplify_if_then_else): Likewise.
19461 (make_field_assignment): Likewise.
19462 (simplify_comparison): Likewise.
19463 * ifcvt.c (noce_try_bitop): Likewise.
19464 * loop-invariant.c (canonicalize_address_mult): Likewise.
19465 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19468 Alan Hayward <alan.hayward@arm.com>
19469 David Sherwood <david.sherwood@arm.com>
19471 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19472 is_a <scalar_int_mode> instead of != BLKmode.
19474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19475 Alan Hayward <alan.hayward@arm.com>
19476 David Sherwood <david.sherwood@arm.com>
19478 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19479 instead of != VOIDmode.
19480 * combine.c (if_then_else_cond): Likewise.
19481 (change_zero_ext): Likewise.
19482 * dwarf2out.c (mem_loc_descriptor): Likewise.
19483 (loc_descriptor): Likewise.
19484 * rtlanal.c (canonicalize_condition): Likewise.
19485 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19488 Alan Hayward <alan.hayward@arm.com>
19489 David Sherwood <david.sherwood@arm.com>
19491 * simplify-rtx.c (simplify_binary_operation_1): Use
19492 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19495 Alan Hayward <alan.hayward@arm.com>
19496 David Sherwood <david.sherwood@arm.com>
19498 * wide-int.h (int_traits<unsigned char>) New class.
19499 (int_traits<unsigned short>) Likewise.
19500 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19501 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19503 * combine.c (set_nonzero_bits_and_sign_copies): Use
19504 is_a <scalar_int_mode>.
19505 (find_split_point): Likewise.
19506 (combine_simplify_rtx): Likewise.
19507 (simplify_logical): Likewise.
19508 (expand_compound_operation): Likewise.
19509 (expand_field_assignment): Likewise.
19510 (make_compound_operation): Likewise.
19511 (extended_count): Likewise.
19512 (change_zero_ext): Likewise.
19513 (simplify_comparison): Likewise.
19514 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19515 (ucompare_loc_descriptor): Likewise.
19516 (minmax_loc_descriptor): Likewise.
19517 (mem_loc_descriptor): Likewise.
19518 (loc_descriptor): Likewise.
19519 * expmed.c (init_expmed_one_mode): Likewise.
19520 * lra-constraints.c (lra_constraint_offset): Likewise.
19521 * optabs.c (prepare_libcall_arg): Likewise.
19522 * postreload.c (move2add_note_store): Likewise.
19523 * reload.c (operands_match_p): Likewise.
19524 * rtl.h (load_extend_op): Likewise.
19525 * rtlhooks.c (gen_lowpart_general): Likewise.
19526 * simplify-rtx.c (simplify_truncation): Likewise.
19527 (simplify_unary_operation_1): Likewise.
19528 (simplify_binary_operation_1): Likewise.
19529 (simplify_const_binary_operation): Likewise.
19530 (simplify_const_relational_operation): Likewise.
19531 (simplify_subreg): Likewise.
19532 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19533 * var-tracking.c (adjust_mems): Likewise.
19534 (prepare_call_arguments): Likewise.
19536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19537 Alan Hayward <alan.hayward@arm.com>
19538 David Sherwood <david.sherwood@arm.com>
19540 * machmode.h (is_int_mode): New fuction.
19541 * combine.c (find_split_point): Use it.
19542 (combine_simplify_rtx): Likewise.
19543 (simplify_if_then_else): Likewise.
19544 (simplify_set): Likewise.
19545 (simplify_shift_const_1): Likewise.
19546 (simplify_comparison): Likewise.
19547 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19548 * cse.c (notreg_cost): Likewise.
19549 (cse_insn): Likewise.
19550 * cselib.c (cselib_lookup_1): Likewise.
19551 * dojump.c (do_jump_1): Likewise.
19552 (do_compare_rtx_and_jump): Likewise.
19553 * dse.c (get_call_args): Likewise.
19554 * dwarf2out.c (rtl_for_decl_init): Likewise.
19555 (native_encode_initializer): Likewise.
19556 * expmed.c (emit_store_flag_1): Likewise.
19557 (emit_store_flag): Likewise.
19558 * expr.c (convert_modes): Likewise.
19559 (store_field): Likewise.
19560 (expand_expr_real_1): Likewise.
19561 * fold-const.c (fold_read_from_constant_string): Likewise.
19562 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19563 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19564 * optabs.c (expand_binop): Likewise.
19565 (expand_unop): Likewise.
19566 (expand_abs_nojump): Likewise.
19567 (expand_one_cmpl_abs_nojump): Likewise.
19568 * simplify-rtx.c (mode_signbit_p): Likewise.
19569 (val_signbit_p): Likewise.
19570 (val_signbit_known_set_p): Likewise.
19571 (val_signbit_known_clear_p): Likewise.
19572 (simplify_relational_operation_1): Likewise.
19573 * tree.c (vector_type_mode): Likewise.
19575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19576 Alan Hayward <alan.hayward@arm.com>
19577 David Sherwood <david.sherwood@arm.com>
19579 * machmode.h (smallest_mode_for_size): Fix formatting.
19580 (smallest_int_mode_for_size): New function.
19581 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19582 instead of smallest_mode_for_size.
19583 * combine.c (make_extraction): Likewise.
19584 * config/arc/arc.c (arc_expand_movmem): Likewise.
19585 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19586 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19587 * config/s390/s390.c (s390_expand_insv): Likewise.
19588 * config/sparc/sparc.c (assign_int_registers): Likewise.
19589 * config/spu/spu.c (spu_function_value): Likewise.
19590 (spu_function_arg): Likewise.
19591 * coverage.c (get_gcov_type): Likewise.
19592 (get_gcov_unsigned_t): Likewise.
19593 * dse.c (find_shift_sequence): Likewise.
19594 * expmed.c (store_bit_field_1): Likewise.
19595 * expr.c (convert_move): Likewise.
19596 (store_field): Likewise.
19597 * internal-fn.c (expand_arith_overflow): Likewise.
19598 * optabs-query.c (get_best_extraction_insn): Likewise.
19599 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19600 * stor-layout.c (layout_type): Likewise.
19601 (initialize_sizetypes): Likewise.
19602 * targhooks.c (default_get_mask_mode): Likewise.
19603 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19606 Alan Hayward <alan.hayward@arm.com>
19607 David Sherwood <david.sherwood@arm.com>
19609 * machmode.h (opt_mode::else_blk): New function.
19610 (int_mode_for_mode): Declare.
19611 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19612 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19614 * cfgexpand.c (expand_debug_expr): Likewise.
19615 * combine.c (gen_lowpart_or_truncate): Likewise.
19616 (gen_lowpart_for_combine): Likewise.
19617 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19618 * config/avr/avr.c (avr_to_int_mode): Likewise.
19619 (avr_out_plus_1): Likewise.
19620 (avr_out_plus): Likewise.
19621 (avr_out_round): Likewise.
19622 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19623 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19624 (s390_expand_vcond): Likewise.
19625 * config/spu/spu.c (spu_split_immediate): Likewise.
19626 (spu_expand_mov): Likewise.
19627 * dse.c (get_stored_val): Likewise.
19628 * expmed.c (store_bit_field_1): Likewise.
19629 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19631 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19632 (extract_low_bits): Likewise.
19633 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19634 handling rather than repeating the check.
19635 (emit_group_store): Likewise.
19636 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19637 * optabs.c (expand_absneg_bit): Likewise.
19638 (expand_copysign_absneg): Likewise.
19639 (expand_copysign_bit): Likewise.
19640 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19641 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19642 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19643 * var-tracking.c (prepare_call_arguments): Likewise.
19644 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19645 int_mode_for_mode instead of mode_for_size.
19646 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19649 Alan Hayward <alan.hayward@arm.com>
19650 David Sherwood <david.sherwood@arm.com>
19652 * machmode.h (int_mode_for_size): New function.
19653 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19654 instead of mode_for_size.
19655 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19657 * combine.c (expand_field_assignment): Use int_mode_for_size
19658 instead of mode_for_size.
19659 (make_extraction): Likewise.
19660 (simplify_shift_const_1): Likewise.
19661 (simplify_comparison): Likewise.
19662 * dojump.c (do_jump): Likewise.
19663 * dwarf2out.c (mem_loc_descriptor): Likewise.
19664 * emit-rtl.c (init_derived_machine_modes): Likewise.
19665 * expmed.c (flip_storage_order): Likewise.
19666 (convert_extracted_bit_field): Likewise.
19667 * expr.c (copy_blkmode_from_reg): Likewise.
19668 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19669 * internal-fn.c (expand_mul_overflow): Likewise.
19670 * lower-subreg.c (simple_move): Likewise.
19671 * optabs-libfuncs.c (init_optabs): Likewise.
19672 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19673 * tree.c (vector_type_mode): Likewise.
19674 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19675 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19676 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19677 * tree-vect-stmts.c (vectorizable_load): Likewise.
19678 (vectorizable_store): Likewise.
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 * coretypes.h (pod_mode): New type.
19685 (scalar_int_mode_pod): New typedef.
19686 * machmode.h (pod_mode): New class.
19687 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19688 * genmodes.c (emit_mode_int_n): Update accordingly.
19689 * lower-subreg.h (target_lower_subreg): Change type to
19690 scalar_int_mode_pod.
19691 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19692 scalar_int_mode_pod.
19694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19695 Alan Hayward <alan.hayward@arm.com>
19696 David Sherwood <david.sherwood@arm.com>
19698 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19699 machine_mode to scalar_int_mode.
19700 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19701 (rs6000_option_override_internal): Remove cast to int.
19702 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19703 machine_mode to scalar_int_mode.
19704 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19705 (rs6000_option_override_internal): Remove cast to int.
19706 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19707 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19709 * config/s390/s390.c (s390_expand_builtin): Likewise.
19710 * coretypes.h (scalar_int_mode): New type.
19711 (opt_scalar_int_mode): New typedef.
19712 * machmode.h (scalar_int_mode): New class.
19713 (scalar_int_mode::includes_p): New function.
19714 (byte_mode): Change type to scalar_int_mode.
19715 (word_mode): Likewise.
19716 (ptr_mode): Likewise.
19717 * emit-rtl.c (byte_mode): Likewise.
19718 (word_mode): Likewise.
19719 (ptr_mode): Likewise.
19720 (init_derived_machine_modes): Update accordingly.
19721 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19722 and MODE_PARTIAL_INT.
19723 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19724 opt_scalar_int_mode.
19726 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19727 Alan Hayward <alan.hayward@arm.com>
19728 David Sherwood <david.sherwood@arm.com>
19730 * target.def (libgcc_floating_mode_supported_p): Take a
19732 * doc/tm.texi: Regenerate.
19733 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19735 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19736 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19739 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19740 Alan Hayward <alan.hayward@arm.com>
19741 David Sherwood <david.sherwood@arm.com>
19743 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19744 * doc/tm.texi: Regenerate.
19745 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19746 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19747 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19748 * targhooks.h (default_floatn_mode): Likewise.
19749 * targhooks.c (default_floatn_mode): Likewise.
19750 * tree.c (build_common_tree_nodes): Update accordingly.
19752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19753 Alan Hayward <alan.hayward@arm.com>
19754 David Sherwood <david.sherwood@arm.com>
19756 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19757 (mode_iterator::iterate_p): Likewise.
19758 (mode_iterator::get_wider): Likewise.
19759 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19762 Alan Hayward <alan.hayward@arm.com>
19763 David Sherwood <david.sherwood@arm.com>
19765 * coretypes.h (opt_scalar_float_mode): New typedef.
19766 * machmode.h (float_mode_for_size): New function.
19767 * emit-rtl.c (double_mode): Delete.
19768 (init_emit_once): Use float_mode_for_size.
19769 * stor-layout.c (layout_type): Likewise.
19770 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19773 Alan Hayward <alan.hayward@arm.com>
19774 David Sherwood <david.sherwood@arm.com>
19776 * output.h (assemble_real): Take a scalar_float_mode.
19777 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19778 * config/arm/arm.md (consttable_4): Likewise.
19779 (consttable_8): Likewise.
19780 (consttable_16): Likewise.
19781 * config/mips/mips.md (consttable_float): Likewise.
19782 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19783 * varasm.c (assemble_real): Take a scalar_float_mode.
19784 (output_constant_pool_2): Update accordingly.
19785 (output_constant): Likewise.
19787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19788 Alan Hayward <alan.hayward@arm.com>
19789 David Sherwood <david.sherwood@arm.com>
19791 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19792 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19793 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19794 (native_encode_real): Likewise.
19795 (native_interpret_real): Likewise.
19796 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19797 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19800 Alan Hayward <alan.hayward@arm.com>
19801 David Sherwood <david.sherwood@arm.com>
19803 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19804 <scalar_float_mode>. Simplify.
19805 (gen_extend_conv_libfunc): Likewise.
19807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19808 Alan Hayward <alan.hayward@arm.com>
19809 David Sherwood <david.sherwood@arm.com>
19811 * coretypes.h (scalar_float_mode): New type.
19812 * machmode.h (mode_traits::from_int): Use machine_mode if
19813 USE_ENUM_MODES is defined.
19814 (is_a): New function.
19816 (dyn_cast): Likewise.
19817 (scalar_float_mode): New class.
19818 (scalar_float_mode::includes_p): New function.
19819 (is_float_mode): Likewise.
19820 * gdbhooks.py (MachineModePrinter): New class.
19821 (build_pretty_printer): Use it for scalar_float_mode.
19822 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19823 (format_helper::format_helper): Turn into a template.
19824 * genmodes.c (get_mode_class): New function.
19825 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19826 or machine_mode if none.
19827 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19828 as_a <scalar_float_mode>.
19829 * dwarf2out.c (mem_loc_descriptor): Likewise.
19830 (insert_float): Likewise.
19831 (add_const_value_attribute): Likewise.
19832 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19833 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19834 (expand_unop): Update accordingly.
19835 (expand_abs_nojump): Likewise.
19836 (expand_copysign_absneg): Take a scalar_float_mode.
19837 (expand_copysign_bit): Likewise.
19838 (expand_copysign): Update accordingly.
19840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19841 Alan Hayward <alan.hayward@arm.com>
19842 David Sherwood <david.sherwood@arm.com>
19844 * coretypes.h (opt_mode): New class.
19845 * machmode.h (opt_mode): Likewise.
19846 (opt_mode::else_void): New function.
19847 (opt_mode::require): Likewise.
19848 (opt_mode::exists): Likewise.
19849 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19850 (GET_MODE_2XWIDER_MODE): Likewise.
19851 (mode_iterator::get_wider): Update accordingly.
19852 (mode_iterator::get_2xwider): Likewise.
19853 (mode_iterator::get_known_wider): Likewise, turning into a template.
19854 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19855 forcing a wider mode to exist.
19856 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19857 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19858 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19859 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19860 * lower-subreg.c (init_lower_subreg): Likewise.
19861 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19862 on the final iteration.
19863 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19864 a wider mode exists before asking for a move pattern.
19865 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19866 forcing a wider mode to exist.
19867 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19868 returning false if no such mode exists.
19869 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19870 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19871 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19872 Avoid checking for a MODE_INT if we already know the mode is not a
19874 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19875 forcing a wider mode to exist.
19876 (expmed_mult_highpart_optab): Likewise.
19877 (expmed_mult_highpart): Likewise.
19878 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19880 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19881 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19882 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19883 * internal-fn.c (expand_mul_overflow): Update use of
19884 GET_MODE_2XWIDER_MODE.
19885 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19886 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19887 GET_MODE_WIDER_MODE.
19888 (convert_plusminus_to_widen): Likewise.
19889 * tree-switch-conversion.c (array_value_type): Likewise.
19890 * var-tracking.c (emit_note_insn_var_location): Likewise.
19891 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19892 Return false inside rather than outside the loop if no wider mode
19894 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19895 and GET_MODE_2XWIDER_MODE
19896 (can_compare_p): Use else_void.
19897 * gdbhooks.py (OptMachineModePrinter): New class.
19898 (build_pretty_printer): Use it for opt_mode.
19900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19901 Alan Hayward <alan.hayward@arm.com>
19902 David Sherwood <david.sherwood@arm.com>
19904 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19905 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19908 Alan Hayward <alan.hayward@arm.com>
19909 David Sherwood <david.sherwood@arm.com>
19911 * machmode.h (mode_traits): New structure.
19912 (get_narrowest_mode): New function.
19913 (mode_iterator::start): Likewise.
19914 (mode_iterator::iterate_p): Likewise.
19915 (mode_iterator::get_wider): Likewise.
19916 (mode_iterator::get_known_wider): Likewise.
19917 (mode_iterator::get_2xwider): Likewise.
19918 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19919 (FOR_EACH_MODE): Likewise.
19920 (FOR_EACH_MODE_FROM): Likewise.
19921 (FOR_EACH_MODE_UNTIL): Likewise.
19922 (FOR_EACH_WIDER_MODE): Likewise.
19923 (FOR_EACH_2XWIDER_MODE): Likewise.
19924 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19925 * combine.c (simplify_comparison): Likewise
19926 * config/i386/i386.c (type_natural_mode): Likewise.
19927 * cse.c (cse_insn): Likewise.
19928 * dse.c (find_shift_sequence): Likewise.
19929 * emit-rtl.c (init_derived_machine_modes): Likewise.
19930 (init_emit_once): Likewise.
19931 * explow.c (hard_function_value): Likewise.
19932 * expmed.c (extract_fixed_bit_field_1): Likewise.
19933 (extract_bit_field_1): Likewise.
19934 (expand_divmod): Likewise.
19935 (emit_store_flag_1): Likewise.
19936 * expr.c (init_expr_target): Likewise.
19937 (convert_move): Likewise.
19938 (alignment_for_piecewise_move): Likewise.
19939 (widest_int_mode_for_size): Likewise.
19940 (emit_block_move_via_movmem): Likewise.
19941 (copy_blkmode_to_reg): Likewise.
19942 (set_storage_via_setmem): Likewise.
19943 (compress_float_constant): Likewise.
19944 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19945 * optabs-query.c (get_best_extraction_insn): Likewise.
19946 * optabs.c (expand_binop): Likewise.
19947 (expand_twoval_unop): Likewise.
19948 (expand_twoval_binop): Likewise.
19949 (widen_leading): Likewise.
19950 (widen_bswap): Likewise.
19951 (expand_parity): Likewise.
19952 (expand_unop): Likewise.
19953 (prepare_cmp_insn): Likewise.
19954 (prepare_float_lib_cmp): Likewise.
19955 (expand_float): Likewise.
19956 (expand_fix): Likewise.
19957 (expand_sfix_optab): Likewise.
19958 * postreload.c (move2add_use_add2_insn): Likewise.
19959 * reg-stack.c (reg_to_stack): Likewise.
19960 * reginfo.c (choose_hard_reg_mode): Likewise.
19961 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19962 * stor-layout.c (mode_for_size): Likewise.
19963 (smallest_mode_for_size): Likewise.
19964 (mode_for_vector): Likewise.
19965 (finish_bitfield_representative): Likewise.
19966 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19967 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19968 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19969 * var-tracking.c (prepare_call_arguments): Likewise.
19971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19972 Alan Hayward <alan.hayward@arm.com>
19973 David Sherwood <david.sherwood@arm.com>
19975 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19976 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19977 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19978 * machmode.h (mode_size): Move earlier in file.
19979 (mode_precision): Likewise.
19980 (mode_inner): Likewise.
19981 (mode_nunits): Likewise.
19982 (mode_unit_size): Likewise.
19983 (unit_unit_precision): Likewise.
19984 (mode_wider): Likewise.
19985 (mode_2xwider): Likewise.
19986 (machine_mode): New class.
19987 (mode_to_bytes): New function.
19988 (mode_to_bits): Likewise.
19989 (mode_to_precision): Likewise.
19990 (mode_to_inner): Likewise.
19991 (mode_to_unit_size): Likewise.
19992 (mode_to_unit_precision): Likewise.
19993 (mode_to_nunits): Likewise.
19994 (GET_MODE_SIZE): Use mode_to_bytes.
19995 (GET_MODE_BITSIZE): Use mode_to_bits.
19996 (GET_MODE_PRECISION): Use mode_to_precision.
19997 (GET_MODE_INNER): Use mode_to_inner.
19998 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19999 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
20000 (GET_MODE_NUNITS): Use mode_to_nunits.
20001 * system.h (ALWAYS_INLINE): New macro.
20002 * config/powerpcspe/powerpcspe-c.c
20003 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
20004 int for arg1_mode and arg2_mode.
20006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20007 Alan Hayward <alan.hayward@arm.com>
20008 David Sherwood <david.sherwood@arm.com>
20010 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20011 Prefix mode names with E_ in case statements.
20012 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20013 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
20014 (aarch64_split_simd_move): Likewise.
20015 (aarch64_gen_storewb_pair): Likewise.
20016 (aarch64_gen_loadwb_pair): Likewise.
20017 (aarch64_gen_store_pair): Likewise.
20018 (aarch64_gen_load_pair): Likewise.
20019 (aarch64_get_condition_code_1): Likewise.
20020 (aarch64_constant_pool_reload_icode): Likewise.
20021 (get_rsqrte_type): Likewise.
20022 (get_rsqrts_type): Likewise.
20023 (get_recpe_type): Likewise.
20024 (get_recps_type): Likewise.
20025 (aarch64_gimplify_va_arg_expr): Likewise.
20026 (aarch64_simd_container_mode): Likewise.
20027 (aarch64_emit_load_exclusive): Likewise.
20028 (aarch64_emit_store_exclusive): Likewise.
20029 (aarch64_expand_compare_and_swap): Likewise.
20030 (aarch64_gen_atomic_cas): Likewise.
20031 (aarch64_emit_bic): Likewise.
20032 (aarch64_emit_atomic_swap): Likewise.
20033 (aarch64_emit_atomic_load_op): Likewise.
20034 (aarch64_evpc_trn): Likewise.
20035 (aarch64_evpc_uzp): Likewise.
20036 (aarch64_evpc_zip): Likewise.
20037 (aarch64_evpc_ext): Likewise.
20038 (aarch64_evpc_rev): Likewise.
20039 (aarch64_evpc_dup): Likewise.
20040 (aarch64_gen_ccmp_first): Likewise.
20041 (aarch64_gen_ccmp_next): Likewise.
20042 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
20043 (alpha_emit_xfloating_libcall): Likewise.
20044 (emit_insxl): Likewise.
20045 (alpha_arg_type): Likewise.
20046 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
20047 (arc_preferred_simd_mode): Likewise.
20048 (arc_secondary_reload): Likewise.
20049 (get_arc_condition_code): Likewise.
20050 (arc_print_operand): Likewise.
20051 (arc_legitimate_constant_p): Likewise.
20052 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20053 * config/arc/arc.md (casesi_load): Likewise.
20054 (casesi_compact_jump): Likewise.
20055 * config/arc/predicates.md (proper_comparison_operator): Likewise.
20056 (cc_use_register): Likewise.
20057 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20058 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20059 (arm_init_iwmmxt_builtins): Likewise.
20060 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
20061 (neon_expand_vector_init): Likewise.
20062 (arm_attr_length_move_neon): Likewise.
20063 (maybe_get_arm_condition_code): Likewise.
20064 (arm_emit_vector_const): Likewise.
20065 (arm_preferred_simd_mode): Likewise.
20066 (arm_output_iwmmxt_tinsr): Likewise.
20067 (thumb1_output_casesi): Likewise.
20068 (thumb2_output_casesi): Likewise.
20069 (arm_emit_load_exclusive): Likewise.
20070 (arm_emit_store_exclusive): Likewise.
20071 (arm_expand_compare_and_swap): Likewise.
20072 (arm_evpc_neon_vuzp): Likewise.
20073 (arm_evpc_neon_vzip): Likewise.
20074 (arm_evpc_neon_vrev): Likewise.
20075 (arm_evpc_neon_vtrn): Likewise.
20076 (arm_evpc_neon_vext): Likewise.
20077 (arm_validize_comparison): Likewise.
20078 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
20079 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
20080 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
20081 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
20082 (c6x_preferred_simd_mode): Likewise.
20083 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
20084 (epiphany_rtx_costs): Likewise.
20085 * config/epiphany/predicates.md (proper_comparison_operator):
20087 * config/frv/frv.c (condexec_memory_operand): Likewise.
20088 (frv_emit_move): Likewise.
20089 (output_move_single): Likewise.
20090 (output_condmove_single): Likewise.
20091 (frv_hard_regno_mode_ok): Likewise.
20092 (frv_matching_accg_mode): Likewise.
20093 * config/h8300/h8300.c (split_adds_subs): Likewise.
20094 (h8300_rtx_costs): Likewise.
20095 (h8300_print_operand): Likewise.
20096 (compute_mov_length): Likewise.
20097 (output_logical_op): Likewise.
20098 (compute_logical_op_length): Likewise.
20099 (compute_logical_op_cc): Likewise.
20100 (h8300_shift_needs_scratch_p): Likewise.
20101 (output_a_shift): Likewise.
20102 (compute_a_shift_length): Likewise.
20103 (compute_a_shift_cc): Likewise.
20104 (expand_a_rotate): Likewise.
20105 (output_a_rotate): Likewise.
20106 * config/i386/i386.c (classify_argument): Likewise.
20107 (function_arg_advance_32): Likewise.
20108 (function_arg_32): Likewise.
20109 (function_arg_64): Likewise.
20110 (function_value_64): Likewise.
20111 (ix86_gimplify_va_arg): Likewise.
20112 (ix86_legitimate_constant_p): Likewise.
20113 (put_condition_code): Likewise.
20114 (split_double_mode): Likewise.
20115 (ix86_avx256_split_vector_move_misalign): Likewise.
20116 (ix86_expand_vector_logical_operator): Likewise.
20117 (ix86_split_idivmod): Likewise.
20118 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
20119 (ix86_build_const_vector): Likewise.
20120 (ix86_build_signbit_mask): Likewise.
20121 (ix86_match_ccmode): Likewise.
20122 (ix86_cc_modes_compatible): Likewise.
20123 (ix86_expand_branch): Likewise.
20124 (ix86_expand_sse_cmp): Likewise.
20125 (ix86_expand_sse_movcc): Likewise.
20126 (ix86_expand_int_sse_cmp): Likewise.
20127 (ix86_expand_vec_perm_vpermi2): Likewise.
20128 (ix86_expand_vec_perm): Likewise.
20129 (ix86_expand_sse_unpack): Likewise.
20130 (ix86_expand_int_addcc): Likewise.
20131 (ix86_split_to_parts): Likewise.
20132 (ix86_vectorize_builtin_gather): Likewise.
20133 (ix86_vectorize_builtin_scatter): Likewise.
20134 (avx_vpermilp_parallel): Likewise.
20135 (inline_memory_move_cost): Likewise.
20136 (ix86_tieable_integer_mode_p): Likewise.
20137 (x86_maybe_negate_const_int): Likewise.
20138 (ix86_expand_vector_init_duplicate): Likewise.
20139 (ix86_expand_vector_init_one_nonzero): Likewise.
20140 (ix86_expand_vector_init_one_var): Likewise.
20141 (ix86_expand_vector_init_concat): Likewise.
20142 (ix86_expand_vector_init_interleave): Likewise.
20143 (ix86_expand_vector_init_general): Likewise.
20144 (ix86_expand_vector_set): Likewise.
20145 (ix86_expand_vector_extract): Likewise.
20146 (emit_reduc_half): Likewise.
20147 (ix86_emit_i387_round): Likewise.
20148 (ix86_mangle_type): Likewise.
20149 (ix86_expand_round_sse4): Likewise.
20150 (expand_vec_perm_blend): Likewise.
20151 (canonicalize_vector_int_perm): Likewise.
20152 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
20153 (expand_vec_perm_1): Likewise.
20154 (expand_vec_perm_interleave3): Likewise.
20155 (expand_vec_perm_even_odd_pack): Likewise.
20156 (expand_vec_perm_even_odd_1): Likewise.
20157 (expand_vec_perm_broadcast_1): Likewise.
20158 (ix86_vectorize_vec_perm_const_ok): Likewise.
20159 (ix86_expand_vecop_qihi): Likewise.
20160 (ix86_expand_mul_widen_hilo): Likewise.
20161 (ix86_expand_sse2_abs): Likewise.
20162 (ix86_expand_pextr): Likewise.
20163 (ix86_expand_pinsr): Likewise.
20164 (ix86_preferred_simd_mode): Likewise.
20165 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
20166 * config/i386/sse.md (*andnot<mode>3): Likewise.
20167 (<mask_codefor><code><mode>3<mask_name>): Likewise.
20168 (*<code><mode>3): Likewise.
20169 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
20170 (ia64_expand_atomic_op): Likewise.
20171 (ia64_arg_type): Likewise.
20172 (ia64_mode_to_int): Likewise.
20173 (ia64_scalar_mode_supported_p): Likewise.
20174 (ia64_vector_mode_supported_p): Likewise.
20175 (expand_vec_perm_broadcast): Likewise.
20176 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
20177 (iq2000_function_arg_advance): Likewise.
20178 (iq2000_function_arg): Likewise.
20179 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
20180 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
20181 (m68k_libcall_value): Likewise.
20182 (m68k_function_value): Likewise.
20183 (sched_attr_op_type): Likewise.
20184 * config/mcore/mcore.c (mcore_output_move): Likewise.
20185 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
20187 (microblaze_function_arg): Likewise.
20188 * config/mips/mips.c (mips16_build_call_stub): Likewise.
20189 (mips_print_operand): Likewise.
20190 (mips_mode_ok_for_mov_fmt_p): Likewise.
20191 (mips_vector_mode_supported_p): Likewise.
20192 (mips_preferred_simd_mode): Likewise.
20193 (mips_expand_vpc_loongson_even_odd): Likewise.
20194 (mips_expand_vec_unpack): Likewise.
20195 (mips_expand_vi_broadcast): Likewise.
20196 (mips_expand_vector_init): Likewise.
20197 (mips_expand_vec_reduc): Likewise.
20198 (mips_expand_msa_cmp): Likewise.
20199 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
20200 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
20201 (cc_flags_for_mode): Likewise.
20202 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20203 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20204 (nds32_output_casesi_pc_relative): Likewise.
20205 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20206 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20207 (nvptx_gen_unpack): Likewise.
20208 (nvptx_gen_pack): Likewise.
20209 (nvptx_gen_shuffle): Likewise.
20210 (nvptx_gen_wcast): Likewise.
20211 (nvptx_preferred_simd_mode): Likewise.
20212 * config/pa/pa.c (pa_secondary_reload): Likewise.
20213 * config/pa/predicates.md (base14_operand): Likewise.
20214 * config/powerpcspe/powerpcspe-c.c
20215 (altivec_resolve_overloaded_builtin): Likewise.
20216 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20218 (rs6000_preferred_simd_mode): Likewise.
20219 (output_vec_const_move): Likewise.
20220 (rs6000_expand_vector_extract): Likewise.
20221 (rs6000_split_vec_extract_var): Likewise.
20222 (reg_offset_addressing_ok_p): Likewise.
20223 (rs6000_legitimate_offset_address_p): Likewise.
20224 (rs6000_legitimize_address): Likewise.
20225 (rs6000_emit_set_const): Likewise.
20226 (rs6000_const_vec): Likewise.
20227 (rs6000_emit_move): Likewise.
20228 (spe_build_register_parallel): Likewise.
20229 (rs6000_darwin64_record_arg_recurse): Likewise.
20230 (swap_selector_for_mode): Likewise.
20231 (spe_init_builtins): Likewise.
20232 (paired_init_builtins): Likewise.
20233 (altivec_init_builtins): Likewise.
20234 (do_load_for_compare): Likewise.
20235 (rs6000_generate_compare): Likewise.
20236 (rs6000_expand_float128_convert): Likewise.
20237 (emit_load_locked): Likewise.
20238 (emit_store_conditional): Likewise.
20239 (rs6000_output_function_epilogue): Likewise.
20240 (rs6000_handle_altivec_attribute): Likewise.
20241 (rs6000_function_value): Likewise.
20242 (emit_fusion_gpr_load): Likewise.
20243 (emit_fusion_p9_load): Likewise.
20244 (emit_fusion_p9_store): Likewise.
20245 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20246 (fusion_gpr_mem_load): Likewise.
20247 (fusion_addis_mem_combo_load): Likewise.
20248 (fusion_addis_mem_combo_store): Likewise.
20249 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20250 (fusion_gpr_mem_load): Likewise.
20251 (fusion_addis_mem_combo_load): Likewise.
20252 (fusion_addis_mem_combo_store): Likewise.
20253 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20255 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20256 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20257 (rs6000_preferred_simd_mode): Likewise.
20258 (output_vec_const_move): Likewise.
20259 (rs6000_expand_vector_extract): Likewise.
20260 (rs6000_split_vec_extract_var): Likewise.
20261 (reg_offset_addressing_ok_p): Likewise.
20262 (rs6000_legitimate_offset_address_p): Likewise.
20263 (rs6000_legitimize_address): Likewise.
20264 (rs6000_emit_set_const): Likewise.
20265 (rs6000_const_vec): Likewise.
20266 (rs6000_emit_move): Likewise.
20267 (rs6000_darwin64_record_arg_recurse): Likewise.
20268 (swap_selector_for_mode): Likewise.
20269 (paired_init_builtins): Likewise.
20270 (altivec_init_builtins): Likewise.
20271 (rs6000_expand_float128_convert): Likewise.
20272 (emit_load_locked): Likewise.
20273 (emit_store_conditional): Likewise.
20274 (rs6000_output_function_epilogue): Likewise.
20275 (rs6000_handle_altivec_attribute): Likewise.
20276 (rs6000_function_value): Likewise.
20277 (emit_fusion_gpr_load): Likewise.
20278 (emit_fusion_p9_load): Likewise.
20279 (emit_fusion_p9_store): Likewise.
20280 * config/rx/rx.c (rx_gen_move_template): Likewise.
20281 (flags_from_mode): Likewise.
20282 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20283 (s390_slb_comparison): Likewise.
20284 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20285 (s390_vector_mode_supported_p): Likewise.
20286 (s390_cc_modes_compatible): Likewise.
20287 (s390_match_ccmode_set): Likewise.
20288 (s390_canonicalize_comparison): Likewise.
20289 (s390_emit_compare_and_swap): Likewise.
20290 (s390_branch_condition_mask): Likewise.
20291 (s390_rtx_costs): Likewise.
20292 (s390_secondary_reload): Likewise.
20293 (__SECONDARY_RELOAD_CASE): Likewise.
20294 (s390_expand_cs): Likewise.
20295 (s390_preferred_simd_mode): Likewise.
20296 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20297 * config/sh/sh.c (sh_print_operand): Likewise.
20298 (dump_table): Likewise.
20299 (sh_secondary_reload): Likewise.
20300 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20301 * config/sh/sh.md (casesi_worker_1): Likewise.
20302 (casesi_worker_2): Likewise.
20303 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20304 (fcc_comparison_operator): Likewise.
20305 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20306 (emit_soft_tfmode_cvt): Likewise.
20307 (sparc_preferred_simd_mode): Likewise.
20308 (output_cbranch): Likewise.
20309 (sparc_print_operand): Likewise.
20310 (sparc_expand_vec_perm_bmask): Likewise.
20311 (vector_init_bshuffle): Likewise.
20312 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20313 (spu_vector_mode_supported_p): Likewise.
20314 (spu_expand_insv): Likewise.
20315 (spu_emit_branch_or_set): Likewise.
20316 (spu_handle_vector_attribute): Likewise.
20317 (spu_builtin_splats): Likewise.
20318 (spu_builtin_extract): Likewise.
20319 (spu_builtin_promote): Likewise.
20320 (spu_expand_sign_extend): Likewise.
20321 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20322 (tilegx_simd_int): Likewise.
20323 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20324 (tilepro_simd_int): Likewise.
20325 * config/v850/v850.c (const_double_split): Likewise.
20326 (v850_print_operand): Likewise.
20327 (ep_memory_offset): Likewise.
20328 * config/vax/vax.c (vax_rtx_costs): Likewise.
20329 (vax_output_int_move): Likewise.
20330 (vax_output_int_add): Likewise.
20331 (vax_output_int_subtract): Likewise.
20332 * config/visium/predicates.md (visium_branch_operator): Likewise.
20333 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20334 (visium_print_operand_address): Likewise.
20335 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20336 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20337 (xtensa_expand_conditional_branch): Likewise.
20338 (xtensa_copy_incoming_a7): Likewise.
20339 (xtensa_output_literal): Likewise.
20340 * dfp.c (decimal_real_maxval): Likewise.
20341 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20344 Alan Hayward <alan.hayward@arm.com>
20345 David Sherwood <david.sherwood@arm.com>
20347 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20348 (mode_nunits_inline): Likewise.
20349 (mode_inner_inline): Likewise.
20350 (mode_unit_size_inline): Likewise.
20351 (mode_unit_precision_inline): Likewise.
20352 (emit_insn_modes_h): Likewise. Also emit a #define of the
20354 (emit_mode_wider): Add an E_ prefix to mode names.
20355 (emit_mode_complex): Likewise.
20356 (emit_mode_inner): Likewise.
20357 (emit_mode_adjustments): Likewise.
20358 (emit_mode_int_n): Likewise.
20359 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20360 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20361 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20362 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20363 (CRC32_BUILTIN, ENTRY): Likewise.
20364 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20365 (aarch64_pop_regs): Likewise.
20366 (aarch64_process_components): Likewise.
20367 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20368 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20369 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20370 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20371 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20372 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20374 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20375 (pcmpistr): Likewise.
20376 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20377 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20378 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20380 * config/rl78/rl78.c (mduc_regs): Likewise.
20381 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20382 (htm_expand_builtin): Likewise.
20383 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20384 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20385 * config/xtensa/xtensa.c (print_operand): Likewise.
20386 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20387 (NUM_MODE_VECTOR_INT): Likewise.
20388 * genoutput.c (null_operand): Likewise.
20389 (output_operand_data): Likewise.
20390 * genrecog.c (print_parameter_value): Likewise.
20391 * lra.c (debug_operand_data): Likewise.
20393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20395 * dwarf2out.c (scompare_loc_descriptor_wide)
20396 (scompare_loc_descriptor_narrow): New functions, split out from...
20397 (scompare_loc_descriptor): ...here.
20398 * expmed.c (emit_store_flag_int): New function, split out from...
20399 (emit_store_flag): ...here.
20401 2017-08-30 Richard Biener <rguenther@suse.de>
20403 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20404 (dwarf2out_early_finish): Move setting of AT_pubnames from
20405 early debug output to early finish.
20407 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20409 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20410 and -mdata-region to the assembler.
20412 2017-08-30 Richard Biener <rguenther@suse.de>
20414 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20416 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20417 (dwarf2out_early_global_decl): Only generate a DIE for the
20418 abstract origin if it doesn't already exist or is a declaration DIE.
20419 (resolve_addr): Do not add the linkage name twice when
20420 generating a stub DIE for the DW_TAG_GNU_call_site target.
20422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20424 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20425 Use machine_mode rather than int for arg1_mode.
20427 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20430 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20431 that the second argument of the built-in functions to unpack
20432 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20433 switch statement instead a lot of if statements.
20434 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20435 Allow 64-bit values to be in Altivec registers as well as
20436 traditional floating point registers.
20437 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20439 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20441 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20442 MAX_REGS_PER_ADDRESS == 1.
20444 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20446 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20447 * config/i386/i386.c (ix86_option_override_internal): Simplify
20448 setting of opts->x_flag_entry.
20450 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20451 Jakub Jelinek <jakub@redhat.com>
20452 Richard Biener <rguenther@suse.de>
20454 PR tree-optimization/81503
20455 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20456 folded constant fits in the target type; reorder tests for clarity.
20458 2017-08-29 Martin Liska <mliska@suse.cz>
20460 * passes.def: Include pass_lower_switch.
20461 * stmt.c (dump_case_nodes): Remove and move to
20462 tree-switch-conversion.
20463 (case_values_threshold): Likewise.
20464 (expand_switch_as_decision_tree_p): Likewise.
20465 (emit_case_decision_tree): Likewise.
20466 (expand_case): Likewise.
20467 (balance_case_nodes): Likewise.
20468 (node_has_low_bound): Likewise.
20469 (node_has_high_bound): Likewise.
20470 (node_is_bounded): Likewise.
20471 (emit_case_nodes): Likewise.
20472 (struct simple_case_node): New struct.
20473 (add_case_node): Remove.
20474 (emit_case_dispatch_table): Use vector instead of case_list.
20475 (reset_out_edges_aux): Remove.
20476 (compute_cases_per_edge): Likewise.
20477 (expand_case): Build list of simple_case_node.
20478 (expand_sjlj_dispatch_table): Use it.
20479 * tree-switch-conversion.c (struct case_node): Moved from
20480 stmt.c and adjusted.
20481 (emit_case_nodes): Likewise.
20482 (node_has_low_bound): Likewise.
20483 (node_has_high_bound): Likewise.
20484 (node_is_bounded): Likewise.
20485 (case_values_threshold): Likewise.
20486 (reset_out_edges_aux): Likewise.
20487 (compute_cases_per_edge): Likewise.
20488 (add_case_node): Likewise.
20489 (dump_case_nodes): Likewise.
20490 (balance_case_nodes): Likewise.
20491 (expand_switch_as_decision_tree_p): Likewise.
20492 (emit_jump): Likewise.
20493 (emit_case_decision_tree): Likewise.
20494 (try_switch_expansion): Likewise.
20495 (do_jump_if_equal): Likewise.
20496 (emit_cmp_and_jump_insns): Likewise.
20497 (fix_phi_operands_for_edge): New function.
20498 (record_phi_operand_mapping): Likewise.
20499 (class pass_lower_switch): New pass.
20500 (pass_lower_switch::execute): New function.
20501 (make_pass_lower_switch): Likewise.
20502 (conditional_probability):
20503 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20504 * tree-pass.h: Add make_pass_lower_switch.
20506 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20509 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20512 2017-08-29 Richard Biener <rguenther@suse.de>
20514 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20515 we do not add a DW_AT_inline attribute twice.
20516 (gen_subprogram_die): Remove code setting DW_AT_inline on
20517 DECL_ABSTRACT_P nodes.
20519 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20521 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20522 calls to internal functions.
20523 (gimplify_modify_expr): Likewise.
20524 * tree-call-cdce.c (use_internal_fn): Likewise.
20525 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20526 (convert_to_divmod): Set the nothrow flag.
20527 * tree-if-conv.c (predicate_mem_writes): Likewise.
20528 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20529 (vectorizable_call): Likewise.
20530 (vectorizable_store): Likewise.
20531 (vectorizable_load): Likewise.
20532 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20533 (vect_recog_mask_conversion_pattern): Likewise.
20535 2017-08-29 Martin Liska <mliska@suse.cz>
20538 * gcc.c (driver_handle_option): Add new argument.
20539 * opts-common.c (handle_option): Pass
20540 target_option_override_hook.
20541 * opts-global.c (lang_handle_option): Add new option.
20542 (set_default_handlers): Add new argument.
20543 (decode_options): Likewise.
20544 * opts.c (target_handle_option): Likewise.
20545 (common_handle_option): Call target_option_override_hook.
20546 * opts.h (struct cl_option_handler_func): Add hook for
20547 target option override.
20548 (struct cl_option_handlers): Likewise.
20549 (set_default_handlers): Add new argument.
20550 (decode_options): Likewise.
20551 (common_handle_option): Likewise.
20552 (target_handle_option): Likewise.
20553 * toplev.c (toplev::main): Pass targetm.target_option.override
20556 2017-08-29 Richard Biener <rguenther@suse.de>
20557 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20559 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20560 life to the active subtree.
20562 2017-08-28 Jeff Law <law@redhat.com>
20564 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20565 derive_equivalences.
20566 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20567 Code moved into....
20568 (edge_info::derive_equivalences): New private member function
20570 * tree-ssa-dom.c (class edge_info): Changed from a struct
20571 to a class. Add ctor/dtor, methods and data members.
20572 (edge_info::edge_info): Renamed from allocate_edge_info.
20573 Initialize additional members.
20574 (edge_info::~edge_info): New.
20575 (free_dom_edge_info): Delete the edge info.
20576 (record_edge_info): Use new class & associated member functions.
20577 Tighten forms for testing for edge equivalences.
20578 (record_temporary_equivalences): Iterate over the simple
20579 equivalences rather than assuming there's only one per edge.
20580 (cprop_into_successor_phis): Iterate over the simple
20581 equivalences rather than assuming there's only one per edge.
20582 (optimize_stmt): Use operand_equal_p rather than pointer
20583 equality for mini-DSE code.
20585 2017-08-28 Nathan Sidwell <nathan@acm.org>
20587 * gcc.c (execute): Fold SIGPIPE handling into switch
20588 statement. Adjust internal error message.
20590 2017-08-28 Richard Biener <rguenther@suse.de>
20593 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20594 Do nothing for removed DIEs.
20596 2017-08-28 Richard Biener <rguenther@suse.de>
20598 PR tree-optimization/81977
20599 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20602 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20605 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20606 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20607 using targetm.gen_mem_thread_fence.
20609 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20612 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20613 predicate to register_operand. Reorder operands.
20614 (*btr<mode>): Ditto.
20615 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20616 (*btr<mode>_mask): Ditto.
20618 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20620 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20621 * config/rs6000/xmmintrin.h: New file.
20622 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20624 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20627 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20628 parameter and_insn and return it.
20629 (recombine_lvx_pattern): Insert a copy to ensure availability of
20630 the base register of the copied masking operation at the point of
20631 the instruction replacement.
20632 (recombine_stvx_pattern): Likewise.
20634 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20636 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20637 undocumented switches.
20638 (-mpower9-dform-vector): Likewise.
20639 (-mpower9-dform): Likewise.
20640 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20641 comments to delete references to -mpower9-dform* switches.
20642 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20643 Delete reference to -mpower9-dform* switches, test for
20644 -mpower9-vector instead.
20645 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20646 (OTHER_P9_VECTOR_MASKS): Likewise.
20647 (POWERPC_MASKS): Likewise.
20648 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20649 tests against -mpower9-dform* to -mpower9-vector. Delete code
20650 that checked for -mpower9-dform* consistancy with other options.
20651 Add test for -mpower9-misc to enable other power9 switches.
20652 (rs6000_init_hard_regno_mode_ok): Likewise.
20653 (rs6000_option_override_internal): Likewise.
20654 (rs6000_emit_prologue): Likewise.
20655 (rs6000_emit_epilogue): Likewise.
20656 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20657 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20658 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20660 (emit_fusion_p9_store): Likewise.
20661 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20662 resetting these macros if the assembler does not support ISA 3.0
20664 (TARGET_P9_DFORM_VECTOR): Likewise.
20665 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20666 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20668 2017-08-25 Alan Modra <amodra@gmail.com>
20670 PR rtl-optimization/81747
20671 * cse.c (cse_extended_basic_block): Don't attempt to record
20672 equivalences for degenerate conditional branches that branch
20673 to their fall-through.
20675 2017-08-24 Martin Sebor <msebor@redhat.com>
20677 PR middle-end/81908
20678 * gimple-fold.c (size_must_be_zero_p): New function.
20679 (gimple_fold_builtin_memory_op): Call it.
20681 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20683 * config/rs6000/mm_malloc.h: New file.
20685 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20687 PR tree-optimization/81913
20688 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20689 analysis when either IVs in condition can wrap.
20691 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20693 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20694 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20696 2017-08-24 Richard Biener <rguenther@suse.de>
20699 * targhooks.c (default_target_can_inline_p): Properly
20700 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20701 is present and always compare.
20702 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20703 infer -mfpmath=sse from TARGET_SSE_P.
20704 (ix86_can_inline_p): Properly use target_option_default_node when
20705 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20707 2017-08-24 Richard Biener <rguenther@suse.de>
20710 * dwarf2out.c (output_die): Handle flag_generate_offload like
20712 (output_comp_unit): Likewise.
20713 (gen_array_type_die): Likewise.
20714 (dwarf2out_early_finish): Likewise.
20715 (note_variable_value_in_expr): Likewise.
20716 (dwarf2out_finish): Likewise. Adjust assert.
20717 * cgraphunit.c (symbol_table::compile): Move setting of
20718 flag_generate_offload earlier ...
20719 (symbol_table::finalize_compilation_unit): ... here, before
20720 early debug finalization.
20722 2017-08-24 Richard Biener <rguenther@suse.de>
20724 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20725 and ipa-fnsummary.h.
20726 (ix86_can_inline_p): When ix86_fpmath flags do not match
20727 check whether the callee uses FP math at all.
20729 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20731 PR middle-end/81931
20732 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20733 instead of TYPE_PRECISION.
20735 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20736 Alan Hayward <alan.hayward@arm.com>
20737 David Sherwood <david.sherwood@arm.com>
20739 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20741 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20743 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20745 (ix86_frame::stack_realign_allocate): New field.
20746 (struct machine_frame_state): Modify comments.
20747 (machine_frame_state::sp_realigned_fp_end): New field.
20748 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20749 layout calculation.
20750 (sp_valid_at): Add assertion to assure no attempt to access invalid
20751 offset of a realigned stack.
20752 (fp_valid_at): Likewise.
20753 (choose_baseaddr): Modify comments.
20754 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20755 ix86_expand_prologue.
20756 (ix86_expand_prologue): Modify stack realignment and allocation.
20757 (ix86_expand_epilogue): Modify comments.
20758 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20759 avx2_runtime, avx512f, and avx512f_runtime.
20761 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20763 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20764 (mstackrealign): Do not init to -1.
20765 * config/i386/i386.c (ix86_option_override_internal):
20766 Check opts_set, not opts when setting default value of
20767 opts->x_ix86_force_align_arg_pointer.
20769 2017-08-23 Richard Biener <rguenther@suse.de>
20771 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20772 lang_hooks.decl_printable_name.
20773 * print-rtl-function.c (print_rtx_function): Likewise.
20774 * tree-pretty-print.c (dump_function_header): Likewise.
20776 2017-08-23 Richard Biener <rguenther@suse.de>
20779 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20780 -g0 at compile-time.
20782 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20784 PR middle-end/19706
20785 * doc/sourcebuild.texi (Other hardware attributes):
20788 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20790 PR middle-end/19706
20791 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20792 Fix single-use check.
20794 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20796 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20798 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20800 * doc/install.texi: Modify to add more details on running selected
20803 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20805 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20806 is combined with -mabi=ms.
20807 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20810 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20812 PR tree-optimization/81488
20813 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20814 and cached_basis fields.
20815 (MAX_SPREAD): New constant.
20816 (alloc_cand_and_find_basis): Initialize new fields.
20817 (clear_visited): New function.
20818 (create_phi_basis_1): Rename from create_phi_basis, set visited
20819 and cached_basis fields.
20820 (create_phi_basis): New wrapper function.
20821 (phi_add_costs_1): Rename from phi_add_costs, add spread
20822 parameter, set visited field, short-circuit when limits reached.
20823 (phi_add_costs): New wrapper function.
20824 (record_phi_increments_1): Rename from record_phi_increments, set
20826 (record_phi_increments): New wrapper function.
20827 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20828 (phi_incr_cost): New wrapper function.
20829 (all_phi_incrs_profitable_1): Rename from
20830 all_phi_incrs_profitable, set visited field.
20831 (all_phi_incrs_profitable): New wrapper function.
20833 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20834 Alan Hayward <alan.hayward@arm.com>
20835 David Sherwood <david.sherwood@arm.com>
20837 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20838 that takes the outer and inner modes.
20839 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20840 comparison as the canonical test for a paradoxical subreg.
20841 * combine.c (simplify_set): Use paradoxical_subreg_p.
20842 (make_extraction): Likewise.
20843 (force_to_mode): Likewise.
20844 (rtx_equal_for_field_assignment_p): Likewise.
20845 (gen_lowpart_for_combine): Likewise.
20846 (simplify_comparison): Likewise.
20847 * cse.c (equiv_constant): Likewise.
20848 * expmed.c (store_bit_field_1): Likewise.
20849 * final.c (alter_subreg): Likewise.
20850 * fwprop.c (propagate_rtx): Likewise.
20851 (forward_propagate_subreg): Likewise.
20852 * ira-conflicts.c (ira_build_conflicts): Likewise.
20853 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20854 * lra-constraints.c (curr_insn_transform): Likewise.
20855 (split_reg): Likewise.
20856 * lra-eliminations.c (move_plus_up): Likewise.
20857 (lra_eliminate_regs_1): Likewise.
20858 * recog.c (general_operand): Likewise.
20859 * ree.c (combine_reaching_defs): Likewise.
20860 * reload.c (push_reload): Likewise.
20861 (find_reloads): Likewise.
20862 * reload1.c (elimination_effects): Likewise.
20863 (compute_reload_subreg_offset): Likewise.
20864 (choose_reload_regs): Likewise.
20865 * rtlanal.c (subreg_lsb_1): Likewise.
20866 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20867 (simplify_subreg): Likewise.
20868 * var-tracking.c (track_loc_p): Likewise.
20869 * emit-rtl.c (byte_lowpart_offset): Likewise.
20870 (paradoxical_subreg_p): Delete out-of-line definition.
20872 2017-08-22 Jeff Law <law@redhat.com>
20874 PR tree-optimization/81741
20875 PR tree-optimization/71947
20876 * tree-ssa-dom.c: Include tree-inline.h.
20877 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20878 equivalences if one is more expensive to compute than the other.
20879 * tree-ssa-scopedtables.h (class const_or_copies): Make
20880 record_const_or_copy_raw method private.
20881 (class avail_exprs_stack): New method simplify_binary_operation.
20882 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20883 avail_exprs_stack::simplify_binary_operation as needed.
20884 (avail_exprs_stack::simplify_binary_operation): New function.
20886 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20888 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20889 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20890 (DOT_SYMBOLS): Likewise.
20891 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20892 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20893 (RS6000_ABI_NAME): Likewise.
20894 (TARGET_CMODEL): Likewise.
20895 (TOC_SECTION_ASM_OP): Likewise.
20896 (SET_CMODEL): New macro.
20897 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20899 2017-08-22 Richard Biener <rguenther@suse.de>
20901 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20902 to allow for free-lang-data replacements similar to verify_type_variant.
20904 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20905 Alan Hayward <alan.hayward@arm.com>
20906 David Sherwood <david.sherwood@arm.com>
20908 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20909 VOIDmode for the LABEL_REF.
20911 2017-08-22 Richard Biener <rguenther@suse.de>
20913 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20915 2017-08-22 Richard Biener <rguenther@suse.de>
20917 * common.opt (feliminate-dwarf2-dups): Ignore.
20918 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20919 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20920 same_die_p_wrap, compute_section_prefix,
20921 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20922 (comdat_symbol_id, comdat_symbol_number): Likewise.
20923 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20925 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20926 (output_die): Mark unreachable path unreachable.
20927 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20928 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20929 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20930 (dwarf2out_early_finish): Likewise.
20932 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20934 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20936 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20939 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20940 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20941 (avr_attribute_table) <io, io_low, address>: Initialize
20942 .decl_required with true.
20944 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20946 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20947 undocumented debugging options.
20948 (-mvsx-scalar-double): Likewise.
20949 (-mallow-df-permute): Likewise.
20950 (-mvectorize-builtins): Likewise.
20951 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20952 (rs6000_builtin_vectorized_function): Likewise.
20953 (rs6000_builtin_md_vectorized_function): Likewise.
20954 (rs6000_opt_vars): Likewise.
20956 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20959 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20960 (*btrq_imm): Rename from *btrq.
20961 (*btcq_imm): Rename from *btcq.
20962 (btsc): New code attribute.
20963 (*<btsc><mode>): New insn pattern.
20964 (*btr<mode>): Ditto.
20965 (*<btsc><mode>_mask): New insn_and_split pattern.
20966 (*btr<mode>_mask): Ditto.
20968 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20969 Alan Hayward <alan.hayward@arm.com>
20970 David Sherwood <david.sherwood@arm.com>
20972 * function.c (pad_below): Simplify padding calculation.
20974 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20975 Alan Hayward <alan.hayward@arm.com>
20976 David Sherwood <david.sherwood@arm.com>
20978 * target.def (function_prologue): Remove frame size argument.
20979 (function_epilogue): Likewise.
20980 * doc/tm.texi: Regenerate.
20981 * final.c (final_start_function): Update call to function_prologue.
20982 (final_end_function): Update call to function_epilogue.
20983 (default_function_pro_epilogue): Remove frame size argument.
20984 * output.h (default_function_pro_epilogue): Likewise.
20985 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20986 (arm_output_function_prologue): Likewise.
20987 * config/frv/frv.c (frv_function_prologue): Likewise.
20988 (frv_function_epilogue): Likewise.
20989 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20990 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20991 (ia64_output_function_epilogue): Likewise.
20992 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20993 (m32r_output_function_epilogue): Likewise.
20994 * config/microblaze/microblaze.c (microblaze_function_prologue)
20995 (microblaze_function_epilogue): Likewise.
20996 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20997 (mips_output_function_epilogue): Likewise.
20998 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20999 (mmix_target_asm_function_epilogue): Likewise.
21000 * config/msp430/msp430.c (msp430_start_function): Likewise.
21001 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
21002 (nds32_asm_function_epilogue): Likewise.
21003 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
21004 * config/pa/pa.c (pa_output_function_prologue): Likewise.
21005 (pa_output_function_epilogue): Likewise.
21006 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
21007 (rs6000_output_function_epilogue): Likewise.
21008 * config/rl78/rl78.c (rl78_start_function): Likewise.
21009 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
21010 (rs6000_output_function_epilogue): Likewise.
21011 * config/rx/rx.c (rx_output_function_prologue): Likewise.
21012 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
21013 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
21014 (sparc_asm_function_epilogue): Likewise.
21016 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21018 * tree.h (type_has_mode_precision_p): New function.
21019 * convert.c (convert_to_integer_1): Use it.
21020 * expr.c (expand_expr_real_2): Likewise.
21021 (expand_expr_real_1): Likewise.
21022 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
21023 * match.pd: Likewise.
21024 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
21025 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
21026 * tree-tailcall.c (process_assignment): Likewise.
21027 * tree-vect-loop.c (vectorizable_reduction): Likewise.
21028 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
21029 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
21030 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
21031 (vectorizable_assignment): Likewise.
21032 (vectorizable_shift): Likewise.
21033 (vectorizable_operation): Likewise.
21034 * tree-vrp.c (register_edge_assert_for_2): Likewise.
21036 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
21038 * match.pd: Add pow (C, x) simplification.
21040 2017-08-21 Richard Biener <rguenther@suse.de>
21042 PR tree-optimization/81900
21043 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
21044 for blocks with abnormal predecessors.
21045 (compute_antic): Do not set visited flag prematurely.
21047 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
21050 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
21052 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21054 * stor-layout.h (vector_type_mode): Move to...
21055 * tree.h (vector_type_mode): ...here.
21056 * stor-layout.c (vector_type_mode): Move to...
21057 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
21059 2017-08-21 Richard Biener <rguenther@suse.de>
21061 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
21062 register_external_die hooks.
21063 (debug_false_tree_charstarstar_uhwistar): Declare.
21064 (debug_nothing_tree_charstar_uhwi): Likewise.
21065 * debug.c (do_nothing_debug_hooks): Adjust.
21066 (debug_false_tree_charstarstar_uhwistar): New do nothing.
21067 (debug_nothing_tree_charstar_uhwi): Likewise.
21068 * dbxout.c (dbx_debug_hooks): Adjust.
21069 (xcoff_debug_hooks): Likewise.
21070 * sdbout.c (sdb_debug_hooks): Likewise.
21071 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21072 * dwarf2out.c (macinfo_label_base): New global.
21073 (dwarf2out_register_external_die): New function for the
21074 register_external_die hook.
21075 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
21076 (dwarf2_debug_hooks): Use them.
21077 (dwarf2_lineno_debug_hooks): Adjust.
21078 (struct die_struct): Add with_offset flag.
21079 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
21080 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
21081 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
21082 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
21083 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
21084 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
21085 defining section names for the early LTO debug variants.
21086 (reset_indirect_string): New helper.
21087 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
21088 (print_dw_val): Add support for offsetted symbol references.
21089 (get_ultimate_context): Split out from is_cxx.
21090 (is_cxx): Use get_ultimate_context.
21091 (is_fortran): Add decl overload.
21092 (compute_comp_unit_symbol): Split out worker from
21093 compute_section_prefix.
21094 (compute_section_prefix): Call compute_comp_unit_symbol and
21095 set comdat_type_p here.
21096 (output_die): Skip DIE symbol output for the LTO added one.
21097 Handle DIE symbol references with offset.
21098 (output_comp_unit): Guard section name mangling properly.
21099 For LTO debug sections emit a symbol at the section beginning
21100 which we use to refer to its DIEs.
21101 (add_abstract_origin_attribute): For DIEs registered via
21102 dwarf2out_register_external_die directly refer to the early
21103 DIE rather than indirectly through the shadow one we created.
21104 Remove obsolete call to dwarf2out_abstract_function for
21105 non-function/block origins.
21106 (gen_array_type_die): When generating early LTO debug do
21107 not emit DW_AT_string_length.
21108 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
21109 late when in LTO. As suggested place a gcc_unreachable for
21110 the DECL_ABSTRACT_P case.
21111 (gen_subprogram_die): Avoid another specification DIE
21112 for early built declarations/definitions for the late LTO case.
21113 (gen_variable_die): Add type references for late duplicated VLA dies
21115 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
21116 we have the abstract instance already.
21117 (process_scope_var): Adjust decl DIE contexts in LTO which
21118 first puts them in limbo.
21119 (gen_decl_die): Do not generate type DIEs late apart from
21120 types for VLAs or for decls we do not yet have a DIE. Do not
21121 call dwarf2out_abstract_function late.
21122 (dwarf2out_early_global_decl): Make sure to create DIEs
21123 for abstract instances of a decl first.
21124 (dwarf2out_late_global_decl): Adjust comment.
21125 (output_macinfo_op): With multiple macro sections use
21126 macinfo_label_base to distinguish labels.
21127 (output_macinfo): Likewise. Update macinfo_label_base.
21128 Pass in the line info label.
21129 (note_variable_value_in_expr): When generating LTO resolve
21130 all variable values here by generating DIEs as needed.
21131 (init_sections_and_labels): Add early LTO debug flag parameter
21132 and generate different sections and names if set. Add generation
21133 counter for the labels so we can have multiple of them.
21134 (reset_dies): Helper to allow DIEs to be output multiple times.
21135 (dwarf2out_finish): When outputting DIEs to the fat part of an
21136 LTO object first reset DIEs.
21137 (dwarf2out_early_finish): Output early DIEs when generating LTO.
21138 (modified_type_die): Check for decl_ultimate_origin being self
21140 (gen_type_die_with_usage): Likewise.
21141 (gen_typedef_die): Allow decl_ultimate_origin being self.
21142 (set_decl_abstract_flags): Remove.
21143 (set_block_abstract_flags): Likewise.
21144 (dwarf2out_abstract_function): Treat the early generated DIEs
21145 as the abstract copy and only add DW_AT_inline and
21146 DW_AT_artificial here and call set_decl_origin_self.
21147 If the DIE has an abstract origin don't do anything.
21148 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
21149 if we have none yet (Go fails to build one, PR78628).
21150 (variably_modified_type_p): Prevent endless recursion for Ada
21151 cyclic pointer types.
21152 * lto-streamer-in.c: Include debug.h.
21153 (dref_queue): New global.
21154 (lto_read_tree_1): Stream in DIE references.
21155 (lto_input_tree): Register DIE references.
21156 (input_function): Stream DECL_DEBUG_ARGS.
21157 * lto-streamer-out.c: Include debug.h.
21158 (lto_write_tree_1): Output DIE references.
21159 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
21160 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
21161 (output_function): Stream DECL_DEBUG_ARGS.
21162 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
21163 Stream DECL_ABSTRACT_ORIGIN.
21164 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
21165 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
21166 DECL_CONTEXT for file-scope decls.
21167 * lto-streamer.h (struct dref_entry): Declare.
21168 (dref_queue): Likewise.
21169 * cfgexpand.c (pass_expand::execute): Do not call the
21170 outlining_inline_function hook here.
21171 * lto-wrapper.c (debug_obj): New global.
21172 (tool_cleanup): Unlink it if required.
21173 (debug_objcopy): New function.
21174 (run_gcc): Handle early debug sections in the IL files by
21175 extracting them to separate files, partially linkin them and
21176 feeding the result back as result to the linker.
21177 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
21178 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
21179 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
21180 sections into a separate segment.
21181 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
21183 (darwin_asm_dwarf_section): Likewise.
21184 (darwin_asm_output_dwarf_offset): Likewise.
21185 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
21187 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21188 Alan Hayward <alan.hayward@arm.com>
21189 David Sherwood <david.sherwood@arm.com>
21191 * read-md.h (md_reader::record_potential_iterator_use): Replace
21192 pointer argument with an rtx and an index.
21193 * read-rtl.c (iterator_group::apply_iterator): Likewise.
21194 (apply_mode_iterator): Likewise.
21195 (apply_code_iterator): Likewise.
21196 (apply_int_iterator): Likewise.
21197 (apply_subst_iterator): Likewise.
21198 (record_iterator_use): Likewise.
21199 (record_attribute_use): Likewise.
21200 (md_reader::record_potential_iterator_use): Likewise. Update calls
21201 to record_iterator_use and apply_iterator.
21202 (iterator_use): Replace ptr with x and index.
21203 (attribute_use): Likewise.
21204 (apply_attribute_uses): Update calls to apply_iterator.
21205 (apply_iterators): Likewise. Update initialization of iterator_use.
21206 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21207 and record_potential_iterator_use.
21208 (rtx_reader::read_rtx_operand): Likewise.
21210 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21211 Alan Hayward <alan.hayward@arm.com>
21212 David Sherwood <david.sherwood@arm.com>
21214 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21217 2017-08-21 Richard Biener <rguenther@suse.de>
21219 PR middle-end/81884
21220 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21221 at struct end conservatively when comparing common bases.
21223 2017-08-21 Richard Biener <rguenther@suse.de>
21225 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21226 (mem_ref_in_stmt): Remove.
21227 (determine_max_movement): Use ref index to get at the reference.
21228 (invariantness_dom_walker::before_dom_children): Deal with
21229 lim data already initialized.
21230 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21232 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21234 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21235 (TARGET_ISA_ROUND): Ditto.
21236 (TARGET_ROUND): Ditto.
21237 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21238 * config/i386/i386.md: Ditto.
21239 * config/i386/sse.md: Ditto.
21240 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21241 with OPTION_MASK_ISA_SSE4_1.
21243 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21246 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21247 __builtin_ia32_lzcnt_u16.
21249 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21252 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21253 (rs6000_set_current_function): Rewrite function to use it.
21255 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21258 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21259 and TYPE_WARN_IF_NOT_ALIGN.
21260 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21261 (handle_warn_if_not_align): New.
21262 (place_union_field): Call handle_warn_if_not_align.
21263 (place_field): Call handle_warn_if_not_align.
21264 Copy TYPE_WARN_IF_NOT_ALIGN.
21265 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21266 (layout_type): Likewise.
21267 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21269 (tree_decl_common): Add warn_if_not_align.
21270 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21271 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21272 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21273 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21274 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21275 * doc/extend.texi: Document warn_if_not_aligned attribute.
21276 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21278 2017-08-17 Martin Liska <mliska@suse.cz>
21281 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21282 (get_data_dependence): Use it as pointer type.
21283 (distribute_loop): Likewise.
21285 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21287 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21288 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21289 (p8_vmrgow_<mode>_direct): New define_insn.
21290 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21291 handle endianness for vmrgew and vmrgow permute patterns.
21293 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21295 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21296 * config/rs6000/rs6000-cpus.def: Remove comment.
21297 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21298 (POWERPC_MASKS): Likewise.
21299 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21300 use of TARGET_VSX_TIMODE.
21301 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21302 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21303 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21304 (rs6000_option_override_internal): Remove dead code.
21305 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21306 (rs6000_legitimize_reload_address): Likewise.
21307 (rs6000_legitimate_address_p): Likewise.
21308 (rs6000_opt_masks): Delete "vsx-timode".
21309 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21310 from function comment.
21311 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21312 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21313 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21315 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21316 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21317 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21318 (VSX_TI): Likewise.
21320 (define_peephole2): Likewise.
21322 2017-08-17 Martin Sebor <msebor@redhat.com>
21325 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21326 past the end of an array.
21327 (test_pp_format): Add test cases.
21329 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21331 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21332 missing ECF_NOTHROW flags.
21334 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21337 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21338 operands residing in integer registers.
21339 (*vsx_le_perm_load_<mode>): Likewise.
21340 (*vsx_le_perm_store_<mode>): Likewise.
21341 (define_peephole2): Add peepholes to optimize the above.
21343 2017-08-17 Marek Polacek <polacek@redhat.com>
21345 PR middle-end/81814
21346 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21347 to mimic what shorten_compare did. Change the return type to bool.
21348 (fold_cond_expr_with_comparison): Update call to
21349 operand_equal_for_comparison_p.
21350 (fold_ternary_loc): Likewise.
21352 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21354 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21356 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21358 2017-08-17 Richard Biener <rguenther@suse.de>
21360 * tree-ssa-structalias.c (solve_graph): When propagating
21361 to successors update the graphs succ edges and avoid duplicate work.
21363 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21366 * config/i386/i386.c (ix86_option_override_internal): Save target
21367 specific options after ix86_stack_protector_guard_reg was changed.
21369 2017-08-17 Richard Biener <rguenther@suse.de>
21371 PR tree-optimization/81827
21372 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21373 (new_var_info): Initialize it conservatively.
21374 (get_call_vi): Mark register vars.
21375 (new_scalar_tmp_constraint_exp): Likewise.
21376 (handle_rhs_call): Likewise.
21377 (handle_const_call): Likewise.
21378 (create_function_info_for): Likewise.
21379 (solve_constraints): Sort varinfos to separate register from
21380 non-register vars to pack points-to solution bitmaps during
21383 2017-08-17 Marek Polacek <polacek@redhat.com>
21385 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21387 2017-08-17 Richard Biener <rguenther@suse.de>
21389 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21390 to true when overflow is undefined and we saturated the result.
21392 2017-08-17 Alan Modra <amodra@gmail.com>
21395 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21396 Don't use store multiple if only one reg needs saving.
21397 (interesting_frame_related_regno): New function.
21398 (rs6000_frame_related): Don't emit frame info for regs that
21400 (rs6000_emit_epilogue): Likewise.
21402 2017-08-16 Nathan Sidwell <nathan@acm.org>
21404 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21405 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21406 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21407 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21408 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21409 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21410 (verify_type): Adjust for TYPE_BINFO move.
21411 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21412 process TYPE_BINFO directly.
21413 (hash_tree): Likewise.
21414 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21416 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21419 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21421 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21423 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21426 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21427 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21428 Add ix86_binary_operator_ok to insn constraint.
21429 (*iordi_1_bts): Ditto.
21430 (*xordi_1_btc): Ditto.
21431 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21432 Update corresponding peephole2 pattern.
21436 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21438 PR tree-optimization/81832
21439 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21440 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21442 2017-08-16 Marek Polacek <polacek@redhat.com>
21445 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21446 regarding size_zero_node.
21448 2017-08-16 Martin Liska <mliska@suse.cz>
21451 * config.gcc: Respect previously set extra_objs in case
21454 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21456 PR tree-optimization/81835
21457 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21458 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21459 not depend on the phi.
21461 2017-08-16 Alan Modra <amodra@gmail.com>
21463 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21466 2017-08-16 Alan Modra <amodra@gmail.com>
21468 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21469 (save_reg_p): ..into this. Update all callers.
21470 (first_reg_to_save): Simplify.
21472 2017-08-16 Alan Modra <amodra@gmail.com>
21474 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21477 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21481 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21482 constant count if that count is less than 32.
21484 2017-08-15 Nathan Sidwell <nathan@acm.org>
21486 * gcc.c (execute): Emit friendlier message if inferior is killed
21487 by an external cause.
21489 2017-08-15 Richard Biener <rguenther@suse.de>
21491 PR tree-optimization/81790
21492 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21493 CONSTRUCTORs from simplifying and VN.
21495 2017-08-14 Martin Sebor <msebor@redhat.com>
21497 * builtin-attrs.def: Add comments.
21499 2017-08-14 Martin Sebor <msebor@redhat.com>
21502 * doc/extend.texi (attribute nonstring): Document new attribute.
21504 2017-08-14 Martin Sebor <msebor@redhat.com>
21507 * tree-diagnostic.c (default_tree_printer): Handle %G.
21508 * gimple-pretty-print.h (percent_G_format): Declare new function.
21509 * gimple-pretty-print.c (percent_G_format): Define.
21510 * tree-pretty-print.c (percent_K_format): Add argument.
21512 2017-08-14 Martin Sebor <msebor@redhat.com>
21514 PR translation/79998
21515 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21516 Remove a stray space.
21518 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21521 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21522 (*iordi_1_bts): Ditto.
21523 (*xordi_1_btc): Ditto.
21525 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21528 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21529 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21531 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21532 quoted strings, and make more translator-friendly.
21533 (darwin_rs6000_override_options): Likewise.
21534 (rs6000_option_override_internal): Likewise.
21535 (rs6000_return_in_memory): Fix overlong line.
21536 (init_cmulative_args): Use quoted strings, and make more
21537 translator-friendly.
21538 (rs6000_pass_by_reference): Fix overlong line.
21539 (def_builtin): Use quoted strings.
21540 (altivec_expand_predicate_builtin): Use quoted strings, and make
21541 more translator-friendly.
21542 (htm_expand_builtin): Use quoted strings.
21543 (cpu_expand_builtin): Use quoted strings, and make more
21544 translator-friendly.
21545 (altivec_expand_builtin): Likewise.
21546 (paired_expand_predicate_builtin): Likewise.
21547 (rs6000_invalid_builtin): Likewise.
21548 (builtin_function_type): Use quoted strings.
21549 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21550 more translator-friendly.
21551 (rs6000_trampoline_init): Likewise.
21552 (rs6000_handle_altivec_attribute): Likewise.
21553 (rs6000_inner_target_options): Use quoted strings.
21554 (rs6000_disable_incompatible_switches): Likewise.
21555 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21556 strings, and make more translator-friendly.
21557 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21559 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21561 PR tree-optimization/81799
21562 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21563 cond_expr to simple gimple operand.
21565 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21567 PR middle-end/46932
21568 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21570 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21574 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21575 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21576 TARGET_GASISR_PROLOGUES.
21577 * config/avr/avr.c (avr_option_override): Same.
21578 (avr_pass_pre_proep::execute): Same.
21580 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21583 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21584 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21586 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21588 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21589 define_insn_and_split. Split to a memory load from 0 in
21590 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21591 using PTR mode iterator.
21592 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21593 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21594 (*add_tp_<mode>): Redefine as define_insn_and_split.
21595 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21596 address space. Merge with *add_tp_x32 using PTR mode iterator.
21597 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21598 Split to an add with a memory load from 0 in
21599 DEFAULT_TLS_SEG_REG address space.
21601 2017-08-12 Andrew Pinski <apinski@cavium.com>
21603 * config/aarch64/aarch64-option-extensions.def (rdma):
21604 Fix feature string to what Linux prints out in /proc/cpuinfo.
21606 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21609 * dwarf2out.c (modified_type_die): For C typedef types that have
21610 an ultimate origin, process the ultimate origin instead of the
21612 (gen_typedef_die): Assert that input DECLs have no ultimate
21614 (gen_type_die_with_usage): For typedef variants that have an
21615 ultimate origin, just call gen_decl_die on the original DECL.
21616 (process_scope_var): Avoid creating DIEs for local typedefs and
21617 concrete static variables.
21619 2017-08-12 Alan Modra <amodra@gmail.com>
21623 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21624 match gnu-user.h startfile.
21625 (ENDFILE_LINUX_SPEC): Similarly.
21627 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21630 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21632 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21634 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21635 * config/aarch64/aarch64.md (mov<mode>): Change.
21636 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21637 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21638 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21640 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21642 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21643 for storage order barriers.
21645 2017-08-11 Martin Liska <mliska@suse.cz>
21648 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21649 variables of void type.
21651 2017-08-11 Martin Liska <mliska@suse.cz>
21653 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21654 TARGET_SUPPORTS_ALIASES.
21655 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21656 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21657 (optimize_weakref): Likewise.
21658 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21659 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21660 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21662 2017-08-11 Martin Liska <mliska@suse.cz>
21665 * config/i386/i386.c (make_resolver_func): Do complete
21666 refactoring of the function.
21668 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21671 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21672 * config/i386/i386.c (ix86_stack_protect_guard): Use
21673 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21674 * doc/invoke.texi (x86 Options): Document
21675 -mstack-protector-guard-symbol= option.
21677 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21679 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21680 * config/i386/i386.c (ix86_split_stack_guard): New function.
21681 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21682 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21683 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21684 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21685 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21686 (split_stack_space_check): Call ix86_split_stack_guard.
21688 2017-08-10 Martin Sebor <msebor@redhat.com>
21690 * print-tree.c (print_node): Print location using the established
21692 Replace spaces with colons.
21693 (debug_raw, debug): Ditto.
21695 2017-08-10 Martin Sebor <msebor@redhat.com>
21698 * pretty-print.c (pp_format): Correct the handling of %s precision.
21700 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21703 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21705 (ix86_finalize_stack_frame_flags): This. Also clear
21706 frame_pointer_needed if -fno-omit-frame-pointer is used without
21708 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21709 with ix86_finalize_stack_frame_flags.
21710 (ix86_expand_epilogue): Likewise.
21711 (ix86_expand_split_stack_prologue): Likewise.
21712 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21714 2017-08-10 Martin Liska <mliska@suse.cz>
21717 * c-attribs.c (handle_target_attribute):
21718 Report warning for an empty string argument of target attribute.
21720 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21723 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21725 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21726 or DECL_NONLOCAL labels.
21727 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21728 or DECL_NONLOCAL labels here.
21730 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21732 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21733 to indicate when early gimple folding has been disabled.
21734 (rs6000_gimple_fold_builtin): Add debug content.
21735 (rs6000_invalid_builtin): Fix whitespace.
21736 (rs6000_expand_builtin): Fix whitespace.
21737 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21739 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21742 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21743 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21745 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21747 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21748 (qdf24xx): Likewise.
21749 * config/aarch64/aarch64.md: Include falkor.md.
21750 * config/aarch64/falkor.md: New.
21752 2017-08-09 Marek Polacek <polacek@redhat.com>
21755 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21756 * diagnostic.c (emit_diagnostic): Add a comment.
21757 (emit_diagnostic_valist): New function.
21759 2017-08-09 Marek Polacek <polacek@redhat.com>
21762 * input.c (make_location): New overload.
21763 * input.h (make_location): Declare.
21765 2017-08-08 Alan Modra <amodra@gmail.com>
21766 H.J. Lu <hongjiu.lu@intel.com>
21769 * gcc.c (NO_PIE_SPEC): Delete.
21770 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21772 (LINK_PIE_SPEC): ..to here.
21773 (LINK_COMMAND_SPEC): Support -no-pie.
21774 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21775 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21777 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21778 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21779 (ENDFILE_CRTEND_SPEC): Similarly.
21781 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21784 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21785 (mstack-protector-guard-offset=): Ditto.
21786 * config/i386/i386.c (ix86_option_override): Handle
21787 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21789 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21790 ix86_stack_protect_guard_offset variables.
21791 (TARGET_STACK_PROTECT_GUARD): Always define.
21792 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21793 and -mstack-protector-guard-offset= options.
21795 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21797 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21798 boundary case for the last candidate.
21800 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21802 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21803 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21805 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21807 PR middle-end/19706
21808 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21809 * config/aarch64/aarch64-builtins.c
21810 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21811 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21812 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21814 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21815 Andrew Pinski <pinskia@gmail.com>
21817 PR middle-end/19706
21818 * internal-fn.def (XORSIGN): New.
21819 * optabs.def (xorsign_optab): New.
21820 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21821 (convert_expand_mult_copysign): New.
21822 (pass_optimize_widening_mul::execute): Call
21823 convert_expand_mult_copysign.
21825 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21827 PR tree-optimization/81354
21828 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21829 Insert on edges rather than explicitly creating landing pads.
21830 (analyze_candidates_and_replace): Commit edge inserts.
21832 2017-08-08 Richard Biener <rguenther@suse.de>
21834 PR middle-end/81719
21835 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21836 (expand_simple_operations): Also look through ADDR_EXPRs with
21837 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21839 2017-08-08 Richard Biener <rguenther@suse.de>
21841 PR tree-optimization/81723
21842 * tree-vect-slp.c (struct bst_traits): New hash traits.
21843 (bst_fail): New global.
21844 (vect_build_slp_tree_2): New worker, split out from ...
21845 (vect_build_slp_tree): ... this now wrapping it with using
21846 bst_fail set to cache SLP tree build fails. Properly handle
21848 (vect_analyze_slp_instance): Allocate and free bst_fail.
21850 2017-08-08 Martin Liska <mliska@suse.cz>
21853 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21854 LABEL_DECLs that can be from a different function.
21856 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21858 PR tree-optimization/81744
21859 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21860 loop's number of iterations.
21862 2017-08-08 Martin Liska <mliska@suse.cz>
21864 * asan.c: Include header files.
21865 * attribs.c (build_decl_attribute_variant): New function moved
21867 (build_type_attribute_qual_variant): Likewise.
21868 (cmp_attrib_identifiers): Likewise.
21869 (simple_cst_list_equal): Likewise.
21870 (omp_declare_simd_clauses_equal): Likewise.
21871 (attribute_value_equal): Likewise.
21872 (comp_type_attributes): Likewise.
21873 (build_type_attribute_variant): Likewise.
21874 (lookup_ident_attribute): Likewise.
21875 (remove_attribute): Likewise.
21876 (merge_attributes): Likewise.
21877 (merge_type_attributes): Likewise.
21878 (merge_decl_attributes): Likewise.
21879 (merge_dllimport_decl_attributes): Likewise.
21880 (handle_dll_attribute): Likewise.
21881 (attribute_list_equal): Likewise.
21882 (attribute_list_contained): Likewise.
21883 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21884 (lookup_attribute_by_prefix): Likewise.
21885 * bb-reorder.c: Include header files.
21886 * builtins.c: Likewise.
21887 * calls.c: Likewise.
21888 * cfgexpand.c: Likewise.
21889 * cgraph.c: Likewise.
21890 * cgraphunit.c: Likewise.
21891 * convert.c: Likewise.
21892 * dwarf2out.c: Likewise.
21893 * final.c: Likewise.
21894 * fold-const.c: Likewise.
21895 * function.c: Likewise.
21896 * gimple-expr.c: Likewise.
21897 * gimple-fold.c: Likewise.
21898 * gimple-pretty-print.c: Likewise.
21899 * gimple.c: Likewise.
21900 * gimplify.c: Likewise.
21901 * hsa-common.c: Likewise.
21902 * hsa-gen.c: Likewise.
21903 * internal-fn.c: Likewise.
21904 * ipa-chkp.c: Likewise.
21905 * ipa-cp.c: Likewise.
21906 * ipa-devirt.c: Likewise.
21907 * ipa-fnsummary.c: Likewise.
21908 * ipa-inline.c: Likewise.
21909 * ipa-visibility.c: Likewise.
21911 * lto-cgraph.c: Likewise.
21912 * omp-expand.c: Likewise.
21913 * omp-general.c: Likewise.
21914 * omp-low.c: Likewise.
21915 * omp-offload.c: Likewise.
21916 * omp-simd-clone.c: Likewise.
21917 * opts-global.c: Likewise.
21918 * passes.c: Likewise.
21919 * predict.c: Likewise.
21920 * sancov.c: Likewise.
21921 * sanopt.c: Likewise.
21922 * symtab.c: Likewise.
21923 * toplev.c: Likewise.
21924 * trans-mem.c: Likewise.
21925 * tree-chkp.c: Likewise.
21926 * tree-eh.c: Likewise.
21927 * tree-into-ssa.c: Likewise.
21928 * tree-object-size.c: Likewise.
21929 * tree-parloops.c: Likewise.
21930 * tree-profile.c: Likewise.
21931 * tree-ssa-ccp.c: Likewise.
21932 * tree-ssa-live.c: Likewise.
21933 * tree-ssa-loop.c: Likewise.
21934 * tree-ssa-sccvn.c: Likewise.
21935 * tree-ssa-structalias.c: Likewise.
21936 * tree-ssa.c: Likewise.
21937 * tree-streamer-in.c: Likewise.
21938 * tree-vectorizer.c: Likewise.
21939 * tree-vrp.c: Likewise.
21940 * tsan.c: Likewise.
21941 * ubsan.c: Likewise.
21942 * varasm.c: Likewise.
21943 * varpool.c: Likewise.
21944 * tree.c: Remove functions moved to attribs.[ch].
21945 * tree.h: Likewise.
21946 * config/aarch64/aarch64.c: Add attrs.h header file.
21947 * config/alpha/alpha.c: Likewise.
21948 * config/arc/arc.c: Likewise.
21949 * config/arm/arm.c: Likewise.
21950 * config/avr/avr.c: Likewise.
21951 * config/bfin/bfin.c: Likewise.
21952 * config/c6x/c6x.c: Likewise.
21953 * config/cr16/cr16.c: Likewise.
21954 * config/cris/cris.c: Likewise.
21955 * config/darwin.c: Likewise.
21956 * config/epiphany/epiphany.c: Likewise.
21957 * config/fr30/fr30.c: Likewise.
21958 * config/frv/frv.c: Likewise.
21959 * config/ft32/ft32.c: Likewise.
21960 * config/h8300/h8300.c: Likewise.
21961 * config/i386/winnt.c: Likewise.
21962 * config/ia64/ia64.c: Likewise.
21963 * config/iq2000/iq2000.c: Likewise.
21964 * config/lm32/lm32.c: Likewise.
21965 * config/m32c/m32c.c: Likewise.
21966 * config/m32r/m32r.c: Likewise.
21967 * config/m68k/m68k.c: Likewise.
21968 * config/mcore/mcore.c: Likewise.
21969 * config/microblaze/microblaze.c: Likewise.
21970 * config/mips/mips.c: Likewise.
21971 * config/mmix/mmix.c: Likewise.
21972 * config/mn10300/mn10300.c: Likewise.
21973 * config/moxie/moxie.c: Likewise.
21974 * config/msp430/msp430.c: Likewise.
21975 * config/nds32/nds32-isr.c: Likewise.
21976 * config/nds32/nds32.c: Likewise.
21977 * config/nios2/nios2.c: Likewise.
21978 * config/nvptx/nvptx.c: Likewise.
21979 * config/pa/pa.c: Likewise.
21980 * config/pdp11/pdp11.c: Likewise.
21981 * config/powerpcspe/powerpcspe.c: Likewise.
21982 * config/riscv/riscv.c: Likewise.
21983 * config/rl78/rl78.c: Likewise.
21984 * config/rx/rx.c: Likewise.
21985 * config/s390/s390.c: Likewise.
21986 * config/sh/sh.c: Likewise.
21987 * config/sol2.c: Likewise.
21988 * config/sparc/sparc.c: Likewise.
21989 * config/spu/spu.c: Likewise.
21990 * config/stormy16/stormy16.c: Likewise.
21991 * config/tilegx/tilegx.c: Likewise.
21992 * config/tilepro/tilepro.c: Likewise.
21993 * config/v850/v850.c: Likewise.
21994 * config/vax/vax.c: Likewise.
21995 * config/visium/visium.c: Likewise.
21996 * config/xtensa/xtensa.c: Likewise.
21998 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22001 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
22002 constraints since the -mupper-regs-* switches have been
22004 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
22005 into a vector from a double word element that was extracted from
22006 another vector, and eliminate extra XXPERMDI instructions.
22007 (vsx_concat_<mode>_2): Likewise.
22008 (vsx_concat_<mode>_3): Likewise.
22009 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
22010 concat to allow optimizing inserts from previous extracts.
22012 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
22014 * config/i386/i386.c (ix86_stack_protect_guard): Generate
22015 memory reference to a SSP offset in TLS address space.
22016 (ix86_print_operand) <case '@'>: Remove.
22017 (ix86_print_operand_punct_valid_p): Remove '@' code.
22018 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
22019 UNSPEC_SP_TLS_TEST.
22020 (stack_tls_protect_set_<mode>): Remove.
22021 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
22022 (stack_tls_protect_test_<mode>): Remove.
22023 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
22025 2017-08-07 Olivier Hainque <hainque@adacore.com>
22028 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
22030 2017-08-07 Douglas Rupp <rupp@adacore.com>
22032 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
22033 variable was referenced as multidir in command.
22035 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22038 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
22041 2017-08-07 Martin Liska <mliska@suse.cz>
22043 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
22044 * config/rl78/rl78.c: Add include of attribs.h.
22045 * config/sh/sh.c: Likewise.
22046 * config/v850/v850.c: Likewise.
22048 2017-08-07 Tom de Vries <tom@codesourcery.com>
22050 PR middle-end/78266
22051 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
22053 2017-08-07 Martin Liska <mliska@suse.cz>
22055 * config/mips/mips.c: Include attribs.h.
22057 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
22060 * doc/invoke.texi: Document change in behvaior for -Ofast for
22063 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
22065 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
22067 (aarch64_pop_regs): Likewise.
22068 (aarch64_gen_load_pair): Likewise.
22069 (aarch64_save_callee_saves): Likewise.
22070 (aarch64_restore_callee_saves): Likewise.
22072 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22074 * config/i386/i386.c: Revert the last change.
22076 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22079 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
22081 (ix86_finalize_stack_frame_flags): This. Also clear
22082 frame_pointer_needed if -fno-omit-frame-pointer is used without
22084 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
22085 with ix86_finalize_stack_frame_flags.
22086 (ix86_expand_epilogue): Likewise.
22087 (ix86_expand_split_stack_prologue): Likewise.
22089 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22092 * config/i386/i386.c (get_builtin_code_for_version): Set priority
22093 to P_AES for Westmere.
22095 2017-08-07 Jonathan Yong <10walls@gmail.com>
22097 * config/i386/mingw.opt (fset-stack-executable): Removed.
22098 * config/i386/cygming.opt (fset-stack-executable): Moved
22100 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
22102 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
22104 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
22106 2017-08-07 Marek Polacek <polacek@redhat.com>
22108 PR middle-end/81737
22109 * fold-const.c (fold_indirect_ref_1): Check type_domain.
22111 2017-08-07 Martin Liska <mliska@suse.cz>
22113 * attribs.h (canonicalize_attr_name): New function.
22114 (cmp_attribs): Move from c-format.c and adjusted.
22115 (is_attribute_p): Moved from tree.h.
22116 * tree-inline.c: Add new includes.
22117 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
22118 (private_is_attribute_p): Remove.
22119 (private_lookup_attribute): Likewise.
22120 (private_lookup_attribute_by_prefix): Simplify.
22121 (remove_attribute): Use is_attribute_p.
22122 * tree.h: Remove removed declarations.
22124 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22126 PR middle-end/81698
22127 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
22128 instead of computing it in the function. Formatting fix.
22129 (expand_case): Don't rely on default_edge being the first edge,
22130 clear it if removing it, pass default_edge to
22131 emit_case_dispatch_table.
22132 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
22135 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
22137 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
22138 insn in the function, emit NOP after the insn.
22140 2017-08-06 Tom de Vries <tom@codesourcery.com>
22142 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
22145 2017-08-06 Tom de Vries <tom@codesourcery.com>
22147 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
22150 2017-08-04 Yury Gribov <tetra2005@gmail.com>
22152 PR tree-optimization/57371
22153 * match.pd: New pattern.
22155 2017-08-04 Marek Polacek <polacek@redhat.com>
22157 PR middle-end/81695
22158 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
22159 perform the computation in offset_int.
22161 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22163 PR tree-optimization/81136
22164 * tree-vectorizer.h: Include tree-hash-traits.h.
22165 (vec_base_alignments): New typedef.
22166 (vec_info): Add a base_alignments field.
22167 (vect_record_base_alignments): Declare.
22168 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
22170 (DR_IS_CONDITIONAL_IN_STMT): New macro.
22171 (create_data_ref): Add an is_conditional_in_stmt argument.
22172 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
22173 the is_conditional_in_stmt field.
22174 (data_ref_loc): Add an is_conditional_in_stmt field.
22175 (get_references_in_stmt): Set the is_conditional_in_stmt field.
22176 (find_data_references_in_stmt): Update call to create_data_ref.
22177 (graphite_find_data_references_in_stmt): Likewise.
22178 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
22179 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
22180 (vect_record_base_alignment): New function.
22181 (vect_record_base_alignments): Likewise.
22182 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
22183 for nested statements even if we fail to compute a misalignment.
22184 Use pooled base alignments for unconditional references.
22185 (vect_find_same_alignment_drs): Compare base addresses instead
22187 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
22188 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
22190 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22192 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
22193 Add an explicit name for the enum. Use auto_vec for slp_instances
22194 and grouped_stores.
22195 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
22197 (_bb_vec_info): Add a constructor and destructor.
22198 (vinfo_for_stmt): Return NULL for uids of -1 as well.
22199 (destroy_loop_vec_info): Delete.
22200 (vect_destroy_datarefs): Likewise.
22201 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
22202 (vec_info::vec_info): New function.
22203 (vec_info::~vec_info): Likewise.
22204 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22205 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22206 destroy_loop_vec_info.
22207 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22208 (_loop_vec_info::_loop_vec_info): ...this.
22209 (destroy_loop_vec_info): Replace with...
22210 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22211 the stmt_vec_infos. Leave handling of vec_info information to its
22212 destructor. Remove explicit vector releases.
22213 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22214 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22215 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22216 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22217 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22218 (destroy_bb_vec_info): Replace with...
22219 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22220 information to its destructor.
22221 (vect_slp_analyze_bb_1): Use new and delete instead of
22222 new_bb_vec_info and destroy_bb_vec_info.
22223 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22226 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22228 * tree-data-ref.h (subscript): Add access_fn field.
22229 (data_dependence_relation): Add could_be_independent_p.
22230 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22231 (same_access_functions): Move to tree-data-ref.c.
22232 * tree-data-ref.c (ref_contains_union_access_p): New function.
22233 (access_fn_component_p): Likewise.
22234 (access_fn_components_comparable_p): Likewise.
22235 (dr_analyze_indices): Add a reference to access_fn_component_p.
22236 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22238 (constant_access_functions): Likewise.
22239 (add_other_self_distances): Likewise.
22240 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22241 (initialize_data_dependence_relation): Use XCNEW and remove
22242 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22243 of access functions that have the same type. Allow the
22244 subsequence to end with different bases in some circumstances.
22245 Record the chosen access functions in SUB_ACCESS_FN.
22246 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22247 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22248 (subscript_dependence_tester_1): Likewise dra and drb.
22249 (build_classic_dist_vector): Update calls accordingly.
22250 (subscript_dependence_tester): Likewise.
22251 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22252 DDR_COULD_BE_INDEPENDENT_P.
22253 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22254 comp_alias_ddrs instead of may_alias_ddrs.
22255 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22257 (vect_analyze_data_ref_dependence): Use it if
22258 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22259 distance vectors if that fails.
22260 (dependence_distance_ge_vf): New function.
22261 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22262 LOOP_VINFO_MAY_ALIAS_DDRS.
22264 2017-08-04 Richard Biener <rguenther@suse.de>
22266 PR middle-end/81705
22267 * fold-const.c (fold_binary_loc): Properly restrict
22268 minus_var0 && minus_var1 case when associating undefined overflow
22271 2017-08-04 Tom de Vries <tom@codesourcery.com>
22273 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22275 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22277 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22278 Don't start diagnostic messages with a capital letter.
22279 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22281 (rs6000_invalid_builtin): Likewise.
22282 (rs6000_trampoline_init): Likewise.
22284 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22287 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22288 after setting changeable df flags.
22290 2017-08-03 Richard Biener <rguenther@suse.de>
22292 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22293 up if the use is in USE - X.
22295 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22297 * toplev.c (dumpfile.h): New include.
22298 (internal_error_reentered): New static function. Use it...
22299 (internal_error_function): ...here to handle reentered internal_error.
22301 2017-08-03 Richard Biener <rguenther@suse.de>
22303 PR middle-end/81148
22304 * fold-const.c (split_tree): Add minus_var and minus_con
22305 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22306 here but always use minus_*.
22307 (associate_trees): Assert we never associate with MINUS_EXPR
22308 and NULL first operand. Do not recurse for PLUS_EXPR operands
22309 when associating as MINUS_EXPR either.
22310 (fold_binary_loc): Track minus_var and minus_con.
22312 2017-08-03 Tom de Vries <tom@codesourcery.com>
22315 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22316 ACCEL_COMPILER, apply finish_options on
22317 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22319 2017-08-03 Tom de Vries <tom@codesourcery.com>
22322 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22323 function_entry_patch_area_size > 0.
22325 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22328 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22329 instead of 10??LU, perform unit multiplication in wide_int,
22330 don't change alloc_object_size_limit if the limit is larger
22333 PR tree-optimization/81655
22334 PR tree-optimization/81588
22335 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22336 the case when ranges[i].low and high are 1 for unsigned type with
22339 PR middle-end/81052
22340 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22341 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22343 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22345 * tree-vrp.h: Add include guard.
22347 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22350 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22351 (ud2): New insn pattern.
22352 * config/i386/i386.c (ix86_expand_epilogue):
22353 For naked functions, generate ud2 instead of trap insn.
22355 2017-08-02 Marek Polacek <polacek@redhat.com>
22358 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22360 2017-08-02 Tom de Vries <tom@codesourcery.com>
22361 Cesar Philippidis <cesar@codesourcery.com>
22363 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22364 Add missing edge probabilities.
22366 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22368 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22369 Correct endianness.
22371 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22373 PR middle-end/79499
22374 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22375 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22376 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22377 prologue_seq sequences - if any.
22379 2017-08-02 Richard Biener <rguenther@suse.de>
22381 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22382 via vectors if supported, integer extracts via punning if supported
22383 or otherwise vector extracts.
22385 2017-08-02 Richard Biener <rguenther@suse.de>
22387 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22389 (bitmap_insert_into_set): ... this.
22391 2017-08-02 Richard Biener <rguenther@suse.de>
22393 PR tree-optimization/81633
22395 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22397 PR tree-optimization/71752
22398 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22400 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22402 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22403 (machine_function::call_ms2sysv_pad_out): Remove field.
22404 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22405 (ix86_compute_frame_layout): Likewise.
22407 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22410 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22411 attribute with interrupt attribute.
22413 2017-08-01 Andrew Pinski <apinski@cavium.com>
22415 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22416 BIT_INSERT_EXPR's operand 1
22417 to see if the types precision matches.
22419 2017-08-01 Martin Liska <mliska@suse.cz>
22421 PR middle-end/70140
22422 * builtins.c (expand_builtin_memcpy_args): Remove.
22423 (expand_builtin_memcpy): Call newly added function
22424 expand_builtin_memory_copy_args.
22425 (expand_builtin_memcpy_with_bounds): Likewise.
22426 (expand_builtin_mempcpy): Remove last argument.
22427 (expand_builtin_mempcpy_with_bounds): Likewise.
22428 (expand_builtin_memory_copy_args): New function created from
22429 expand_builtin_mempcpy_args with small modifications.
22430 (expand_builtin_mempcpy_args): Remove.
22431 (expand_builtin_stpcpy): Remove unused argument.
22432 (expand_builtin): Likewise.
22433 (expand_builtin_with_bounds): Likewise.
22435 2017-08-01 Martin Liska <mliska@suse.cz>
22438 Make mempcpy more optimal (PR middle-end/70140).
22440 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22443 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22444 __builtin_vec_cmpne verify both arguments are compatible vectors
22445 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22446 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22447 move computation of aligned to after checking the argument types.
22451 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22452 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22455 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22456 Doug Rupp <rupp@adacore.com>
22457 Olivier Hainque <hainque@adacore.com>
22459 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22460 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22462 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22463 for TARGET_OS_CPP_BUILTIN.
22464 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22465 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22467 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22468 passing required abi options to the assembler for EABI configurations.
22469 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22470 of .text.hot and .text.unlikely sections for kernel modules when
22471 using ARM style exceptions.
22472 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22473 options. Add EXTRA_CC1_SPEC.
22474 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22476 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22478 (ARM_TARGET2_DWARF_FORMAT): Define.
22479 * config/arm/t-vxworks: Adjust multilib control to removal of the
22480 Diab command line options.
22482 2017-08-01 Martin Liska <mliska@suse.cz>
22484 PR gcov-profile/81561
22485 * gcov.c (unblock): Make unblocking safe as we need to preserve
22486 index correspondence of blocks and block_lists.
22488 2017-08-01 Richard Biener <rguenther@suse.de>
22490 PR tree-optimization/81181
22491 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22492 (compute_antic): ... end of iteration here.
22494 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22496 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22497 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22498 (ftree-slp-vectorize): Likewise.
22499 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22500 can no longer be set independent of flag_tree_loop_vectorize.
22501 * omp-general.c (emp_max_vf): Likewise.
22502 * opts.c (enable_fdo_optimizations): Remove references to
22503 flag_tree_vectorize, these are now implicit.
22504 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22505 and leave it for the options machinery.
22507 2017-08-01 Martin Liska <mliska@suse.cz>
22509 PR middle-end/70140
22510 * builtins.c (expand_builtin_memcpy_args): Remove.
22511 (expand_builtin_memcpy): Call newly added function
22512 expand_builtin_memory_copy_args.
22513 (expand_builtin_memcpy_with_bounds): Likewise.
22514 (expand_builtin_mempcpy): Remove last argument.
22515 (expand_builtin_mempcpy_with_bounds): Likewise.
22516 (expand_builtin_memory_copy_args): New function created from
22517 expand_builtin_mempcpy_args with small modifications.
22518 (expand_builtin_mempcpy_args): Remove.
22519 (expand_builtin_stpcpy): Remove unused argument.
22520 (expand_builtin): Likewise.
22521 (expand_builtin_with_bounds): Likewise.
22523 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22526 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22527 print "ds:" only for immediates in generic address space.
22529 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22532 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22533 (ix86_function_ok_for_sibcall): Return false for naked functions.
22535 2017-08-01 Richard Biener <rguenther@suse.de>
22537 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22538 (compute_antic): Seed worklist with exit block predecessors.
22539 * cfganal.c (dfs_find_deadend): For a cycle return the source
22540 of the edge closing it.
22542 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22544 * config/aarch64/aarch64.c
22545 (aarch64_can_const_movi_rtx_p): Move 0 check.
22547 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22549 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22550 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22552 * match.pd: Ditto in address comparison pattern.
22554 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22556 PR tree-optimization/81627
22557 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22558 closed ssa form for store-store chain.
22560 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22562 PR tree-optimization/81620
22563 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22564 for store-store chain.
22566 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22568 PR tree-optimization/81588
22569 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22570 ranges[i].in_p, invert comparison code ccode. For >/>=,
22571 swap rhs1 and rhs2 and comparison code unconditionally,
22572 for </<= don't do that. Don't swap rhs1/rhs2 again if
22573 ranges[i].in_p, instead invert comparison code ccode if
22574 opcode or oe->rank is BIT_IOR_EXPR.
22577 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22578 a direct optab to conversion optab.
22579 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22580 with GET_MODE_INNER as last argument instead of optab_handler.
22581 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22582 vector extraction if possible and optab is available.
22583 * expr.c (store_constructor): Use convert_optab_handler instead
22584 of optab_handler. Use vector initialization from smaller
22585 vectors if possible and optab is available.
22586 * tree-vect-stmts.c (vectorizable_load): Likewise.
22587 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22588 now have two modes.
22589 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22590 of vec_init from half-sized vectors with the same element mode.
22591 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22592 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22593 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22594 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22595 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22596 after mode in gen_vec_extract* calls.
22597 (vec_extract<mode>): Renamed to ...
22598 (vec_extract<mode><ssescalarmodelower>): ... this.
22599 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22600 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22601 element mode after mode in gen_vec_init* calls.
22602 (VEC_INIT_HALF_MODE): New mode iterator.
22603 (vec_init<mode>): Renamed to ...
22604 (vec_init<mode><ssescalarmodelower>): ... this.
22605 (vec_init<mode><ssehalfvecmodelower>): New expander.
22606 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22607 (vec_extractv2sfsf): ... this.
22608 (vec_initv2sf): Renamed to ...
22609 (vec_initv2sfsf): ... this.
22610 (vec_extractv2si): Renamed to ...
22611 (vec_extractv2sisi): ... this.
22612 (vec_initv2si): Renamed to ...
22613 (vec_initv2sisi): ... this.
22614 (vec_extractv4hi): Renamed to ...
22615 (vec_extractv4hihi): ... this.
22616 (vec_initv4hi): Renamed to ...
22617 (vec_initv4hihi): ... this.
22618 (vec_extractv8qi): Renamed to ...
22619 (vec_extractv8qiqi): ... this.
22620 (vec_initv8qi): Renamed to ...
22621 (vec_initv8qiqi): ... this.
22622 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22623 (vec_init<mode>): Renamed to ...
22624 (vec_init<mode><VEC_base_l>): ... this.
22625 (vec_extract<mode>): Renamed to ...
22626 (vec_extract<mode><VEC_base_l>): ... this.
22627 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22628 (vec_initv2sfsf): ... this.
22629 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22630 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22631 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22632 element mode after mode in gen_vec_init* calls.
22633 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22634 (vec_init<mode><Vel>): ... this.
22635 (vec_extract<mode>): Renamed to ...
22636 (vec_extract<mode><Vel>): ... this.
22637 * config/aarch64/iterators.md (Vel): New mode attribute.
22638 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22639 Add element mode after mode in gen_vec_extract* calls.
22640 * config/s390/vector.md (non_vec_l): New mode attribute.
22641 (vec_extract<mode>): Renamed to ...
22642 (vec_extract<mode><non_vec_l>): ... this.
22643 (vec_init<mode>): Renamed to ...
22644 (vec_init<mode><non_vec_l>): ... this.
22645 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22646 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22648 * config/arm/iterators.md (V_elem_l): New mode attribute.
22649 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22650 (vec_extract<mode><V_elem_l>): ... this.
22651 (vec_extractv2di): Renamed to ...
22652 (vec_extractv2didi): ... this.
22653 (vec_init<mode>): Renamed to ...
22654 (vec_init<mode><V_elem_l>): ... this.
22655 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22656 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22657 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22658 Add element mode after gen_vec_extract* calls.
22659 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22660 (vec_init<mode><unitmode>): ... this.
22661 (vec_extract<mode>): Renamed to ...
22662 (vec_extract<mode><unitmode>): ... this.
22663 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22664 (vec_init<mode><unitmode>): ... this.
22665 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22666 (vec_initv2sfsf): ... this.
22667 (vec_extractv2sf): Renamed to ...
22668 (vec_extractv2sfsf): ... this.
22669 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22670 Add element mode after gen_vec_extract* calls.
22671 * config/mips/mips.md (unitmode): New mode iterator.
22672 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22673 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22674 * config/spu/spu.md (inner_l): New mode attribute.
22675 (vec_init<mode>): Renamed to ...
22676 (vec_init<mode><inner_l>): ... this.
22677 (vec_extract<mode>): Renamed to ...
22678 (vec_extract<mode><inner_l>): ... this.
22679 * config/sparc/sparc.md (veltmode): New mode iterator.
22680 (vec_init<VMALL:mode>): Renamed to ...
22681 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22682 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22683 (vec_initv2sisi): ... this.
22684 (vec_initv2sf): Renamed to ...
22685 (vec_initv2sfsf): ... this.
22686 (vec_extractv2sf): Renamed to ...
22687 (vec_extractv2sfsf): ... this.
22688 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22689 (vec_init<mode>): Renamed to ...
22690 (vec_init<mode><VEC_base_l>): ... this.
22691 (vec_extract<mode>): Renamed to ...
22692 (vec_extract<mode><VEC_base_l>): ... this.
22693 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22694 (vec_initv2sfsf): ... this.
22695 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22696 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22697 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22698 gen_vec_init* calls.
22700 2017-08-01 Richard Biener <rguenther@suse.de>
22702 PR tree-optimization/81297
22703 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22704 TREE_OVERFLOW from INTEGER_CSTs.
22706 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22708 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22710 2017-07-31 Carl Love <cel@us.ibm.com>
22712 * config/rs6000/rs6000-c: Add support for built-in functions
22713 vector signed char vec_xl_be (signed long long, signed char *);
22714 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22715 vector signed int vec_xl_be (signed long long, signed int *);
22716 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22717 vector signed long long vec_xl_be (signed long long, signed long long *);
22718 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22719 vector signed short vec_xl_be (signed long long, signed short *);
22720 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22721 vector double vec_xl_be (signed long long, double *);
22722 vector float vec_xl_be (signed long long, float *);
22723 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22724 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22725 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22727 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22728 (altivec_expand_builtin): Add switch statement to call
22729 altivec_expand_xl_be for each builtin.
22730 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22731 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22732 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22733 __builtin_vsx_le_be_v16qi.
22734 * doc/extend.texi: Update the built-in documentation file for the
22735 new built-in functions.
22737 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22740 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22742 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22744 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22746 * config.gcc: Add z14.
22747 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22748 CPU model numbers for z13s and z14.
22749 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22751 * config/s390/s390-opts.h (enum processor_type): Rename
22752 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22753 * config/s390/s390.c (processor_table): Add field for CPU name to
22754 be passed to Binutils.
22755 (s390_asm_output_machine_for_arch): Use the new field in
22756 processor_table for Binutils.
22757 (s390_expand_builtin): Replace arch12 with z14.
22758 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22759 (s390_get_sched_attrmask): Likewise.
22760 (s390_get_unit_mask): Likewise.
22761 * config/s390/s390.opt: Add z14 to processor_type enum.
22763 2017-07-31 Martin Jambor <mjambor@suse.cz>
22766 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22767 regardless of optimization level.
22769 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22770 Martin Liska <mliska@suse.cz>
22772 * predict.def: Remove old comment and adjust probability.
22773 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22774 PREDICT statements.
22776 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22779 * config/i386/i386.c (ix86_function_naked): New function.
22780 (ix86_can_use_return_insn_p): Return false for naked functions.
22781 (ix86_expand_prologue): Skip prologue for naked functions.
22782 (ix86_expand_epilogue): Skip epilogue for naked functions
22783 and emit trap instruction.
22784 (ix86_warn_func_return): New function.
22785 (ix86_attribute_table): Add "naked" attribute specification.
22786 (TARGET_WARN_FUNC_RETURN): Define.
22787 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22789 2017-07-31 Martin Liska <mliska@suse.cz>
22791 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22792 (dump_gimple_bb_header): Always dump BB info.
22793 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22795 2017-07-31 Martin Liska <mliska@suse.cz>
22798 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22799 also with current_function_decl non-null equality.
22801 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22804 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22805 change type to the element type, instead add eltype variable and
22806 use it where we are interested in the element type.
22808 PR tree-optimization/81603
22809 * ipa-polymorphic-call.c
22810 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22811 offset arithmetic in offset_int, bail out if the resulting bit offset
22812 doesn't fit into shwi.
22814 2017-07-31 Martin Liska <mliska@suse.cz>
22816 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22817 (gimplify_save_expr): Fix comment.
22819 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22822 * config/i386/i386.c (ix86_function_arg): Update arguments for
22824 (ix86_compute_frame_layout): Set the initial stack offset to
22825 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22826 INCOMING_FRAME_SP_OFFSET.
22827 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22828 stack before exception handler returns.
22829 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22830 the 'ERROR_CODE' for exception handler.
22832 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22834 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22835 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22836 (ASM_OUTPUT_REG_POP): Ditto.
22837 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22838 instead of asm_fprintf to output pure string.
22840 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22842 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22843 to imported_module_or_decl hook.
22844 (debug_nothing_tree_tree_tree_bool): Remove.
22845 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22846 * debug.c (do_nothing_debug_hooks): Use
22847 debug_nothing_tree_tree_tree_bool_bool instead of
22848 debug_nothing_tree_tree_tree_bool.
22849 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22850 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22851 * sdbout.c (sdb_debug_hooks): Likewise.
22852 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22853 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22855 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22856 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22857 attribute, don't add anything.
22859 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22861 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22862 (fold_build2_stat_loc): Likewise.
22863 (fold_build3_stat_loc): Likewise.
22864 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22865 (fold_build1_loc): Remove macro.
22866 (fold_build2_loc): Likewise.
22867 (fold_build3_loc): Likewise.
22869 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22871 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22872 (gimple_build_debug_bind_source_stat): Likewise.
22873 * gimple.h (gimple_build_debug_bind): Remove macro.
22874 (gimple_build_debug_bind_source): Likewise.
22876 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22878 * bitmap.c (bitmap_alloc): Adjust.
22879 (bitmap_gc_alloc): Likewise.
22880 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22882 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22884 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22885 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22886 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22887 (bitmap_gc_alloc_stat): Likewise.
22888 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22890 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22892 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22893 * rtl.h (shallow_copy_rtx): Remove macro.
22895 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22897 * emit-rtl.c (gen_raw_REG): Adjust.
22898 * gengenrtl.c (gendef): Likewise.
22899 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22900 * rtl.h (rtx_alloc): Remove macro.
22902 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22904 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22905 (build_tree_list_stat): Likewise.
22906 * tree.h (build_tree_list): Remove macro.
22907 (build_tree_list_vec): Likewise.
22909 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22911 * tree.c (make_vector_stat): Remove _stat from name.
22912 (build_vector_stat): Likewise.
22913 * tree.h (make_vector_stat): Remove macro.
22914 (build_vector_stat): Likewise.
22916 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22918 * tree.h (build_var_debug_value): Remove prototype.
22920 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22922 * tree.c (tree_cons_stat): Remove _stat from name.
22923 * tree.h (tree_cons): Remove macro.
22925 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22927 * tree.c (build_vl_exp_stat): Remove _stat from name.
22928 * tree.h (build_vl_exp): Remove macro.
22930 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22932 * tree.c (build_decl_stat): Remove _stat from name.
22933 * tree.h (build_decl): Remove macro.
22935 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22937 * gimple.c (gimple_build_with_ops_stat): Adjust.
22938 (gimple_alloc_stat): Remove _stat from name.
22939 * gimple.h (gimple_alloc): Remove macro.
22941 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22943 * tree.c (make_tree_vec_stat): Remove _stat from name.
22944 (grow_tree_vec_stat): Likewise.
22945 * tree.h (make_tree_vec_stat): Adjust prototype.
22946 (grow_tree_vec_stat): Likewise.
22947 (make_tree_vec): Remove macro.
22948 (grow_tree_vec): Likewise.
22950 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22952 * fold-const.c (fold_build1_stat_loc): Adjust.
22953 (fold_build2_stat_loc): Likewise.
22954 (fold_build3_stat_loc): Likewise.
22955 * tree.c (build0_stat): Remove _stat from name.
22956 (build1_stat): Likewise.
22957 (build2_stat): Likewise.
22958 (build3_stat): Likewise.
22959 (build4_stat): Likewise.
22960 (build5_stat): Likewise.
22961 * tree.h (build1_loc): Remove macro, and rename _stat function
22963 (build2_loc): Likewise.
22964 (build3_loc): Likewise.
22965 (build4_loc): Likewise.
22966 (build5_loc): Likewise.
22968 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22970 * tree.c (make_int_cst_stat): Remove _stat from name.
22971 * tree.h (make_int_cst_stat): Adjust prototype.
22972 (make_int_cst): Remove macro.
22974 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22976 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22977 * tree.h (make_tree_binfo_stat): Adjust prototype.
22978 (make_tree_binfo): Remove.
22980 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22982 * tree.c (copy_node_stat): Rename to copy_node.
22983 (build_distinct_type_copy): Adjust.
22984 * tree.h (copy_node_stat): Adjust prototype.
22985 (copy_node): Remove macro.
22987 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22989 * tree.c (make_node_stat): rename to make_node.
22990 (build_tree_list_stat): Adjust.
22991 (build0_stat): Likewise.
22992 (build2_stat): Likewise.
22993 (build3_stat): Likewise.
22994 (build4_stat): Likewise.
22995 (build5_stat): Likewise.
22996 (build_decl_stat): Likewise.
22997 * tree.h (make_node_stat): Adjust prototype.
22998 (make_node): remove macro.
23000 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
23002 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
23003 (PPC_FEATURE2_SCV): Likewise.
23004 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
23006 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23008 * config/aarch64/aarch64.c
23009 (aarch64_internal_mov_immediate): Add new special pattern.
23010 * config/aarch64/aarch64.md (*movdi_aarch64):
23011 Add reg/32bit const mov case.
23013 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23014 Richard Sandiford <richard.sandiford@linaro.org>
23016 * config/aarch64/aarch64.md (mov<mode>): Generalize.
23017 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
23018 Add integer and movi cases.
23019 (movi-split-hf-df-sf split, fp16): New.
23020 (enabled): Added TARGET_FP_F16INST.
23021 * config/aarch64/iterators.md (GPF_HF): New.
23022 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
23024 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23026 * config/aarch64/aarch64.c
23027 (aarch64_simd_container_mode): Add prototype.
23028 (aarch64_expand_mov_immediate): Add HI support.
23029 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
23030 (aarch64_can_const_movi_rtx_p): New.
23031 (aarch64_preferred_reload_class):
23032 Remove restrictions of using FP registers for certain SIMD operations.
23033 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
23034 (aarch64_valid_floating_const): Add integer move validation.
23035 (aarch64_simd_imm_scalar_p): Remove.
23036 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
23037 (aarch64_legitimate_constant_p): Expand list of supported cases.
23038 * config/aarch64/aarch64-protos.h
23039 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
23040 (aarch64_reinterpret_float_as_int): New.
23041 (aarch64_simd_imm_scalar_p): Remove.
23042 * config/aarch64/constraints.md (Uvi): New.
23043 (Dd): Split into Ds and new Dd.
23044 * config/aarch64/aarch64.md (*movsi_aarch64):
23046 (*movdi_aarch64): Add SIMD mov case.
23048 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23050 * tree-predcom.c: (struct chain): Handle store-store chain in which
23051 stores for elimination only store loop invariant values.
23052 (execute_pred_commoning_chain): Ditto.
23053 (prepare_initializers_chain_store_elim): Ditto.
23054 (prepare_finalizers): Ditto.
23055 (is_inv_store_elimination_chain): New function.
23056 (initialize_root_vars_store_elim_1): New function.
23058 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23060 * tree-predcom.c: Revise general description of the pass.
23061 (enum chain_type): New enum type for store elimination.
23062 (struct chain): New field supporting store elimination.
23063 (struct component): Ditto.
23064 (dump_chain): Dump store-stores chain.
23065 (release_chain): Release resources.
23066 (split_data_refs_to_components): Compute and create component
23067 contains only stores for elimination.
23068 (get_chain_last_ref_at): New function.
23069 (make_invariant_chain): Initialization.
23070 (make_rooted_chain): Specify chain type in parameter and record it.
23071 (add_looparound_copies): Skip for store-stores chain.
23072 (determine_roots_comp): Compute type of chain and pass it to
23074 (initialize_root_vars_store_elim_2): New function.
23075 (finalize_eliminated_stores): New function.
23076 (remove_stmt): Handle store for elimination.
23077 (execute_pred_commoning_chain): Execute predictive commoning on
23078 store-store chains.
23079 (determine_unroll_factor): Skip unroll for store-stores chain.
23080 (prepare_initializers_chain_store_elim): New function.
23081 (prepare_initializers_chain): Hanlde store-store chain.
23082 (prepare_finalizers_chain, prepare_finalizers): New function.
23083 (tree_predictive_commoning_loop): Return integer value indicating
23084 if loop is unrolled or lcssa form is corrupted.
23085 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
23087 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23089 * tree-predcom.c (initialize_root): Delete.
23090 (execute_pred_commoning_chain): Initialize root vars and replace
23091 reference of non-combined chain directly, rather than call above
23094 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23096 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
23097 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
23099 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23101 * tree-predcom.c (struct chain): New field init_seq.
23102 (release_chain): Release init_seq.
23103 (prepare_initializers_chain): Record intialization stmts in above
23105 (insert_init_seqs): New function.
23106 (tree_predictive_commoning_loop): Call insert_init_seqs.
23108 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23110 * tree-predcom.c (determine_roots_comp): Skip trivial components.
23112 2017-07-28 Richard Biener <rguenther@suse.de>
23114 * match.pd: Remove superfluous :c.
23115 * genmatch.c (simplify::id): Add member.
23116 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
23118 (current_id): New global.
23119 (dt_node::parent): Move from ...
23120 (dt_operand::parent): ... here. Add for_id member.
23121 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
23122 (decision_tree::find_node): Relax order requirement when
23123 merging DT_TRUE nodes to ones inbetween the current simplify
23124 and the one we try to merge with. Add diagnostic whenever
23125 we need to enforce pattern order by not merging.
23126 (decision_tree::insert): Set current_id.
23127 (decision_tree::print_node): Dump parent node and for_id.
23128 (parser::last_id): Add member.
23129 (parser::push_simplify): Assign unique id.
23130 (parser::parser): Initialize last_id.
23132 2017-07-28 Martin Liska <mliska@suse.cz>
23135 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
23136 gimple_build_debug_bind.
23138 2017-07-28 Richard Biener <rguenther@suse.de>
23140 PR tree-optimization/81502
23141 * match.pd: Add pattern combining BIT_INSERT_EXPR with
23143 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
23145 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
23146 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
23147 for BIT_FIELD_REF args.
23148 * fold-const.c (make_bit_field_ref): Likewise.
23149 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
23151 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23154 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
23155 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
23156 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
23157 Or it into SANITIZER_UNDEFINED.
23158 * ubsan.c: Include gimple-fold.h and varasm.h.
23159 (ubsan_expand_ptr_ifn): New function.
23160 (instrument_pointer_overflow): New function.
23161 (maybe_instrument_pointer_overflow): New function.
23162 (instrument_object_size): Formatting fix.
23163 (pass_ubsan::execute): Call instrument_pointer_overflow
23164 and maybe_instrument_pointer_overflow.
23165 * internal-fn.c (expand_UBSAN_PTR): New function.
23166 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
23167 * sanitizer.def (__ubsan_handle_pointer_overflow,
23168 __ubsan_handle_pointer_overflow_abort): New builtins.
23169 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
23170 * internal-fn.def (UBSAN_PTR): New internal function.
23171 * opts.c (sanitizer_opts): Add pointer-overflow.
23172 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
23173 * fold-const.c (build_range_check): Compute pointer range check in
23174 integral type if pointer arithmetics would be needed. Formatting
23177 2017-07-28 Martin Liska <mliska@suse.cz>
23180 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
23181 parameters that are of a variable-length.
23183 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
23185 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
23187 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
23188 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
23189 (CRT_CALL_STATIC_FUNCTION): Likewise.
23190 (ASM_DEFAULT_SPEC): New define.
23191 (ASM_SPEC32): Likewise.
23192 (ASM_SPEC64): Likewise.
23193 (ASM_SPEC_COMMON): Likewise.
23194 (ASM_SPEC): Likewise.
23195 (INVALID_64BIT): Likewise.
23196 (LINK_OS_DEFAULT_SPEC): Likewise.
23197 (LINK_OS_SPEC32): Likewise.
23198 (LINK_OS_SPEC64): Likewise.
23199 (POWERPC_LINUX): Likewise.
23200 (PTRDIFF_TYPE): Likewise.
23201 (RESTORE_FP_PREFIX): Likewise.
23202 (RESTORE_FP_SUFFIX): Likewise.
23203 (SAVE_FP_PREFIX): Likewise.
23204 (SAVE_FP_SUFFIX): Likewise.
23205 (SIZE_TYPE): Likewise.
23206 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23207 (TARGET_64BIT): Likewise.
23208 (TARGET_64BIT): Likewise.
23209 (TARGET_AIX): Likewise.
23210 (WCHAR_TYPE_SIZE): Likewise.
23211 (WCHAR_TYPE): Undefine.
23212 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23213 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23214 (CPP_OS_RTEMS_SPEC): Delete.
23215 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23216 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23218 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23220 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23222 PR tree-optimization/81578
23223 * tree-parloops.c (build_new_reduction): Bail out if
23224 reduction_code isn't one of the standard OpenMP reductions.
23225 Move the details printing after that decision.
23227 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23229 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23230 related to reload_in_progress.
23231 (splat_input_operand): Likewise.
23232 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23234 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23236 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23237 (TARGET_INSTANTIATE_DECLS): Likewise.
23238 (legitimate_indexed_address_p): Delete reload_in_progress code.
23239 (rs6000_debug_legitimate_address_p): Likewise.
23240 (rs6000_eliminate_indexed_memrefs): Likewise.
23241 (rs6000_emit_le_vsx_store): Likewise.
23242 (rs6000_emit_move_si_sf_subreg): Likewise.
23243 (rs6000_emit_move): Likewise.
23244 (register_to_reg_type): Likewise.
23245 (rs6000_pre_atomic_barrier): Likewise.
23246 (rs6000_machopic_legitimize_pic_address): Likewise.
23247 (rs6000_allocate_stack_temp): Likewise.
23248 (rs6000_address_for_fpconvert): Likewise.
23249 (rs6000_address_for_altivec): Likewise.
23250 (rs6000_secondary_memory_needed_rtx): Delete function.
23251 (rs6000_check_sdmode): Likewise.
23252 (rs6000_alloc_sdmode_stack_slot): Likewise.
23253 (rs6000_instantiate_decls): Likewise.
23254 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23255 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23256 Delete reload_in_progress.
23257 (*vec_reload_and_plus_<mptrsize>): Likewise.
23258 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23259 (vsx_div_v2di): Likewise.
23260 (vsx_udiv_v2di): Likewise.
23262 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23264 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23265 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23266 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23267 (rs6000_debug_reg_global): Delete print of LRA status.
23268 (rs6000_option_override_internal): Delete dead LRA related code.
23269 (rs6000_lra_p): Delete function.
23270 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23272 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23274 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23275 * config/riscv/rtems.h: New file.
23277 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23278 Sudakshina Das <sudi.das@arm.com>
23280 * config/aarch64/aarch64.md
23281 (define_split for and<mode>3nr_compare): Move
23282 non aarch64_logical_operand to a register.
23283 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23284 register immediate operand to a register.
23285 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23287 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23289 PR middle-end/81564
23290 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23292 2017-07-27 Richard Biener <rguenther@suse.de>
23294 PR tree-optimization/81573
23295 PR tree-optimization/81494
23296 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23297 multi defuse cycle case.
23299 2017-07-27 Richard Biener <rguenther@suse.de>
23301 PR tree-optimization/81571
23302 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23305 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23307 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23308 earlier and only if MASK_FPU is set. Adjust formatting.
23310 2017-07-27 Martin Liska <mliska@suse.cz>
23312 * opt-functions.awk: Add validation of value of Init.
23313 * optc-gen.awk: Pass new argument.
23315 2017-07-27 Martin Liska <mliska@suse.cz>
23317 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23318 Fix wrong condition.
23320 2017-07-27 Martin Liska <mliska@suse.cz>
23322 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23323 BBs and edges seen by autoFDO.
23325 2017-07-27 Richard Biener <rguenther@suse.de>
23327 PR tree-optimization/81502
23328 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23329 with incompatible but same sized type.
23330 (execute_update_addresses_taken): Likewise.
23332 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23334 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23335 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23337 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23340 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23341 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23342 Change s_operand to memory_operand.
23344 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23346 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23347 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23348 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23349 Emit instructions rather than returning an expression. Handle TFmode
23350 and KFmode by casting to TImode.
23351 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23352 (rs6000_emit_le_vsx_store): Likewise.
23353 * config/rs6000/vsx.md (VSX_TI): New iterator.
23354 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23355 (*vsx_le_undo_permute_<mode>): Likewise.
23356 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23357 emit the split sequence.
23358 (*vsx_le_perm_store_<mode>): Likewise.
23360 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23362 PR tree-optimization/81555
23363 PR tree-optimization/81556
23364 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23365 if true, force CHANGED for the recursive invocation.
23366 (reassociate_bb): Remember original length of ops array, pass
23367 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23369 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23370 attributes for noipa attribute. For naked attribute use
23371 lookup_attribute first before lookup_attribute_spec.
23372 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23374 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23375 for functions with noipa attribute.
23376 (cgraph_externally_visible_p): Return true for functions with noipa
23378 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23379 for functions with noipa attribute.
23380 * doc/extend.texi: Document noipa function attribute.
23381 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23382 also for functions with noipa attribute.
23383 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23385 2017-07-26 Andrew Pinski <apinski@cavium.com>
23387 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23388 vec_unalign_load_cost and vec_unalign_store_cost.
23390 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23392 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23393 -mvsx-small-integer option.
23394 (ISA_3_0_MASKS_IEEE): Likewise.
23395 (OTHER_VSX_VECTOR_MASKS): Likewise.
23396 (POWERPC_MASKS): Likewise.
23397 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23398 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23399 code, only testing for DImode being allowed in non-VSX floating
23401 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23402 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23404 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23405 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23406 TARGET_P8_VECTOR test.
23407 (rs6000_secondary_reload_simple_move): Likewise.
23408 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23409 since TARGET_P9_VECTOR was already tested.
23410 (rs6000_opt_masks): Remove -mvsx-small-integer.
23411 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23412 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23414 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23415 test for TARGET_VEXTRACTUB was used, and that uses
23417 (p9 extract splitter): Likewise.
23418 (vsx_extract_<mode>_di_p9): Likewise.
23419 (vsx_extract_<mode>_store_p9): Likewise.
23420 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23421 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23422 the elimination of TARGET_VSX_SMALL_INTEGER.
23423 (vsx_extract_<mode>_p8): Likewise.
23424 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23425 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23426 (vsx_set_<mode>_p9): Likewise.
23427 (vsx_set_v4sf_p9): Likewise.
23428 (vsx_set_v4sf_p9_zero): Likewise.
23429 (vsx_insert_extract_v4sf_p9): Likewise.
23430 (vsx_insert_extract_v4sf_p9_2): Likewise.
23431 * config/rs6000/rs6000.md (sign extend splitter): Change
23432 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23433 (floatsi<mode>2_lfiwax_mem): Likewise.
23434 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23435 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23436 since a test for TARGET_P9_VECTOR was used.
23437 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23438 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23439 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23440 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23441 TARGET_P8_VECTOR test.
23442 (fix_trunc<mode>si2_stfiwx): Likewise.
23443 (fix_trunc<mode>si2_internal): Likewise.
23444 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23445 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23447 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23448 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23449 TARGET_P8_VECTOR test.
23450 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23451 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23452 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23454 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23455 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23456 since a test for TARGET_P9_VECTOR was used.
23457 (splitter for loading small constants): Likewise.
23459 2017-07-26 Andrew Pinski <apinski@cavium.com>
23461 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23464 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23467 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23468 (fp_valid_at): Likewise.
23470 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23472 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23473 (qdf24xx_addrcost_table): Likewise.
23474 (cortexa57_tunings): Update to use generic_branch_cost.
23475 (cortexa72_tunings): Likewise.
23476 (cortexa73_tunings): Likewise.
23477 (qdf24xx_tunings): Likewise.
23479 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23481 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23482 (thunderx2t99_branch_cost): Likewise.
23483 (cortexa35_tunings): Update to use generic_branch_cost.
23484 (cortexa53_tunings): Likewise.
23485 (cortexa57_tunings): Likewise.
23486 (cortexa72_tunings): Likewise.
23487 (cortexa73_tunings): Likewise.
23488 (thunderx2t99_tunings): Likewise.
23490 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23492 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23493 (sparc_option_override): Honour MASK_FSMULD.
23494 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23495 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23496 * config/sparc/sparc.opt (mfsmuld): New option.
23497 * doc/invoke.texi (mfsmuld): Document option.
23499 2017-07-26 Marek Polacek <polacek@redhat.com>
23501 PR middle-end/70992
23502 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23504 2017-07-26 Richard Biener <rguenther@suse.de>
23506 * gimple-match-head.c (do_valueize): Return OP if valueize
23508 (get_def): New helper to get at the def stmt of a SSA name
23509 if valueize allows.
23510 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23511 do_valueize to get at the def stmt.
23512 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23514 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23516 PR middle-end/46932
23517 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23519 2017-07-26 Martin Liska <mliska@suse.cz>
23522 * function.c (expand_function_start): Make expansion of
23523 nonlocal_goto_save_area after parm_birth_insn.
23525 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23527 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23528 from all CPU target flags enable members.
23530 2017-07-26 Richard Biener <rguenther@suse.de>
23532 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23533 (decision_tree::gen): Make 'type' const.
23534 (write_predicate): Likewise.
23536 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23538 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23539 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23540 (rs6000_option_override_internal): Likewise.
23541 (rs6000_expand_vector_set): Likewise.
23542 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23543 (TARGET_UPPER_REGS_SF): Likewise.
23544 (TARGET_UPPER_REGS_DI): Likewise.
23545 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23546 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23547 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23548 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23549 (Splitters for DI constants in Altivec registers): Likewise.
23550 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23551 (vsx_set_v4sf_p9): Likewise.
23552 (vsx_set_v4sf_p9_zero): Likewise.
23553 (vsx_insert_extract_v4sf_p9): Likewise.
23554 (vsx_insert_extract_v4sf_p9_2): Likewise.
23556 2017-07-25 Carl Love <cel@us.ibm.com>
23558 * doc/extend.texi: Update the built-in documentation file for the
23559 existing built-in functions
23560 vector signed char vec_cnttz (vector signed char);
23561 vector unsigned char vec_cnttz (vector unsigned char);
23562 vector signed short vec_cnttz (vector signed short);
23563 vector unsigned short vec_cnttz (vector unsigned short);
23564 vector signed int vec_cnttz (vector signed int);
23565 vector unsigned int vec_cnttz (vector unsigned int);
23566 vector signed long long vec_cnttz (vector signed long long);
23567 vector unsigned long long vec_cnttz (vector unsigned long long);
23569 2017-07-25 Andrew Pinski <apinski@cavium.com>
23571 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23572 accesses where the use is for the first operand of a BIT_INSERT.
23574 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23577 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23578 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23580 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23582 * config/i386/gstabs.h: Delete.
23583 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23585 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23587 * config/i386/i386.c (ix86_decompose_address): Do not check for
23588 register RTX when looking at index_reg or base_reg.
23589 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23591 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23593 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23594 to update EH info here.
23596 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23598 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23600 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23602 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23604 2017-07-25 Torsten Duwe <duwe@suse.de>
23606 * common.opt: Introduce -fpatchable-function-entry
23607 command line option, and its variables function_entry_patch_area_size
23608 and function_entry_patch_area_start.
23609 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23610 including a two-value parser.
23611 * target.def (print_patchable_function_entry): New target hook.
23612 * targhooks.h (default_print_patchable_function_entry): New function.
23613 * targhooks.c (default_print_patchable_function_entry): Likewise.
23614 * toplev.c (process_options): Switch off IPA-RA if
23615 patchable function entries are being generated.
23616 * varasm.c (assemble_start_function): Look at the
23617 patchable-function-entry command line switch and current
23618 function attributes and maybe generate NOP instructions by
23619 calling the print_patchable_function_entry hook.
23620 * doc/extend.texi: Document patchable_function_entry attribute.
23621 * doc/invoke.texi: Document -fpatchable_function_entry
23622 command line option.
23623 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23625 * doc/tm.texi: Re-generate.
23627 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23630 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23631 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23633 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23635 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23637 2017-07-25 Richard Biener <rguenther@suse.de>
23639 PR tree-optimization/81455
23640 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23641 not walk in cycles when looking for guards.
23643 2017-07-25 Richard Biener <rguenther@suse.de>
23645 PR tree-optimization/81529
23646 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23647 when optimizing backedge uses.
23649 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23651 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23653 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23654 to dl_section_ref. On AIX, append an expression to subtract
23655 the size of the section length to dl_section_ref.
23657 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23659 * configure.ac: If any of the config.* scripts fail, exit 1.
23660 * configure: Regenerate.
23662 2017-07-25 Richard Biener <rguenther@suse.de>
23664 PR middle-end/81546
23665 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23666 of immediate uses, be more verbose on errors.
23668 2017-07-25 Richard Biener <rguenther@suse.de>
23670 PR tree-optimization/81510
23671 * tree-vect-loop.c (vect_is_simple_reduction): When the
23672 reduction stmt is not inside the loop bail out.
23674 2017-07-25 Richard Biener <rguenther@suse.de>
23676 PR tree-optimization/81303
23677 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23678 profitability check against LOOP_VINFO_NITERSM1.
23680 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23682 * domwalk.c (cmp_bb_postorder): Simplify.
23683 (sort_bbs_postorder): New function. Use it...
23684 (dom_walker::walk): ...here to optimize common cases.
23686 2017-07-25 Martin Liska <mliska@suse.cz>
23689 * ipa-visibility.c (function_and_variable_visibility): Make the
23690 redirection just on target that supports aliasing.
23691 Fix GNU coding style.
23693 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23696 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23698 * config/arm/rtems.h: Likewise.
23699 * config/bfin/rtems.h: Likewise.
23700 * config/i386/rtemself.h: Likewise.
23701 * config/lm32/rtems.h: Likewise.
23702 * config/m32c/rtems.h: Likewise.
23703 * config/m68k/rtemself.h: Likewise.
23704 * config/microblaze/rtems.h: Likewise.
23705 * config/mips/rtems.h: Likewise.
23706 * config/moxie/rtems.h: Likewise.
23707 * config/nios2/rtems.h: Likewise.
23708 * config/powerpcspe/rtems.h: Likewise.
23709 * config/rs6000/rtems.h: Likewise.
23710 * config/rtems.h: Likewise.
23711 * config/sh/rtems.h: Likewise.
23712 * config/sh/rtemself.h: Likewise.
23713 * config/sparc/rtemself.h: Likewise.
23715 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23718 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23719 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23720 * tree-ssa-structalias.c (alias_get_name): Same.
23722 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23725 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23726 instructions if no du chain is found.
23728 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23730 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23732 2017-07-25 Richard Biener <rguenther@suse.de>
23734 PR middle-end/81505
23735 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23738 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23740 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23741 upper-regs options.
23742 (ISA_2_7_MASKS_SERVER): Likewise.
23743 (ISA_3_0_MASKS_IEEE): Likewise.
23744 (OTHER_P8_VECTOR_MASKS): Likewise.
23745 (OTHER_VSX_VECTOR_MASKS): Likewise.
23746 (POWERPC_MASKS): Likewise.
23747 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23748 duplicate list of options.
23749 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23750 explicit -mupper-regs options.
23751 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23752 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23753 alias for -mupper-regs-df.
23754 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23755 (rs6000_init_hard_regno_mode_ok): Likewise.
23756 (rs6000_option_override_internal): Likewise.
23757 (rs6000_opt_masks): Likewise.
23758 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23759 options in terms of whether -mvsx or -mpower8-vector was used.
23760 (TARGET_UPPER_REGS_DI): Likewise.
23761 (TARGET_UPPER_REGS_SF): Likewise.
23762 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23763 -mupper-regs-* options.
23765 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23767 * passes.c (emergency_dump_function): Print some empty lines and a
23768 header before the RTL dump.
23770 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23772 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23774 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23777 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23778 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23780 2017-07-24 Carl Love <cel@us.ibm.com>
23782 * config/rs6000/rs6000-c.c: Add support for built-in functions
23783 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23784 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23785 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23786 vec_extract_fp_from_shortl): Add defines for the two builtins.
23787 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23788 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23790 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23791 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23792 * doc/extend.texi: Update the built-in documentation file for the
23793 new built-in function.
23795 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23798 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23800 * doc/generic.texi: Likewise.
23801 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23802 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23804 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23806 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23807 (aarch64_mls_elt_merge<mode>): Likewise.
23809 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23811 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23812 having __cxa_atexit.
23814 2017-07-23 Michael Collison <michael.collison@arm.com>
23816 * config/arm/arm.c (arm_option_override): Deprecate
23817 use of -mstructure-size-boundary.
23818 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23819 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23821 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23824 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23825 Reduce cost estimate for direct moves.
23827 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23830 * config/i386/i386.c (ix86_option_override_internal): Disable
23831 BMI, BMI2 and TBM instructions for -m16.
23833 2017-07-21 Carl Love <cel@us.ibm.com>
23835 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23836 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23837 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23838 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23839 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23840 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23841 VMULOSW): New enum "unspec" values.
23842 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23843 altivec_vmulosw): New patterns.
23844 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23845 VMULOSW): Add definitions.
23847 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23849 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23850 (qdf24xx): Likewise.
23851 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23852 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23853 (AARCH64_FL_V8_1): Renumber.
23854 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23855 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23856 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23857 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23858 rdma to feature modifiers list.
23860 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23862 PR middle-end/56727
23863 * ipa-visibility (function_and_variable_visibility): Convert
23864 recursive PLT call to direct call if appropriate.
23866 2017-07-21 Andrew Pinski <apinski@cavium.com>
23868 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23869 operand 1 to see if the types precision matches.
23870 * fold-const.c (operand_equal_p): Likewise.
23872 2017-07-21 Richard Biener <rguenther@suse.de>
23874 PR tree-optimization/81303
23875 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23876 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23877 (vect_peeling_hash_get_lowest_cost): Adjust.
23878 (vect_enhance_data_refs_alignment): Likewise. Use
23879 vect_get_peeling_costs_all_drs to compute the penalty for no
23880 peeling to match up costs.
23882 2017-07-21 Richard Biener <rguenther@suse.de>
23884 PR tree-optimization/81500
23885 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23886 we didn't identify a reduction path.
23888 2017-07-21 Tom de Vries <tom@codesourcery.com>
23889 Cesar Philippidis <cesar@codesourcery.com>
23891 PR gcov-profile/81442
23892 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23895 2017-07-21 Tom de Vries <tom@codesourcery.com>
23898 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23900 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23901 nvptx_override_options_after_change.
23903 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23905 * dwarf2out.c (output_file_names): Avoid double testing for
23906 dwarf_version >= 5.
23908 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23910 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23912 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23914 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23916 (try_crossjump_to_edge): Do not punt on partitioned functions.
23918 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23920 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23921 Put all BBs reachable only via paths crossing cold region to cold
23923 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23925 2016-07-21 Richard Biener <rguenther@suse.de>
23927 PR tree-optimization/81303
23928 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23929 into account prologue and epilogue iterations when raising
23930 min_profitable_iters to sth at least covering one vector iteration.
23932 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23934 * config/arm/arm.c (arm_test_cpu_arch_dat):
23937 2017-07-20 Nathan Sidwell <nathan@acm.org>
23939 Remove TYPE_METHODS.
23940 * tree.h (TYPE_METHODS): Delete.
23941 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23942 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23943 (dbxout_type_methods): Scan TYPE_FIELDS.
23944 (dbxout_type): Don't check TYPE_METHODS here.
23945 * function.c (use_register_for_decl): Always ignore register for
23946 class types when not optimizing.
23947 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23948 * tree.c (free_lang_data_in_type): Stitch out member functions and
23949 templates from TYPE_FIELDS.
23950 (build_distinct_type_copy, verify_type_variant,
23951 verify_type): Member fns are on TYPE_FIELDS.
23952 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23953 * tree-pretty-print.c (dump_generic_node): Likewise.
23955 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23958 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23959 V2TImode and V4TImode.
23960 (ix86_expand_vector_extract): Likewise.
23961 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23962 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23963 (ssescalarmode): Handle V4TImode and V2TImode.
23964 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23965 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23966 (VEXTRACTI128_MODE): New mode iterator.
23967 (splitter for *vec_extractv?ti first element): New.
23968 (VEC_INIT_MODE): New mode iterator.
23969 (vec_init<mode>): Consolidate 3 expanders into one using
23970 VEC_INIT_MODE mode iterator.
23972 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23974 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23975 non_spilled_static_chain_regno_p.
23977 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23979 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23981 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23983 * bb-reorder.c (connect_traces): Allow copying of blocks within
23986 2017-07-20 Richard Biener <rguenther@suse.de>
23988 * gimple.h (gimple_phi_result): Add gphi * overload.
23989 (gimple_phi_result_ptr): Likewise.
23990 (gimple_phi_arg): Likewise. Adjust index assert to only
23991 allow actual argument accesses rather than all slots available
23993 (gimple_phi_arg_def): Add gphi * overload.
23994 * tree-phinodes.c (make_phi_node): Initialize only actual
23996 (resize_phi_node): Clear memory not covered by old node,
23997 do not initialize excess argument slots.
23998 (reserve_phi_args_for_new_edge): Initialize new argument slot
24001 2017-07-20 Bin Cheng <bin.cheng@arm.com>
24003 PR tree-optimization/81388
24005 2016-07-21 Bin Cheng <bin.cheng@arm.com>
24007 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
24008 by removing computation of may_be_zero.
24010 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24011 Tom de Vries <tom@codesourcery.com>
24013 PR middle-end/81030
24014 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
24015 when gimple level profile disagrees with what RTL expander did.
24017 2017-07-20 Richard Biener <rguenther@suse.de>
24019 PR tree-optimization/61171
24020 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
24021 (vect_analyze_stmt): Add slp instance parameter.
24022 (vectorizable_reduction): Likewise.
24023 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
24024 (vect_is_simple_reduction): Deal with chains not detected
24025 as SLP reduction chain, specifically not properly associated
24026 chains containing a mix of plus/minus.
24027 (get_reduction_op): Remove.
24028 (get_initial_defs_for_reduction): Simplify, pass in whether
24029 this is a reduction chain, pass in the SLP node for the PHIs.
24030 (vect_create_epilog_for_reduction): Get the SLP instance as
24032 (vectorizable_reduction): Get the SLP instance as arg.
24033 During analysis remember the SLP node with the PHIs in the
24034 instance. Simplify getting at the vectorized reduction PHIs.
24035 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
24036 through SLP instance.
24037 (vect_slp_analyze_operations): Likewise.
24038 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
24039 (vect_transform_stmt): Likewise.
24041 2017-07-20 Tom de Vries <tom@codesourcery.com>
24043 PR tree-optimization/81489
24044 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
24045 read of phi arg location to before loop that modifies phi.
24047 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24049 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
24052 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24054 PR middle-end/81331
24055 * except.c (execute): Fix ordering issue.
24057 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24059 PR rtl-optimization/81423
24060 * combine.c (make_compound_operation_int): Don't try to optimize
24061 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
24063 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24065 PR rtl-optimization/81423
24066 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
24067 with a constant that is -1 in the truncated to mode.
24069 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24071 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
24072 (determine_unlikely_bbs): ... here.
24073 * predict.h (propagate_unlikely_bbs_forward): Declare.
24074 * cfgexpand.c (pass_expand::execute): Use it.
24075 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
24077 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
24078 propagate_unlikely_bbs_forward.
24080 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24082 PR middle-end/81331
24083 * except.c (maybe_add_nop_after_section_switch): New function.
24086 2017-07-19 Tom de Vries <tom@codesourcery.com>
24088 * gimple.h (gimple_phi_set_arg): Make assert more strict.
24090 2017-07-19 Tom de Vries <tom@codesourcery.com>
24092 * gimple.h (gimple_phi_arg): Make assert more strict.
24094 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
24096 * config.gcc (powerpc*-*-*): Add mmintrin.h.
24097 * config/rs6000/mmintrin.h: New file.
24098 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
24100 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24102 PR tree-optimization/81346
24103 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
24105 2017-07-19 Tom de Vries <tom@codesourcery.com>
24107 * config/nvptx/nvptx.md (VECIM): Add V2DI.
24109 2017-07-19 Tom de Vries <tom@codesourcery.com>
24111 * config/nvptx/nvptx-modes.def: Add V2DImode.
24112 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
24113 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
24114 (nvptx_output_mov_insn): Handle lack of mov.b128.
24115 (nvptx_print_operand): Handle 'H' and 'L' codes.
24116 (nvptx_vector_mode_supported): Allow V2DImode.
24117 (nvptx_preferred_simd_mode): New function.
24118 (nvptx_data_alignment): New function.
24119 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
24120 nvptx_preferred_simd_mode.
24121 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
24123 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
24125 2017-07-19 Tom de Vries <tom@codesourcery.com>
24127 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
24128 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
24129 (nvptx_vector_mode_supported): New function. Allow V2SImode.
24130 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
24131 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
24132 (mov<VECIM>_insn): New define_insn.
24133 (define_expand "mov<VECIM>): New define_expand.
24135 2017-07-19 Tom de Vries <tom@codesourcery.com>
24137 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
24139 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24141 PR tree-optimization/81346
24142 * fold-const.h (fold_div_compare, range_check_type): Declare.
24143 * fold-const.c (range_check_type): New function.
24144 (build_range_check): Use range_check_type.
24145 (fold_div_compare): No longer static, rewritten into
24146 a match.pd helper function.
24147 (fold_comparison): Don't call fold_div_compare here.
24148 * match.pd (X / C1 op C2): New optimization using fold_div_compare
24149 as helper function.
24151 2017-07-19 Nathan Sidwell <nathan@acm.org>
24153 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
24154 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
24155 * tree.c (find_decls_types_r, verify_type): Use
24156 TYPE_{MIN,MAX}_VALUE_RAW.
24157 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
24158 (hash_tree): Likewise.
24159 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
24161 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
24164 2017-07-18 Tom de Vries <tom@codesourcery.com>
24166 PR middle-end/81464
24167 * omp-expand.c (expand_omp_for_static_chunk): Handle
24168 equal-argument loop exit phi.
24170 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
24173 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
24174 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
24175 operand 2 predicate.
24176 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
24177 operand 2 predicate.
24178 (ror,rol -> rorx splitters): Use const_int_operand as
24179 operand 2 predicate.
24181 2017-06-18 Richard Biener <rguenther@suse.de>
24183 PR tree-optimization/81410
24184 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
24185 the gap in the ! slp_perm SLP case after each group.
24187 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24189 PR middle-end/81463
24190 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
24193 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24195 PR middle-end/81462
24196 * predict.c (set_even_probabilities): Cleanup; do not affect
24197 probabilities that are already known.
24198 (combine_predictions_for_bb): Call even when count is set.
24200 2017-07-18 Nathan Sidwell <nathan@acm.org>
24202 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24205 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24208 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24209 optimization for loop niter analysis.
24211 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24214 * config/avr/avr.c (avr_optimize_casesi): Don't use
24215 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24217 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24219 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24220 body_cost_vec from _vect_peel_extended_info.
24221 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24222 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24225 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24227 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24229 2017-07-18 Richard Biener <rguenther@suse.de>
24231 PR tree-optimization/80620
24232 PR tree-optimization/81403
24233 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24234 info when re-using a VN table entry.
24236 2017-07-18 Richard Biener <rguenther@suse.de>
24238 PR tree-optimization/81418
24239 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24240 vectype_in. Verify that with lane-reducing reduction operations
24241 we have a single def-use cycle.
24243 2017-07-17 Carl Love <cel@us.ibm.com>
24245 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24247 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24248 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24249 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24250 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24251 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24252 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24253 VMULOSW): New enum "unspec" values.
24254 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24255 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24256 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24257 altivec_vmulosw): New patterns.
24258 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24259 VMULOSW): Add definitions.
24261 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24263 * config/alpha/alpha.c: Include predict.h.
24265 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24267 * tree-vrp.c (compare_assert_loc): Fix comparison function
24268 to return predictable results.
24270 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24272 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24274 (subdi3): Likewise.
24275 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24276 * doc/invoke.texi (mexpand-adddi): Update text.
24278 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24280 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24281 that also clobbers the CC register. The old expand code is moved
24283 (*arc_clzsi2): ... here.
24284 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24285 the CC register. The old expand code is moved to ...
24286 (arc_ctzsi2): ... here.
24288 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24290 * config/arc/arc.opt (mindexed-loads): Use initial value
24291 TARGET_INDEXED_LOADS_DEFAULT.
24292 (mauto-modify-reg): Use initial value
24293 TARGET_AUTO_MODIFY_REG_DEFAULT.
24294 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24295 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24296 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24297 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24299 2017-07-17 Martin Liska <mliska@suse.cz>
24302 * opts.c (finish_options): Do not allow -fgnu-tm
24303 w/ -fsanitize={kernel-,}address. Say sorry.
24305 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24308 * tree-loop-distribution.c (classify_partition): Only assert on
24309 numer of iterations.
24310 (merge_dep_scc_partitions): Delete prameter. Update function call.
24311 (distribute_loop): Remove code handling loop with unknown niters.
24312 (pass_loop_distribution::execute): Skip loop with unknown niters.
24314 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24317 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24318 function sort_partitions_by_post_order.
24320 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24322 PR tree-optimization/81374
24323 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24324 the max index of basic blocks, rather than number of basic blocks.
24326 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24328 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24330 (arc_legitimate_pic_operand_p): Likewise.
24331 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24333 (arc_needs_pcl_p): Likewise.
24334 (arc_legitimate_pc_offset_p): Likewise.
24335 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24336 function is also used in constrains.md.
24337 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24338 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24339 PLUS. Only return true/false in known cases, otherwise assert.
24340 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24341 is already called in arc_legitimate_constant_p.
24342 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24344 (LEGITIMATE_PIC_OPERAND_P): Use
24345 arc_raw_symbolic_reference_mentioned_p function.
24346 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24351 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24352 Andrew Burgess <andrew.burgess@embecosm.com>
24354 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24355 (arc_return_address_register): New function.
24356 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24357 (arc_handle_fndecl_attribute): Add naked attribute.
24358 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24359 (TARGET_WARN_FUNC_RETURN): Likewise.
24360 (arc_allocate_stack_slots_for_args): New function.
24361 (arc_warn_func_return): Likewise.
24362 (machine_function): Change type fn_type.
24363 (arc_compute_function_type): Consider new naked function type,
24364 change function return type.
24365 (arc_must_save_register): Adapt to handle new
24366 arc_compute_function_type's return type.
24367 (arc_expand_prologue): Likewise.
24368 (arc_expand_epilogue): Likewise.
24369 (arc_return_address_regs): Delete.
24370 (arc_return_address_register): New function.
24371 (arc_epilogue_uses): Use above function.
24372 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24373 (arc_function_type): Change encoding, add naked type.
24374 (ARC_INTERRUPT_P): Change to handle the new encoding.
24375 (ARC_FAST_INTERRUPT_P): Likewise.
24376 (ARC_NORMAL_P): Define.
24377 (ARC_NAKED_P): Likewise.
24378 (arc_compute_function_type): Delete prototype.
24379 * config/arc/arc.md (in_ret_delay_slot): Use
24380 arc_return_address_register function.
24381 (simple_return): Likewise.
24382 (p_return_i): Likewise.
24384 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24386 PR tree-optimization/81428
24387 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24388 can't be built for those types.
24390 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24392 Remove stuff dead since r239246.
24394 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24395 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24396 (avr_inform_devices): Remove dead stuff.
24398 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24400 * config/arm/arm_neon.h: Fix softp typo.
24402 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24404 PR tree-optimization/81365
24405 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24406 aggregate moves onto bb predecessor edges, make sure there are no
24407 loads that could alias the lhs in between the start of bb and the
24410 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24413 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24414 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24415 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24417 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24419 PR tree-optimization/81396
24420 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24421 (init_symbolic_number): Initialize it to 1.
24422 (perform_symbolic_merge): Add n_ops from both operands into the new
24424 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24425 without base_addr as useless if they need more than one operation.
24426 (bswap_replace): Handle !bswap case for NULL base_addr.
24428 2017-07-17 Tom de Vries <tom@codesourcery.com>
24431 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24434 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24436 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24437 conditional builtin define __FIX_LEON3FT_B2BST.
24439 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24441 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24442 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24445 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24447 PR rtl-optimization/81424
24448 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24449 to remove potential trapping from operands if -fnon-call-exceptions.
24451 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24453 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24454 profile_proability for scalling.
24455 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24457 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24459 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24461 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24463 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24464 fixpoint arithmetics.
24466 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24468 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24469 fixpoint arithmetics.
24471 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24473 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24474 fixpoint arithmetics.
24476 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24478 * profile-count.h (profile_probability::from_reg_br_prob_note,
24479 profile_probability::to_reg_br_prob_note): New functions.
24480 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24481 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24482 * predict.c (probability_reliable_p): Update.
24483 (edge_probability_reliable_p): Update.
24484 (br_prob_note_reliable_p): Update.
24485 (invert_br_probabilities): Update.
24486 (add_reg_br_prob_note): New function.
24487 (combine_predictions_for_insn): Update.
24488 * asan.c (asan_clear_shadow): Update.
24489 * cfgbuild.c (compute_outgoing_frequencies): Update.
24490 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24491 (update_br_prob_note): Update.
24492 (rtl_verify_edges): Update.
24493 (purge_dead_edges): Update.
24494 (fixup_reorder_chain): Update.
24495 * emit-rtl.c (try_split): Update.
24496 * ifcvt.c (cond_exec_process_insns): Update.
24497 (cond_exec_process_if_block): Update.
24498 (dead_or_predicable): Update.
24499 * internal-fn.c (expand_addsub_overflow): Update.
24500 (expand_neg_overflow): Update.
24501 (expand_mul_overflow): Update.
24502 * loop-doloop.c (doloop_modify): Update.
24503 * loop-unroll.c (compare_and_jump_seq): Update.
24504 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24505 * predict.h: Update.
24506 * reorg.c (mostly_true_jump): Update.
24508 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24509 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24510 * config/arc/arc.c: (emit_unlikely_jump): Update.
24511 * config/arm/arm.c: (emit_unlikely_jump): Update.
24512 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24513 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24514 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24515 (ix86_print_operand): Update.
24516 (ix86_split_fp_branch): Update.
24517 (predict_jump): Update.
24518 * config/ia64/ia64.c (ia64_print_operand): Update.
24519 * config/mmix/mmix.c (mmix_print_operand): Update.
24520 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24521 (rs6000_expand_split_stack_prologue): Update.
24522 * config/rs6000/rs6000.c: Update.
24523 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24524 (s390_expand_vec_movstr): Update.
24525 (s390_expand_cs_tdsi): Update.
24526 (s390_expand_split_stack_prologue): Update.
24527 * config/sh/sh.c (sh_print_operand): Update.
24528 (expand_cbranchsi4): Update.
24529 (expand_cbranchdi4): Update.
24530 * config/sparc/sparc.c (output_v9branch): Update.
24531 * config/spu/spu.c (get_branch_target): Update.
24532 (ea_load_store_inline): Update.
24533 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24534 * config/tilepro/tilepro.c: Update.
24536 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24538 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24539 (gimplify_save_expr): Likewise.
24541 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24543 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24545 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24547 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24549 * timevar.def (TV_IPA_FNSUMMARY): Define.
24551 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24553 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24554 to back store errata sensitive sequence from being generated.
24555 (sqrtdf2_fix): Likewise.
24557 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24559 * tree-ssa-threadupdate.c (compute_path_counts,
24560 update_joiner_offpath_counts): Use profile_probability.
24562 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24565 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24567 * config/arm/arm-c.c (arm_cpu_builtins): Define
24568 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24570 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24572 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24573 array entries to represent __ieee128 versions of the
24574 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24575 scalar_extract_sig, and scalar_insert_exp built-in functions.
24576 (altivec_resolve_overloaded_builtin): Add special case handling
24577 for the __builtin_scalar_insert_exp function, as represented by
24578 the P9V_BUILTIN_VEC_VSIEDP constant.
24579 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24580 exponent support for __ieee128 argument.
24581 (VSESQP): Add scalar extract signature support for __ieee128
24583 (VSTDCNQP): Add scalar test negative support for __ieee128
24585 (VSIEQP): Add scalar insert exponent support for __int128 argument
24586 with __ieee128 result.
24587 (VSIEQPF): Add scalar insert exponent support for __ieee128
24588 argument with __ieee128 result.
24589 (VSTDCQP): Add scalar test data class support for __ieee128
24591 (VSTDCNQP): Add overload support for scalar test negative with
24592 __ieee128 argument.
24593 (VSTDCQP): Add overload support for scalar test data class
24594 __ieee128 argument.
24595 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24596 UNSPEC_VSX_SXSIGDP.
24597 (UNSPEC_VSX_SIEXPQP): New constant.
24598 (xsxexpqp): New insn for VSX scalar extract exponent quad
24600 (xsxsigqp): New insn for VSX scalar extract significand quad
24602 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24603 precision with floating point argument.
24604 (xststdcqp): New expand for VSX scalar test data class quad
24606 (xststdcnegqp): New expand for VSX scalar test negative quad
24608 (xststdcqp): New insn to match expansions for VSX scalar test data
24609 class quad precision and VSX scalar test negative quad precision.
24610 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24611 special case operand checking to enforce that second operand of
24612 VSX scalar test data class with quad precision argument is a 7-bit
24614 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24615 prototypes and descriptions of __ieee128 versions of
24616 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24617 scalar_test_data_class, and scalar_test_neg built-in functions.
24619 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24621 PR tree-optimization/81162
24622 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24623 replace a negate with an add.
24625 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24627 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24629 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24631 * config/arm/arm-c.c (arm_cpu_builtins): Define
24632 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24634 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24636 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24637 (armv8-r): Set ARM Cortex-R52 as default CPU.
24638 * config/arm/arm-tables.opt: Regenerate.
24639 * config/arm/arm-tune.md: Regenerate.
24640 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24642 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24643 extension for -mcpu=cortex-r52.
24645 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24647 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24648 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24649 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24650 (fp-armv8): Define it as FP_ARMv8 only.
24651 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24652 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24654 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24655 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24656 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24657 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24658 than TARGET_FPU_ARMV8.
24659 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24660 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24661 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24663 * config/arm/neon.md (neon_vrint): Likewise.
24664 (neon_vcvt): Likewise.
24665 (neon_<fmaxmin_op><mode>): Likewise.
24666 (<fmaxmin><mode>3): Likewise.
24667 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24668 * config/arm/predicates.md (arm_cond_move_operator): Check against
24669 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24671 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24673 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24674 to top of function.
24676 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24678 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24679 loop in comment with memset.
24681 2017-07-14 Martin Liska <mliska@suse.cz>
24683 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24684 * dwarf2out.c (is_java): Remove the function.
24685 (output_pubname): Remove usage of the function.
24686 (lower_bound_default): Remove usage of DW_LANG_Java.
24687 (gen_compile_unit_die): Likewise.
24688 * gcc.c: Remove compiler defaults for .java and .zip files.
24689 * gimple-expr.c (remove_suffix): Change as there's no longer
24690 extension than 4-letter one.
24691 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24692 (gimplify_save_expr): Likewise.
24693 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24694 as it's possible even for other languages than Java.
24695 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24696 * lto-opts.c (lto_write_options): Remove reference to Java.
24697 * opts.c (strip_off_ending): Update file extension handling.
24698 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24699 * tree-eh.c (lower_resx): Likewise.
24700 * tree.c (free_lang_data_in_type): Remove dead code.
24701 (find_decls_types_r): Likewise.
24702 (build_common_builtin_nodes): Remove Java from a comment.
24703 (verify_type): Remove dead code.
24704 * varasm.c (assemble_external): Remove Java from a comment.
24706 2017-07-14 Martin Liska <mliska@suse.cz>
24708 * opts.c (finish_options): Add quotes.
24709 (common_handle_option): Likewise.
24711 2017-07-14 Martin Liska <mliska@suse.cz>
24713 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24714 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24715 Remove N_SO_PASCAL.
24716 * dwarf2out.c (lower_bound_default): Do not handle
24718 (gen_compile_unit_die): Likewise.
24719 * gcc.c: Remove default extension binding for GNU Pascal.
24720 * stmt.c: Remove Pascal language from a comment.
24721 * xcoffout.c: Likewise.
24723 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24726 * diagnostic-show-locus.c (fixit_cmp): New function.
24727 (layout::layout): Sort m_fixit_hints.
24728 (column_range::column_range): Assert that the values are valid.
24729 (struct char_span): New struct.
24730 (correction::overwrite): New method.
24731 (struct source_line): New struct.
24732 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24733 calls in terms of classes source_line and char_span, and
24734 correction::overwrite.
24735 (selftest::test_overlapped_fixit_printing_2): New function.
24736 (selftest::diagnostic_show_locus_c_tests): Call it.
24738 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24740 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24741 early if there is no lhs.
24743 2017-07-13 Martin Liska <mliska@suse.cz>
24745 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24746 (gen_reference_type_die): Likewise.
24747 * stor-layout.c: Remove Pascal-related comment.
24749 2017-07-13 Martin Liska <mliska@suse.cz>
24751 * opts.c (finish_options): Add quotes to error messages.
24752 (parse_sanitizer_options): Likewise.
24754 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24756 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24758 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24760 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24762 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24764 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24766 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24768 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24771 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24772 provides the hardware capability bits, define the macro
24773 __BUILTIN_CPU_SUPPORTS__.
24774 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24775 if GLIBC does not provide the hardware capability bits. Add a
24776 gcc_unreachable call if the built-in cpu function is neither
24777 __builtin_cpu_is nor __builtin_cpu_supports.
24778 (rs6000_get_function_versions_dispatcher): Change the warning
24779 that an old GLIBC is used which does not export the capability
24780 bits to be an error.
24781 * doc/extend.texi (target_clones attribute): Document the
24782 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24783 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24784 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24785 the macros defined by GCC if the newer GLIBC is available.
24787 2017-07-12 Jeff Law <law@redhat.com>
24789 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24790 remaining includes slightly.
24791 * config/riscv/riscv-builtins.c: Include profile-count.h.
24793 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24796 * config/avr/avr.c (avr_set_current_function): In diagnostic
24797 messages: Quote keywords and (parts of) identifiers.
24798 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24801 2017-07-12 Carl Love <cel@us.ibm.com>
24803 * config/rs6000/rs6000-c.c: Add support for built-in functions
24804 vector bool char vec_revb (vector bool char);
24805 vector bool short vec_revb (vector short char);
24806 vector bool int vec_revb (vector bool int);
24807 vector bool long long vec_revb (vector bool long long);
24808 * doc/extend.texi: Update the built-in documentation file for the
24809 new built-in functions.
24811 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24813 * config/s390/s390.md: Remove movcc splitter.
24815 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24817 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24818 load/store on condition.
24820 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24823 * config/avr/avr.c (avr_encode_section_info)
24824 [progmem && !TREE_READONLY]: Error if progmem object needs
24827 2017-07-11 Michael Collison <michael.collison@arm.com>
24829 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24832 2017-07-11 Carl Love <cel@us.ibm.com>
24834 * config/rs6000/rs6000-c.c: Add support for builtins
24835 vector unsigned int vec_parity_lsbb (vector signed int);
24836 vector unsigned int vec_parity_lsbb (vector unsigned int);
24837 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24838 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24839 vector unsigned long long vec_parity_lsbb (vector signed long long);
24840 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24841 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24842 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24843 * doc/extend.texi: Update the built-in documentation file for the
24844 new built-in functions.
24846 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24848 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24849 (layout::m_primary_loc): New field.
24850 (layout::layout): Initialize new field. Move location filtering
24851 logic from here to...
24852 (layout::maybe_add_location_range): ...this new method. Add
24853 support for filtering to just the lines already specified by other
24855 (layout::will_show_line_p): New method.
24856 (gcc_rich_location::add_location_if_nearby): New method.
24857 (selftest::test_add_location_if_nearby): New test function.
24858 (selftest::diagnostic_show_locus_c_tests): Call it.
24859 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24862 2017-07-11 Tom de Vries <tom@codesourcery.com>
24864 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24865 (bb_first_real_insn): New function.
24866 (nvptx_single): Add extra initialization of broadcasted condition
24869 2017-07-11 Nathan Sidwell <nathan@acm.org>
24871 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24873 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24875 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24876 Binutils doc as TEXI will mess them up.
24877 * doc/invoke.texi (AVR Options): Same here.
24879 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24881 * config/sparc/sparc.opt (mfix-ut700): New option.
24882 (mfix-gr712rc): Likewise.
24883 (sparc_fix_b2bst): New variable.
24884 * doc/invoke.texi (SPARC options): Document them.
24885 (ARM options): Fix warnings.
24886 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24887 instructions to prevent sequences that can trigger the store-store
24888 errata for certain LEON3FT processors.
24889 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24890 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24891 * config/sparc/sparc.md (fix_b2bst): New attribute.
24892 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24894 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24897 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24899 (*rsqrtsf2_sse): Ditto.
24901 (div<mode>3): Macroize insn from divdf3 and divsf3
24902 using MODEF mode iterator.
24904 2017-07-10 Martin Sebor <msebor@redhat.com>
24906 PR tree-optimization/80397
24907 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24908 instead of testing for equality to INTEGER_TYPE.
24910 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24912 * config.gcc: Remove uclibc from arc target spec.
24914 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24916 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24918 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24921 * lto-wrapper.c (remove_option): New function.
24922 (merge_and_complain): Merge PIC/PIE options more realistically.
24924 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24926 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24930 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24931 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24932 * config.in: Regenerate.
24933 * configure: Regenerate.
24934 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24935 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24936 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24937 (TARGET_GASISR_PROLOGUES): ...target mask.
24938 * common/config/avr/avr-common.c
24939 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24940 Set -mgas-isr-prologues.
24941 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24942 INSERT_PASS_BEFORE for it.
24943 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24944 * config/avr/avr.c (avr_option_override)
24945 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24946 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24947 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24948 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24949 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24951 (make_avr_pass_pre_proep): New function.
24952 (emit_push_sfr) <treg>: Add argument to function and use it
24953 instead of TMP_REG.
24954 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24955 and set machine->gasisr.yes.
24956 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24957 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24958 __gcc_isr.n_pushed to .L__stack_usage.
24959 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24960 (avr_asm_final_postscan_insn): ...this new static function.
24961 * config/avr/avr.h (machine_function)
24962 <is_no_gccisr, use_L__stack_usage>: New fields.
24963 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24964 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24965 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24966 (gasisr, *gasisr): New expander and insn.
24967 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24968 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24969 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24971 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24973 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24976 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24978 Move jump-tables out of .text again.
24981 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24982 (ASM_OUTPUT_ADDR_VEC): New function.
24983 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24984 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24985 INSN_ADDRESSes as asm comment.
24986 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24987 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24988 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24989 * config/avr/avr.md (*tablejump): Adjust comment.
24990 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24991 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24993 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24994 (avr_output_addr_vec): New proto.
24995 (avr_log_t) <insn_addresses>: New field.
24997 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
25000 * config/i386/i386.c (ix86_function_arg_advance): Set
25001 outgoing_args_on_stack to true if there are outgoing arguments
25003 (ix86_function_arg): Likewise.
25004 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
25005 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
25006 * config/i386/i386.h (machine_function): Add
25007 outgoing_args_on_stack.
25009 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
25011 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
25012 supporting pthreds.
25013 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
25015 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
25017 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
25018 (REAL_H): Remove $(MACHMODE_H).
25019 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
25021 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
25022 $(MACHMODE_H) and double-int.h.
25023 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
25025 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
25028 2017-07-07 Andrew Pinski <apinski@cavium.com>
25030 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
25031 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
25033 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
25035 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
25036 Add warning if GCC was not configured to link against a GLIBC that
25037 exports the hardware capability bits.
25038 (make_resolver_func): Make resolver function private and not a
25039 COMDAT function. Create the name with clone_function_name instead
25040 of make_unique_name.
25043 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
25044 correct operand in doing the split.
25046 2017-07-07 Carl Love <cel@us.ibm.com>
25048 * config/rs6000/rs6000-c: Add support for built-in function
25049 vector unsigned short vec_pack_to_short_fp32 (vector float,
25051 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
25052 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
25053 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
25054 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
25055 (convert_4f32_8i16): Add define_expand.
25056 * doc/extend.texi: Update the built-in documentation file for the
25057 new built-in function.
25059 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25061 * config/sparc/m8.md: New file.
25062 * config/sparc/sparc.md: Include m8.md.
25064 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25066 * config/sparc/sparc.opt: New option -mvis4b.
25067 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
25068 (sparc_option_override): Handle VIS4B.
25069 (enum sparc_builtins): Define
25070 SPARC_BUILTIN_DICTUNPACK{8,16,32},
25071 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
25072 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
25073 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
25074 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
25075 (check_constant_argument): New function.
25076 (sparc_vis_init_builtins): Define builtins
25077 __builtin_vis_dictunpack{8,16,32},
25078 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
25079 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
25080 __builtin_vis_fpcmpde{8,16,32}shl and
25081 __builtin_vis_fpcmpur{8,16,32}shl.
25082 (sparc_expand_builtin): Check that the constant operands to
25083 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
25084 constant and in range.
25085 * config/sparc/sparc-c.c (sparc_target_macros): Handle
25087 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
25088 (SPARC_IMM5_P): Likewise.
25089 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
25090 (enabled): Handle vis4b.
25091 (UNSPEC_DICTUNPACK): New unspec.
25092 (UNSPEC_FPCMPSHL): Likewise.
25093 (UNSPEC_FPUCMPSHL): Likewise.
25094 (UNSPEC_FPCMPDESHL): Likewise.
25095 (UNSPEC_FPCMPURSHL): Likewise.
25096 (cpu_feature): New CPU feature `vis4b'.
25097 (dictunpack{8,16,32}): New insns.
25098 (FPCSMODE): New mode iterator.
25099 (fpcscond): New code iterator.
25100 (fpcsucond): Likewise.
25101 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
25102 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
25103 (fpcmpde{8,16,32}{si,di}shl): Likewise.
25104 (fpcmpur{8,16,32}{si,di}shl): Likewise.
25105 * config/sparc/constraints.md: Define constraints `q' for unsigned
25106 2-bit integer constants and `t' for unsigned 5-bit integer
25108 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
25110 (imm5_operand_dictunpack16): Likewise.
25111 (imm5_operand_dictunpack32): Likewise.
25112 (imm2_operand): Likewise.
25113 * doc/invoke.texi (SPARC Options): Document -mvis4b.
25114 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
25115 ditunpack* and fpcmp*shl builtins.
25117 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25119 * config.gcc: Handle m8 in --with-{cpu,tune} options.
25120 * config.in: Add HAVE_AS_SPARC6 define.
25121 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
25123 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
25125 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
25126 (CPP_CPU_SPEC): Handle m8.
25127 (ASM_CPU_SPEC): Likewise.
25128 * config/sparc/sparc-opts.h (enum processor_type): Add
25130 * config/sparc/sparc.c (m8_costs): New struct.
25131 (sparc_option_override): Handle TARGET_CPU_m8.
25132 (sparc32_initialize_trampoline): Likewise.
25133 (sparc64_initialize_trampoline): Likewise.
25134 (sparc_issue_rate): Likewise.
25135 (sparc_register_move_cost): Likewise.
25136 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
25137 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
25138 (ASM_CPU64_DEFAULT_SPEC): Likewise.
25139 (CPP_CPU_SPEC): Handle M8.
25140 (ASM_CPU_SPEC): Likewise.
25141 (AS_M8_FLAG): Define.
25142 * config/sparc/sparc.md: Add m8 to the cpu attribute.
25143 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
25144 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
25146 * configure: Regenerate.
25147 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
25150 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25152 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
25154 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
25155 ("*movdi_insn_sp32"): Do not set v3pipe.
25156 ("*movsi_insn"): Likewise.
25157 ("*movdi_insn_sp64"): Likewise.
25158 ("*movsf_insn"): Likewise.
25159 ("*movdf_insn_sp32"): Likewise.
25160 ("*movdf_insn_sp64"): Likewise.
25161 ("*zero_extendsidi2_insn_sp64"): Likewise.
25162 ("*sign_extendsidi2_insn"): Likewise.
25163 ("*mov<VM32:mode>_insn"): Likewise.
25164 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25165 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25166 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25167 ("<vlop:code><VL:mode>3"): Likewise.
25168 ("*not_<vlop:code><VL:mode>3"): Likewise.
25169 ("*nand<VL:mode>_vis"): Likewise.
25170 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
25171 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
25172 ("one_cmpl<VL:mode>2"): Likewise.
25173 ("faligndata<VM64:mode>_vis"): Likewise.
25174 ("alignaddrsi_vis"): Likewise.
25175 ("alignaddrdi_vis"): Likweise.
25176 ("alignaddrlsi_vis"): Likewise.
25177 ("alignaddrldi_vis"): Likewise.
25178 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25179 ("bmaskdi_vis"): Likewise.
25180 ("bmasksi_vis"): Likewise.
25181 ("bshuffle<VM64:mode>_vis"): Likewise.
25182 ("cmask8<P:mode>_vis"): Likewise.
25183 ("cmask16<P:mode>_vis"): Likewise.
25184 ("cmask32<P:mode>_vis"): Likewise.
25185 ("pdistn<P:mode>_vis"): Likewise.
25186 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25188 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25190 * config/sparc/sparc.md ("subtype"): New insn attribute.
25191 ("*wrgsr_sp64"): Set insn subtype.
25192 ("*rdgsr_sp64"): Likewise.
25193 ("alignaddrsi_vis"): Likewise.
25194 ("alignaddrdi_vis"): Likewise.
25195 ("alignaddrlsi_vis"): Likewise.
25196 ("alignaddrldi_vis"): Likewise.
25197 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25198 ("fexpand_vis"): Likewise.
25199 ("fpmerge_vis"): Likewise.
25200 ("faligndata<VM64:mode>_vis"): Likewise.
25201 ("bshuffle<VM64:mode>_vis"): Likewise.
25202 ("cmask8<P:mode>_vis"): Likewise.
25203 ("cmask16<P:mode>_vis"): Likewise.
25204 ("cmask32<P:mode>_vis"): Likewise.
25205 ("fchksm16_vis"): Likewise.
25206 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25207 ("fmean16_vis"): Likewise.
25208 ("fp<plusminus_insn>64_vis"): Likewise.
25209 ("<plusminus_insn>v8qi3"): Likewise.
25210 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25211 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25212 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25213 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25214 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25215 ("*movqi_insn"): Likewise.
25216 ("*movhi_insn"): Likewise.
25217 ("*movsi_insn"): Likewise.
25218 ("movsi_pic_gotdata_op"): Likewise.
25219 ("*movdi_insn_sp32"): Likewise.
25220 ("*movdi_insn_sp64"): Likewise.
25221 ("movdi_pic_gotdata_op"): Likewise.
25222 ("*movsf_insn"): Likewise.
25223 ("*movdf_insn_sp32"): Likewise.
25224 ("*movdf_insn_sp64"): Likewise.
25225 ("*zero_extendhisi2_insn"): Likewise.
25226 ("*zero_extendqihi2_insn"): Likewise.
25227 ("*zero_extendqisi2_insn"): Likewise.
25228 ("*zero_extendqidi2_insn"): Likewise.
25229 ("*zero_extendhidi2_insn"): Likewise.
25230 ("*zero_extendsidi2_insn_sp64"): Likewise.
25231 ("ldfsr"): Likewise.
25232 ("prefetch_64"): Likewise.
25233 ("prefetch_32"): Likewise.
25234 ("tie_ld32"): Likewise.
25235 ("tie_ld64"): Likewise.
25236 ("*tldo_ldub_sp32"): Likewise.
25237 ("*tldo_ldub1_sp32"): Likewise.
25238 ("*tldo_ldub2_sp32"): Likewise.
25239 ("*tldo_ldub_sp64"): Likewise.
25240 ("*tldo_ldub1_sp64"): Likewise.
25241 ("*tldo_ldub2_sp64"): Likewise.
25242 ("*tldo_ldub3_sp64"): Likewise.
25243 ("*tldo_lduh_sp32"): Likewise.
25244 ("*tldo_lduh1_sp32"): Likewise.
25245 ("*tldo_lduh_sp64"): Likewise.
25246 ("*tldo_lduh1_sp64"): Likewise.
25247 ("*tldo_lduh2_sp64"): Likewise.
25248 ("*tldo_lduw_sp32"): Likewise.
25249 ("*tldo_lduw_sp64"): Likewise.
25250 ("*tldo_lduw1_sp64"): Likewise.
25251 ("*tldo_ldx_sp64"): Likewise.
25252 ("*mov<VM32:mode>_insn"): Likewise.
25253 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25254 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25256 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25258 * config/sparc/sparc.md ("type"): New insn type viscmp.
25259 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25261 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25262 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25263 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25264 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25266 ("n7_vis_logical_11cycle"): Likewise.
25267 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25268 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25269 * config/sparc/niagara.md ("niag_vis"): Likewise.
25270 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25271 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25273 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25275 * config/sparc/sparc.md: New instruction type `bmask'.
25276 (bmaskdi_vis): Use the `bmask' type.
25277 (bmasksi_vis): Likewise.
25278 * config/sparc/ultra3.md (us3_array): Likewise.
25279 * config/sparc/niagara7.md (n7_array): Likewise.
25280 * config/sparc/niagara4.md (n4_array): Likewise.
25281 * config/sparc/niagara2.md (niag2_vis): Likewise.
25282 (niag3_vis): Likewise.
25283 * config/sparc/niagara.md (niag_vis): Likewise.
25285 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25287 * ipa-comdats.c: Remove optimize check from gate.
25288 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25289 for functions not optimized.
25290 (ipa_fn_summary_read): Skip optimize check.
25291 (ipa_fn_summary_write): Likewise.
25292 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25294 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25296 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25297 (check_callers): Give up on caller which is not optimized.
25298 (inline_small_functions): Likewise.
25299 (ipa_inline): Do not give up when not optimizing.
25300 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25301 away unoptimizes cdtors.
25302 (whole_program_function_and_variable_visibility): Do
25303 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25304 * ipa.c (process_references): Do not check optimize.
25305 (symbol_table::remove_unreachable_nodes): Update optimize check.
25306 (set_writeonly_bit): Update optimize check.
25307 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25308 (pass_ipa_single_use::gate): Remove.
25310 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25312 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25313 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25314 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25315 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25316 permute_load, permute_store, adjust_extract, adjust_splat,
25317 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25318 replace_swap_with_copy, dump_swap_insn_table,
25319 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25320 recombine_lvx_pattern, recombine_stvx_pattern,
25321 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25322 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25323 to file rs6000-p8swap.c.
25324 * config/rs6000/rs6000-p8swap.c: New file.
25325 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25326 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25327 and rs6000*-*-* targets.
25329 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25331 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25333 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25335 * lto-wrapper.c (merge_and_complain): Do not merge
25336 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25337 fsigned_zeros, ftrapping_math, fwrapv.
25338 (append_compiler_options): Do not track these options.
25339 (append_linker_options): Likewie
25341 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25343 * cgraphunit.c (cgraph_node::finalize_function): When
25344 !flag_toplevel_reorde set no_reorder flag.
25345 (varpool_node::finalize_decl): Likewise.
25346 (symbol_table::compile): Drop no toplevel reorder path.
25348 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25350 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25351 edges; zero probability is not better than uninitialized.
25353 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25355 * asan.h (asan_sanitize_allocas_p): Declare.
25356 * asan.c (asan_sanitize_allocas_p): New function.
25357 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25358 (handle_builtin_alloca): Likewise.
25359 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25360 if !asan_sanitize_allocas_p.
25361 * params.def (asan-instrument-allocas): Add new option.
25362 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25363 * opts.c (common_handle_option): Disable allocas sanitization for
25366 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25368 * asan.c: Include gimple-fold.h.
25369 (get_last_alloca_addr): New function.
25370 (handle_builtin_stackrestore): Likewise.
25371 (handle_builtin_alloca): Likewise.
25372 (asan_emit_allocas_unpoison): Likewise.
25373 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25374 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25375 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25376 (instrument_builtin_call): Pass gimple iterator to
25377 get_mem_refs_of_builtin_call.
25378 (last_alloca_addr): New global.
25379 * asan.h (asan_emit_allocas_unpoison): Declare.
25380 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25381 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25382 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25383 if function calls alloca.
25384 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25385 * gimple-fold.h (replace_call_with_value): Declare.
25386 * internal-fn.c: Include asan.h.
25387 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25388 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25390 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25392 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25393 (C_SELFTEST_FLAGS): New.
25394 (CPP_SELFTEST_FLAGS): New.
25395 (SELFTEST_DEPS): New, from deps of s-selftest.
25396 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25397 (CPP_SELFTEST_DEPS): New.
25398 (selftest): Add dependency on s-selftest-c++.
25399 (s-selftest): Rename to...
25400 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25401 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25402 than SELFTEST_FLAGS.
25403 (selftest-gdb): Rename to...
25404 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25406 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25407 (selftest-valgrind): Rename to...
25408 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25410 (selftest-valgrind): Reintroduce as an alias for
25411 selftest-c-valgrind.
25412 (s-selftest-c++): New.
25413 (selftest-c++-gdb): New.
25414 (selftest-c++-valgrind): New.
25416 2017-07-06 Olivier Hainque <hainque@adacore.com>
25418 * gcc.c (process_command): When deciding if undefined variables
25419 should be ignored when processing specs, accept "gcc -v" as well.
25421 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25423 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25424 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25426 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25428 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25429 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25430 * config/arm/arm-tables.opt: Regenerate.
25431 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25433 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25435 2017-07-06 Carl Love <cel@us.ibm.com>
25437 * ChangeLog: Clean up from mid air collision
25439 2017-07-06 Carl Love <cel@us.ibm.com>
25441 * config/rs6000/rs6000-c.c: Add support for built-in functions
25442 vector signed int vec_subc (vector signed int, vector signed int);
25443 vector signed __int128 vec_subc (vector signed __int128,
25444 vector signed __int128);
25445 vector unsigned __int128 vec_subc (vector unsigned __int128,
25446 vector unsigned __int128);
25447 vector signed int vec_sube (vector signed int, vector signed int,
25448 vector signed int);
25449 vector unsigned int vec_sube (vector unsigned int,
25450 vector unsigned int,
25451 vector unsigned int);
25452 vector signed __int128 vec_sube (vector signed __int128,
25453 vector signed __int128,
25454 vector signed__int128);
25455 vector unsigned __int128 vec_sube (vector unsigned __int128,
25456 vector unsigned __int128,
25457 vector unsigned __int128);
25458 vector signed int vec_subec (vector signed int, vector signed int,
25459 vector signed int);
25460 vector unsigned int vec_subec (vector unsigned int,
25461 vector unsigned int,
25462 vector unsigned int);
25463 vector signed __int128 vec_subec (vector signed __int128,
25464 vector signed __int128,
25465 vector signed__int128);
25466 vector unsigned __int128 vec_subec (vector unsigned __int128,
25467 vector unsigned __int128,
25468 vector unsigned __int128);
25469 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25470 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25471 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25472 BU_ALTIVEC_OVERLOAD_X definitions.
25473 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25474 * doc/extend.texi: Update the built-in documentation file for the new
25475 built-in functions.
25477 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25480 * diagnostic-show-locus.c (layout::layout): Use start and finish
25481 spelling location for the start and finish of each range.
25482 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25483 Add unused aspect param.
25484 * input.c (expand_location_1): Add "aspect" param, and use it
25485 to access the correct part of the location.
25486 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25488 (expand_location_to_spelling_point): Likewise.
25489 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25490 param, and pass it to expand_location_1.
25492 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25494 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25495 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25496 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25497 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25498 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25499 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25500 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25501 _mm_maskz_getmant_ss): New intrinsics.
25502 (__builtin_ia32_getexpss128_mask): Changed to ...
25503 __builtin_ia32_getexpss128_round ... this.
25504 (__builtin_ia32_getexpsd128_mask): Changed to ...
25505 __builtin_ia32_getexpsd128_round ... this.
25506 * config/i386/i386-builtin-types.def
25507 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25508 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25509 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25510 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25511 __builtin_ia32_getmantss_mask_round): New builtins.
25512 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25513 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25514 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25515 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25516 * config/i386/sse.md
25517 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25518 avx512f_sgetexp<mode><mask_scalar_name>
25519 <round_saeonly_scalar_name> ... this.
25520 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25521 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25522 vgetexp<ssescalarmodesuffix>
25523 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25524 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25525 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25526 avx512f_vgetmant<mode><mask_scalar_name>
25527 <round_saeonly_scalar_name> ... this.
25528 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25529 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25530 vgetmant<ssescalarmodesuffix>
25531 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25532 %0<mask_scalar_operand4>, %1, %2
25533 <round_saeonly_scalar_mask_op4>, %3} ... this.
25534 * config/i386/subst.md (mask_scalar_operand4,
25535 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25536 round_saeonly_scalar_nimm_predicate): New subst attributes.
25538 2017-07-06 Julia Koval <julia.koval@intel.com>
25540 * config/i386/i386.c (ix86_erase_embedded_rounding):
25541 Remove code for old rounding pattern.
25543 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25545 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25547 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25549 * doc/sourcebuild.texi (Test Directives, Variants of
25550 dg-require-support): Add documentation for dg-require-stack-check.
25552 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25554 * config/i386/subst.md (mask_scalar, round_scalar,
25555 round_saeonly_scalar): New meta-templates.
25556 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25557 round_scalar_mask_operand3, round_scalar_mask_op3,
25558 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25559 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25560 round_saeonly_scalar_constraint,
25561 round_saeonly_scalar_prefix): New subst attribute.
25562 * config/i386/sse.md
25563 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25564 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25565 <round_scalar_name> ... this.
25566 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25567 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25568 <round_scalar_name> ... this.
25569 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25570 <sse>_vm<code><mode>3<mask_scalar_name>
25571 <round_saeonly_scalar_name> ... this.
25572 (v<plusminus_mnemonic><ssescalarmodesuffix>
25573 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25574 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25575 v<plusminus_mnemonic><ssescalarmodesuffix>
25576 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25577 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25578 (v<multdiv_mnemonic><ssescalarmodesuffix>
25579 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25580 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25581 v<multdiv_mnemonic><ssescalarmodesuffix>
25582 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25583 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25584 (v<maxmin_float><ssescalarmodesuffix>
25585 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25586 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25587 v<maxmin_float><ssescalarmodesuffix>
25588 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25589 %0<mask_scalar_operand3>, %1, %<iptr>2
25590 <round_saeonly_scalar_mask_op3>} ... this.
25592 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25594 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25595 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25597 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25598 Alan Hayward <alan.hayward@arm.com>
25599 David Sherwood <david.sherwood@arm.com>
25601 * combine.c (simplify_if_then_else): Remove "enum" before
25603 * compare-elim.c (can_eliminate_compare): Likewise.
25604 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25606 (aarch64_lookup_simd_builtin_type): Likewise.
25607 (aarch64_simd_builtin_type): Likewise.
25608 (aarch64_init_simd_builtin_types): Likewise.
25609 (aarch64_simd_expand_args): Likewise.
25610 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25612 (aarch64_reverse_mask): Likewise.
25613 (aarch64_simd_emit_reg_reg_move): Likewise.
25614 (aarch64_gen_adjusted_ldpstp): Likewise.
25615 (aarch64_ccmp_mode_to_code): Likewise.
25616 (aarch64_operands_ok_for_ldpstp): Likewise.
25617 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25618 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25620 (aarch64_min_divisions_for_recip_mul): Likewise.
25621 (aarch64_reassociation_width): Likewise.
25622 (aarch64_get_condition_code_1): Likewise.
25623 (aarch64_simd_emit_reg_reg_move): Likewise.
25624 (aarch64_simd_attr_length_rglist): Likewise.
25625 (aarch64_reverse_mask): Likewise.
25626 (aarch64_operands_ok_for_ldpstp): Likewise.
25627 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25628 (aarch64_gen_adjusted_ldpstp): Likewise.
25629 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25631 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25632 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25633 (arm_lookup_simd_builtin_type): Likewise.
25634 (arm_simd_builtin_type): Likewise.
25635 (arm_init_simd_builtin_types): Likewise.
25636 (arm_expand_builtin_args): Likewise.
25637 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25638 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25639 (ft32_setup_incoming_varargs): Likewise.
25640 (ft32_function_arg): Likewise.
25641 (ft32_function_arg_advance): Likewise.
25642 (ft32_pass_by_reference): Likewise.
25643 (ft32_arg_partial_bytes): Likewise.
25644 (ft32_valid_pointer_mode): Likewise.
25645 (ft32_addr_space_pointer_mode): Likewise.
25646 (ft32_addr_space_legitimate_address_p): Likewise.
25647 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25649 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25650 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25651 (iamcu_alignment): Likewise.
25652 (canonicalize_vector_int_perm): Likewise.
25653 (ix86_noce_conversion_profitable_p): Likewise.
25654 (ix86_mpx_bound_mode): Likewise.
25655 (ix86_operands_ok_for_move_multiple): Likewise.
25656 * config/microblaze/microblaze-protos.h
25657 (microblaze_expand_conditional_branch_reg): Likewise.
25658 * config/microblaze/microblaze.c
25659 (microblaze_expand_conditional_branch_reg): Likewise.
25660 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25662 (rs6000_reassociation_width): Likewise.
25663 (rs6000_invalid_binary_op): Likewise.
25664 (fusion_p9_p): Likewise.
25665 (emit_fusion_p9_load): Likewise.
25666 (emit_fusion_p9_store): Likewise.
25667 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25669 (riscv_hard_regno_mode_ok_p): Likewise.
25670 (riscv_address_insns): Likewise.
25671 (riscv_split_symbol): Likewise.
25672 (riscv_legitimize_move): Likewise.
25673 (riscv_function_value): Likewise.
25674 (riscv_hard_regno_nregs): Likewise.
25675 (riscv_expand_builtin): Likewise.
25676 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25677 (riscv_build_integer): Likewise.
25678 (riscv_split_integer): Likewise.
25679 (riscv_legitimate_constant_p): Likewise.
25680 (riscv_cannot_force_const_mem): Likewise.
25681 (riscv_regno_mode_ok_for_base_p): Likewise.
25682 (riscv_valid_base_register_p): Likewise.
25683 (riscv_valid_offset_p): Likewise.
25684 (riscv_valid_lo_sum_p): Likewise.
25685 (riscv_classify_address): Likewise.
25686 (riscv_legitimate_address_p): Likewise.
25687 (riscv_address_insns): Likewise.
25688 (riscv_load_store_insns): Likewise.
25689 (riscv_force_binary): Likewise.
25690 (riscv_split_symbol): Likewise.
25691 (riscv_force_address): Likewise.
25692 (riscv_legitimize_address): Likewise.
25693 (riscv_move_integer): Likewise.
25694 (riscv_legitimize_const_move): Likewise.
25695 (riscv_legitimize_move): Likewise.
25696 (riscv_address_cost): Likewise.
25697 (riscv_subword): Likewise.
25698 (riscv_output_move): Likewise.
25699 (riscv_canonicalize_int_order_test): Likewise.
25700 (riscv_emit_int_order_test): Likewise.
25701 (riscv_function_arg_boundary): Likewise.
25702 (riscv_pass_mode_in_fpr_p): Likewise.
25703 (riscv_pass_fpr_single): Likewise.
25704 (riscv_pass_fpr_pair): Likewise.
25705 (riscv_get_arg_info): Likewise.
25706 (riscv_function_arg): Likewise.
25707 (riscv_function_arg_advance): Likewise.
25708 (riscv_arg_partial_bytes): Likewise.
25709 (riscv_function_value): Likewise.
25710 (riscv_pass_by_reference): Likewise.
25711 (riscv_setup_incoming_varargs): Likewise.
25712 (riscv_print_operand): Likewise.
25713 (riscv_elf_select_rtx_section): Likewise.
25714 (riscv_save_restore_reg): Likewise.
25715 (riscv_for_each_saved_reg): Likewise.
25716 (riscv_register_move_cost): Likewise.
25717 (riscv_hard_regno_mode_ok_p): Likewise.
25718 (riscv_hard_regno_nregs): Likewise.
25719 (riscv_class_max_nregs): Likewise.
25720 (riscv_memory_move_cost): Likewise.
25721 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25722 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25723 (rl78_addr_space_address_mode): Likewise.
25724 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25726 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25727 (rs6000_reassociation_width): Likewise.
25728 (rs6000_invalid_binary_op): Likewise.
25729 (fusion_p9_p): Likewise.
25730 (emit_fusion_p9_load): Likewise.
25731 (emit_fusion_p9_store): Likewise.
25732 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25733 (ok_for_simple_move_operands): Likewise.
25734 (ok_for_simple_move_strict_operands): Likewise.
25735 (ok_for_simple_arith_logic_operands): Likewise.
25736 (visium_legitimize_reload_address): Likewise.
25737 (visium_select_cc_mode): Likewise.
25738 (output_cbranch): Likewise.
25739 (visium_split_double_move): Likewise.
25740 (visium_expand_copysign): Likewise.
25741 (visium_expand_int_cstore): Likewise.
25742 (visium_expand_fp_cstore): Likewise.
25743 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25744 (visium_function_arg): Likewise.
25745 (visium_function_arg_advance): Likewise.
25746 (visium_libcall_value): Likewise.
25747 (visium_setup_incoming_varargs): Likewise.
25748 (visium_legitimate_constant_p): Likewise.
25749 (visium_legitimate_address_p): Likewise.
25750 (visium_legitimize_address): Likewise.
25751 (visium_secondary_reload): Likewise.
25752 (visium_register_move_cost): Likewise.
25753 (visium_memory_move_cost): Likewise.
25754 (prepare_move_operands): Likewise.
25755 (ok_for_simple_move_operands): Likewise.
25756 (ok_for_simple_move_strict_operands): Likewise.
25757 (ok_for_simple_arith_logic_operands): Likewise.
25758 (visium_function_value_1): Likewise.
25759 (rtx_ok_for_offset_p): Likewise.
25760 (visium_legitimize_reload_address): Likewise.
25761 (visium_split_double_move): Likewise.
25762 (visium_expand_copysign): Likewise.
25763 (visium_expand_int_cstore): Likewise.
25764 (visium_expand_fp_cstore): Likewise.
25765 (visium_split_cstore): Likewise.
25766 (visium_select_cc_mode): Likewise.
25767 (visium_split_cbranch): Likewise.
25768 (output_cbranch): Likewise.
25769 (visium_print_operand_address): Likewise.
25770 * expmed.c (flip_storage_order): Likewise.
25771 * expmed.h (emit_cstore): Likewise.
25772 (flip_storage_order): Likewise.
25773 * genrecog.c (validate_pattern): Likewise.
25774 * hsa-gen.c (gen_hsa_addr): Likewise.
25775 * internal-fn.c (expand_arith_overflow): Likewise.
25776 * ira-color.c (allocno_copy_cost_saving): Likewise.
25777 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25778 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25779 (process_invariant_for_inheritance): Likewise.
25780 * lra-eliminations.c (move_plus_up): Likewise.
25781 * omp-low.c (lower_oacc_reductions): Likewise.
25782 * simplify-rtx.c (simplify_subreg): Likewise.
25783 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25784 (TARGET_CHKP_BOUND_MODE): Likewise..
25785 * targhooks.c (default_chkp_bound_mode): Likewise.
25786 (default_setup_incoming_vararg_bounds): Likewise.
25787 * targhooks.h (default_chkp_bound_mode): Likewise.
25788 (default_setup_incoming_vararg_bounds): Likewise.
25789 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25790 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25791 (have_whole_vector_shift): Likewise.
25792 * tree-vect-stmts.c (vectorizable_load): Likewise.
25793 * doc/tm.texi: Regenerate.
25795 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25797 Graceful degrade if Binutils PR21472 is not available.
25800 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25801 .rodata in flash test fails.
25802 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25803 * confgure: Regenerate.
25804 * config.in: Regenerate.
25805 * config/avr/avr.c (avr_asm_named_section)
25806 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25807 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25808 (avr_asm_init_sections): Same.
25810 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25812 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25813 (fma<VH:mode>4_intrinsic): Likewise.
25814 (*fmsub<VCVTF:mode>4): Likewise.
25815 (*fmsub<VH:mode>4_intrinsic): Likewise.
25817 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25820 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25821 Don't depend on "optimize > 0".
25822 (out_movhi_r_mr, out_movqi_mr_r): Same.
25823 (out_movhi_mr_r, out_movqi_r_mr): Same.
25824 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25825 io_address_operand on "optimize > 0".
25827 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25829 * tree-loop-distribution.c: Add general explanantion on the pass.
25830 (generate_loops_for_partition): Mark distributed loop.
25831 (pg_add_dependence_edges): New parameter. Handle alias data
25832 dependence specially and record it in the parameter if asked.
25833 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25834 (init_partition_graph_vertices, add_partition_graph_edge): New.
25835 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25836 (free_partition_graph_vdata, build_partition_graph): New.
25837 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25838 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25839 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25840 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25841 (version_for_distribution_p, finalize_partitions): New.
25842 (distribute_loop): Handle alias data dependence specially. Factor
25843 out loop fusion code as functions and call these functions.
25845 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25847 * tree-loop-distribution.c (classify_partition): New parameter and
25848 better handle reduction statement.
25849 (rdg_build_partitions): Revise comment.
25850 (distribute_loop): Compute statements in all partitions and pass it
25851 to classify_partition.
25853 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25855 * tree-loop-distribution.c (enum partition_type): New.
25856 (struct partition): New field type.
25857 (partition_merge_into): Add parameter. Update partition type.
25858 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25859 (build_rdg_partition_for_vertex): Compute partition type.
25860 (rdg_build_partitions): Dump partition type.
25861 (distribute_loop): Update calls to partition_merge_into.
25863 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25865 * tree-loop-distribution.c (struct ddr_hasher): New.
25866 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25868 (classify_partition): Call get_data_dependence.
25869 (pg_add_dependence_edges): Ditto.
25870 (distribute_loop): Release data dependence hash table.
25872 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25874 * tree-loop-distribution.c (ref_base_address): Delete.
25875 (similar_memory_accesses): Rename ...
25876 (share_memory_accesses): ... to this. Check if partitions access
25877 the same memory reference.
25878 (distribute_loop): Call share_memory_accesses.
25880 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25882 * tree-loop-distribution.c (struct partition): New field recording
25883 its data reference.
25884 (partition_alloc, partition_free): Init and release data refs.
25885 (partition_merge_into): Merge data refs.
25886 (build_rdg_partition_for_vertex): Collect data refs for partition.
25887 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25889 (distribute_loop): Remve data refs from vertice data of partition
25892 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25894 * tree-loop-distribution.c (params.h): Include header file.
25895 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25896 (datarefs_vec): New global var.
25897 (create_rdg_vertices): Use datarefs_vec directly.
25898 (free_rdg): Don't free data references.
25899 (build_rdg): Update use. Don't free data references.
25900 (distribute_loop): Compute global variable for data references.
25901 Bail out if there are too many data references.
25903 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25905 * tree-loop-distribution.c (loop_nest): New global var.
25906 (build_rdg): Use loop directly, rather than loop nest.
25907 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25909 (distribute_loop): Compute global variable loop nest. Update use.
25911 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25913 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25914 (partition_merge_into): New parameter. Dump reason for fusion.
25915 (distribute_loop): Update use of partition_merge_into.
25917 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25919 * tree-loop-distribution.c (bb_top_order_index): New.
25920 (bb_top_order_index_size, bb_top_order_cmp): New.
25921 (stmts_from_loop): Use topological order.
25922 (pass_loop_distribution::execute): Compute and release topological
25923 order for basic blocks.
25925 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25927 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25930 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25932 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25933 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25934 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25935 * internal-fn.def (LOOP_DIST_ALIAS): New.
25936 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25937 (fold_loop_internal_call): ... this.
25938 (vect_loop_dist_alias_call): New function.
25939 (set_uid_loop_bbs): Call fold_loop_internal_call.
25940 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25943 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25946 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25947 Require dead FLAGS_REG at the beginning of a peephole.
25949 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25952 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25953 arguments in the call to __builtin_ia32_sbb_u32.
25954 (_subborrow_u64): Swap _X and _Y arguments in the call to
25955 __builtin_ia32_sbb_u64.
25957 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25960 * tree-vrp.c (compare_assert_loc): Turn into a function template
25961 with stable template parameter. Only test if a->e is NULL,
25962 !a->e == !b->e has been verified already. Use e == NULL or
25963 e != NULL instead of e or ! e tests. If stable is true, don't use
25964 iterative_hash_expr, on the other side allow a or b or both NULL
25965 and sort the NULLs last.
25966 (process_assert_insertions): Sort using compare_assert_loc<false>
25967 instead of compare_assert_loc, later sort using
25968 compare_assert_loc<true> before calling process_assert_insertions_for
25969 in a loop. Use break instead of continue once seen NULL pointer.
25971 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25973 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25974 Cortex-R7 and Cortex-R8 processors.
25976 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25978 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25979 uninitialized while src is not.
25981 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25983 * common/config/arm/arm-common.c: Adjust include path for
25985 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25986 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25987 (arm-cpu-data.h): Likewise.
25988 (arm-cpu-cdata.h): Likewise.
25989 * config/arm/arm-cpu.h: Delete.
25990 * config/arm/arm-cpu-cdata.h: Delete.
25991 * config/arm/arm-cpu-data.h: Delete.
25993 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25995 * config/arm/arm-cpus.in (cortex-a55): New.
25996 (cortex-a75): Likewise.
25997 (cortex-a75.cortex-a55): Likewise.
25998 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
26000 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
26001 * config/arm/arm-cpu-cdata.h: Regenerate.
26002 * config/arm/arm-cpu-data.h: Regenerate.
26003 * config/arm/arm-cpu.h: Regenerate.
26004 * config/arm/arm-tables.opt: Regenerate.
26005 * config/arm/arm-tune.md: Regenerate.
26007 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
26009 * haifa-sched.c (sched_create_recovery_edges): Update profile.
26011 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
26013 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
26016 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
26018 PR tree-optimization/81292
26019 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
26020 full_string_p, also call adjust_related_strinfos if the adjustment
26021 is simple, otherwise invalidate related strinfos.
26023 2017-07-04 Martin Liska <mliska@suse.cz>
26026 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
26027 newly created variable as DECL_IGNORED_P.
26029 2017-07-04 Martin Liska <mliska@suse.cz>
26032 * ipa-inline.c (inline_small_functions):
26033 Use xstrdup_for_dump.
26035 2017-07-04 Tom de Vries <tom@codesourcery.com>
26037 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
26039 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
26042 * config/darwin.c (darwin_function_switched_text_sections):
26045 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
26047 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
26049 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26051 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
26053 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26055 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
26056 min_profitable_iters, and th as inclusive lower bounds.
26057 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
26058 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
26059 for min_profitable_iters and min_profitable_estimate.
26060 (vect_transform_loop): Treat th as an inclusive lower bound.
26061 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26063 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
26066 * config/darwin.c (darwin_function_switched_text_sections):
26067 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
26068 in two pieces, and suppress the use of buf.
26070 2017-07-03 Nathan Sidwell <nathan@acm.org>
26072 * hash-table.h (hash_table_mod1): Fix indentation.
26074 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26076 PR middle-end/81290
26077 * predict.c (force_edge_cold): Be more careful about propagation
26079 * profile-count.h (profile_probability::guessed,
26080 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
26082 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
26084 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
26086 * doc/invoke.texi (rcpc architecture extension): Document it.
26088 2017-07-03 Richard Biener <rguenther@suse.de>
26090 PR tree-optimization/60510
26091 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
26092 the scalar reduction PHI and use it.
26093 (vectorizable_reduction): Properly guard the single_defuse_cycle
26094 path for non-SLP reduction chains where we cannot use it.
26095 Rework reduc_def/index and vector type deduction. Rework
26096 vector operand gathering during reduction op code-gen.
26097 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
26098 chains dissolve the chain and leave it to non-SLP reduction
26101 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26103 * tree-data-ref.h (dr_alignment): Declare.
26104 * tree-data-ref.c (dr_alignment): New function.
26105 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
26106 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
26108 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
26110 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26112 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
26113 and base_misalignment fields.
26114 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
26115 * tree-data-ref.c: Include builtins.h.
26116 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
26117 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
26118 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
26119 * tree-vect-data-refs.c: Include tree-cfg.h.
26120 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
26121 fields instead of calculating an alignment here.
26122 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
26123 innermost_loop_behavior fields.
26125 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26127 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
26129 (DR_STEP_ALIGNMENT): New macro.
26130 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
26131 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
26132 (create_data_ref): Print it.
26133 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
26134 to tell whether the step preserves vector (mis)alignment.
26135 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26136 Move the check for an integer step and generalise to all INTEGER_CST.
26137 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
26138 Print the outer step alignment.
26140 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26142 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
26143 with offset_alignment.
26144 (DR_ALIGNED_TO): Delete.
26145 (DR_OFFSET_ALIGNMENT): New macro.
26146 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
26147 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
26148 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
26149 (create_data_ref): Likewise.
26150 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26151 (vect_analyze_data_refs): Likewise.
26152 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
26153 creating dummy innermost behavior.
26155 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26157 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
26158 with a "innermost_loop_behavior *" and refeence tree.
26159 * tree-data-ref.c (dr_analyze_innermost): Likewise.
26160 (create_data_ref): Update call accordingly.
26161 * tree-predcom.c (find_looparound_phi): Likewise.
26163 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26165 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
26166 fields with dr_wrt_vec_loop.
26167 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
26168 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
26169 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
26170 (vect_dr_behavior): New function.
26171 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26172 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
26173 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
26174 track whether the step preserves the misalignment.
26175 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26176 Use vect_dr_behavior.
26177 (vect_setup_realignment): Update call accordingly.
26178 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
26179 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
26180 call to vect_create_addr_base_for_vector_ref.
26181 (vect_create_cond_for_align_checks): Likewise.
26182 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
26183 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
26184 (vect_recog_mask_conversion_pattern): Likewise.
26185 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
26186 (new_stmt_vec_info): Remove redundant zeroing.
26188 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26190 * common/config/arm/arm-common.c (arm_be8_option): New function.
26191 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
26192 (ISA_ARMv6): Add isa_bit_be8.
26193 * config/arm/arm.h (arm_be8_option): Add prototype.
26194 (BE8_SPEC_FUNCTION): New define.
26195 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
26196 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
26197 (mlittle-endian): Similarly.
26198 (mbe8, mbe32): New options.
26199 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
26200 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
26202 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26204 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26206 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26208 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26209 (cleanup_tree_cfg_bb): Use it.
26210 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26212 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26214 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26217 * loop-doloop.c (add_test): Update profile.
26219 2017-07-03 Martin Liska <mliska@suse.cz>
26222 * sanopt.c (rewrite_usage_of_param): New function.
26223 (sanitize_rewrite_addressable_params): Likewise.
26224 (pass_sanopt::execute): Call rewrite_usage_of_param.
26226 2017-07-03 Richard Biener <rguenther@suse.de>
26228 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26229 back to using VIEW_CONVERT_EXPR.
26231 2017-07-03 Martin Liska <mliska@suse.cz>
26234 * doc/extend.texi: Document when a resolver function is
26235 generated for target_clones.
26237 2017-07-03 Martin Liska <mliska@suse.cz>
26239 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26240 and shadow memory of auto variables which are subject of
26241 use-after-scope sanitization.
26242 (asan_expand_mark_ifn): Add do set only when is_poison.
26244 2016-07-03 Richard Biener <rguenther@suse.de>
26246 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26248 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26249 (vectorizable_reduction): Perform reduction PHI creation when
26250 visiting a reduction PHI and adjust and simplify code generation
26251 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26252 (vect_transform_loop): Visit reduction PHIs.
26253 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26254 defs into the SLP tree.
26255 (vect_build_slp_tree): Reduction defs terminate the recursion.
26256 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26258 (vect_get_vec_defs_for_stmt_copy): Export.
26259 (vect_get_vec_defs): Likewise.
26260 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26262 (vect_get_vec_defs_for_stmt_copy): Declare.
26263 (vect_get_vec_defs): Likewise.
26265 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26267 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26268 parameter with a "loop" parameter and use it instead of the
26269 loop containing DR_STMT. Don't check simple_iv when doing
26270 BB analysis. Describe the two analysis modes in the comment.
26272 2017-07-03 Tom de Vries <tom@codesourcery.com>
26274 PR tree-optimization/69468
26275 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26276 (find_same_succ_bb): Handle ignore_edge_flags.
26278 2017-07-03 Tom de Vries <tom@codesourcery.com>
26280 PR tree-optimization/81192
26281 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26283 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26285 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26287 2017-07-03 Tom de Vries <tom@codesourcery.com>
26289 PR tree-optimization/81192
26290 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26291 BB_SAME_SUCC (bb) == NULL.
26293 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26295 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26298 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26300 * dumpfile.c: Include profile-count.h
26301 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26303 (insert_cond_bb): Update profile.
26304 * tree-cfg.h (insert_cond_bb): Update prototype.
26305 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26306 * tree-dump.c: Do not include tree-cfg.
26308 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26310 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26312 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26314 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26317 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26319 * tree-complex.c (expand_complex_div_wide): update profile.
26321 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26322 Alan Hayward <alan.hayward@arm.com>
26323 David Sherwood <david.sherwood@arm.com>
26325 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26326 (CORETYPES_H): New define.
26327 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26328 (insn-modes-inline.h, s-modes-inline-h): New rules.
26329 (generated_files): Add insn-modes-inline.h.
26330 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26331 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26332 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26333 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26334 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26335 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26336 (build/gencodes.o, build/genconditions.o): Likewise.
26337 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26338 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26339 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26340 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26341 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26342 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26343 * coretypes.h: Include everything up to real.h for generators.
26344 Include insn-modes.h first. Include wide-int-print.h after
26345 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26346 * machmode.h: Don't include insn-modes.h here.
26347 * function-tests.c: Remove includes of signop.h, machmode.h,
26348 double-int.h and wide-int.h.
26350 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26352 * optc-save-gen.awk: Likewise.
26353 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26354 * godump.c: Remove include of wide-int-print.h.
26355 * pretty-print.h: Likewise.
26356 * wide-int-print.cc: Likewise.
26357 * wide-int.cc: Likewise.
26358 * hash-map-tests.c: Remove include of signop.h.
26359 * hash-set-tests.c: Likewise.
26360 * rtl-tests.c: Likewise.
26361 * mkconfig.sh: Remove include of machmode.h.
26362 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26364 (emit_insn_modes_inline_h): ...this new function. Emit the code
26365 into an insn-modes-inline.h header file, adding appropriate
26366 include guards and end comments.
26367 (emit_insn_modes_c_header): Remove include of machmode.h.
26368 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26370 (main): Handle -i flag and call emit_insn_modes_inline_h when
26373 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26375 * tree-ssa-strlen.c (strinfo): Rename the length field to
26376 nonzero_chars. Add a full_string_p field.
26377 (compare_nonzero_chars, zero_length_string_p): New functions.
26378 (get_addr_stridx): Add an offset_out parameter.
26379 Use compare_nonzero_chars.
26380 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26381 (new_strinfo): Update after above changes to strinfo.
26382 (set_endptr_and_length): Set full_string_p.
26383 (get_string_length): Update after above changes to strinfo.
26384 (unshare_strinfo): Update call to new_strinfo.
26385 (maybe_invalidate): Likewise.
26386 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26387 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26388 as a uhwi instead of an shwi. Update after above changes to
26389 strinfo and new_strinfo.
26390 (zero_length_string): Assert that chainsi contains full strings.
26391 Use zero_length_string_p. Update call to new_strinfo.
26392 (adjust_related_strinfos): Update after above changes to strinfo.
26393 Copy full_string_p from origsi.
26394 (adjust_last_stmt): Use zero_length_string_p.
26395 (handle_builtin_strlen): Update after above changes to strinfo and
26396 new_strinfo. Install the lhs as the string length if the previous
26397 entry didn't describe a full string.
26398 (handle_builtin_strchr): Update after above changes to strinfo
26400 (handle_builtin_strcpy): Likewise.
26401 (handle_builtin_strcat): Likewise.
26402 (handle_builtin_malloc): Likewise.
26403 (handle_pointer_plus): Likewise.
26404 (handle_builtin_memcpy): Likewise. Track nonzero characters
26405 that aren't necessarily followed by a nul terminator.
26406 (handle_char_store): Likewise.
26408 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26410 PR tree-optimization/80769
26411 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26412 for malloc and calloc. Document the new invariant that all related
26413 strinfos have delayed lengths or none do.
26414 (verify_related_strinfos): Move earlier in file.
26415 (set_endptr_and_length): New function, split out from...
26416 (get_string_length): ...here. Also set the lengths of related
26418 (zero_length_string): Assert that chainsi has known (rather than
26420 (adjust_related_strinfos): Likewise.
26422 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26424 PR tree-optimization/81136
26425 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26426 assert that two references with the same misalignment have the same
26427 compile-time misalignment if those compile-time misalignments
26430 2017-07-01 Andi Kleen <ak@linux.intel.com>
26432 * print-tree.c (print_node): Print all attributes.
26434 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26436 * cfg.c (scale_bbs_frequencies): New function.
26437 * cfg.h (scale_bbs_frequencies): Declare it.
26438 * cfgloopanal.c (single_likely_exit): Cleanup.
26439 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26441 (scale_loop_profile): Likewise.
26442 (loop_version): Likewise.
26443 (create_empty_loop_on_edge): Update.
26444 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26445 scale_loop_frequencies, scale_loop_profile, loopify,
26446 loop_version): Update prototypes.
26447 * modulo-sched.c (sms_schedule): Update.
26448 * predict.c (unlikely_executed_edge_p): Also check probability.
26449 (probably_never_executed_edge_p): Fix typo.
26450 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26451 * tree-parloops.c (gen_parallel_loop): Update.
26452 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26453 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26454 * tree-ssa-loop-split.c (split_loop): Update.
26455 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26456 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26457 (vect_loop_versioning): Update.
26458 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26460 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26462 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26464 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26466 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26467 to keep profile consistent.
26469 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26471 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26472 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26473 * profile-count.h (max_safe_multiplier): Make unsigned.
26474 (profile_count::guessed_zero): New.
26476 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26478 * bb-reorder.c (fix_up_crossing_landing_pad,
26479 fix_crossing_conditional_branches): Use make_single_succ_edge
26480 to keep profile consistent.
26482 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26484 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26487 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26490 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26491 the right scopes, make sure cond_jump isn't preserved between multiple
26492 iterations. Search for fallthru edge whenever there are 3+ edges and
26493 use find_fallthru_edge for it.
26495 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26497 Patch by Alexander Monakov <amonakov@ispras.ru>
26498 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26499 probabilities consistently.
26501 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26503 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26504 emit_cmp_and_jump_insns.
26506 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26509 * tree-inline.c (expand_call_inline): Combine profile statuses.
26511 2017-06-30 Andrew Pinski <apinski@cavium.com>
26513 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26514 fold_stmt returned true.
26516 2017-06-30 Nathan Sidwell <nathan@acm.org>
26518 * ggc.h (empty_string): Delete.
26519 * cfgexpand.c (expand_asm_stmt): Use plain "".
26520 * optabs.c (expand_asm_memory_barrier): Likewise.
26521 * stringpool.c (empty_string): Delete.
26522 (digit_vector, digit_string): Delete.
26523 (ggc_alloc_string): Use plain "", don't optimize single digit
26524 strings. Use ggc_alloc_atomic.
26526 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26528 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26529 comparison set and one other set, use the cost of the non-comparison
26532 2017-06-30 Nathan Sidwell <nathan@acm.org>
26534 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26537 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26539 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26540 loops. Remove now unneeded calls to gimple_switch_set_label() that
26541 just set removed labels to NULL_TREE.
26543 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26545 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26546 (set_range_info): ...here. Only call set_range_info_raw if domain
26548 (set_nonzero_bits): Call set_range_info_raw.
26549 * tree-ssanames.h (set_range_info_raw): New.
26551 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26554 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26555 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26556 of nonimmediate_operand and <store_mask_constraint> instead of m
26557 for the input operand. For V8FI iterator, always split if input
26558 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26559 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26560 <store_mask_predicate> instead of register_operand and
26561 <store_mask_constraint> instead of v for the input operand. Make
26562 sure both operands aren't MEMs for if not <mask_applied>.
26564 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26566 * lto-wrapper.c (copy_file) Close both file descriptors before
26569 2017-06-30 Martin Liska <mliska@suse.cz>
26572 * multiple_target.c (create_dispatcher_calls): Make ifunc
26573 also for function that don't have calls or are not referenced.
26575 2017-06-30 Richard Biener <rguenther@suse.de>
26577 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26578 analyze the first scalar stmt. Move vector type computation
26579 for the BB case here from ...
26580 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26581 live operation processing in the SLP case properly.
26583 2017-06-30 Richard Biener <rguenther@suse.de>
26585 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26587 2017-06-30 Martin Liska <mliska@suse.cz>
26590 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26591 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26593 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26595 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26598 2017-06-30 Richard Biener <rguenther@suse.de>
26600 PR tree-optimization/81249
26601 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26602 condition reduction result to original scalar type.
26604 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26606 * profile-count.h (enum profile_quality): Fix typos and whitespace
26609 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26611 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26612 type for branch probabilities.
26614 2017-06-29 Julian Brown <julian@codesourcery.com>
26615 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26617 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26618 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26619 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26620 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26622 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26624 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26625 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26626 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26627 CC usage from generic code to here.
26628 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26629 CC usage into the target macros.
26631 2017-06-29 Maya Rashish <coypu@sdf.org>
26633 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26636 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26638 * arm/arm-builtins.c: Include profile-count.h
26639 * except.c (sjlj_emit_function_enter): Use
26640 profile_probability::unlikely.
26642 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26644 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26645 and tocrel_offset be pointer args rather than implicitly using
26647 (legitimate_constant_pool_address_p, rs6000_emit_move,
26648 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26649 tocrel_offset and use in toc_relative_expr_p call.
26650 (print_operand, print_operand_address): Use static tocrel_base_oac
26651 and tocrel_offset_oac.
26652 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26655 2017-06-29 Maya Rashish <coypu@sdf.org>
26657 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26659 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26661 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26662 objects, take into account only the alignment of 'op0' and 'mode1' if
26665 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26667 * ccmp.c (ccmp_tree_comparison_p): New function.
26668 (ccmp_candidate_p): Update to use above function.
26669 (get_compare_parts): New function.
26670 (expand_ccmp_next): Update to use new functions.
26671 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26673 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26674 take mode as argument.
26675 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26676 * expr.c (expand_expr_real_1): Pass mode as argument.
26678 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26680 * combine.c (combine_instructions): Print insns to dump_file, together
26683 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26685 * asan.c (asan_emit_stack_protection): Update.
26686 (create_cond_insert_point): Update.
26687 * auto-profile.c (afdo_propagate_circuit): Update.
26688 * basic-block.h (struct edge_def): Turn probability to
26689 profile_probability.
26690 (EDGE_FREQUENCY): Update.
26691 * bb-reorder.c (find_traces_1_round): Update.
26692 (better_edge_p): Update.
26693 (sanitize_hot_paths): Update.
26694 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26695 (make_single_succ_edge): Update.
26696 (check_bb_profile): Update.
26697 (dump_edge_info): Update.
26698 (update_bb_profile_for_threading): Update.
26699 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26700 probabilitycount to 0.
26701 * cfgbuild.c (compute_outgoing_frequencies): Update.
26702 * cfgcleanup.c (try_forward_edges): Update.
26703 (outgoing_edges_match): Update.
26704 (try_crossjump_to_edge): Update.
26705 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26706 (expand_gimple_tailcall): Update.
26707 (construct_init_block): Use make_single_succ_edge.
26708 (construct_exit_block): Use make_single_succ_edge.
26709 * cfghooks.c (verify_flow_info): Update.
26710 (redirect_edge_succ_nodup): Update.
26711 (split_edge): Update.
26712 (account_profile_record): Update.
26713 * cfgloopanal.c (single_likely_exit): Update.
26714 * cfgloopmanip.c (scale_loop_profile): Update.
26715 (set_zero_probability): Remove.
26716 (duplicate_loop_to_header_edge): Update.
26717 * cfgloopmanip.h (loop_version): Update prototype.
26718 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26719 (force_nonfallthru_and_redirect): Update.
26720 (update_br_prob_note): Update.
26721 (rtl_verify_edges): Update.
26722 (purge_dead_edges): Update.
26723 (rtl_lv_add_condition_to_bb): Update.
26724 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26725 * cgraphunit.c (init_lowered_empty_function): Update.
26726 (cgraph_node::expand_thunk): Update.
26727 * cilk-common.c: Include profile-count.h
26728 * dojump.c (inv): Remove.
26729 (jumpifnot): Update.
26730 (jumpifnot_1): Update.
26731 (do_jump_1): Update.
26733 (do_jump_by_parts_greater_rtx): Update.
26734 (do_compare_rtx_and_jump): Update.
26735 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26736 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26737 * dwarf2cfi.c: Include profile-count.h
26738 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26739 (sjlj_emit_dispatch_table): Likewise.
26740 * explow.c: Include profile-count.h
26741 * expmed.c (emit_store_flag_force): Update.
26742 (do_cmp_and_jump): Update.
26743 * expr.c (compare_by_pieces_d::generate): Update.
26744 (compare_by_pieces_d::finish_mode): Update.
26745 (emit_block_move_via_loop): Update.
26746 (store_expr_with_bounds): Update.
26747 (store_constructor): Update.
26748 (expand_expr_real_2): Update.
26749 (expand_expr_real_1): Update.
26750 * expr.h (try_casesi, try_tablejump): Update prototypes.
26751 * gimple-pretty-print.c (dump_probability): Update.
26752 (dump_profile): New.
26753 (dump_gimple_label): Update.
26754 (dump_gimple_bb_header): Update.
26755 * graph.c (draw_cfg_node_succ_edges): Update.
26756 * hsa-gen.c (convert_switch_statements): Update.
26757 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26758 (find_if_case_1): Update.
26759 (find_if_case_2): Update.
26760 * internal-fn.c (expand_arith_overflow_result_store): Update.
26761 (expand_addsub_overflow): Update.
26762 (expand_neg_overflow): Update.
26763 (expand_mul_overflow): Update.
26764 (expand_vector_ubsan_overflow): Update.
26765 * ipa-cp.c (good_cloning_opportunity_p): Update.
26766 * ipa-split.c (split_function): Use make_single_succ_edge.
26767 * ipa-utils.c (ipa_merge_profiles): Update.
26768 * loop-doloop.c (add_test): Update.
26769 (doloop_modify): Update.
26770 * loop-unroll.c (compare_and_jump_seq): Update.
26771 (unroll_loop_runtime_iterations): Update.
26772 * lra-constraints.c (lra_inheritance): Update.
26773 * lto-streamer-in.c (input_cfg): Update.
26774 * lto-streamer-out.c (output_cfg): Update.
26775 * mcf.c (adjust_cfg_counts): Update.
26776 * modulo-sched.c (sms_schedule): Update.
26777 * omp-expand.c (expand_omp_for_init_counts): Update.
26778 (extract_omp_for_update_vars): Update.
26779 (expand_omp_ordered_sink): Update.
26780 (expand_omp_for_ordered_loops): Update.
26781 (expand_omp_for_generic): Update.
26782 (expand_omp_for_static_nochunk): Update.
26783 (expand_omp_for_static_chunk): Update.
26784 (expand_cilk_for): Update.
26785 (expand_omp_simd): Update.
26786 (expand_omp_taskloop_for_outer): Update.
26787 (expand_omp_taskloop_for_inner): Update.
26788 * omp-simd-clone.c (simd_clone_adjust): Update.
26789 * optabs.c (expand_doubleword_shift): Update.
26790 (expand_abs): Update.
26791 (emit_cmp_and_jump_insn_1): Update.
26792 (expand_compare_and_swap_loop): Update.
26793 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26794 * predict.c (predictable_edge_p): Update.
26795 (edge_probability_reliable_p): Update.
26796 (set_even_probabilities): Update.
26797 (combine_predictions_for_insn): Update.
26798 (combine_predictions_for_bb): Update.
26799 (propagate_freq): Update.
26800 (estimate_bb_frequencies): Update.
26801 (force_edge_cold): Update.
26802 * profile-count.c (profile_count::dump): Add missing space into dump.
26803 (profile_count::debug): Add newline.
26804 (profile_count::differs_from_p): Explicitly convert to unsigned.
26805 (profile_count::stream_in): Update.
26806 (profile_probability::dump): New member function.
26807 (profile_probability::debug): New member function.
26808 (profile_probability::differs_from_p): New member function.
26809 (profile_probability::differs_lot_from_p): New member function.
26810 (profile_probability::stream_in): New member function.
26811 (profile_probability::stream_out): New member function.
26812 * profile-count.h (profile_count_quality): Rename to ...
26813 (profile_quality): ... this one.
26814 (profile_probability): New.
26815 (profile_count): Update.
26816 * profile.c (compute_branch_probabilities): Update.
26817 * recog.c (peep2_attempt): Update.
26818 * sched-ebb.c (schedule_ebbs): Update.
26819 * sched-rgn.c (find_single_block_region): Update.
26820 (compute_dom_prob_ps): Update.
26821 (schedule_region): Update.
26822 * sel-sched-ir.c (compute_succs_info): Update.
26823 * stmt.c (struct case_node): Update.
26824 (do_jump_if_equal): Update.
26825 (get_outgoing_edge_probs): Update.
26826 (conditional_probability): Update.
26827 (emit_case_dispatch_table): Update.
26828 (expand_case): Update.
26829 (expand_sjlj_dispatch_table): Update.
26830 (emit_case_nodes): Update.
26831 * targhooks.c: Update.
26832 * tracer.c (better_p): Update.
26833 (find_best_successor): Update.
26834 * trans-mem.c (expand_transaction): Update.
26835 * tree-call-cdce.c: Update.
26836 * tree-cfg.c (gimple_split_edge): Upate.
26837 (move_sese_region_to_fn): Upate.
26838 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26839 * tree-eh.c (lower_resx): Upate.
26840 (cleanup_empty_eh_move_lp): Upate.
26841 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26842 * tree-inline.c (copy_edges_for_bb): Update.
26843 (copy_cfg_body): Update.
26844 * tree-parloops.c (gen_parallel_loop): Update.
26845 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26846 (gimple_gen_time_profiler): Update.
26847 * tree-ssa-dce.c (remove_dead_stmt): Update.
26848 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26849 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26850 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26851 (unloop_loops): Update.
26852 (try_peel_loop): Update.
26853 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26854 * tree-ssa-loop-split.c (connect_loops): Update.
26855 (split_loop): Update.
26856 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26857 (hoist_guard): Update.
26858 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26859 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26860 (value_replacement): Update.
26861 * tree-ssa-reassoc.c (branch_fixup): Update.
26862 * tree-ssa-tail-merge.c (replace_block_by): Update.
26863 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26864 (create_edge_and_update_destination_phis): Update.
26865 (compute_path_counts): Update.
26866 (recompute_probabilities): Update.
26867 (update_joiner_offpath_counts): Update.
26868 (freqs_to_counts_path): Update.
26869 (duplicate_thread_path): Update.
26870 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26871 (struct switch_conv_info): Update.
26872 (gen_inbound_check): Update.
26873 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26874 (vect_do_peeling): Update.
26875 (vect_loop_versioning): Update.
26876 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26877 (optimize_mask_stores): Update.
26878 * ubsan.c (ubsan_expand_null_ifn): Update.
26879 * value-prof.c (gimple_divmod_fixed_value): Update.
26880 (gimple_divmod_fixed_value_transform): Update.
26881 (gimple_mod_pow2): Update.
26882 (gimple_mod_pow2_value_transform): Update.
26883 (gimple_mod_subtract): Update.
26884 (gimple_mod_subtract_transform): Update.
26885 (gimple_ic): Update.
26886 (gimple_stringop_fixed_value): Update.
26887 (gimple_stringops_transform): Update.
26888 * value-prof.h: Update.
26890 2017-06-29 Carl Love <cel@us.ibm.com>
26892 * config/rs6000/rs6000-c.c: Add support for built-in functions
26893 vector signed int vec_signed (vector float);
26894 vector signed long long vec_signed (vector double);
26895 vector signed int vec_signed2 (vector double, vector double);
26896 vector signed int vec_signede (vector double);
26897 vector signed int vec_signedo (vector double);
26898 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26899 instruction generator.
26900 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26901 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26902 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26904 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26905 vunsignede_v2df): Add define_expands.
26906 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26907 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26908 VEC_UNSIGNEDO): Add definitions.
26909 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26910 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26911 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26912 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26913 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26914 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26915 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26916 vec_unsignede, vec_unsignedo): Add builtin defines.
26917 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26919 * doc/extend.texi: Update the built-in documentation file for the
26920 new built-in functions.
26922 2017-06-29 Richard Biener <rguenther@suse.de>
26924 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26925 reduction chains to LOOP_VINFO_REDUCTIONS.
26926 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26927 SLP reductions after processing reduction chains.
26929 2017-06-29 Nathan Sidwell <nathan@acm.org>
26931 * builtins.c (fold_builtin_FUNCTION): Use
26932 lang_hooks.decl_printable_name.
26934 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26936 PR middle-end/81194
26937 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26938 with only one label.
26939 * stmt.c (expand_case): Assert NCASES is greater than one.
26941 2017-06-29 Richard Biener <rguenther@suse.de>
26943 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26945 (group_case_labels): Likewise.
26946 (find_taken_edge): Push sanity checking on val to workers...
26947 (find_taken_edge_cond_expr): ... here
26948 (find_taken_edge_switch_expr): ... and here, handle cases
26949 with just a default label.
26950 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26951 (group_case_labels): Likewise.
26952 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26953 group_case_labels does anything cleanup the CFG again.
26955 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26957 PR tree-optimization/81196
26958 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26959 exit condition comparing two IVs.
26961 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26963 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26964 profile to the dummy entry at the end of the list of architectures.
26965 * config/arm/arm-cpu-cdata.h: Regenerated.
26967 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26968 Michael Collison <michael.collison@arm.com>
26971 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26973 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26974 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26975 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26976 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26977 of shift when the shift amount is masked with constant equal to
26978 the size of the mode.
26979 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26982 2017-06-29 Martin Liska <mliska@suse.cz>
26984 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26986 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26989 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26992 (edge_growth_cache_entry::edge_growth_cache_entry): New
26994 (reset_edge_growth_cache): Update to use constructor.
26996 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26998 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26999 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
27000 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
27002 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
27004 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
27005 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
27007 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
27009 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
27010 (*-linux-uclibc*): Add t-uclibc tmake_file.
27011 * config/t-musl: New.
27012 * config/t-uclibc: New.
27014 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
27016 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
27018 (gen_comm_data): Emit architectural setting of arch_prof.
27019 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
27021 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
27022 (armv8-m.base, armv8-m.main): Likewise.
27023 * arm-protos.h (arm_build_target): Add profile field.
27024 (arch_option): Likewise.
27025 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
27027 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
27028 arm_active_target.profile.
27030 2017-06-28 Richard Biener <rguenther@suse.de>
27032 PR middle-end/81227
27033 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
27034 TYPE_OVERFLOW_WRAPS.
27035 * match.pd (negate_expr_p): Likewise.
27036 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
27037 fold_build2, not fold_binary.
27039 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
27041 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
27042 Convert memory address to Pmode.
27043 (aarch64_print_operand): Assert MEM operands are always Pmode.
27045 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
27048 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
27049 Remove redundant if.
27050 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
27051 * config/arm/aarch-common-protos.h
27052 (aarch_forward_to_shift_is_not_shifted_re): Remove.
27053 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
27055 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
27058 * multiple_target.c (create_dispatcher_calls): Set the default
27059 clone to be static, not public.
27061 2017-06-28 Richard Biener <rguenther@suse.de>
27063 * tree-vect-loop.c (vectorizable_reduction): Move special
27064 cond reduction IV var creation ...
27065 (vect_create_epilog_for_reduction): ... here. Remove induction_index
27066 parameter. Use STMT_VINFO_VECTYPE.
27067 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
27070 2017-06-28 Martin Liska <mliska@suse.cz>
27073 * ipa-visibility.c (non_local_p): Handle visibility.
27075 2017-06-28 Martin Liska <mliska@suse.cz>
27078 * common.opt: Add IntegerRange to various options.
27079 * opt-functions.awk (integer_range_info): New function.
27080 * optc-gen.awk: Add integer_range_info to cl_options struct.
27081 * opts-common.c (decode_cmdline_option): Handle
27082 CL_ERR_INT_RANGE_ARG.
27083 (cmdline_handle_error): Likewise.
27084 * opts.c (print_filtered_help): Show valid interval in
27085 when --help is provided.
27086 * opts.h (struct cl_option): Add range_min and range_max fields.
27087 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
27089 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
27091 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
27092 (x * C EQ/NE y * C): New transformation.
27094 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
27096 * genmultilib (combination_space): Accept '+' in option names.
27098 2017-06-28 Martin Liska <mliska@suse.cz>
27101 * asan.c (instrument_derefs): Bail out inner references
27102 that are hard register variables.
27104 2017-06-28 Jakub Jelinek <jakub@redhat.com>
27107 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
27108 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
27110 2017-06-28 Richard Biener <rguenther@suse.de>
27112 * tree-vectorizer.h (vect_get_vec_defs): Remove.
27113 (vect_get_slp_defs): Adjust.
27114 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
27116 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
27118 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
27119 get_initial_defs_for_reduction instead of vect_get_vec_defs.
27120 (vectorizable_reduction): Adjust.
27121 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
27123 (vect_get_slp_defs): Likewise.
27124 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
27125 (vectorizable_bswap): Adjust.
27126 (vectorizable_call): Likewise.
27127 (vectorizable_conversion): Likewise.
27128 (vectorizable_assignment): Likewise.
27129 (vectorizable_shift): Likewise.
27130 (vectorizable_operation): Likewise.
27131 (vectorizable_store): Likewise.
27132 (vectorizable_condition): Likewise.
27133 (vectorizable_comparison): Likewise.
27135 2017-06-28 Michael Collison <michael.collison@arm.com>
27138 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
27140 (arm_gen_movmemqi): Removed unused variable 'i'.
27141 Convert 'for' loop into 'while' loop.
27142 (arm_expand_prologue): Remove last unnecessary set of insn.
27143 (thumb_pop): Remove unused variable 'pushed_words'.
27144 (thumb_exit): Remove last unnecessary set of regs_to_pop.
27146 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27148 * config/s390/predicates.md: Use s390_rel_address_ok_p.
27149 * config/s390/s390-protos.h: Add prototype of
27150 s390_rel_address_ok_p.
27151 * config/s390/s390.c (s390_got_symbol): New function.
27152 (s390_rel_address_ok_p): New function.
27153 (legitimize_pic_address): Use s390_rel_address_ok_p.
27154 (s390_load_got): Use s390_got_symbol.
27155 (s390_option_override): Issue error if
27156 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
27157 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
27159 * config/s390/s390.opt: New option mpic-data-is-text-relative.
27161 2017-06-27 Andrew Pinski <apinski@cavium.com>
27163 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
27164 (X * copysign (1.0, X)): New pattern.
27165 (X * copysign (1.0, -X)): New pattern.
27166 (copysign (-1.0, CST)): New pattern.
27168 2017-06-27 Joseph Myers <joseph@codesourcery.com>
27170 * genmultilib (combination_space): Remove variable.
27171 Validate reuse rules against regular expression for any sequence
27172 of multilib options in any order.
27174 2017-06-27 Michael Collison <michael.collison@arm.com>
27176 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
27177 call aarch64_split_simd_combine.
27178 * (aarch64_combine_internal<mode>): Delete pattern.
27179 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
27180 Allow register and subreg operands.
27182 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27184 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
27185 specific need, just fallback on defaults.
27186 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
27188 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27189 Olivier Hainque <hainque@adacore.com>
27191 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
27193 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
27194 targets. Pick a default if no particular attempt applied.
27195 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
27198 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27200 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
27201 (x86_64-wrs-vxworks7): Likewise.
27203 2017-06-27 Marek Polacek <polacek@redhat.com>
27206 * ubsan.c (instrument_null): Check get_base_address's result for null.
27208 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27210 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27212 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27214 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27215 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27216 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27217 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27218 New function types.
27219 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27220 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27221 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27222 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27223 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27224 BUILT_IN_FEUPDATEENV): New builtins.
27225 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27226 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27227 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27228 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27230 (builtin_structptr_types): Adjust size.
27231 * tree.c (builtin_structptr_types): Add four entries.
27233 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27234 Olivier Hainque <hainque@adacore.com>
27236 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27237 (TLS_SYM): New local macro, forcing reference to __tls__ on
27238 link command lines for VxWorks 7 RTPs, triggering initialization
27240 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27241 OS features TLS support, true for RTPs on VxWorks 7.
27242 * config/vxworks.c (vxworks_override_options): Setup emutls
27245 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27247 * predict.c (test_prediction_value_range): Use -1U instead of -1
27248 to avoid narrowing conversion warning.
27249 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27250 to avoid narrowing conversion warning.
27251 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27253 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27255 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27257 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27258 64bit configurations.
27259 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27260 (SIZE_TYPE): Likewise.
27261 * config/vxworks.c (vxworks_emutls_var_fields): Use
27262 long_unsigned_type_node instead of unsigned_type_node as the offset
27263 field type, which is "pointer" mode in emutls.c.
27265 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27268 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27270 PR middle-end/81207
27271 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27272 gimple_vuse copying separately from gimple_vdef copying.
27274 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27276 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27279 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27280 Olivier Hainque <hainque@adacore.com>
27282 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27283 for all vxworks7 targets.
27284 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27285 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27286 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27287 variations for VX6/VX7 and 32/64bits later on in ...
27288 (VXWORKS_LIB_SPEC): Leverage new macros.
27289 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27290 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27292 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27294 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27295 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27297 2017-06-26 Carl Love <cel@us.ibm.com>
27299 * config/rs6000/rs6000-c.c: Add support for built-in functions
27300 vector bool char vec_reve (vector bool char);
27301 vector signed char vec_reve (vector signed char);
27302 vector unsigned char vec_reve (vector unsigned char);
27303 vector bool int vec_reve (vector bool int);
27304 vector signed int vec_reve (vector signed int);
27305 vector unsigned int vec_reve (vector unsigned int);
27306 vector bool long long vec_reve (vector bool long long);
27307 vector signed long long vec_reve (vector signed long long);
27308 vector unsigned long long vec_reve (vector unsigned long long);
27309 vector bool short vec_reve (vector bool short);
27310 vector signed short vec_reve (vector signed short);
27311 vector double vec_reve (vector double);
27312 vector float vec_reve (vector float);
27313 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27314 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27315 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27316 (altivec_vreve): New pattern.
27317 * config/rs6000/altivec.h (vec_reve): New define.
27318 * doc/extend.texi (vec_rev): Update the built-in documentation file
27319 for the new built-in functions.
27321 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27323 PR tree-optimization/71815
27324 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27326 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27328 (slsr_process_phi): Likewise.
27329 (replace_uncond_cands_and_profitable_phis): Don't replace a
27330 multiply candidate with a stride of 1 (copy or cast).
27331 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27333 (lowest_cost_path): Likewise.
27334 (total_savings): Likewise.
27336 2017-06-26 Richard Biener <rguenther@suse.de>
27339 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27340 Use def_builtin_pure for all gather builtins.
27342 2017-06-26 Richard Biener <rguenther@suse.de>
27344 PR tree-optimization/81203
27345 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27346 non-dominating BBs.
27348 2017-06-26 Marek Polacek <polacek@redhat.com>
27351 * doc/invoke.texi: Document -Wmultistatement-macros.
27353 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27355 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27356 arm_neon_ok_no_float_abi effective target.
27358 2017-06-26 Richard Biener <rguenther@suse.de>
27360 PR tree-optimization/80928
27361 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27362 (copy_bbs): Set BB_DUPLICATED flag early.
27363 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27365 (execute_on_shrinking_pred): Likewise.
27366 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27367 BB_DUPLICATED blocks.
27368 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27369 iterate over all PHIs considering removal of *gsi.
27371 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27373 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27376 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27378 * config/rs6000/rs6000-string.c: (expand_block_clear,
27379 do_load_for_compare, select_block_compare_mode,
27380 compute_current_alignment, expand_block_compare,
27381 expand_strncmp_align_check, expand_strn_compare,
27382 expand_block_move, rs6000_output_load_multiple)
27383 Move functions related to string/block move/compare
27384 to a separate file.
27385 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27386 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27387 for this function which is now used in two files.
27388 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27389 * config.gcc: Add rs6000-string.o to extra_objs for
27390 targets powerpc*-*-* and rs6000*-*-*.
27392 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27395 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27396 32-bit, since indexed is not valid for DImode.
27397 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27398 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27399 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27400 (define_peephole2 for Altivec d-form store): Likewise.
27403 * multiple_target.c (create_dispatcher_calls): Only create the
27404 dispatcher call if the function is the default clone of a
27405 versioned function.
27407 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27409 PR middle-end/80902
27410 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27411 a call, force the call to not be a tail call.
27413 2017-06-23 Jeff Law <law@redhat.com>
27415 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27418 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27420 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27421 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27422 handling for early expansion of vector shifts (sl,sr,sra,rl).
27423 (builtin_function_type): Add vector shift right instructions
27424 to the unsigned argument list.
27426 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27428 rtl-optimizatoin/79286
27429 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27430 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27431 trap. PIC register plus a const unspec without offset can never trap.
27433 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27435 * tree.h (builtin_structptr_type): New type.
27436 (builtin_structptr_types): Declare new array.
27437 * tree.c (builtin_structptr_types): New array.
27438 (free_lang_data, build_common_tree_nodes): Use it.
27440 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27443 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27446 2017-06-22 Matt Turner <mattst88@gmail.com>
27448 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27449 Lake models to skylake case. Assume skylake for unknown
27450 models with clflushopt.
27452 2017-06-22 Jeff Law <law@redhat.com>
27454 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27455 frame sizes that do not satisfy aarch64_uimm12_shift.
27457 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27459 * profile-count.h (apply_probability,
27460 apply_scale, probability_in): Fix checks for zero.
27462 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27464 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27465 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27467 2016-06-22 Richard Biener <rguenther@suse.de>
27469 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27470 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27471 REDUC_MAX_EXPR support.
27472 (vectorizable_reduction): Likewise.
27473 (vect_create_epilog_for_reduction): Likewise.
27475 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27477 * match.pd (A / (1 << B) -> A >> B): New.
27478 * generic-match-head.c: Include optabs-tree.h.
27479 * gimple-match-head.c: Likewise.
27480 * optabs-tree.h (target_supports_op_p): New.
27481 * optabs-tree.c (target_supports_op_p): New.
27483 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27485 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27486 $gcc_cv_ld --help output.
27487 (gcc_cv_ld_demangle): Likewise.
27488 (gcc_cv_ld_eh_frame_hdr): Likewise.
27489 (gcc_cv_ld_pie): Likewise.
27490 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27491 (gcc_cv_ld_buildid): Likewise.
27492 (gcc_cv_ld_sysroot): Likewise.
27493 (ld_bndplt_support): Likewise.
27494 (ld_pushpopstate_support): Likewise.
27495 * configure: Regenerate.
27496 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27498 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27501 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27502 operands indexes to avoid gap between operands and match_dups.
27504 2017-06-21 Andrew Pinski <apinski@cavium.com>
27506 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27507 Increment Arith_shift and Arith_shift_reg by 1.
27508 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27510 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27511 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27512 (aarch64_strip_extend): Add new argument and test for it.
27513 (aarch64_cheap_mult_shift_p): New function.
27514 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27515 add a cost if it is true.
27516 Update calls to aarch64_strip_extend.
27517 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27519 2017-06-21 Andrew Pinski <apinski@cavium.com>
27521 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27523 (thunderxt88): Likewise.
27524 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27525 (thunderx_prefetch_tune): New variable.
27526 (thunderx2t99_prefetch_tune): Update for the correct values.
27527 (thunderxt88_tunings): New variable.
27528 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27529 generic_prefetch_tune.
27530 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27532 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27534 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27535 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27536 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27537 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27538 (aarch64_atomic_cas<mode>, GPI): Likewise.
27540 2017-06-21 Martin Liska <mliska@suse.cz>
27542 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27543 statements on cold and hot labels.
27544 * predict.c (tree_estimate_probability_bb): Remove the
27545 prediction from this place.
27547 2017-06-21 Martin Liska <mliska@suse.cz>
27549 PR tree-optimization/79489
27550 * gimplify.c (maybe_add_early_return_predict_stmt): New
27552 (gimplify_return_expr): Call the function.
27553 * predict.c (tree_estimate_probability_bb): Remove handling
27555 * predict.def: Update comment about early return predictor.
27556 * gimple-predict.h (is_gimple_predict): New function.
27557 * predict.def: Change default value of early return to 66.
27558 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27560 * passes.def: Put pass_strip_predict_hints to the beginning of
27563 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27565 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27566 FUNCTION_DECL declarations.
27567 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27569 (dwaf2out_decl): Likewise.
27570 * godump.c (go_early_global_decl): Skip call to the real debug hook
27571 for FUNCTION_DECL declarations.
27572 * passes.c (rest_of_decl_compilation): Skip call to the
27573 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27574 -fdump-go-spec is passed.
27576 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27578 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27579 Reorder for compactness.
27580 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27581 (def_builtin_pure, def_builtin_pure2): New functions.
27582 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27584 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27586 * match.pd (nop_convert): New predicate.
27587 ((A +- CST1) +- CST2): Allow some NOP conversions.
27589 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27592 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27593 with ctors/dtors if GOVD_SHARED is set.
27595 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27597 * config/aarch64/aarch64.md (movti_aarch64):
27598 Emit mov rather than orr.
27599 (movtf_aarch64): Likewise.
27600 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27601 Emit mov rather than orr.
27603 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27605 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27606 Swap alternatives, make integer dup more expensive.
27608 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27610 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27611 Return true for non-tls symbols.
27613 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27615 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27616 (cortex-a75): Likewise.
27617 (cortex-a75.cortex-a55): Likewise.
27618 * config/aarch64/aarch64-tune.md: Regenerate.
27619 * doc/invoke.texi (-mtune): Document new values for -mtune.
27621 2017-06-21 Tom de Vries <tom@codesourcery.com>
27623 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27624 stack_size feature.
27625 (Effective-Target Keywords, Other attributes): Suggest using
27626 dg-add-options stack_size feature to get stack limit in stack_size
27627 effective target documentation.
27629 2017-06-21 Julian Brown <julian@codesourcery.com>
27630 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27632 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27633 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27634 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27636 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27637 attribute type list for neon_multiply.
27638 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27639 attribute type list for neon_multiply.
27640 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27641 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27642 attribute type list for neon_multiply.
27643 * config/arm/types.md (crypto_pmull): Add.
27644 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27645 attribute type list.
27647 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27649 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27652 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27654 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27655 to make sure not to dereference a NULL cost_classes_ptr pointer.
27657 2017-06-20 Carl Love <cel@us.ibm.com>
27659 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27660 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27661 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27662 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27663 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27664 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27665 VMULOSW): New enum "unspec" values.
27666 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27667 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27668 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27669 altivec_vmulosw): New patterns.
27670 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27671 VMULOSW): Add definitions.
27673 2017-06-20 Julia Koval <julia.koval@intel.com>
27675 * config/i386/i386.c: Fix rounding expand for new pattern.
27676 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27678 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27680 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27681 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27683 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27685 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27688 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27690 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27691 architecture, then by implementer ID.
27692 * config/aarch64/aarch64-tune.md: Regenerate.
27694 2017-06-20 Richard Biener <rguenther@suse.de>
27696 PR middle-end/81097
27697 * fold-const.c (split_tree): Fold to type before negating.
27699 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27701 * diagnostic-show-locus.c
27702 (selftest::test_fixit_deletion_affecting_newline): New function.
27703 (selftest::diagnostic_show_locus_c_tests): Call it.
27705 2017-06-20 Andreas Schwab <schwab@suse.de>
27708 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27711 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27713 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27714 __ARM_FEATURE_COPROC according to support.
27716 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27718 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27719 Rewritten to avoid overflow for > 32-bit pointers.
27722 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27723 by removing enum keyword.
27724 (ubsan_type_descriptor): Likewise. Formatting fix.
27727 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27728 splitter): Require TARGET_SSE2 in the condition.
27730 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27733 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27734 for doing vector set of SFmode on ISA 3.0.
27735 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27736 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27738 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27739 SFmode value into a V4SF variable that was extracted from another
27740 V4SF variable without converting the element to double precision
27741 and back to single precision vector format.
27742 (vsx_insert_extract_v4sf_p9_2): Likewise.
27744 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27746 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27747 in UWHI to avoid undefined overflow.
27750 * ubsan.h (enum ubsan_encode_value_phase): New.
27751 (ubsan_encode_value): Change second argument to
27752 enum ubsan_encode_value_phase with default value of
27753 UBSAN_ENCODE_VALUE_GENERIC.
27754 * ubsan.c (ubsan_encode_value): Change second argument to
27755 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27756 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27757 create_tmp_var_raw instead of create_tmp_var and use a
27759 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27760 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27761 ubsan_encode_value callers.
27764 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27765 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27766 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27768 2017-06-19 Richard Biener <rguenther@suse.de>
27770 PR middle-end/81118
27771 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27772 estimates if we changed anything.
27774 2017-06-19 Richard Biener <rguenther@suse.de>
27776 PR tree-optimization/80887
27777 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27778 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27779 simplified lookups, then reset mprts_hook.
27780 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27782 (try_to_simplify): Likewise.
27784 2017-06-19 Martin Liska <mliska@suse.cz>
27787 * gimplify.c (gimplify_switch_expr):
27788 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27790 2017-06-19 Martin Liska <mliska@suse.cz>
27792 * doc/install.texi: Document that PGO runs in 4 stages.
27794 2017-06-19 Martin Liska <mliska@suse.cz>
27797 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27798 to dispatcher function name.
27799 * multiple_target.c (replace_function_decl): New function.
27800 (create_dispatcher_calls): Redirect both edges and references.
27802 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27804 * profile-count.c (profile_count::dump): Dump quality.
27805 (profile_count::differs_from_p): Update for unsigned val.
27806 * profile-count.h (profile_count_quality): New enum.
27807 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27809 2017-06-19 Richard Biener <rguenther@suse.de>
27811 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27812 struct function as arg.
27813 (estimate_numbers_of_iterations): Export overload with loop arg.
27814 (free_numbers_of_iterations_estimates_loop): Use an overload of
27815 free_numbers_of_iterations_estimates instead.
27816 * tree-cfg.c (remove_bb): Adjust.
27817 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27818 * tree-parloops.c (gen_parallel_loop): Likewise.
27819 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27821 (tree_unroll_loops_completely): Likewise.
27822 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27823 Use an overload instead and export.
27824 (estimated_loop_iterations): Adjust.
27825 (max_loop_iterations): Likewise.
27826 (likely_max_loop_iterations): Likewise.
27827 (estimate_numbers_of_iterations): Take struct function as arg
27829 (loop_exits_before_overflow): Adjust.
27830 (free_numbers_of_iterations_estimates_loop): Use an overload.
27831 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27832 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27834 2017-06-19 Richard Biener <rguenther@suse.de>
27837 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27838 RANGE_EXPR conservatively.
27840 2017-06-16 Carl Love <cel@us.ibm.com>
27842 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27843 definitions for vec_float, vec_float2, vec_floato,
27844 vec_floate built-ins.
27845 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27846 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27848 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27849 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27850 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27851 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27852 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27853 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27854 vec_floato): Add builtin defines.
27855 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27856 Update the built-in documentation file for the new built-in
27859 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27861 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27862 (mthumb): Mark as the negative of -marm.
27864 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27866 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27868 (ARM Options, -mtune): Document that this accepts the same
27869 extension options as -mcpu.
27870 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27872 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27874 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27875 permitted extensions.
27877 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27879 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27880 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27881 (armv8-m.main): Add option +nodsp.
27882 * config/arm/arm-cpu-cdata.h: Regenerated.
27884 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27886 * config/arm/t-fuchsia: New file.
27887 * config.gcc (arm*-*-fuchsia*): Use it.
27889 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27891 * config/arm/t-symbian: Rewrite for new option infrastructure.
27893 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27895 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27896 (MULTILIB_REQUIRED): Likewise.
27898 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27900 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27901 (MULTILIB_RESUE): Likewise.
27902 (MULTILIB_MATCHES): Likewise.
27903 (MULTLIB_REQUIRED): Likewise.
27905 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27907 * config/arm/t-rtems: Rewrite for new option framework.
27909 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27911 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27912 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27913 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27914 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27915 * config/arm/t-multilib: ... here.
27916 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27917 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27918 armv7-a and armv8*-a when A-profile libraries have not been built.
27919 * config/arm/t-rmprofile: Rewrite.
27921 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27923 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27924 with a backslash. Remove the backslash after substituting unescaped
27926 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27928 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27930 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27931 the driver to pass through the default setting of -mfloat-abi.
27932 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27934 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27935 (all_feat_combs): New rule.
27936 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27938 * config/arm/t-aprofile: Rewrite.
27940 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27942 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27943 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27944 fpu is not specified by the user/command-line.
27945 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27946 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27947 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27948 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27949 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27950 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27952 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27954 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27955 * config/arm/t-arm-elf: Rewritten.
27957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27959 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27960 have some floating-point instructions.
27961 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27962 (TARGET_MAYBE_HARD_FLOAT): New macro.
27963 * config/arm/arm-builtins.c (arm_init_builtins): Use
27964 TARGET_MAYBE_HARD_FLOAT.
27965 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27967 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27969 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27970 (configargs.h): Include it.
27971 (arm_print_hint_for_fpu_option): New function.
27972 (arm_parse_fpu_option): New function.
27973 (candidate_extension): New class.
27974 (arm_canon_for_multilib): New function.
27975 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27976 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27977 (ARCH_CANONICAL_SPECS): New macro.
27978 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27980 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27982 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27983 are set after handling multilib fragments. Set target_cpu_default2
27986 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27988 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27990 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27991 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27993 (check_arch): Likewise.
27994 * config/arm/arm.c (arm_configure_build_target): Handle
27995 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27996 options in the default.
27998 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28000 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
28001 when an option is an alias of another.
28002 * config/arm/parsecpu.awk (optalias): New parser token.
28003 (gen_comm_data): Mark non-alias options as such. Emit entries
28004 for extension aliases.
28005 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
28006 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
28007 (armv6kz, armv6zk, armv6t2): Likewise.
28008 (armv7): Make vfpv3-d16 an alias.
28009 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
28011 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
28012 Sort in canonical order.
28013 (armv8-a): Sort in canonical order.
28014 (armv8.1-a, armv8.2-a): Likewise.
28015 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
28017 (cortex-a9): Sort in canonical order.
28018 * config/arm/arm.c (selftests.h): Include it.
28019 (arm_test_cpu_arch_data): New function.
28020 (arm_run_self_tests): New function.
28021 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
28022 (targetm): Move declaration to the end of the file.
28023 * arm-cpu-cdata.h: Regenerated.
28025 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28027 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
28028 call to target_mode_check describing the type of option passed.
28029 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
28030 (arm_target_thumb_only): Use arm_parse_arch_option_name or
28031 arm_parse_cpu_option_name to match parameters against list of
28033 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
28034 arm_arch_core_flags data structure.
28035 * config/arm/arm-cpu_cdata.h: Regenerated.
28037 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28039 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
28041 (arm_print_hint_for_cpu_option): Likewise.
28042 (arm_print_hint_for_arch_option): Likewise.
28043 (arm_parse_cpu_option_name): Likewise.
28044 (arm_parse_arch_option_name): Likewise.
28045 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
28046 of entries in the all_fpus list.
28047 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
28048 (arm_parse_cpu_option_name): Declare.
28049 (arm_parse_arch_option_name): Declare.
28050 (arm_parse_option_features): Declare.
28051 (arm_intialize_isa): Declare.
28052 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
28054 (gen_comm_data): ... here. Make definitions non-static.
28055 * config/arm/arm-cpu-data.h: Regenerated.
28056 * config/arm/arm-cpu-cdata.h: Regenerated.
28058 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28060 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
28061 (cpu_arch_extension): New structure.
28062 (cpu_arch_option, arch_option, cpu_option): New structures.
28063 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
28064 architecture types.
28065 (gen_data): Generate new format data tables.
28066 * config/arm/arm.c (cpu_tune): New structure.
28067 (cpu_option, processors): Delete.
28068 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
28069 (arm_print_hint_for_cpu_option): ... this and ...
28070 (arm_print_hint_for_arch_option): ... this.
28071 (arm_parse_arch_cpu_name): Delete. Replace with ...
28072 (arm_parse_cpu_option_name): ... this and ...
28073 (arm_parse_arch_option_name): ... this.
28074 (arm_unrecognized_feature): Change type of target parameter to
28076 (arm_parse_arch_cpu_features): Delete. Replace with ...
28077 (arm_parse_option_features): ... this.
28078 (arm_configure_build_target): Rework to use new configuration data
28080 (arm_print_tune_info): Rework for new configuration data tables.
28081 * config/arm/arm-cpu-data.h: Regenerated.
28082 * config/arm/arm-cpu.h: Regenerated.
28084 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28086 * Makefile.in (OBJS): Move sbitmap.o from here ...
28087 (OBJS-libcommon): ... to here.
28089 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28091 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
28092 (ISA_ALL_CRYPTO): New macro.
28093 (ISA_ALL_SIMD): New macro
28094 (ISA_ALL_FP): New macro.
28095 * config/arm/arm.c (fpu_bitlist): Update initializer.
28096 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
28098 (arm9e): Add fpu. Add option for nofp
28099 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
28100 (arm926ej-s, arm1026ej-s): Likewise.
28101 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
28102 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
28103 neon-fp16, neon-vfpv4, nofp and nosimd.
28104 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
28105 (cortex-a8): Add fpu. Add option for nofp.
28106 (cortex-a9): Add fpu. Add options for nosimd and nofp.
28107 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
28108 (cortex-r4f): Add fpu.
28109 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
28110 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
28112 (cortex-r8): Likewise.
28113 (cortex-m4): Add fpu. Add option for nofp.
28114 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
28115 (cortex-a17.cortex-a7): Likewise.
28116 (cortex-a32): Add fpu. Add options for crypto and nofp.
28117 (cortex-a35, cortex-a53): Likewise.
28118 (cortex-a57): Add fpu. Add option for crypto.
28119 (cortex-a72, cortex-a73): Likewise.
28120 (exynos-m1): Likewise.
28121 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
28122 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
28123 (cortex-m33): Add fpu. Add option for nofp.
28124 * config/arm/arm-cpu-cdata.h: Regenerated
28125 * config/arm/arm-cpu-data.h: Regenerated.
28127 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28129 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
28130 (armv5te, armv5tej): Likewise.
28131 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
28132 (armv7): Add options fp and vfpv3-d16.
28133 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
28134 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
28136 (armv7ve): Likewise.
28137 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
28138 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
28139 (armv8-a): Add nocrypto option.
28140 (armv8.1-a, armv8.2-a): Likewise.
28141 (armv8-m.main): add options fp, fp.dp and nofp.
28143 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28145 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
28147 (armv8-a+crc): Delete.
28148 (armv8.1-a): Add options simd, crypto and nofp.
28149 (armv8.2-a): Add options fp16, simd, crypto and nofp.
28150 (armv8.2-a+fp16): Delete.
28151 (armv8-m.main): Add option dsp.
28152 (armv8-m.main+dsp): Delete.
28153 (cortex-a8): Add fpu. Add nofp option.
28154 (cortex-a9): Add fpu. Add nofp and nosimd options.
28155 * config/arm/parsecpu.awk (gen_data): Generate option tables and
28156 link to main cpu and architecture data structures.
28157 (gen_comm_data): Only put isa attributes from the main architecture
28159 (option): New statement for architecture and CPU entries.
28160 * arm.c (struct cpu_option): New structure.
28161 (struct processors): Add entry for options.
28162 (arm_unrecognized_feature): New function.
28163 (arm_parse_arch_cpu_name): Ignore any characters after the first
28165 (arm_parse_arch_cpu_feature): New function.
28166 (arm_configure_build_target): Separate out any CPU and architecture
28167 features and parse separately. Don't error out if -mfpu=auto is
28168 used with only an architecture string.
28169 (arm_print_asm_arch_directives): New function.
28170 (arm_file_start): Call it.
28171 * config/arm/arm-cpu-cdata.h: Regenerated.
28172 * config/arm/arm-cpu-data.h: Likewise.
28173 * config/arm/arm-tables.opt: Likewise.
28175 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28177 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
28178 assembler when it is not -mfpu=auto.
28180 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28182 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
28183 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
28184 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
28185 (ASM_CPU_SPEC): Rewrite.
28186 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
28187 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
28188 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
28189 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
28190 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
28191 copied string is NUL-terminated. Also strip any characters prefixed
28193 (arm_rewrite_selected_arch): New function.
28194 (arm_rewrite_march): New function.
28196 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28198 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
28199 (x_arm_cpu_string, x_arm_tune_string): Likewise.
28200 (march, mcpu, mtune): Convert to string-based options.
28201 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
28202 (arm_parse_arch_cpu_name): New function.
28203 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28204 identify selected architecture or CPU.
28205 (arm_option_save): New function.
28206 (TARGET_OPTION_SAVE): Redefine.
28207 (arm_option_restore): Restore string options.
28208 (arm_option_print): Print string options.
28210 2017-06-16 Martin Sebor <msebor@redhat.com>
28212 PR tree-optimization/80933
28213 PR tree-optimization/80934
28214 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28215 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28216 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28217 (gimple_fold_builtin): Call them.
28219 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28221 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28222 as unlikely; update profile.
28224 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28226 * predict.c (force_edge_cold): Handle declaring edges impossible
28229 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28231 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28233 (try_unroll_loop_completely): Fix reporting.
28235 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28237 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28239 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28242 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28243 if given a non-constant argument for an intrinsic which requires a
28246 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28248 * profile.c (compare_freqs): New function.
28249 (branch_prob): Sort edge list.
28250 (find_spanning_tree): Assume that the list is priority sorted.
28252 2017-06-16 Richard Biener <rguenther@suse.de>
28254 PR tree-optimization/81090
28255 * passes.def (pass_record_bounds): Remove.
28256 * tree-pass.h (make_pass_record_bounds): Likewise.
28257 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28258 make_pass_record_bounds): Likewise.
28259 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28260 not free niter estimates at the beginning but at the end.
28261 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28263 2017-06-16 Richard Biener <rguenther@suse.de>
28265 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28266 initializer to workaround ICE in host GCC 4.8.
28268 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28270 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28272 (clone_inlined_nodes): Update.
28274 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28276 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28277 prefetch settings, and enable prefetching by default at -O3.
28279 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28281 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28282 Set flag_prefetch_loop_arrays according to tuning data.
28284 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28286 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28287 New tune structure.
28288 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28289 [Unrelated to main purpose of the patch] Place the pointer field last
28290 to enable type checking errors when tune structure are wrongly merged.
28291 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28292 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28293 (thunderx2t99_prefetch_tune): New tune constants.
28294 (tune_params *_tunings): Update all tunings (no functional change).
28295 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28296 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28297 from tunings structures.
28299 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28302 * ubsan.c (instrument_null): Add T argument, use it instead
28303 of computing it based on IS_LHS.
28304 (instrument_object_size): Likewise.
28305 (pass_ubsan::execute): Adjust instrument_null and
28306 instrument_object_size callers to pass gimple_get_lhs or
28307 gimple_assign_rhs1 result to it. Use instrument_null instead of
28308 calling get_base_address and instrument_mem_ref. Handle
28309 aggregate call arguments for object-size sanitization.
28311 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28313 PR tree-optimization/81089
28314 * tree-vrp.c (is_masked_range_test): Validate operands of
28317 2017-06-15 Martin Sebor <msebor@redhat.com>
28320 * dumpfile.c (dump_register): Avoid calling memset to initialize
28321 a class with a default ctor.
28322 * gcc.c (struct compiler): Remove const qualification.
28323 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28324 * hash-table.h: Ditto.
28325 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28327 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28328 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28330 * params.h (struct param_info): Make struct members non-const.
28331 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28332 with default initialization.
28333 * vec.h (vec_copy_construct, vec_default_construct): New helper
28335 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28336 with vec_copy_construct.
28337 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28338 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28339 * doc/invoke.texi (-Wclass-memaccess): Document.
28341 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28343 * emit-rtl.h (is_leaf): Update comment about local
28344 register allocator.
28346 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28349 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28350 for a variable to have a section before checking if the section has a
28352 Set section to.persistent if persistent attribute is set.
28353 Warn if .persistent attribute is used on an automatic variable.
28355 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28357 PR rtl-optimization/80474
28358 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28360 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28362 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28365 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28367 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28368 (casesi): Emit bounds checking as RTL.
28369 (casesi_internal_mips16_<mode>): Remove bounds checking.
28371 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28373 * config/xtensa/xtensa.c (xtensa_option_override): Append
28374 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28375 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28376 xtensa_doloop_hooks): Define unconditionally.
28377 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28379 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28380 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28381 for it in xtensa_option_override.
28382 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28383 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28385 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28387 * doc/cppopts.texi: Document '-' special value to -MF.
28389 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28391 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28392 (cortex_a53_fconst): Likewise.
28393 (cortex_a53_fpmul): Likewise.
28394 (cortex_a53_f_load_64): Likewise.
28395 (cortex_a53_f_load_many): Likewise.
28396 (cortex_a53_advsimd_alu): Likewise.
28397 (cortex_a53_advsimd_alu_q): Likewise.
28398 (cortex_a53_advsimd_mul): Likewise.
28399 (cortex_a53_advsimd_mul_q): Likewise.
28400 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28401 Add missing fmul, r2f_cvt and fconst cases.
28403 2017-06-14 Richard Biener <rguenther@suse.de>
28405 PR middle-end/81088
28406 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28408 (fold_binary_loc): When associating do not treat pre-existing
28409 TREE_OVERFLOW on literal constants as a reason to allow
28410 TREE_OVERFLOW on associated literal constants.
28412 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28414 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28415 (MASK_FEATURES): New macro.
28416 * config/sparc/sparc.c (sparc_option_override): Remove the special
28417 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28419 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28421 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28422 a division of 0 if non-call exceptions are enabled.
28424 2017-06-14 Andrew Pinski <apinski@cavium.com>
28425 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28428 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28429 Improve vector initialization code gen for only variable case.
28431 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28433 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28435 2017-06-14 Richard Biener <rguenther@suse.de>
28437 PR tree-optimization/81083
28438 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28441 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28443 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28444 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28445 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28446 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28447 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28448 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28450 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28452 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28453 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28455 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28457 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28459 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28461 * config/rs6000/t-rtems: Don't handle SPE.
28463 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28465 * config/rs6000/t-linux: Don't handle SPE.
28467 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28469 * config/rs6000/eabispe.h: Delete file.
28471 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28473 * config/rs6000/t-spe: Delete file.
28475 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28477 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28478 (rs6000_legitimate_offset_address_p): Return false for anything in
28479 V2SImode or V2SFmode.
28481 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28483 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28484 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28485 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28487 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28488 (rs6000_legitimate_offset_address_p): Ditto.
28489 (rs6000_emit_move): Ditto.
28490 (rs6000_init_builtins): Remove V4HI_type_node.
28492 2017-06-13 Martin Liska <mliska@suse.cz>
28495 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28496 (gate_asan): Likewise.
28497 * asan.h (asan_no_sanitize_address_p): Remove the function.
28498 (sanitize_flags_p): New function.
28499 * builtins.def: Fix coding style.
28500 * common.opt: Use renamed enum value.
28501 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28502 * doc/extend.texi: Document no_sanitize attribute.
28503 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28504 to SANITIZE_UNDEFINED_NONDEFAULT.
28505 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28506 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28507 Use sanitize_flags_p.
28508 * gimplify.c (gimplify_function_tree): Likewise.
28509 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28510 * opts.c (parse_no_sanitize_attribute): New function.
28511 (common_handle_option): Use renamed enum value.
28512 * opts.h (parse_no_sanitize_attribute): Declare.
28513 * tree.c (sanitize_flags_p): New function.
28514 * tree.h: Declared here.
28515 * tsan.c: Use sanitize_flags_p.
28516 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28517 (instrument_mem_ref): Likewise.
28518 (instrument_bool_enum_load): Likewise.
28519 (do_ubsan_in_current_function): Remove the function.
28520 (pass_ubsan::execute): Use sanitize_flags_p.
28521 * ubsan.h: Remove do_ubsan_in_current_function
28522 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28523 (dump_function_to_file): Use it here.
28525 2017-06-13 Martin Jambor <mjambor@suse.cz>
28527 PR tree-optimization/80803
28528 PR tree-optimization/81063
28529 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28530 (propagate_subaccesses_across_link): Enqueue subtree whenever
28531 necessary instead of relying on the caller.
28533 2017-06-13 Martin Jambor <mjambor@suse.cz>
28535 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28536 that have a first_link.
28537 (sort_and_splice_var_accesses): Do not check first_link before
28539 (subtree_mark_written_and_enqueue): Likewise.
28540 (propagate_all_subaccesses): Likewise and do not stop at first
28541 parent with a first_link.
28543 2017-06-13 Martin Jambor <mjambor@suse.cz>
28545 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28548 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28550 * match.pd: New pattern.
28552 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28554 * tree-vrp.c (is_masked_range_test): New function.
28555 (register_edge_assert_for): Determine ranges for
28558 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28560 PR tree-optimization/67328
28561 * fold-const.c (maskable_range_p): New function.
28562 (build_range_check): Generate bittests if possible.
28564 2017-06-13 Martin Liska <mliska@suse.cz>
28566 * gimple-pretty-print.c (dump_probability): Add new argument.
28567 (dump_edge_probability): Dump both probability and count.
28568 (dump_gimple_label): Likewise.
28569 (dump_gimple_bb_header): Likewise.
28571 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28574 * config/avr/avr-devices.c: Fix indentation.
28575 * config/avr/gen-avr-mmcu-specs.c: Dito.
28577 2017-06-13 Richard Biener <rguenther@suse.de>
28579 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28580 instead get vector type from stmt_info.
28581 (vectorizable_reduction): Adjust. Remove dead code.
28583 2017-06-13 Richard Biener <rguenther@suse.de>
28585 PR middle-end/81065
28586 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28587 case of C * (x * C2 + C3).
28588 (fold_addr_of_array_ref_difference): Properly fold index difference.
28590 2017-06-12 David S. Miller <davem@davemloft.net>
28593 * config/sparc/sparc.md (return expander): Emit frame blockage if
28594 function uses alloca.
28596 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28598 * combine.c (make_field_assignment): Check len rather than the mode
28599 precision when calling force_to_mode.
28601 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28603 Support multilibs and devices that see flash in RAM address range.
28606 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28607 (avr_mcu_t) <flash_pm_offset>: New field.
28608 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28609 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28610 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28611 (AVR_TINY_PM_OFFSET): Remove macro.
28612 * config/avr/avr.opt (-mshort-calls): New option.
28613 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28614 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28615 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28616 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28617 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28618 instead of avr_arch->have_jmp_call.
28619 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28620 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28621 avr_arch->flash_pm_offset to define.
28622 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28623 new field flash_pm_offset. Add entry for avrxmega3.
28624 (avr_texinfo): Add entry for avrxmega3.
28625 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28626 attiny212, attiny214,
28627 attiny412, attiny414, attiny416, attiny417,
28628 attiny814, attiny816, attiny817,
28629 attiny1614, attiny1616, attiny1617,
28630 attiny3214, attiny3216, attiny3217.
28631 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28632 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28633 (avr_print_operand_address) [AVR_TINY]: Same.
28634 (avr_asm_init_sections) <readonly_data_section>: Only patch
28635 callback if avr_arch->flash_pm_offset = 0.
28636 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28637 for rodata if avr_arch->flash_pm_offset != 0.
28638 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28639 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28640 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28641 (m_options): Append opt_rcall.
28642 (m_dirnames): Append dir_rcall.
28643 * config/avr/t-multilib: Regenerate.
28645 * configure.ac [target=avr]: Check whether avrxmega3 default
28646 linker description file works as needed.
28647 * configure: Regenerate.
28648 * doc/avr-mmcu.texi: Regenerate.
28649 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28650 <__AVR_ARCH__>: Document avrxmega3 and 103.
28651 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28652 <__AVR_SHORT_CALLS__>: Document it.
28653 <__AVR_PM_BASE_ADDRESS__>: Document it.
28654 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28655 (AVR Variable Attributes) <progmem>: Document this is
28656 not needed for avrxmega3.
28657 (AVR Named Address Spaces) <__flash>: Dito.
28659 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28661 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28663 2017-06-12 Doug Rupp <rupp@adacore.com>
28665 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28666 Append vxworks-stdint.h to the tm_file list.
28667 * config/vxworks-stdint.h: New file.
28669 2017-06-12 Martin Liska <mliska@suse.cz>
28671 PR tree-optimization/81041
28672 * tree-profile.c (gimple_gen_ic_func_profiler):
28673 Create an extra BB in profile-generate
28674 (gimple_gen_time_profiler): Likewise.
28676 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28678 PR tree-optimization/81003
28679 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28680 (update_range_test): Use it instead of force_gimple_operand_gsi.
28682 2017-06-12 Richard Biener <rguenther@suse.de>
28684 PR tree-optimization/81053
28685 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28686 with backedge value not defined in loop. Simplify def stmt
28689 2017-06-11 Tom de Vries <tom@codesourcery.com>
28692 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28694 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28695 nvptx_cannot_force_const_mem.
28697 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28699 * opts.c (finish_options): Move test for flag_split_stack after
28700 it has been initialized.
28702 2017-06-11 Jason Merrill <jason@redhat.com>
28704 * tree.h (id_equal): New.
28705 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28706 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28707 instead of strcmp of IDENTIFIER_POINTER.
28709 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28711 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28712 (mark_all_inlined_calls_cdtor): Fix formating.
28713 (inline_transform): Rescale profile before inlining.
28715 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28717 * cgraph.h (cgraph_edge::clone): Update prototype.
28718 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28719 (cgraph_node::create_clone): Update.
28720 (cgraph_node::create_version_clone): Update.
28721 * tree-inline.c (copy_bb): Update.
28722 (expand_call_inline): Update.
28724 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28726 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28727 factored out from ...
28728 (rs6000_emit_prologue): ... here.
28730 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28732 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28733 factored out from ...
28734 (rs6000_emit_prologue): ... here.
28736 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28738 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28740 (handle_missing_profiles): Fix computation of tp_first_run.
28741 (counts_to_freqs): Do not touch freqs when count is 0.
28743 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28745 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28748 2017-06-10 Tom de Vries <tom@codesourcery.com>
28750 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28751 attributes): Document signal effective target.
28753 2017-06-10 Tom de Vries <tom@codesourcery.com>
28755 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28756 Document effective target stack_size.
28758 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28760 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28761 to the edit_context if they can be auto-applied.
28763 2017-06-9 Ian Lance Taylor <iant@golang.org>
28765 * opts.c (finish_options): If -fsplit-stack, disable implicit
28766 -forder-blocks-and-partition.
28767 * doc/invoke.texi (Optimize Options): Document that when using
28768 -fsplit-stack -forder-blocks-and-partition is not implicitly
28771 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28773 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28774 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28775 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28776 * builtins.def (abort, trap, unreachable): Declare cold.
28777 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28778 * tree-core.h (ECF_COLD): New.
28779 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28780 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28782 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28784 * predict.c (unlikely_executed_stmt_p): Cleanup.
28786 2017-06-09 Richard Biener <rguenther@suse.de>
28788 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28789 model if the ref is always written to.
28791 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28793 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28795 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28797 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28800 2017-06-09 Tom de Vries <tom@codesourcery.com>
28803 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28804 "target cannot support label values" when encountering LABEL_REF.
28806 2017-06-09 Martin Liska <mliska@suse.cz>
28808 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28809 (gimple_gen_ic_func_profiler): Emit direct comparison
28810 of __gcov_indirect_call_callee with NULL.
28811 (gimple_gen_time_profiler): Change probability from
28812 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28814 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28816 * profile.c (edge_gcov_counts): Turn to pointer.
28817 (compute_branch_probabilities, compute_branch_probabilities): Update.
28818 (branch_prob): Do not clear edge_gcov_count.
28819 * profile.h (edge_gcov_counts): Turn to pointer.
28820 (edge_gcov_count): Update.
28822 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28824 * gimple.h (gimple_check_failed): Mark cold.
28826 2017-06-09 Richard Biener <rguenther@suse.de>
28828 PR tree-optimization/66623
28829 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28830 refactor check_reduction into two parts, properly computing
28831 whether we have to check reduction validity for outer loop
28834 2017-06-09 Richard Biener <rguenther@suse.de>
28836 PR tree-optimization/79483
28837 * graphite-scop-detection.c (order): New global.
28838 (get_order): Compute bb to order mapping that satisfies code
28839 generation constraints.
28840 (cmp_pbbs): New helper.
28841 (build_scops): Start domwalk at entry block, sort generated
28844 2017-06-09 Richard Biener <rguenther@suse.de>
28846 PR middle-end/81007
28847 * ipa-polymorphic-call.c
28848 (ipa_polymorphic_call_context::restrict_to_inner_class):
28849 Skip FIELD_DECLs with error_mark_node type.
28850 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28853 2017-06-09 Martin Liska <mliska@suse.cz>
28855 * predict.c (struct branch_predictor): New struct.
28856 (test_prediction_value_range): New test.
28857 (predict_c_tests): New function.
28858 * selftest-run-tests.c (selftest::run_tests): Run the function.
28859 * selftest.h: Declare new tests.
28861 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28864 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28865 gen_add3_insn did not fail.
28866 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28867 r0, construct that number in a temporary reg and add that reg to r0.
28868 If asked to put the result in r0 as well, fail.
28870 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28872 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28873 for early expansion of vec_eqv.
28875 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28877 PR middle-end/81005
28878 * ubsan.c (instrument_null): Avoid pointless code temporary.
28879 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28881 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28885 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28888 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28891 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28893 * predict.c (unlikely_executed_edge_p): Move ahead.
28894 (probably_never_executed_edge_p): Use it.
28896 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28898 PR middle-end/79988
28899 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28900 gimple_call_builtin_p call.
28902 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28904 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28905 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28906 rtl_check_failed_type2, rtl_check_failed_code1,
28907 rtl_check_failed_code2, rtl_check_failed_code_mode,
28908 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28909 rtvec_check_failed_bounds, rtl_check_failed_flag,
28910 _fatal_insn_not_found, _fatal_insn): Likewise.
28911 * tree.h (tree_contains_struct_check_failed,
28912 tree_check_failed, tree_not_check_failed,
28913 tree_class_check_failed, tree_range_check_failed,
28914 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28915 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28916 tree_operand_check_failed, omp_clause_check_failed,
28917 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28920 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28922 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28923 flag_branch_probabilities.
28924 * ipa-inline.c (edge_badness): Likewise.
28925 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28926 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28927 * predict.c (maybe_hot_frequency_p): Likewise.
28928 (probably_never_executed): Likewise.
28929 * sched-ebb.c (schedule_ebbs): Likewise.
28930 * sched-rgn.c (find_single_block_region): Likewise.
28931 * tracer.c (tail_duplicate): Likewise.
28933 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28935 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28936 longer requires x_flag_profile_use.
28938 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28940 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28941 instead of flag_reorder_blocks_and_partition.
28942 * dbxout.c (dbxout_function_end): Likewise.
28943 * dwarf2out.c (gen_subprogram_die): Likewise.
28944 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28945 * hw-doloop.c (reorg_loops): Likewise.
28946 * varasm.c (assemble_start_function,
28947 assemble_end_function): Likewise.
28948 (decide_function_section): Do not check for
28949 flag_reorder_blocks_and_partition.
28951 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28953 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28955 (chkp_get_hard_register_fake_addr_expr): Ditto.
28956 (chkp_build_addr_expr): Add check for hard reg case.
28957 (chkp_parse_array_and_component_ref): Ditto.
28958 (chkp_find_bounds_1): Ditto.
28959 (chkp_process_stmt): Don't generate bounds store for
28962 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28964 * predict.c (maybe_hot_bb_p): Do not check profile status.
28965 (maybe_hot_edge_p): Likewise.
28966 (probably_never_executed): Check for zero counts even if profile
28968 (unlikely_executed_edge_p): New function.
28969 (unlikely_executed_stmt_p): New function.
28970 (unlikely_executed_bb_p): New function.
28971 (set_even_probabilities): Use unlikely predicates.
28972 (combine_predictions_for_bb): Likewise.
28973 (predict_paths_for_bb): Likewise.
28974 (predict_paths_leading_to_edge): Likewise.
28975 (determine_unlikely_bbs): New function.
28976 (estimate_bb_frequencies): Use it.
28977 (compute_function_frequency): Use zero counts even if profile is
28979 * profile-count.h: Fix typo.
28981 2017-08-08 Julia Koval <julia.koval@intel.com>
28983 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28984 _mm512_mask_cvtsepi16_storeu_epi8,
28985 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28986 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28987 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28988 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28989 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28990 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28991 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28992 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28993 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28994 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28995 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28996 __builtin_ia32_pmovuswb256mem_mask,
28997 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28998 __builtin_ia32_pmovwb512mem_mask): New builtins.
29000 2017-08-08 Julia Koval <julia.koval@intel.com>
29002 PR target/73350,80862
29003 * config/i386/subst.md (round): Fix round pattern.
29004 * config/i386/i386.c (ix86_erase_embedded_rounding):
29005 Fix erasing rounding for the fixed pattern.
29007 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
29009 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
29011 2017-06-08 Martin Liska <mliska@suse.cz>
29013 PR gcov-profile/80911
29014 * gcov.c (block_info::block_info): New constructor.
29016 2017-06-07 Carl Love <cel@us.ibm.com>
29018 * config/rs6000/rs6000-c: The return type of the following
29019 built-in functions was implemented as int not long long. Fix sign
29020 of return value for the unsigned version of vec_mulo and vec_mule.
29021 vector unsigned long long vec_bperm (vector unsigned long long,
29022 vector unsigned char)
29023 vector signed long long vec_mule (vector signed int,
29025 vector unsigned long long vec_mule (vector unsigned int,
29026 vector unsigned int)
29027 vector signed long long vec_mulo (vector signed int,
29029 vector unsigned long long vec_mulo (vector unsigned int,
29030 vector unsigned int)
29031 * doc/extend.texi: Fix the documentation for the built-in
29034 2017-06-07 Carl Love <cel@us.ibm.com>
29037 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
29040 2017-06-07 Carl Love <cel@us.ibm.com>
29042 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
29043 support, Generate doublehv for signed int/float for BE case only.
29045 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
29047 * doc/invoke.texi (mcx16): Rewrite.
29049 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29051 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
29052 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
29053 *mov<mode>_softfloat, and an anonymous splitter): Use
29054 nonimmediate_operand instead of rs6000_nonimmediate_operand.
29056 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29058 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
29060 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29061 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
29062 (rs6000_debug_reg_global): Adjust.
29063 (rs6000_init_hard_regno_mode_ok): Adjust.
29064 (rs6000_dbx_register_number): Adjust.
29065 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
29066 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29067 Remove SPE_ACC and SPEFSCR.
29068 (REG_ALLOC_ORDER): Ditto.
29069 (FRAME_POINTER_REGNUM): Change to 111.
29070 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
29071 (REG_CLASS_NAMES): Ditto.
29072 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
29073 (REGISTER_NAMES): Ditto.
29074 (ADDITIONAL_REG_NAMES): Ditto.
29075 (rs6000_reg_names): Ditto.
29076 * config/rs6000/rs6000.md: Renumber some register number
29079 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29081 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
29083 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29084 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
29086 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
29087 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29088 Delete the SPE high registers.
29089 (REG_ALLOC_ORDER): Ditto.
29090 (enum reg_class): Remove SPE_HIGH_REGS.
29091 (REG_CLASS_NAMES): Ditto.
29092 (REG_CLASS_CONTENTS): Delete the SPE high registers.
29093 (REGISTER_NAMES): Ditto.
29094 (rs6000_reg_names): Ditto.
29095 * doc/tm.texi.in: Remove SPE as example.
29096 * doc/tm.texi: Regenerate.
29098 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29100 * config/rs6000/8540.md (ppc8540_brinc): Delete.
29101 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
29102 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
29103 * config/rs6000/rs6000.md (type): Remove "brinc".
29105 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29107 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
29108 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
29109 * config/rs6000/linuxspe.h: Delete file.
29110 * config/rs6000/rs6000.md: Don't include spe.md.
29111 * config/rs6000/spe.h: Delete file.
29112 * config/rs6000/spe.md: Delete file.
29113 * config/rs6000/t-rs6000: Remove spe.md.
29115 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29117 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
29118 (reg_or_none500mem_operand): Delete.
29119 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
29120 instead of reg_or_none500mem_operand.
29122 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29124 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29125 handling of SPE flags.
29126 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
29128 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29130 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
29132 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
29133 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
29134 paired_divv2sf3): Similar.
29135 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
29136 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
29137 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
29139 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
29140 Rename the paired_* instruction patterns.
29141 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
29143 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
29144 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
29145 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
29146 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
29147 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
29148 PAIRED_VECTOR_MODE.
29149 (struct machine_function): Delete field spe_insn_chain_scanned_p.
29150 (spe_func_has_64bit_regs_p): Delete.
29151 (spe_expand_predicate_builtin): Delete.
29152 (spe_expand_evsel_builtin): Delete.
29153 (TARGET_DWARF_REGISTER_SPAN): Do not define.
29154 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
29155 (invalid_e500_subreg): Delete.
29156 (rs6000_legitimize_address): Always force_reg op2 as well, for
29157 paired single memory accesses.
29158 (rs6000_member_type_forces_blk): Delete.
29159 (rs6000_spe_function_arg): Delete.
29160 (rs6000_expand_unop_builtin): Delete SPE handling.
29161 (rs6000_expand_binop_builtin): Ditto.
29162 (spe_expand_stv_builtin): Delete.
29163 (bdesc_2arg_spe): Delete.
29164 (spe_expand_builtin): Delete.
29165 (spe_expand_predicate_builtin): Delete.
29166 (spe_expand_evsel_builtin): Delete.
29167 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
29168 (spe_init_builtins): Delete.
29169 (spe_func_has_64bit_regs_p): Delete.
29170 (savres_routine_name): Delete "info" parameter. Adjust callers.
29171 (rs6000_emit_stack_reset): Ditto.
29172 (rs6000_dwarf_register_span): Delete.
29173 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
29174 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
29175 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
29177 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
29179 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
29180 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
29181 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
29182 mulv2sf3, divv2sf3): Delete expanders.
29184 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29186 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
29188 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29190 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
29191 * config/rs6000/rs6000.c: Ditto.
29193 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29195 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
29196 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
29197 comparison_operator.
29199 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29201 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
29202 * config/rs6000/rs6000.opt: Ditto.
29203 * config/rs6000/t-rtems: Ditto.
29205 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29207 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29208 TARGET_E500_SINGLE by 0, simplify.
29209 * config/rs6000/rs6000.c: Ditto.
29210 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29211 (spe_build_register_parallel): Delete.
29212 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29213 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29214 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29215 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29216 (E500_CONVERT): Delete.
29217 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29219 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29221 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29222 * config/rs6000/dfp.md: Ditto.
29223 (negdd2, *negdd2_fpr): Merge.
29224 (absdd2, *absdd2_fpr): Merge.
29225 (negtd2, *negtd2_fpr): Merge.
29226 (abstd2, *abstd2_fpr): Merge.
29227 * config/rs6000/e500.h: Delete file.
29228 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29229 TARGET_FPRS by 1 and simplify.
29230 * config/rs6000/rs6000-c.c: Ditto.
29231 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29232 TARGET_DF_SPE by 0.
29233 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29235 * config/rs6000/rs6000.md: Ditto.
29236 (floatdidf2, *floatdidf2_fpr): Merge.
29237 (move_from_CR_gt_bit): Delete.
29238 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29239 (E500_CR_IOR_COMPARE): Delete.
29240 (All patterns that require !TARGET_FPRS): Delete.
29241 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29243 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29245 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29247 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29249 * graphds.c (add_edge): Intitialize edge's attached data.
29250 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29251 pointer parameter. Call pointed function on each edge during
29252 graph traversing. Skip traversing the edge when the function
29254 (graphds_dfs, graphds_scc): Ditto.
29255 (for_each_edge): New parameter. Pass the new parameter to callback
29257 * graphds.h (skip_edge_callback): New function pointer type.
29258 (graphds_dfs, graphds_scc): New function pointer parameter.
29259 (graphds_edge_callback, for_each_edge): New parameter.
29261 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29263 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29264 out code checking if runtime alias check is possible to below ...
29265 Call the new function.
29266 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29267 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29269 2017-06-07 Marek Polacek <polacek@redhat.com>
29272 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29273 TYPE_OVERFLOW_WRAPS check.
29275 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29277 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29278 if versioning is required.
29279 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29280 peeling with the check for versioning.
29282 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29284 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29285 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29286 Set true to new parameter if new ssa variable is defined.
29287 (vect_gen_vector_loop_niters): Refactor. Set range information
29288 for the new vector loop bound variable.
29289 (vect_do_peeling): Ditto.
29291 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29293 * tree-affine.c (ssa.h): Include header file.
29294 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29295 has wrapping overflow behavior.
29297 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29299 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29301 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29303 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29304 (tree_to_aff_combination): ... here.
29306 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29308 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29309 reg_pressure model function.
29310 (ivopts_global_cost_for_size): Delete.
29311 (determine_set_costs, iv_ca_recount_cost): Call new model function
29312 ivopts_estimate_reg_pressure.
29314 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29316 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29317 expensive than udiv. Remove floating point cases from mod.
29319 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29321 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29322 Increase idiv cost.
29324 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29326 * config/aarch64/aarch64.md
29327 (copysignsf3): Fix mask generation.
29329 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29331 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29333 (class dump_manager): Add register_dumps method.
29334 * dumpfile.c: Include langhooks.h.
29335 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29336 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29337 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29338 (dump_manager::dump_register): Start with 512 entries instead of 32.
29339 (dump_manager::register_dumps): New method.
29340 * toplev.c (general_init): Instead of invoking register_dumps
29341 langhook, invoke register_dumps method on the dump manager.
29342 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29345 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29347 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29348 Say that # requires an associated define_split to exist, and that
29349 the define_split must be suitable for use after register allocation.
29351 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29353 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29354 (compute_outgoing_frequencies): Also initialize zero counts.
29355 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29356 around loops; preserve more of profile when nothing changes.
29358 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29360 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29362 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29363 * config/arm/arm-cpu-cdata.h: Regenerate.
29364 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29365 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29366 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29367 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29368 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29370 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29372 * config/arm/t-rmprofile: Likewise.
29373 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29375 2017-06-06 David S. Miller <davem@davemloft.net>
29378 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29379 blockage if function uses alloca.
29381 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29383 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29384 New "uid" fields to hold pretty-print IDs of group and ref.
29385 Memory references are now identified as <group_id>:<ref_id>
29386 instead of using [random] addresses.
29387 (dump_mem_details): Simplify, no functional change.
29388 (dump_mem_ref): Simplify and make output more concise.
29389 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29390 (find_or_create_group): Initialize group uid.
29391 (record_ref): Initialize ref uid. Improve debug output.
29392 (prune_group_by_reuse, should_issue_prefetch_p,)
29393 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29394 (mark_nontemporal_store, determine_loop_nest_reuse):
29395 Improve debug output.
29397 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29399 * dbgcnt.def (prefetch): New debug counter.
29400 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29401 (schedule_prefetches): Stop issueing prefetches if debug counter
29404 2017-06-06 Tom de Vries <tom@codesourcery.com>
29406 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29407 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29409 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29411 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29412 Use aarch64_reg_or_zero predicate for operand 4.
29413 (aarch64_compare_and_swap<mode> define_insn_and_split):
29414 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29415 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29417 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29419 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29420 (arm_compute_save_core_reg_mask): This.
29421 (thumb1_compute_save_reg_mask): Rename into ...
29422 (thumb1_compute_save_core_reg_mask): This.
29423 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29424 (arm_compute_frame_layout): Likewise.
29426 2017-06-06 Richard Biener <rguenther@suse.de>
29428 PR tree-optimization/80974
29429 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29430 keep or clear leaders SSA info.
29432 2017-06-06 Tom de Vries <tom@codesourcery.com>
29434 * config/nvptx/nvptx.c (split_mode_p): New function.
29435 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29437 2017-06-06 Tom de Vries <tom@codesourcery.com>
29439 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29441 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29444 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29447 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29449 * shrink-wrap.c (handle_simple_exit): Update profile.
29450 (try_shrink_wrapping): Upate profile.
29452 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29454 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29455 (tree_guess_outgoing_edge_probabilities): New.
29456 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29457 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29459 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29461 * ipa-split.c (split_function): Initialize return bb profile.
29463 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29465 * profile.c (compute_branch_probabilities): Also initialize
29466 EXIT_BLOCK profile.
29468 2017-06-06 Richard Biener <rguenther@suse.de>
29470 PR tree-optimization/80928
29471 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29472 (vect_analyze_loop_operations): Properly guard analysis for
29474 (vect_transform_loop): Likewise.
29475 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29476 (vect_model_induction_cost): Do not cost for pure SLP.
29477 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29478 of induction in inner loop vectorization.
29479 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29480 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29481 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29483 (vect_analyze_slp_cost_1): Cost induction.
29484 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29485 (vect_get_slp_vect_defs): Likewise.
29486 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29487 (vect_transform_stmt): Handle SLP reductions.
29488 * tree-vectorizer.h (vectorizable_induction): Adjust.
29490 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29492 * config/rs6000/rs6000.c (make_resolver_func): Update
29493 init_lowered_empty_function call.
29495 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29497 * doc/invoke.texi: Document the -fprofile-abs-path option.
29498 * common.opt (fprofile-abs-path): New option.
29499 * gcov-io.h (gcov_write_filename): Declare.
29500 * gcov-io.c (gcov_write_filename): New function.
29501 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29502 * profile.c (output_location): Likewise.
29504 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29506 * shring-wrap.c: Revert accidental commit.
29508 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29510 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29512 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29514 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29516 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29517 profile in callgraph edge.
29518 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29519 (apply_scale): Likewise.
29520 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29521 Also scale profile when inlining function with zero profile.
29522 (initialize_cfun): Update exit block profile even when it is zero.
29523 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29524 when profile is read.
29526 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29528 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29529 (CLONE_*): New constants to define the processors we can generate
29530 code for with the target_clone attribute.
29531 (rs6000_clone_map): New array to identify which clone processors
29532 the current program is running on.
29533 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29534 target_clone attribute.
29535 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29536 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29537 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29538 (cpu_expand_builtin): Add support for target_clone attribute.
29539 (rs6000_valid_attribute_p): Allow "default" attribute.
29540 (get_decl_name): New debug function to simplify printing the
29541 current function name in debugging statements.
29542 (rs6000_clone_priority): New functions to support the target_clone
29543 attribute, and be able to generate code to switch between ISA 2.05
29544 through ISA 3.0 (power6 through power9).
29545 (rs6000_compare_version_priority): Likewise.
29546 (rs6000_get_function_versions_dispatcher): Likewise.
29547 (make_resolver_func): Likewise.
29548 (add_condition_to_bb): Likewise.
29549 (dispatch_function_versions): Likewise.
29550 (rs6000_generate_version_dispatcher_body): Likewise.
29551 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29552 (fusion_gpr_load_p): Fix a spacing issue.
29553 * doc/extend.texi (Common Function Attributes): Document that the
29554 PowerPC supports the target_clone attribute.
29556 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29558 * config/arm/arm.h: explain F symbol found in description of ARM
29559 register allocation in its legend.
29561 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29563 * config/mips/frame-header-opt.c: Include profile-count.h.
29564 * config/riscv/riscv.c: Include profile-count.h
29566 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29568 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29570 (sm_set_flag_if_changed): Add bbs field.
29571 (execute_sm_if_changed_flag_set): Pass BBS.
29572 (execute_sm): Update.
29574 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29576 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29579 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29581 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29582 (peephole2): New peephole2 to emit the above.
29583 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29585 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29587 * config/aarch64/aarch64.c (define_peephole2 above
29588 *sub_<shift>_<mode>): New peephole.
29590 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29592 * config/i386/i386.c (make_resolver_func): Update.
29593 * Makefile.in: Add profile-count.h and profile-count.o
29594 * auto-profile.c (afdo_indirect_call): Update to new API.
29595 (afdo_set_bb_count): Update.
29596 (afdo_propagate_edge): Update.
29597 (afdo_propagate_circuit): Update.
29598 (afdo_calculate_branch_prob): Update.
29599 (afdo_annotate_cfg): Update.
29600 * basic-block.h: Include profile-count.h
29601 (struct edge_def): Turn count to profile_count.
29602 (struct basic_block_def): Likewie.
29603 (REG_BR_PROB_BASE): Move to profile-count.h
29604 (RDIV): Move to profile-count.h
29605 * bb-reorder.c (max_entry_count): Turn to profile_count.
29606 (find_traces): Update.
29607 (rotate_loop):Update.
29608 (connect_traces):Update.
29609 (sanitize_hot_paths):Update.
29610 * bt-load.c (migrate_btr_defs): Update.
29611 * cfg.c (RDIV): Remove.
29612 (init_flow): Use alloc_block.
29613 (alloc_block): Uninitialize count.
29614 (unchecked_make_edge): Uninitialize count.
29615 (check_bb_profile): Update.
29616 (dump_edge_info): Update.
29617 (dump_bb_info): Update.
29618 (update_bb_profile_for_threading): Update.
29619 (scale_bbs_frequencies_int): Update.
29620 (scale_bbs_frequencies_gcov_type): Update.
29621 (scale_bbs_frequencies_profile_count): New.
29622 * cfg.h (update_bb_profile_for_threading): Update.
29623 (scale_bbs_frequencies_profile_count): Declare.
29624 * cfgbuild.c (compute_outgoing_frequencies): Update.
29625 (find_many_sub_basic_blocks): Update.
29626 * cfgcleanup.c (try_forward_edges): Update.
29627 (try_crossjump_to_edge): Update.
29628 * cfgexpand.c (expand_gimple_tailcall): Update.
29629 (construct_exit_block): Update.
29630 * cfghooks.c (verify_flow_info): Update.
29631 (dump_bb_for_graph): Update.
29632 (split_edge): Update.
29633 (make_forwarder_block): Update.
29634 (duplicate_block): Update.
29635 (account_profile_record): Update.
29636 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29637 (get_estimated_loop_iterations): Update.
29638 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29639 (single_likely_exit): Update.
29640 * cfgloopmanip.c (scale_loop_profile): Update.
29642 (set_zero_probability): Update.
29643 (lv_adjust_loop_entry_edge): Update.
29644 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29645 (purge_dead_edges): Update.
29646 (rtl_account_profile_record): Update.
29647 * cgraph.c (cgraph_node::create): Uninitialize count.
29648 (symbol_table::create_edge): Uninitialize count.
29649 (cgraph_update_edges_for_call_stmt_node): Update.
29650 (cgraph_edge::dump_edge_flags): Update.
29651 (cgraph_node::dump): Update.
29652 (cgraph_edge::maybe_hot_p): Update.
29653 * cgraph.h: Include profile-count.h
29654 (create_clone), create_edge, create_indirect_edge): Update.
29655 (cgraph_node): Turn count to profile_count.
29656 (cgraph_edge0: Likewise.
29657 (make_speculative, clone): Update.
29658 (create_edge): Update.
29659 (init_lowered_empty_function): Update.
29660 * cgraphclones.c (cgraph_edge::clone): Update.
29661 (duplicate_thunk_for_node): Update.
29662 (cgraph_node::create_clone): Update.
29663 * cgraphunit.c (cgraph_node::analyze): Update.
29664 (cgraph_node::expand_thunk): Update.
29665 * final.c (dump_basic_block_info): Update.
29666 * gimple-streamer-in.c (input_bb): Update.
29667 * gimple-streamer-out.c (output_bb): Update.
29668 * graphite.c (print_global_statistics): Update.
29669 (print_graphite_scop_statistics): Update.
29670 * hsa-brig.c: Include basic-block.h.
29671 * hsa-dump.c: Include basic-block.h.
29672 * hsa-gen.c (T sum_slice): Update.
29673 (convert_switch_statements):Update.
29674 * hsa-regalloc.c: Include basic-block.h.
29675 * ipa-chkp.c (chkp_produce_thunks): Update.
29676 * ipa-cp.c (struct caller_statistics): Update.
29677 (init_caller_stats): Update.
29678 (gather_caller_stats): Update.
29679 (ipcp_cloning_candidate_p): Update.
29680 (good_cloning_opportunity_p): Update.
29681 (get_info_about_necessary_edges): Update.
29682 (dump_profile_updates): Update.
29683 (update_profiling_info): Update.
29684 (update_specialized_profile): Update.
29685 (perhaps_add_new_callers): Update.
29686 (decide_about_value): Update.
29687 (ipa_cp_c_finalize): Update.
29688 * ipa-devirt.c (struct odr_type_warn_count): Update.
29689 (struct decl_warn_count): Update.
29690 (struct final_warning_record): Update.
29691 (possible_polymorphic_call_targets): Update.
29692 (ipa_devirt): Update.
29693 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29694 * ipa-icf.c (sem_function::merge): Update.
29695 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29696 * ipa-inline.c (compute_uninlined_call_time): Update.
29697 (compute_inlined_call_time): Update.
29698 (want_inline_small_function_p): Update.
29699 (want_inline_self_recursive_call_p): Update.
29700 (edge_badness): Update.
29701 (lookup_recursive_calls): Update.
29702 (recursive_inlining): Update.
29703 (inline_small_functions): Update.
29704 (dump_overall_stats): Update.
29705 (dump_inline_stats): Update.
29706 * ipa-profile.c (ipa_profile_generate_summary): Update.
29707 (ipa_propagate_frequency): Update.
29708 (ipa_profile): Update.
29709 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29710 * ipa-utils.c (ipa_merge_profiles): Update.
29711 * loop-doloop.c (doloop_modify): Update.
29712 * loop-unroll.c (report_unroll): Update.
29713 (unroll_loop_runtime_iterations): Update.
29714 * lto-cgraph.c (lto_output_edge): Update.
29715 (lto_output_node): Update.
29716 (input_node): Update.
29717 (input_edge): Update.
29718 (merge_profile_summaries): Update.
29719 * lto-streamer-in.c (input_cfg): Update.
29720 * lto-streamer-out.c (output_cfg): Update.
29721 * mcf.c (create_fixup_graph): Update.
29722 (adjust_cfg_counts): Update.
29723 (sum_edge_counts): Update.
29724 * modulo-sched.c (sms_schedule): Update.
29725 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29726 * predict.c (maybe_hot_count_p): Update.
29727 (probably_never_executed): Update.
29728 (dump_prediction): Update.
29729 (combine_predictions_for_bb): Update.
29730 (propagate_freq): Update.
29731 (handle_missing_profiles): Update.
29732 (counts_to_freqs): Update.
29733 (rebuild_frequencies): Update.
29734 (force_edge_cold): Update.
29735 * predict.h: Include profile-count.h
29736 (maybe_hot_count_p, counts_to_freqs): UPdate.
29737 * print-rtl-function.c: Do not include cfg.h
29738 * print-rtl.c: Include basic-block.h
29739 * profile-count.c: New file.
29740 * profile-count.h: New file.
29741 * profile.c (is_edge_inconsistent): Update.
29742 (correct_negative_edge_counts): Update.
29743 (is_inconsistent): Update.
29744 (set_bb_counts): Update.
29745 (read_profile_edge_counts): Update.
29746 (compute_frequency_overlap): Update.
29747 (compute_branch_probabilities): Update; Initialize and deinitialize
29749 (branch_prob): Update.
29750 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29751 (edge_gcov_count): New.
29752 (bb_gcov_count): New.
29753 * shrink-wrap.c (try_shrink_wrapping): Update.
29754 * tracer.c (better_p): Update.
29755 * trans-mem.c (expand_transaction): Update.
29756 (ipa_tm_insert_irr_call): Update.
29757 (ipa_tm_insert_gettmclone_call): Update.
29758 * tree-call-cdce.c: Update.
29759 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29760 (gimple_duplicate_sese_tail): Update.
29761 (gimple_account_profile_record): Update.
29762 (execute_fixup_cfg): Update.
29763 * tree-inline.c (copy_bb): Update.
29764 (copy_edges_for_bb): Update.
29765 (initialize_cfun): Update.
29766 (freqs_to_counts): Update.
29767 (copy_cfg_body): Update.
29768 (expand_call_inline): Update.
29769 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29770 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29771 (try_unroll_loop_completely): Update.
29772 (try_peel_loop): Update.
29773 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29774 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29775 * tree-ssa-loop-split.c (connect_loops): Update.
29776 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29777 * tree-ssa-reassoc.c (branch_fixup): Update.
29778 * tree-ssa-tail-merge.c (replace_block_by): Update.
29779 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29780 (compute_path_counts): Update.
29781 (update_profile): Update.
29782 (recompute_probabilities): Update.
29783 (update_joiner_offpath_counts): Update.
29784 (estimated_freqs_path): Update.
29785 (freqs_to_counts_path): Update.
29786 (clear_counts_path): Update.
29787 (ssa_fix_duplicate_block_edges): Update.
29788 (duplicate_thread_path): Update.
29789 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29790 (struct switch_conv_info): Update.
29791 * tree-tailcall.c (decrease_profile): Update.
29792 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29793 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29794 * value-prof.c (check_counter): Update.
29795 (gimple_divmod_fixed_value): Update.
29796 (gimple_mod_pow2): Update.
29797 (gimple_mod_subtract): Update.
29798 (gimple_ic_transform): Update.
29799 (gimple_stringop_fixed_value): Update.
29800 * value-prof.h (gimple_ic): Update.
29802 2017-06-02 Carl Love <cel@us.ibm.com>
29804 * config/rs6000/rs6000-c: Add support for built-in functions
29805 vector double vec_doublee (vector signed int);
29806 vector double vec_doublee (vector unsigned int);
29807 vector double vec_doublee (vector float);
29808 vector double vec_doubleh (vector signed int);
29809 vector double vec_doubleh (vector unsigned int);
29810 vector double vec_doubleh (vector float);
29811 vector double vec_doublel (vector signed int);
29812 vector double vec_doublel (vector unsigned int);
29813 vector double vec_doublel (vector float);
29814 vector double vec_doubleo (vector signed int);
29815 vector double vec_doubleo (vector unsigned int);
29816 vector double vec_doubleo (vector float);.
29817 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29818 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29820 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29821 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29822 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29824 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29825 vec_doublel, vec_doubleh.
29826 * doc/extend.texi: Update the built-in documentation file for the
29827 new built-in functions.
29829 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29832 * ipa-inline-analysis.c (free_growth_caches): Set
29833 edge_removal_hook_holder to NULL after removing it.
29835 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29837 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29838 comparision with zero.
29840 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29841 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29842 for early expansion of vec_min and vec_max builtins.
29843 (builtin_function_type): Add min/max unsigned variants to those
29844 identified as having unsigned arguments.
29846 2017-06-02 Olivier Hainque <hainque@adacore.com>
29848 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29850 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29852 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29853 Use VALL_F16 iterator rather than VALL.
29855 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29857 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29858 Emit CBNZ inside loop when doing a strong exchange and comparing
29859 against zero. Generate the CC flags after the loop.
29861 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29863 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29864 (dl_section_ref): New.
29865 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29866 On AIX, append an expression to subtract the size of the
29867 section length to dl_section_ref.
29869 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29871 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29872 for early expansion of vector absolute builtins.
29874 2017-06-02 Richard Biener <rguenther@suse.de>
29876 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29877 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29879 2017-06-02 Richard Biener <rguenther@suse.de>
29881 PR tree-optimization/80948
29882 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29883 stmt order as well.
29885 2017-06-02 Richard Biener <rguenther@suse.de>
29887 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29889 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29890 for inductions as relevant.
29892 2017-06-02 Richard Biener <rguenther@suse.de>
29894 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29895 (vectorizable_induction): ... this. Remove dead code.
29897 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29899 * builtins. (expand_builtin_alloca): Remove second parameter and
29900 infer its value from the first parameter instead.
29901 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29903 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29905 PR rtl-optimization/80903
29906 * loop-doloop.c (add_test): Unshare sequence.
29908 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29910 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29912 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29914 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29916 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29917 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29918 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29919 (xlogue_layout::get_stub_rtx): Make static.
29920 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29921 (xlogue_layout::compute_stub_managed_regs): Rename to...
29922 (xlogue_layout::count_stub_managed_regs): ...this.
29923 (xlogue_layout::is_stub_managed_reg): New function.
29924 (xlogue_layout::m_stub_names): Rename to...
29925 (xlogue_layout::s_stub_names): ...this, make static.
29926 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29927 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29928 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29929 xlogue_layout::s_stub_names): Instantiate statics.
29930 (stub_managed_regs): Remove.
29931 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29932 (disable_call_ms2sysv_xlogues): Rename to...
29933 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29934 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29936 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29937 change after reload_completed.
29938 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29940 (ix86_expand_prologue): Likewise.
29941 (ix86_expand_epilogue): Likewise.
29942 (ix86_expand_split_stack_prologue): Likewise.
29943 (ix86_compute_frame_layout): Remove frame parameter ...
29944 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29945 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29947 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29948 (ix86_frame): Move from here ...
29949 * config/i386/i386.h (ix86_frame): ... to here.
29950 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29951 complete ix86_frame data structure instead. Remove some_ld_name.
29953 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29955 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29956 symbols that hold a DECL_VALUE_EXPR.
29958 2017-06-01 Martin Jambor <mjambor@suse.cz>
29960 PR tree-optimization/80898
29961 * tree-sra.c (process_subtree_disqualification): Removed.
29962 (disqualify_candidate): Do not acll
29963 process_subtree_disqualification.
29964 (subtree_mark_written_and_enqueue): New function.
29965 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29966 RHS has been disqualified and re-queue LHS if necessary. Apart
29967 from that, ignore disqualified RHS.
29969 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29971 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29972 address fetch for z10 or later.
29974 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29976 * config/arc/arc.md (tst_movb): Add guard when splitting.
29978 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29980 * config/arc/arc.c (arc_can_eliminate): Test against
29981 arc_frame_pointer_needed.
29983 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29985 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29986 to prevent store reordering.
29987 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29988 (type): Add block type.
29989 (stack_tie): Define special instruction to be used in
29992 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29994 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29995 constraint. It is not valid for the pattern.
29996 (noncommutative_binary_comparison): Likewise.
29998 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30000 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
30003 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30005 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
30006 be used by the reg-alloc.
30008 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30010 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
30011 reg-alloc when having mul64 or mul32x16 instructions.
30012 (mulsidi3): Likewise.
30013 (umulsidi3): Likewise.
30014 (mulsi32x16): New pattern.
30015 (mulsi64): Likewise.
30016 (mulsidi64): Likewise.
30017 (umulsidi64): Likewise.
30018 (MUL32x16_REG): Define.
30019 (mul64_600): Use MUL32x16_REG.
30020 (mac64_600): Likewise.
30021 (umul64_600): Likewise.
30022 (umac64_600): Likewise.
30024 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30026 * config/arc/arc.md (mulsi3_700): Make it commutative.
30028 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
30030 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
30032 (*sign_extendsidi2_insn): Likewise for movstosw.
30034 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
30036 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
30037 the type of the input discriminant value. Convert the
30038 discriminant value of signedness vary.
30040 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
30042 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
30043 Add to -Wall section.
30045 2017-06-01 Richard Biener <rguenther@suse.de>
30047 PR middle-end/66313
30048 * fold-const.c (fold_plusminus_mult_expr): If the factored
30049 factor may be zero use a wrapping type for the inner operation.
30050 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
30051 and handle moved defs.
30052 (process_assignment): Properly guard the unary op case. Return a
30053 tri-state indicating that moving the stmt before the call may allow
30054 to continue. Pass through to_move.
30055 (find_tail_calls): Handle moving unrelated defs before
30058 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
30061 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
30062 splitter result in the canonical way.
30064 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
30066 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
30067 also for 32bit target. Update insn attributes.
30068 (zero-extendsidi2 splitter): Allow all registers for operand 1.
30070 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
30072 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
30073 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
30074 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
30075 (_mm_maskz_min_ss): New intrinsics.
30077 2017-05-31 Martin Liska <mliska@suse.cz>
30079 * tree-vect-loop.c (vect_create_epilog_for_reduction):
30080 Change comment style to one we normally use.
30081 (vectorizable_reduction): Likewise.
30082 (vectorizable_induction): Likewise.
30083 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
30084 (vectorizable_call): Likewise.
30085 (vectorizable_simd_clone_call): Likewise.
30086 (vectorizable_conversion): Likewise.
30087 (vectorizable_assignment): Likewise.
30088 (vectorizable_shift): Likewise.
30089 (vectorizable_operation): Likewise.
30090 (vectorizable_store): Likewise.
30091 (vectorizable_load): Likewise.
30092 * tree-vectorizer.h: Likewise.
30094 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
30096 * passes.c (emergency_dump_function): New.
30097 * tree-pass.h (emergency_dump_function): Declare.
30098 * plugin.c (plugins_internal_error_function): Remove.
30099 * plugin.h (plugins_internal_error_function): Remove declaration.
30100 * toplev.c (internal_error_function): New static function. Use it...
30101 (general_init): ...here.
30103 2017-05-31 Graham Markall <graham.markall@embecosm.com>
30105 * config/arc/arc.c (arc_print_operand): Handle constant operands.
30106 (arc_rtx_costs): Add costs for new patterns.
30107 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
30108 * config/arc/predicates.md: Add _1_2_3_operand predicate.
30110 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30112 * tree-ssa-strlen.c (get_next_strinfo): New function.
30113 (get_stridx_plus_constant): Use it.
30114 (zero_length_string): Likewise.
30115 (adjust_related_strinfos): Likewise.
30116 (adjust_last_stmt): Likewise.
30118 2017-05-31 Richard Biener <rguenther@suse.de>
30121 * config/i386/i386.c (ix86_expand_builtin): Remove assert
30122 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
30124 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30126 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
30127 loop_vinfo argument and use of dependence distance vectors.
30128 Check instead whether the two references differ only in their
30129 initial value and assume that they have the same alignment if the
30130 difference is a multiple of the vector alignment.
30131 (vect_analyze_data_refs_alignment): Update call accordingly.
30133 2017-05-31 Martin Liska <mliska@suse.cz>
30136 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
30138 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30140 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
30141 (create_intersect_range_checks): Move from ...
30142 * tree-data-ref.c (create_intersect_range_checks_index)
30143 (create_intersect_range_checks): ... to here.
30144 (create_runtime_alias_checks): New function factored from ...
30145 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
30146 here. Call above function.
30147 * tree-data-ref.h (create_runtime_alias_checks): New function.
30149 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30151 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
30152 segment length for dr_b and compute it in wide_int.
30154 2017-05-31 Richard Biener <rguenther@suse.de>
30156 PR tree-optimization/80906
30157 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
30158 and pass through iv_map.
30159 (copy_bb_and_scalar_dependences): Adjust.
30160 (translate_pending_phi_nodes): Likewise.
30161 (copy_loop_close_phi_args): Handle code-generating IVs instead
30164 2017-05-30 David Malcolm <dmalcolm@redhat.com>
30166 * diagnostic-color.c (color_dict): Add "type-diff".
30167 (parse_gcc_colors): Update comment.
30168 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
30169 -fdiagnostics-show-template-tree and -fno-elide-type.
30170 (GCC_COLORS): Add type-diff to example.
30172 (-fdiagnostics-show-template-tree): New.
30173 (-fno-elide-type): New.
30174 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
30175 the pp_format_decoder callback. Call any m_format_postprocessor's
30177 (pretty_printer::pretty_printer): Initialize
30178 m_format_postprocessor.
30179 (pretty_printer::~pretty_printer): Delete any
30180 m_format_postprocessor.
30181 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
30182 (class format_postprocessor): New class.
30183 (struct pretty_printer::format_decoder): Document the new parameters.
30184 (struct pretty_printer::m_format_postprocessor): New field.
30185 * tree-diagnostic.c (default_tree_printer): Update for new
30186 bool and const char ** params.
30187 * tree-diagnostic.h (default_tree_printer): Likewise.
30189 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
30191 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
30192 (lwa_operand): Delete rs6000_gen_cell_microcode test.
30193 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30194 rs6000_gen_cell_microcode code.
30195 (rs6000_final_prescan_insn): Delete.
30196 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
30197 "warn-cell-microcode" entries.
30198 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
30199 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
30200 throughout. Change cc_reg_not_micro_cr0_operand to
30201 cc_reg_not_cr0_operand throughout.
30202 (*extendhi<mode>2_noload): Delete.
30203 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30204 (mwarn-cell-microcode): Delete.
30205 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30206 -mgen-cell-microcode and -mwarn-cell-microcode.
30208 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30211 * config/i386/constraints.md (Yd): New constraint.
30213 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30214 and (?Yd, r) alternatives. Update insn attributes.
30215 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30216 and (?*Yd, r) alternatives. Update insn attributes.
30217 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30219 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30221 * gimplify.c (gimplify_modify_expr): Don't create a
30222 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30225 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30227 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30229 2017-05-30 Richard Biener <rguenther@suse.de>
30231 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30232 and reduc_def fields.
30233 (STMT_VINFO_REDUC_TYPE): New define.
30234 (STMT_VINFO_REDUC_DEF): Likewise.
30235 (vect_force_simple_reduction): Adjust prototype.
30236 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30237 (vect_is_simple_reduction): Remove check_reduction argument.
30238 (vect_force_simple_reduction): Adjust and set
30239 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30240 (vectorizable_reduction): Do not re-do reduction analysis
30241 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30242 * tree-parloops.c (gather_scalar_reductions): Adjust.
30244 2017-05-30 Richard Biener <rguenther@suse.de>
30246 PR middle-end/80901
30247 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30250 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30252 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30253 Introduce unknown_misalignment parameter and remove vf.
30254 (vect_peeling_hash_get_lowest_cost):
30255 Pass unknown_misalignment parameter.
30256 (vect_enhance_data_refs_alignment):
30257 Fix unsupportable data ref treatment.
30259 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30261 * tree-vect-data-refs.c (vect_get_data_access_cost):
30262 Workaround for SLP handling.
30263 (vect_enhance_data_refs_alignment):
30264 Compute costs for doing no peeling at all, compare to the best
30265 peeling costs so far and avoid peeling if cheaper.
30267 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30269 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30270 Return peeling info and set costs to zero for unlimited cost
30272 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30273 with unknown misalignment. Compute and costs for unknown
30274 misalignment, compare them to the costs for known misalignment
30275 and choose the cheapest for peeling.
30277 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30279 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30280 (vect_get_peeling_costs_all_drs): Create function.
30281 (vect_peeling_hash_get_lowest_cost):
30282 Use vect_get_peeling_costs_all_drs.
30283 (vect_peeling_supportable): Create function.
30284 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30286 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30288 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30289 DR_HAS_NEGATIVE_STEP.
30290 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30291 (vect_enhance_data_refs_alignment): Use.
30292 (vect_duplicate_ssa_name_ptr_info): Use.
30293 * tree-vectorizer.h (dr_misalignment): Use.
30294 (known_alignment_for_access_p): Use.
30296 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30299 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30301 (has_section_name): New function.
30303 2017-05-30 Martin Liska <mliska@suse.cz>
30306 * auto-profile.c (get_function_decl_from_block): Fix
30309 2017-05-30 Richard Biener <rguenther@suse.de>
30311 PR middle-end/80876
30312 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30314 2017-05-30 Martin Liska <mliska@suse.cz>
30316 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30317 * dumpfile.h (struct dump_file_info): Remove ctors.
30319 2017-05-30 Martin Liska <mliska@suse.cz>
30321 * predict.def: Fix GNU coding style.
30323 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30325 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30326 Mark 'to' argument with ATTRIBUTE_UNUSED.
30328 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30330 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30331 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30332 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30335 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30337 * doc/install.texi (Options specification): Restore entry of
30338 --enable-sjlj-exceptions.
30340 2017-05-27 Michael Eager <eager@eagercon.com>
30343 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30345 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30347 * config/microblaze/microblaze.h
30348 (FIXED_REGISTERS): Update in macro.
30349 (CALL_USED_REGISTERS): Update in macro.
30351 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30353 * doc/install.texi: Add links to macOS binary distributions.
30355 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30359 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30361 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30363 2017-05-26 Martin Liska <mliska@suse.cz>
30365 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30367 2017-05-26 Martin Liska <mliska@suse.cz>
30369 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30370 always leading ';; '.
30371 (dump_bb_info): Likewise.
30372 (brief_dump_cfg): Likewise.
30373 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30374 * dumpfile.c: Remove usage of TDF_VERBOSE.
30375 * dumpfile.h (enum dump_kind): Likewise.
30376 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30377 * print-tree.c (debug_verbose): Remove.
30378 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30379 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30380 * tree-diagnostic.c (default_tree_printer): Replace
30381 TDF_DIAGNOSTIC with TDF_SLIM.
30383 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30385 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30386 in parameter loop, rather than loop_vinfo.
30387 (create_intersect_range_checks): Ditto.
30388 (vect_create_cond_for_alias_checks): Update call to above functions.
30390 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30392 PR tree-optimization/80815
30393 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30394 for merging runtime alias checks. Handle negative DR_STEPs.
30396 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30398 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30400 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30401 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30402 out code pruning runtime alias checks.
30403 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30404 factored out from above.
30405 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30407 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30409 (prune_runtime_alias_test_list): New decalaration.
30411 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30413 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30414 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30415 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30416 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30417 (operator==, comp_dr_with_seg_len_pair): Ditto.
30418 (vect_prune_runtime_alias_test_list): Ditto.
30420 2017-05-26 Martin Liska <mliska@suse.cz>
30423 * params.def: Bound partial-inlining-entry-probability param.
30425 2017-05-26 Marek Polacek <polacek@redhat.com>
30428 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30431 2017-05-26 Richard Biener <rguenther@suse.de>
30433 PR tree-optimization/80842
30434 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30437 2017-05-26 Richard Biener <rguenther@suse.de>
30439 PR tree-optimization/80844
30440 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30442 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30444 * doc/md.texi (Machine Constraints): Update x86 family
30445 machine constraints section to match 'config/i386/constraints.md'.
30447 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30449 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30451 2017-05-25 Nathan Sidwell <nathan@acm.org>
30453 * doc/invoke.texi (--enable-languages): Update documentation.
30455 2017-05-25 Martin Liska <mliska@suse.cz>
30457 * dumpfile.c: Add TDF_FOLDING.
30458 * dumpfile.h (enum dump_kind): Likewise.
30459 * genmatch.c (dt_simplify::gen_1): Use it.
30461 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30463 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30465 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30467 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30468 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30470 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30472 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30473 * match.pd (X == C): Rewrite it here.
30474 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30475 with_certain_nonzero_bits2): New predicates.
30476 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30478 2017-05-24 Nathan Sidwell <nathan@acm.org>
30480 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30483 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30486 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30488 * config/powerpcspe: New port. Files are copied from the rs6000
30489 port, with "rs6000" in filenames replaced by "powerpcspe".
30491 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30493 PR rtl-optimization/80754
30494 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30496 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30498 * config/sparc/sparc.md (length): Return the correct value for -mflat
30499 sibcalls to match output_sibcall.
30501 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30505 * config/rs6000/rs6000.c (struct machine_function): Add new field
30507 (rs6000_get_separate_components): Init that field, use it.
30508 (rs6000_components_for_bb): Use the field.
30510 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30512 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30514 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30516 PR middle-end/80823
30517 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30519 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30522 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30523 address against address_operand predicate.
30524 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30526 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30528 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30529 parameters passed indirectly.
30531 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30533 * config/i386/i386.md (*movdi_internal): Remove SSE4
30534 alternative 18 (?r, *v). Update insn attributes.
30535 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30536 Update insn attributes.
30537 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30538 Update insn attributes.
30539 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30540 alternative 1 (r, v). Remove isa attribute.
30541 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30542 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30543 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30545 2017-05-23 Tom de Vries <tom@codesourcery.com>
30547 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30550 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30552 * cgraphunit.c (symbol_table::process_new_functions): Update.
30553 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30554 (inline_generate_summary): Rename to ...
30555 (ipa_fn_summary_generate): ... this one.
30556 (inline_read_summary): Rename to ...
30557 (ipa_fn_summary_read): ... this one.
30558 (inline_write_summary): Rename to ...
30559 (ipa_fn_summary_write): ... this one.
30560 (inline_free_summary): Rename to ...
30561 (ipa_free_fn_summary): ... this one.
30562 (pass_data_local_fn_summary, pass_local_fn_summary,
30563 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30564 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30565 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30566 make_pass_ipa_fn_summary): New.
30567 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30568 inline_write_summary, inline_free_summary): Remove.
30569 (ipa_free_fn_summary) : New.
30570 * ipa-inline.c (ipa_inline): Update.
30571 (pass_ipa_inline): Do not generate summaries.
30572 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30574 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30575 and add pass_ipa_fn_summary.
30576 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30578 (make_pass_inline_parameters): Remove.
30580 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30582 * omp-low.c (struct omp_context): Remove "default_kind" member.
30585 * omp-offload.c (execute_oacc_device_lower): Remove the
30586 parallelism dimensions function attributes for unparallelized
30587 OpenACC kernels constructs.
30589 2017-05-23 Martin Liska <mliska@suse.cz>
30591 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30593 (cgraph_edge::make_speculative): Likewise.
30594 (cgraph_edge::resolve_speculation): Likewise.
30595 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30596 (cgraph_node::dump): Likewise.
30597 * cgraph.h: Likewise.
30598 * cgraphunit.c (analyze_functions): Likewise.
30599 (symbol_table::compile): Likewise.
30600 * ipa-cp.c (print_all_lattices): Likewise.
30601 (determine_versionability): Likewise.
30602 (initialize_node_lattices): Likewise.
30603 (ipcp_verify_propagated_values): Likewise.
30604 (estimate_local_effects): Likewise.
30605 (update_profiling_info): Likewise.
30606 (create_specialized_node): Likewise.
30607 (perhaps_add_new_callers): Likewise.
30608 (decide_about_value): Likewise.
30609 (decide_whether_version_node): Likewise.
30610 (identify_dead_nodes): Likewise.
30611 (ipcp_store_bits_results): Likewise.
30612 * ipa-devirt.c (dump_targets): Likewise.
30613 (ipa_devirt): Likewise.
30614 * ipa-icf.c (sem_item::dump): Likewise.
30615 (sem_function::equals): Likewise.
30616 (sem_variable::equals): Likewise.
30617 (sem_item_optimizer::read_section): Likewise.
30618 (sem_item_optimizer::execute): Likewise.
30619 (congruence_class::dump): Likewise.
30620 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30621 (dump_inline_summary): Likewise.
30622 (estimate_node_size_and_time): Likewise.
30623 (inline_analyze_function): Likewise.
30624 * ipa-inline-transform.c (inline_call): Likewise.
30625 * ipa-inline.c (report_inline_failed_reason): Likewise.
30626 (want_early_inline_function_p): Likewise.
30627 (edge_badness): Likewise.
30628 (update_edge_key): Likewise.
30629 (inline_small_functions): Likewise.
30630 * ipa-profile.c (ipa_profile): Likewise.
30631 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30632 (ipa_make_edge_direct_to_target): Likewise.
30633 (remove_described_reference): Likewise.
30634 (ipa_impossible_devirt_target): Likewise.
30635 (propagate_controlled_uses): Likewise.
30636 (ipa_print_node_params): Likewise.
30637 (ipcp_transform_function): Likewise.
30638 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30639 (propagate_pure_const): Likewise.
30640 * ipa-reference.c (generate_summary): Likewise.
30641 (read_write_all_from_decl): Likewise.
30642 (propagate): Likewise.
30643 (ipa_reference_read_optimization_summary): Likewise.
30644 * ipa-utils.c (ipa_merge_profiles): Likewise.
30645 * ipa.c (walk_polymorphic_call_targets): Likewise.
30646 (symbol_table::remove_unreachable_nodes): Likewise.
30647 (ipa_single_use): Likewise.
30648 * passes.c (execute_todo): Likewise.
30649 * predict.c (drop_profile): Likewise.
30650 * symtab.c (symtab_node::get_dump_name): New function.
30651 (symtab_node::dump_name): Likewise.
30652 (symtab_node::dump_asm_name): Likewise.
30653 (symtab_node::dump_references): Likewise.
30654 (symtab_node::dump_referring): Likewise.
30655 (symtab_node::dump_base): Likewise.
30656 (symtab_node::debug_symtab): Likewise.
30657 * tree-sra.c (convert_callers_for_node): Likewise.
30658 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30659 * value-prof.c (init_node_map): Likewise.
30661 2017-05-23 Martin Liska <mliska@suse.cz>
30663 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30664 and symtab_node::debug_symtab to symbol_table::debug.
30665 * cgraphunit.c (analyze_functions): Use the renamed function.
30666 (symbol_table::compile): Likewise.
30667 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30668 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30669 * passes.c (execute_todo): Likewise.
30670 * symtab.c (symbol_table::dump): New function.
30671 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30673 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30675 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30676 that nonconst implies exec.
30678 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30680 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30681 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30682 (inline_edge_summary_vec): Turn into ...
30683 (ipa_call_summaries): ... this one.
30684 (redirect_to_unreachable, edge_set_predicate,
30685 evaluate_properties_for_edge, inline_summary_alloc,
30686 reset_ipa_call_summary, reset_inline_summary,
30687 inline_summary_t::duplicate): Update.
30688 (inline_edge_duplication_hook): Turn to ...
30689 (ipa_call_summary_t::duplicate): ... this one.
30690 (inline_edge_removal_hook): Turn to ...
30691 (ipa_call_summary_t::remove): ... this one.
30692 (dump_inline_edge_summary): Turn to ...
30693 (dump_ipa_call_summary): ... this one.
30694 (estimate_function_body_sizes): Update.
30695 (inline_update_callee_summaries): Update.
30696 (remap_edge_change_prob): Update.
30697 (remap_edge_summaries): Update.
30698 (inline_merge_summary): Update.
30699 (do_estimate_edge_time): Update.
30700 (inline_generate_summary): Update.
30701 (inline_read_section): Update.
30702 (inline_read_summary): Update.
30703 (inline_free_summary): Update.
30704 * ipa-inline.c (can_inline_edge_p): Update.
30705 (compute_inlined_call_time): Update.
30706 (want_inline_small_function_p): Update.
30707 (edge_badness): Update.
30708 (early_inliner): Update.
30709 * ipa-inline.h (inline_edge_summary): Turn to ...
30710 (ipa_call_summary): ... this one.
30711 (ipa_call_summary_t): New class.
30712 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30713 (ipa_call_summaries): New.
30714 (inline_edge_summary): Remove.
30715 (estimate_edge_growth): Update.
30716 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30717 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30718 * ipa-split.c (execute_split_functions): Update.
30719 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30721 2017-05-23 Tom de Vries <tom@codesourcery.com>
30723 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30724 attributes): Document rdrand effective target.
30726 2017-05-23 Tom de Vries <tom@codesourcery.com>
30728 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30729 attributes): Sort alphabetically.
30731 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30733 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30735 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30738 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30739 V2DF/V2DI splat into two separate patterns, one that handles
30740 registers, and the other that only handles memory. Drop support
30741 for splatting from a GPR on ISA 2.07 and then splitting the
30742 splat into direct move and splat.
30743 (vsx_splat_<mode>_reg): Likewise.
30744 (vsx_splat_<mode>_mem): Likewise.
30746 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30748 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30750 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30752 PR middle-end/80809
30753 * omp-low.c (finish_taskreg_remap): New function.
30754 (finish_taskreg_scan): If unit size of ctx->record_type
30755 is non-constant, unshare the size expression and replace
30756 decls in it with possible outer var refs.
30758 PR middle-end/80809
30759 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30760 GOVD_SHARED rather than GOVD_PRIVATE with it.
30761 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30762 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30764 PR middle-end/80853
30765 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30766 as last argument to build_outer_var_ref for pointer bases of array
30767 section reductions.
30769 2017-05-19 Martin Sebor <msebor@redhat.com>
30771 * print-tree.c (print_node): Print DECL_READ_P flag.
30773 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30775 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30776 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30777 * cgraph.c: Likewise.
30778 * cgraphunit.c: Likewise.
30779 * gengtype.c: Likewise.
30780 * ipa-cp.c: Likewise.
30781 * ipa-devirt.c: Likewise.
30782 * ipa-icf.c: Likewise.
30783 * ipa-predicate.c: Likewise.
30784 * ipa-profile.c: Likewise.
30785 * ipa-prop.c: Likewise.
30786 * ipa-split.c: Likewise.
30788 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30789 edge_predicate_pool, dump_inline_hints,
30790 inline_summary::account_size_time, redirect_to_unreachable,
30791 edge_set_predicate, set_hint_predicate,
30792 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30793 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30794 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30795 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30796 ipa_call_summary_t::remove, initialize_growth_caches,
30797 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30798 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30799 mark_modified, unmodified_parm_1, unmodified_parm,
30800 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30801 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30802 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30803 will_be_nonconstant_predicate, record_modified_bb_info,
30804 get_minimal_bb, record_modified, param_change_prob,
30805 phi_result_unknown_predicate, predicate_for_phi_result,
30806 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30807 estimate_function_body_sizes, compute_inline_parameters,
30808 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30809 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30810 inline_update_callee_summaries, remap_edge_change_prob,
30811 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30812 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30813 inline_analyze_function, inline_summary_t::insert,
30814 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30815 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30816 inline_free_summary): Move to ipa-fnsummary.h
30817 (predicate_t): Remove.
30818 * ipa-fnsummary.c: New file.
30819 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30820 (enum inline_hints_vals, inline_hints, agg_position_info,
30821 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30822 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30823 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30824 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30825 inline_read_summary, inline_write_summary, inline_free_summary,
30826 inline_analyze_function, initialize_inline_failed,
30827 inline_merge_summary, inline_update_overall_summary,
30828 compute_inline_parameters): Move to ipa-fnsummary.h
30829 * ipa-fnsummary.h: New file.
30830 * ipa-inline-transform.h: Include ipa-inline.h.
30831 * ipa-inline.c: LIkewise.
30833 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30835 * ipa-inline.c (edge_badness): Use inlined_time instead of
30836 inline_summaries->get.
30838 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30840 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30842 2017-05-22 Nathan Sidwell <nathan@acm.org>
30844 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30845 (fdump-lang): Document 'raw' option.
30846 * dumpfile.h (TDI_tu): Delete.
30847 * dumpfile.c (dump_files): Remove translation-unit.
30848 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30850 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30852 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30853 command option from $(AWK) call.
30854 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30855 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30856 [FORMAT]: Remove handling of variable.
30857 * config/avr/t-multilib: Regenerate.
30859 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30861 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30863 (dump_inline_summary): Do not print self_time.
30864 (estimate_function_body_sizes): Do not set self_time.
30865 (compute_inline_parameters): Likewise.
30866 (inline_read_section, inline_write_summary): Do not stream self_time.
30867 * ipa-inline.h (inline_summary): Drop self_time.
30869 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30871 * ipa-inline-analysis.c (account_size_time): Rename to ...
30872 (inline_summary::account_size_time): ... this one.
30873 (reset_ipa_call_summary): Turn to ...
30874 (ipa_call_summary::reset): ... this one.
30875 (reset_inline_summary): Turn to ...
30876 (inline_summary::reset): ... this one.
30877 (inline_summary_t::remove): Update.
30878 (inline_summary_t::duplicate): Update.
30879 (ipa_call_summary_t::remove): Update.
30880 (dump_inline_summary): Update.
30881 (estimate_function_body_sizes): Update.
30882 (compute_inline_parameters): Update.
30883 (estimate_node_size_and_time): Update.
30884 (inline_merge_summary): Update.
30885 (inline_update_overall_summary): Update.
30886 (inline_read_section): Update.
30887 (inline_write_summary): Update.
30888 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30889 add account_size_time and reset member functions.
30890 (ipa_call_summary): Add reset function.
30891 * ipa-predicate.h (predicate::operator &): Constify.
30893 2017-05-22 Richard Biener <rguenther@suse.de>
30895 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30897 2017-05-19 Jason Merrill <jason@redhat.com>
30899 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30901 2017-05-19 Marek Polacek <polacek@redhat.com>
30904 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30905 TYPE_OVERFLOW_WRAPS checks.
30907 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30909 * tree-core.h (enum omp_clause_default_kind): Add
30910 "OMP_CLAUSE_DEFAULT_PRESENT".
30911 * tree-pretty-print.c (dump_omp_clause): Handle it.
30912 * gimplify.c (enum gimplify_omp_var_data): Add
30913 "GOVD_MAP_FORCE_PRESENT".
30914 (gimplify_adjust_omp_clauses_1): Map it to
30915 "GOMP_MAP_FORCE_PRESENT".
30916 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30918 * gimplify.c (oacc_default_clause): Clarify.
30920 2017-05-19 Nathan Sidwell <nathan@acm.org>
30922 LANG_HOOK_REGISTER_DUMPS
30923 * toplev.c (general_init): Call register dump lang hook.
30924 * doc/invoke.texi: Document -fdump-lang option family.
30925 * dumpfile.c (dump_files): Remove class dump here.
30926 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30927 * dumpfile.h (tree_dump_index): Remove TDI_class.
30928 * langhooks-def.h (lhd_register_dumps): Declare.
30929 (LANG_HOOKS_REGISTER_DUMPS): Define.
30930 (LANG_HOOKS_INITIALIZER): Add it.
30931 * langhooks.c (lhd_register_dumps): Define.
30932 * langhooks.h (struct lang_hooks): Add register_dumps.
30934 2017-05-19 Nathan Sidwell <nathan@acm.org>
30936 * context.h (context::set_passes): New.
30937 * context.c (context::context): Do not create pass manager.
30938 * toplev.c (general_init): Create pass manager here.
30940 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30942 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30943 use this splitter if two add or or instructions would also work for
30944 the constant we want to generate.
30946 2017-05-19 Richard Biener <rguenther@suse.de>
30949 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30950 predicate evaluation.
30952 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30954 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30956 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30959 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30961 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30962 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30963 (inline_edge_summary_vec): Turn into ...
30964 (ipa_call_summaries): ... this one.
30965 (redirect_to_unreachable, edge_set_predicate,
30966 evaluate_properties_for_edge, inline_summary_alloc,
30967 reset_ipa_call_summary, reset_inline_summary,
30968 inline_summary_t::duplicate): Update.
30969 (inline_edge_duplication_hook): Turn to ...
30970 (ipa_call_summary_t::duplicate): ... this one.
30971 (inline_edge_removal_hook): Turn to ...
30972 (ipa_call_summary_t::remove): ... this one.
30973 (dump_inline_edge_summary): Turn to ...
30974 (dump_ipa_call_summary): ... this one.
30975 (estimate_function_body_sizes): Update.
30976 (inline_update_callee_summaries): Update.
30977 (remap_edge_change_prob): Update.
30978 (remap_edge_summaries): Update.
30979 (inline_merge_summary): Update.
30980 (do_estimate_edge_time): Update.
30981 (inline_generate_summary): Update.
30982 (inline_read_section): Update.
30983 (inline_read_summary): Update.
30984 (inline_free_summary): Update.
30985 * ipa-inline.c (can_inline_edge_p): Update.
30986 (compute_inlined_call_time): Update.
30987 (want_inline_small_function_p): Update.
30988 (edge_badness): Update.
30989 (early_inliner): Update.
30990 * ipa-inline.h (inline_edge_summary): Turn to ...
30991 (ipa_call_summary): ... this one.
30992 (ipa_call_summary_t): New class.
30993 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30994 (ipa_call_summaries): New.
30995 (inline_edge_summary): Remove.
30996 (estimate_edge_growth): Update.
30997 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30998 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30999 * ipa-split.c (execute_split_functions): Update.
31000 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
31002 2017-05-19 Richard Biener <rguenther@suse.de>
31004 PR middle-end/80764
31005 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
31007 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
31009 * config/rs6000/rs6000.c (struct machine_function): Add field
31010 fpr_is_wrapped_separately.
31011 (rs6000_get_separate_components): Use 64 components. Handle the
31012 new FPR components.
31013 (rs6000_components_for_bb): Handle the FPR components.
31014 (rs6000_emit_prologue_components): Handle the FPR components.
31015 (rs6000_emit_epilogue_components): Handle the FPR components.
31016 (rs6000_set_handled_components): Handle the FPR components.
31017 (rs6000_emit_prologue): Don't output prologue code for those FPRs
31018 that are already separately shrink-wrapped.
31019 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
31020 that are already separately shrink-wrapped.
31022 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31025 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
31028 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
31029 (define_peephole2 for Altivec d-form load): Add peepholes to catch
31030 cases where the register allocator uses a move and an offsettable
31031 memory operation to/from a FPR register on ISA 2.06/2.07.
31032 (define_peephole2 for Altivec d-form store): Likewise.
31034 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
31037 * config/i386/mmx.md (*mov<mode>_internal): Enable
31038 alternatives 11, 12, 13 and 14 also for 32bit targets.
31039 Remove alternatives 15, 16, 17 and 18.
31040 * config/i386/sse.md (vec_concatv2di): Change
31041 alternative (!x, *y) to (x, ?!*Yn).
31043 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
31045 * dumpfile.h (enum dump_kind): Remove stray comma.
31047 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31049 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
31050 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
31051 predicate::num_conditions
31052 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
31053 (CHANGED): turn into predicate::changed.
31054 (agg_position_info): Move to ipa-predicate.h
31055 (add_condition, predicate::add_clause, predicate::operator &=,
31056 predicate::or_with, predicate::evaluate, predicate::probability,
31057 dump_condition, dump_clause, predicate::dump,
31058 predicate::remap_after_duplication, predicate::remap_after_inlining,
31059 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
31060 (evaluate_conditions_for_known_args): Update.
31061 (set_cond_stmt_execution_predicate): Update.
31062 * ipa-inline.h: Include ipa-predicate.h
31063 (condition, inline_param_summary, conditions, agg_position_info,
31064 predicate): Move to ipa-predicate.h
31065 * ipa-predicate.c: New file.
31066 * ipa-predicate.h: New file.
31068 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
31070 * final.c (leaf_function_p): Check we are not in a sequence.
31072 2017-05-18 Martin Liska <mliska@suse.cz>
31074 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
31075 * dumpfile.c (dump_register): Use new enum dump_kind.
31076 (get_dump_file_name): Likewise.
31077 (dump_enable_all): Likewise.
31078 (dump_switch_p_1): Likewise.
31079 (enable_rtl_dump_file): Remove usage of TDF_RTL.
31080 * dumpfile.h (enum dump_kind): New enum type.
31081 (struct dump_file_info): Create constructor and
31082 format fields and comments.
31083 * passes.c (pass_manager::register_one_dump_file):
31085 * statistics.c (statistics_early_init): Likewise.
31086 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
31087 TDF_TREE with TDF_SLIM.
31088 (gather_memory_references_ref): Likewise.
31090 2017-05-18 Martin Liska <mliska@suse.cz>
31092 * vec.h (struct vnull): Use it.
31094 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31096 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
31097 (true_predicate, false_predicate, true_predicate_p,
31098 false_predicate_p): Remove.
31099 (single_cond_predicate, not_inlined_predicate): Turn to member function
31101 (add_condition): Update.
31102 (add_clause): Turn to...
31103 (predicate::add_clause): ... this one; update; allow passing NULL
31105 (and_predicates): Turn to ...
31106 (predicate::operator &=): ... this one.
31107 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
31108 (or_predicates): Turn to ...
31109 (predicate::or_with): ... this one.
31110 (evaluate_predicate): Turn to ...
31111 (predicate::evaluate): ... this one.
31112 (predicate_probability): Turn to ...
31113 (predicate::probability): ... this one.
31114 (dump_condition): Update.
31115 (dump_predicate): Turn to ...
31116 (predicate::dump): ... this one.
31117 (account_size_time): Update.
31118 (edge_set_predicate): Update.
31119 (set_hint_predicate): UPdate.
31120 (evaluate_conditions_for_known_args): Update.
31121 (evaluate_properties_for_edge): Update.
31122 (remap_predicate_after_duplication): Turn to...
31123 (predicate::remap_after_duplication): ... this one.
31124 (remap_hint_predicate_after_duplication): Update.
31125 (inline_summary_t::duplicate): UPdate.
31126 (dump_inline_edge_summary): Update.
31127 (dump_inline_summary): Update.
31128 (set_cond_stmt_execution_predicate): Update.
31129 (set_switch_stmt_execution_predicate): Update.
31130 (compute_bb_predicates): Update.
31131 (will_be_nonconstant_expr_predicate): Update.
31132 (will_be_nonconstant_predicate): Update.
31133 (phi_result_unknown_predicate): Update.
31134 (predicate_for_phi_result): Update.
31135 (array_index_predicate): Update.
31136 (estimate_function_body_sizes): Update.
31137 (estimate_node_size_and_time): Update.
31138 (estimate_ipcp_clone_size_and_time): Update.
31139 (remap_predicate): Rename to ...
31140 (predicate::remap_after_inlining): ... this one.
31141 (remap_hint_predicate): Update.
31142 (inline_merge_summary): Update.
31143 (inline_update_overall_summary): Update.
31144 (estimate_size_after_inlining): Update.
31145 (read_predicate): Rename to ...
31146 (predicate::stream_in): ... this one.
31147 (read_inline_edge_summary): Update.
31148 (write_predicate): Rename to ...
31149 (predicate::stream_out): ... this one.
31150 (write_inline_edge_summary): Update.
31151 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
31152 (clause_t): Turn to uint32_t
31153 (predicate): Turn to class; implement constructor and operators
31155 (size_time_entry): Update.
31156 (inline_summary): Update.
31157 (inline_edge_summary): Update.
31159 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
31161 * fold-const.c (fold_binary_loc): Move transformation...
31162 * match.pd (C - X CMP X): ... here.
31164 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
31166 * config/sparc/sparc.c (sparc_option_override): Set function
31167 alignment for -mcpu=niagara7 to 64 to match the I$ line.
31168 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
31170 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
31172 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
31174 2017-05-18 Marek Polacek <polacek@redhat.com>
31177 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
31178 (pass_ubsan::execute): Call gimple_assign_single_p instead of
31179 gimple_assign_load_p.
31181 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
31183 PR middle-end/80692
31184 * real.c (do_compare): Give decimal_do_compare preference over
31185 comparing just the signs.
31187 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
31189 * doc/md.texi (Canonicalization of Instructions): Describe the
31190 canonical form of instructions that inherently set a condition
31193 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
31195 PR middle-end/80775
31196 * tree-cfg.c: Move deletion of unreachable case statements to after
31197 the merging of consecutive case labels.
31199 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
31201 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
31202 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31203 restoring of callee-saved registers.
31205 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31207 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31208 * config/visium/visium.c (single_set_and_flags): Likewise.
31209 * config/visium/visium.md (Substitutions): Likewise.
31211 2017-05-17 Martin Liska <mliska@suse.cz>
31213 * cfg.c: Introduce dump_flags_t type and
31214 use it instead of int type.
31216 * cfghooks.c: Likewise.
31217 * cfghooks.h (struct cfg_hooks): Likewise.
31218 * cfgrtl.c: Likewise.
31219 * cfgrtl.h: Likewise.
31220 * cgraph.c (cgraph_node::get_body): Likewise.
31221 * coretypes.h: Likewise.
31222 * domwalk.c: Likewise.
31223 * domwalk.h: Likewise.
31224 * dumpfile.c (struct dump_option_value_info): Likewise.
31225 (dump_enable_all): Likewise.
31226 (dump_switch_p_1): Likewise.
31227 (opt_info_switch_p): Likewise.
31228 * dumpfile.h (enum tree_dump_index): Likewise.
31229 (struct dump_file_info): Likewise.
31230 * genemit.c: Likewise.
31231 * generic-match-head.c: Likewise.
31232 * gengtype.c (open_base_files): Likewise.
31233 * gimple-pretty-print.c: Likewise.
31234 * gimple-pretty-print.h: Likewise.
31235 * graph.c (print_graph_cfg): Likewise.
31236 * graphite-scop-detection.c (dot_all_sese): Likewise.
31237 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31238 * loop-unroll.c (report_unroll): Likewise.
31239 * passes.c (pass_manager::register_one_dump_file): Likewise.
31240 * print-tree.c: Likewise.
31241 * statistics.c: Likewise.
31242 * tree-cfg.c: Likewise.
31243 * tree-cfg.h: Likewise.
31244 * tree-dfa.c: Likewise.
31245 * tree-dfa.h: Likewise.
31246 * tree-dump.c (dump_function): Likewise.
31247 * tree-dump.h (struct dump_info): Likewise.
31248 * tree-pretty-print.c: Likewise.
31249 * tree-pretty-print.h: Likewise.
31250 * tree-ssa-live.c: Likewise.
31251 * tree-ssa-live.h: Likewise.
31252 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31253 * tree-vect-loop.c: Likewise.
31254 * tree-vect-slp.c: Likewise.
31256 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31257 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31259 PR tree-optimization/80457
31260 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31261 of all arguments to a statement as scalar_to_vec operations.
31262 (vectorizable_call): Adjust call to vect_model_simple_cost for
31264 (vectorizable_conversion): Likewise.
31265 (vectorizable_assignment): Likewise.
31266 (vectorizable_shift): Likewise.
31267 (vectorizable_operation): Likewise.
31268 (vectorizable_comparison): Likewise.
31269 (vect_is_simple_cond): Record the def types for operands.
31270 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31271 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31272 for statement argument count.
31274 2017-05-16 Carl Love <cel@us.ibm.com>
31276 * config/rs6000/rs6000-c: Add support for built-in functions
31277 vector unsigned long long vec_bperm (vector unsigned long long,
31278 vector unsigned char)
31279 vector signed long long vec_mule (vector signed int,
31281 vector unsigned long long vec_mule (vector unsigned int,
31282 vector unsigned int)
31283 vector signed long long vec_mulo (vector signed int,
31285 vector unsigned long long vec_mulo (vector unsigned int,
31286 vector unsigned int)
31287 vector signed char vec_sldw (vector signed char,
31288 vector signed char,
31290 vector unsigned char vec_sldw (vector unsigned char,
31291 vector unsigned char,
31293 vector signed short vec_sldw (vector signed short,
31294 vector signed short,
31296 vector unsigned short vec_sldw (vector unsigned short,
31297 vector unsigned short,
31299 vector signed int vec_sldw (vector signed int,
31302 vector unsigned int vec_sldw (vector unsigned int,
31303 vector unsigned int,
31305 vector signed long long vec_sldw (vector signed long long,
31306 vector signed long long,
31308 vector unsigned long long vec_sldw (vector unsigned long long,
31309 vector unsigned long long,
31311 * config/rs6000/rs6000-c: Add support for built-in functions
31312 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31313 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31314 * doc/extend.texi: Update the built-in documentation for the
31315 new built-in functions.
31317 2017-05-16 Marek Polacek <polacek@redhat.com>
31321 * tree.c (save_expr): Don't fold the expression.
31323 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31325 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31326 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31327 and (?*y,m). Update insn attributes.
31329 2017-05-16 Martin Liska <mliska@suse.cz>
31331 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31332 flags argument of print_gimple_stmt, print_gimple_expr,
31333 print_generic_stmt and print_generic_expr.
31334 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31335 * coretypes.h: Likewise.
31336 * except.c (dump_eh_tree): Likewise.
31337 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31338 * gimple-pretty-print.h: Likewise.
31339 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31340 (backprop::push_to_worklist): Likewise.
31341 (backprop::pop_from_worklist): Likewise.
31342 (backprop::process_use): Likewise.
31343 (backprop::intersect_uses): Likewise.
31344 (note_replacement): Likewise.
31345 * gimple-ssa-store-merging.c
31346 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31347 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31348 (pass_store_merging::execute): Likewise.
31349 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31350 (ssa_base_cand_dump_callback): Likewise.
31351 (dump_incr_vec): Likewise.
31352 (replace_refs): Likewise.
31353 (replace_mult_candidate): Likewise.
31354 (create_add_on_incoming_edge): Likewise.
31355 (create_phi_basis): Likewise.
31356 (insert_initializers): Likewise.
31357 (all_phi_incrs_profitable): Likewise.
31358 (introduce_cast_before_cand): Likewise.
31359 (replace_one_candidate): Likewise.
31360 * gimplify.c (gimplify_expr): Likewise.
31361 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31362 (set_rename): Likewise.
31363 (rename_uses): Likewise.
31364 (copy_loop_phi_nodes): Likewise.
31365 (add_close_phis_to_merge_points): Likewise.
31366 (copy_loop_close_phi_args): Likewise.
31367 (copy_cond_phi_args): Likewise.
31368 (graphite_copy_stmts_from_block): Likewise.
31369 (translate_pending_phi_nodes): Likewise.
31370 * graphite-poly.c (print_pdr): Likewise.
31371 (dump_gbb_cases): Likewise.
31372 (dump_gbb_conditions): Likewise.
31373 (print_scop_params): Likewise.
31374 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31375 (build_cross_bb_scalars_use): Likewise.
31376 (gather_bbs::before_dom_children): Likewise.
31377 * hsa-dump.c (dump_hsa_immed): Likewise.
31378 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31379 (get_replacement_map): Likewise.
31380 * ipa-inline-analysis.c (dump_condition): Likewise.
31381 (estimate_function_body_sizes): Likewise.
31382 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31383 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31384 * ipa-prop.c (ipa_dump_param): Likewise.
31385 (ipa_print_node_jump_functions_for_edge): Likewise.
31386 (ipa_modify_call_arguments): Likewise.
31387 (ipa_modify_expr): Likewise.
31388 (ipa_dump_param_adjustments): Likewise.
31389 (ipa_dump_agg_replacement_values): Likewise.
31390 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31391 * ipa-pure-const.c (check_stmt): Likewise.
31392 (pass_nothrow::execute): Likewise.
31393 * ipa-split.c (execute_split_functions): Likewise.
31394 * omp-offload.c (dump_oacc_loop_part): Likewise.
31395 (dump_oacc_loop): Likewise.
31396 * trans-mem.c (tm_log_emit): Likewise.
31397 (tm_memopt_accumulate_memops): Likewise.
31398 (dump_tm_memopt_set): Likewise.
31399 (dump_tm_memopt_transform): Likewise.
31400 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31401 (print_loop): Likewise.
31402 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31403 (chkp_gather_checks_info): Likewise.
31404 (chkp_get_check_result): Likewise.
31405 (chkp_remove_check_if_pass): Likewise.
31406 (chkp_use_outer_bounds_if_possible): Likewise.
31407 (chkp_reduce_bounds_lifetime): Likewise.
31408 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31409 (chkp_mark_completed_bounds): Likewise.
31410 (chkp_register_incomplete_bounds): Likewise.
31411 (chkp_mark_invalid_bounds): Likewise.
31412 (chkp_maybe_copy_and_register_bounds): Likewise.
31413 (chkp_build_returned_bound): Likewise.
31414 (chkp_get_bound_for_parm): Likewise.
31415 (chkp_build_bndldx): Likewise.
31416 (chkp_get_bounds_by_definition): Likewise.
31417 (chkp_generate_extern_var_bounds): Likewise.
31418 (chkp_get_bounds_for_decl_addr): Likewise.
31419 * tree-chrec.c (chrec_apply): Likewise.
31420 * tree-data-ref.c (dump_data_reference): Likewise.
31421 (dump_subscript): Likewise.
31422 (dump_data_dependence_relation): Likewise.
31423 (analyze_overlapping_iterations): Likewise.
31424 * tree-inline.c (expand_call_inline): Likewise.
31425 (tree_function_versioning): Likewise.
31426 * tree-into-ssa.c (dump_defs_stack): Likewise.
31427 (dump_currdefs): Likewise.
31428 (dump_names_replaced_by): Likewise.
31429 (dump_update_ssa): Likewise.
31430 (update_ssa): Likewise.
31431 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31432 * tree-parloops.c (build_new_reduction): Likewise.
31433 (try_create_reduction_list): Likewise.
31434 (ref_conflicts_with_region): Likewise.
31435 (oacc_entry_exit_ok_1): Likewise.
31436 (oacc_entry_exit_single_gang): Likewise.
31437 * tree-pretty-print.h: Likewise.
31438 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31439 (get_scalar_evolution): Likewise.
31440 (add_to_evolution): Likewise.
31441 (get_loop_exit_condition): Likewise.
31442 (analyze_evolution_in_loop): Likewise.
31443 (analyze_initial_condition): Likewise.
31444 (analyze_scalar_evolution): Likewise.
31445 (instantiate_scev): Likewise.
31446 (number_of_latch_executions): Likewise.
31447 (gather_chrec_stats): Likewise.
31448 (final_value_replacement_loop): Likewise.
31449 (scev_const_prop): Likewise.
31450 * tree-sra.c (dump_access): Likewise.
31451 (disqualify_candidate): Likewise.
31452 (create_access): Likewise.
31453 (reject): Likewise.
31454 (maybe_add_sra_candidate): Likewise.
31455 (create_access_replacement): Likewise.
31456 (analyze_access_subtree): Likewise.
31457 (analyze_all_variable_accesses): Likewise.
31458 (sra_modify_assign): Likewise.
31459 (initialize_constant_pool_replacements): Likewise.
31460 (find_param_candidates): Likewise.
31461 (decide_one_param_reduction): Likewise.
31462 (replace_removed_params_ssa_names): Likewise.
31463 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31464 * tree-ssa-copy.c (dump_copy_of): Likewise.
31465 (copy_prop_visit_cond_stmt): Likewise.
31466 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31467 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31468 (record_equivalences_from_stmt): Likewise.
31469 * tree-ssa-dse.c (compute_trims): Likewise.
31470 (delete_dead_call): Likewise.
31471 (delete_dead_assignment): Likewise.
31472 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31473 (forward_propagate_into_cond): Likewise.
31474 (pass_forwprop::execute): Likewise.
31475 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31476 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31478 (move_computations_worker): Likewise.
31479 (execute_sm): Likewise.
31480 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31481 (remove_exits_and_undefined_stmts): Likewise.
31482 (remove_redundant_iv_tests): Likewise.
31483 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31484 (adjust_iv_update_pos): Likewise.
31485 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31486 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31487 (value_replacement): Likewise.
31488 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31489 * tree-ssa-pre.c (print_pre_expr): Likewise.
31490 (get_representative_for): Likewise.
31491 (create_expression_by_pieces): Likewise.
31492 (insert_into_preds_of_block): Likewise.
31493 (eliminate_insert): Likewise.
31494 (eliminate_dom_walker::before_dom_children): Likewise.
31495 (eliminate): Likewise.
31496 (remove_dead_inserted_code): Likewise.
31497 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31498 * tree-ssa-reassoc.c (get_rank): Likewise.
31499 (eliminate_duplicate_pair): Likewise.
31500 (eliminate_plus_minus_pair): Likewise.
31501 (eliminate_not_pairs): Likewise.
31502 (undistribute_ops_list): Likewise.
31503 (eliminate_redundant_comparison): Likewise.
31504 (update_range_test): Likewise.
31505 (optimize_range_tests_var_bound): Likewise.
31506 (optimize_vec_cond_expr): Likewise.
31507 (rewrite_expr_tree): Likewise.
31508 (rewrite_expr_tree_parallel): Likewise.
31509 (linearize_expr): Likewise.
31510 (break_up_subtract): Likewise.
31511 (linearize_expr_tree): Likewise.
31512 (attempt_builtin_powi): Likewise.
31513 (attempt_builtin_copysign): Likewise.
31514 (transform_stmt_to_copy): Likewise.
31515 (transform_stmt_to_multiply): Likewise.
31516 (dump_ops_vector): Likewise.
31517 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31518 (print_scc): Likewise.
31519 (set_ssa_val_to): Likewise.
31520 (visit_reference_op_store): Likewise.
31521 (visit_use): Likewise.
31522 (sccvn_dom_walker::before_dom_children): Likewise.
31523 (run_scc_vn): Likewise.
31524 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31526 (expr_hash_elt::print): Likewise.
31527 (const_and_copies::pop_to_marker): Likewise.
31528 (const_and_copies::record_const_or_copy_raw): Likewise.
31529 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31530 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31531 (dump_predicates): Likewise.
31532 (find_uninit_use): Likewise.
31533 (warn_uninitialized_phi): Likewise.
31534 (pass_late_warn_uninitialized::execute): Likewise.
31535 * tree-ssa.c (verify_vssa): Likewise.
31536 (verify_ssa): Likewise.
31537 (maybe_optimize_var): Likewise.
31538 * tree-vrp.c (dump_value_range): Likewise.
31539 (dump_all_value_ranges): Likewise.
31540 (dump_asserts_for): Likewise.
31541 (register_edge_assert_for_2): Likewise.
31542 (vrp_visit_cond_stmt): Likewise.
31543 (vrp_visit_switch_stmt): Likewise.
31544 (vrp_visit_stmt): Likewise.
31545 (vrp_visit_phi_node): Likewise.
31546 (simplify_cond_using_ranges_1): Likewise.
31547 (fold_predicate_in): Likewise.
31548 (evrp_dom_walker::before_dom_children): Likewise.
31549 (evrp_dom_walker::push_value_range): Likewise.
31550 (evrp_dom_walker::pop_value_range): Likewise.
31551 (execute_early_vrp): Likewise.
31553 2017-05-16 Richard Biener <rguenther@suse.de>
31555 * dwarf2out.c (loc_list_from_tree_1): Do not create
31556 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31558 2017-05-16 Richard Biener <rguenther@suse.de>
31560 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31562 (note_variable_value_in_expr): If we resolved the decl ref
31563 do not push to the stack.
31565 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31567 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31568 operations in fast-math mode.
31569 (vaddq_f16): Likewise.
31570 (vmul_f16): Likewise.
31571 (vmulq_f16): Likewise.
31572 (vsub_f16): Likewise.
31573 (vsubq_f16): Likewise.
31574 * config/arm/neon.md (add<mode>3): New.
31576 (fma:<VH:mode>3): New. Also remove outdated comment.
31579 2017-05-16 Martin Liska <mliska@suse.cz>
31583 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31584 (odr_types_equivalent_p): Likewise.
31586 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31588 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31590 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31593 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31594 non-interunit SSE move alternatives with '?'.
31595 (zero-extendsidi peephole2): New peephole to skip intermediate
31596 general register in SSE zero-extend sequence.
31598 2017-05-15 Jeff Law <law@redhat.com>
31600 * reorg.c (relax_delay_slots): Create a new variable to hold
31601 the temporary target rather than clobbering TARGET_LABEL.
31603 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31604 missing argument to extract_bit_field call.
31605 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31607 2017-05-15 Martin Liska <mliska@suse.cz>
31610 * gcc.c (process_command): Do not allow empty argument of -o option.
31612 2017-05-15 Renlin Li <renlin.li@arm.com>
31614 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31615 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31616 * config/aarch64/constraints.md (Usf): Add long call check.
31617 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31618 (call_value): Likewise.
31619 (sibcall): Likewise.
31620 (sibcall_value): Likewise.
31622 (call_value_insn): New.
31623 (sibcall_insn): Update rtx pattern.
31624 (sibcall_value_insn): Likewise.
31625 (call_internal): Remove.
31626 (call_value_internal): Likewise.
31627 (sibcall_internal): Likewise.
31628 (sibcall_value_internal): Likewise.
31629 (call_reg): Likewise.
31630 (call_symbol): Likewise.
31631 (call_value_reg): Likewise.
31632 (call_value_symbol): Likewise.
31634 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31637 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31639 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31641 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31642 compatible with CCGOCmode and with CCZmode.
31644 2017-05-14 Martin Sebor <msebor@redhat.com>
31646 PR middle-end/77671
31647 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31648 (gimple_fold_builtin_snprintf): Same.
31649 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31650 (gimple_fold_builtin_snprintf): Same.
31651 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31652 of character types.
31653 (is_call_safe): New function.
31654 (try_substitute_return_value): Call it.
31655 (try_simplify_call): New function.
31656 (pass_sprintf_length::handle_gimple_call): Call it.
31658 2017-05-14 Martin Sebor <msebor@redhat.com>
31660 PR middle-end/80669
31661 * builtins.c (expand_builtin_stpncpy): Simplify.
31663 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31665 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31666 * config/i386/i386.h
31667 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31668 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31669 (struct machine_function): Add new members call_ms2sysv,
31670 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31671 (struct machine_frame_state): New fields sp_realigned and
31672 sp_realigned_offset.
31673 * config/i386/i386.c
31674 (enum xlogue_stub): New enum.
31675 (enum xlogue_stub_sets): New enum.
31676 (class xlogue_layout): New class.
31677 (struct ix86_frame): New fields stack_realign_allocate_offset,
31678 stack_realign_offset and outlined_save_offset. Modify comments to
31679 detail stack layout when using out-of-line stubs.
31680 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31681 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31682 -mcall-ms2sysv-xlogues.
31683 (stub_managed_regs): New static variable.
31684 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31685 registers managed by out-of-line stub.
31686 (disable_call_ms2sysv_xlogues): New function.
31687 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31688 m->call_ms2sysv when appropriate and compute frame layout for
31690 (sp_valid_at, fp_valid_at): New inline functions.
31691 (choose_basereg): New function.
31692 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31694 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31695 Use align parameter of choose_baseaddr to generated aligned SSE movs
31697 (pro_epilogue_adjust_stack): Modify to track
31698 machine_frame_state::sp_realigned.
31699 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31700 (ix86_nsaved_sseregs): Likewise.
31701 (ix86_emit_save_regs): Likewise.
31702 (ix86_emit_save_regs_using_mov): Likewise.
31703 (ix86_emit_save_sse_regs_using_mov): Likewise.
31704 (get_scratch_register_on_entry): Likewise.
31705 (gen_frame_set): New function.
31706 (gen_frame_load): Likewise.
31707 (gen_frame_store): Likewise.
31708 (emit_outlined_ms2sysv_save): Likewise.
31709 (emit_outlined_ms2sysv_restore): Likewise.
31710 (ix86_expand_prologue): Modify stack re-alignment code and call
31711 emit_outlined_ms2sysv_save when appropriate.
31712 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31713 parameter rtx_insn *insn, which allows the function to be used to only
31714 generate the notes.
31715 (ix86_expand_epilogue): Modify validity checks of frame and stack
31716 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31717 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31718 * config/i386/predicates.md
31719 (save_multiple): New predicate.
31720 (restore_multiple): Likewise.
31721 * config/i386/sse.md
31722 (save_multiple<mode>): New pattern.
31723 (save_multiple_realign<mode>): Likewise.
31724 (restore_multiple<mode>): Likewise.
31725 (restore_multiple_and_return<mode>): Likewise.
31726 (restore_multiple_leave_return<mode>): Likewise.
31727 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31729 2017-05-14 Julia Koval <julia.koval@intel.com>
31731 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31732 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31733 (__builtin_ia32_xsetbv): New builtins.
31734 * config/i386/i386.c (ix86_expand_special_args_builtin):
31736 (ix86_expand_builtin): Special expand for new intrinsics.
31737 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31738 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31739 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31741 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31743 * cfganal.c (inverted_post_order_compute): Change argument type
31745 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31746 * df-core.c (rest_of_handle_df_initialize): Adjust.
31747 (rest_of_handle_df_finish): Likewise.
31748 (df_analyze_1): Likewise.
31749 (df_analyze): Likewise.
31750 (loop_inverted_post_order_compute): Change argument to be a vec *.
31751 (df_analyze_loop): Adjust.
31752 (df_get_n_blocks): Likewise.
31753 (df_get_postorder): Likewise.
31754 * df.h (struct df_d): Change field to be a vec.
31755 * lcm.c (compute_laterin): Adjust.
31756 (compute_available): Likewise.
31757 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31758 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31759 * tree-ssa-pre.c (compute_antic): Likewise.
31761 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31763 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31764 (depth_first_search::depth_first_search): Change structure init
31765 function to this constructor.
31766 (depth_first_search::add_bb): Rename function to this member.
31767 (depth_first_search::execute): Likewise.
31768 (flow_dfs_compute_reverse_finish): Adjust.
31770 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31772 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31773 (longest_simple_path): Likewise.
31774 * shrink-wrap.c (spread_components): Likewise.
31775 (disqualify_problematic_components): Likewise.
31776 (emit_common_heads_for_components): Likewise.
31777 (emit_common_tails_for_components): Likewise.
31778 (insert_prologue_epilogue_for_components): Likewise.
31780 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31782 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31785 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31787 * df-core.c (df_set_blocks): Start using auto_bitmap.
31788 (df_compact_blocks): Likewise.
31789 * df-problems.c (df_rd_confluence_n): Likewise.
31790 * df-scan.c (df_insn_rescan_all): Likewise.
31791 (df_process_deferred_rescans): Likewise.
31792 (df_update_entry_block_defs): Likewise.
31793 (df_update_exit_block_uses): Likewise.
31794 (df_entry_block_bitmap_verify): Likewise.
31795 (df_exit_block_bitmap_verify): Likewise.
31796 (df_scan_verify): Likewise.
31797 * lra-constraints.c (lra_constraints): Likewise.
31798 (undo_optional_reloads): Likewise.
31799 (lra_undo_inheritance): Likewise.
31800 * lra-remat.c (calculate_gen_cands): Likewise.
31801 (do_remat): Likewise.
31802 * lra-spills.c (assign_spill_hard_regs): Likewise.
31803 (spill_pseudos): Likewise.
31804 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31805 (bitmap_set_subtract_values): Likewise.
31807 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31809 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31810 management with auto_bitmap.
31811 (fix_inter_tick): Likewise.
31812 (fix_recovery_deps): Likewise.
31813 * ira.c (add_store_equivs): Likewise.
31814 (find_moveable_pseudos): Likewise.
31815 (split_live_ranges_for_shrink_wrap): Likewise.
31816 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31817 (rtx_reuse_manager::seen_def_p): Likewise.
31818 (rtx_reuse_manager::set_seen_def): Likewise.
31819 * print-rtl.h (class rtx_reuse_manager): Likewise.
31821 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31823 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31825 (migrate_btr_def): Likewise.
31826 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31827 * df-core.c (loop_post_order_compute): Likewise.
31828 (loop_inverted_post_order_compute): Likewise.
31829 * hsa-common.h: Likewise.
31830 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31831 * init-regs.c (initialize_uninitialized_regs): Likewise.
31832 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31833 (inline_small_functions): Likewise.
31834 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31835 * ira.c (combine_and_move_insns): Likewise.
31836 (build_insn_chain): Likewise.
31837 * loop-invariant.c (find_invariants): Likewise.
31838 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31839 * predict.c (tree_predict_by_opcode): Likewise.
31840 (predict_paths_leading_to): Likewise.
31841 (predict_paths_leading_to_edge): Likewise.
31842 (estimate_loops_at_level): Likewise.
31843 (estimate_loops): Likewise.
31844 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31845 (spread_components): Likewise.
31846 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31847 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31848 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31849 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31850 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31851 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31852 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31853 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31854 (mark_threaded_blocks): Likewise.
31855 (thread_through_all_blocks): Likewise.
31856 * tree-ssa.c (verify_ssa): Likewise.
31857 (execute_update_addresses_taken): Likewise.
31858 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31860 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31862 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31864 (post_order_compute): Likewise.
31865 (inverted_post_order_compute): Likewise.
31866 (pre_and_rev_post_order_compute_fn): Likewise.
31868 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31870 * genrecog.c (int_set::int_set): Explicitly construct our
31871 auto_vec base class.
31872 * vec.h (auto_vec::auto_vec): New constructor.
31874 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31876 * bitmap.h (class auto_bitmap): New constructor taking
31877 bitmap_obstack * argument.
31879 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31881 * bitmap.h (class auto_bitmap): Change type of m_bits to
31882 bitmap_head, and adjust ctor / dtor and member operators.
31884 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31886 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31887 when returned register mode doesn't match original mode.
31889 2017-05-12 Jeff Law <law@redhat.com>
31890 Jakub Jelinek <jakub@redhat.com>
31892 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31893 we look for cc setter after the compare-elim changes.
31894 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31895 within the vector to match what compare-elim now expects.
31896 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31897 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31899 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31900 after the compare-elim changes.
31901 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31902 the vector to match what compare-elim now expects.
31903 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31904 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31905 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31906 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31907 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31909 * config/visium/visium.c (single_set_and_flags): Fix where
31910 we look for cc setter after the compare-elim changes.
31911 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31912 with the vector to match what compare-elim now expects.
31913 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31914 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31915 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31916 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31917 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31918 (neg<mode>2_insn_set_overflow): Likewise.
31920 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31922 PR middle-end/79794
31923 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31924 maybe_expand_insn call, set ops[0].target. If still set after call,
31925 set alt_rtl. Add extra arg to recursive calls.
31926 (extract_bit_field): Add alt_rtl argument. Pass to
31928 * expmed.h (extract_bit_field): Fix prototype.
31929 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31930 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31931 to extract_bit_field_calls.
31932 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31933 Pass alt_rtl to extract_bit_field calls.
31934 * calls.c (store_unaligned_arguments_into_psuedos)
31935 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31936 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31938 * optabs.h (struct expand_operand): Add target bitfield.
31940 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31942 * compare-elim.c (try_eliminate_compare): Canonicalize
31943 operation with embedded compare to
31944 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31945 (set (reg) (operation)].
31947 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31949 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31952 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31953 cost of adding a carry flag for ADC instruction.
31954 [case MINUS]: Ignore the cost of subtracting a carry flag
31955 for SBB instruction.
31957 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31959 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31961 * config/rs6000/bmiintrin.h: New file.
31962 * config/rs6000/bmi2intrin.h: New file.
31963 * config/rs6000/x86intrin.h: New file.
31965 2017-05-12 Jeff Law <law@redhat.com>
31967 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31970 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31972 PR middle-end/80707
31973 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31975 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31977 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31978 early expansion of vector divide builtins.
31979 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31980 builtins identified as having unsigned arguments.
31982 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31984 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31985 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31986 expansion of vector logical operations (and, andc, or, xor,
31989 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31991 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31992 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31994 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31996 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31997 early GIMPLE expansion of vector multiplies.
31999 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32001 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
32002 TARGET_HAVE_MOVT conditional.
32003 (movt splitter): Likewise.
32005 2017-05-12 Richard Biener <rguenther@suse.de>
32007 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
32008 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32009 Fold all stmts not inplace.
32011 2017-05-12 Richard Biener <rguenther@suse.de>
32013 PR tree-optimization/80713
32014 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
32015 inserted_exprs bit for not removed stmts.
32017 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
32019 PR middle-end/69921
32020 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
32021 parallelized" attribute for parallelized OpenACC kernels.
32022 * omp-offload.c (execute_oacc_device_lower): Use it.
32024 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
32025 Set "oacc kernels" attribute.
32026 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
32027 parameter. Adjust all users.
32028 (oacc_fn_attrib_kernels_p): Remove function.
32029 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
32030 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
32031 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
32032 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
32033 assert "oacc kernels" attribute is set.
32035 2017-05-11 Carl Love <cel@us.ibm.com>
32037 * config/rs6000/rs6000-c: Add support for built-in functions
32038 vector unsigned char vec_popcnt (vector signed char)
32039 vector unsigned char vec_popcnt (vector unsigned char)
32040 vector unsigned short vec_popcnt (vector signed short)
32041 vector unsigned short vec_popcnt (vector unsigned short)
32042 vector unsigned int vec_popcnt (vector signed int)
32043 vector unsigned int vec_popcnt (vector unsigned int)
32044 vector unsigned long long vec_popcnt (vector signed long long)
32045 vector unsigned long long vec_popcnt (vector unsigned long long)
32046 vector signed long long vec_slo (vector signed long long,
32047 vector signed char)
32048 vector signed long long vec_slo (vector signed long long,
32049 vector unsigned char)
32050 vector unsigned long long vec_slo (vector unsigned long long,
32051 vector signed char)
32052 vector unsigned long long vec_slo (vector unsigned long long,
32053 vector unsigned char)
32054 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
32055 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
32056 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
32057 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
32058 * doc/extend.texi: Update the built-in documentation file for the
32059 new built-in functions.
32061 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
32063 * attribs.h (sorted_attr_string): Move machine independent
32064 functions for target clone support from the i386 port to common
32065 code. Rename ix86_function_versions to common_function_versions.
32066 Rename make_name to make_unique_name.
32067 (common_function_versions): Likewise.
32068 (make_unique_name): Likewise.
32069 (make_dispatcher_decl): Likewise.
32070 (is_function_default_version): Likewise.
32071 * attribs.c (attr_strcmp): Likewise.
32072 (sorted_attr_string): Likewise.
32073 (common_function_versions): Likewise.
32074 (make_unique_name): Likewise.
32075 (make_dispatcher_decl): Likewise.
32076 (is_function_default_version): Likewise.
32077 * config/i386/i386.c (attr_strcmp): Likewise.
32078 (sorted_attr_string): Likewise.
32079 (ix86_function_versions): Likewise.
32080 (make_name): Likewise.
32081 (make_dispatcher_decl): Likewise.
32082 (is_function_default_version): Likewise.
32083 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
32085 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32088 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
32089 Account for direct move costs for vec_construct of integer
32092 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
32095 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
32096 (UNSPEC_STX_ATOMIC): Ditto.
32097 (loaddi_via_sse): New insn.
32098 (storedi_via_sse): Ditto.
32099 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
32100 Update corresponding peephole2 patterns.
32101 (atomic_storedi_fpu): Ditto.
32103 2017-05-11 Julia Koval <julia.koval@intel.com>
32105 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
32106 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
32108 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
32109 (__builtin_ia32_rsqrt14ss_mask): New builtins.
32110 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
32112 2017-05-11 Nathan Sidwell <nathan@acm.org>
32114 * graphite-poly.c: Include dumpfile.h.
32116 * dumpfle.h (dump_function): Declare here ...
32117 * tree-dump.h (dump_function): ... not here.
32118 * dumpfile.c: #include tree-cfg.h.
32119 (dump_function): Move here from ...
32120 * tree-dump.c (dump_function): ... here.
32121 * gimplify.c: #include splay-tree.h, not tree-dump.h.
32122 * graphite-poly.c: Don't include tree-dump.h.
32123 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
32124 * print-tree.c: Likewise.
32125 * stor-layout.c: Likewise.
32126 * tree-nested.c: Likewise.
32128 * dumpfile.c (dump_start): Use TDF_FLAGS.
32129 (dump_enable_all): Fix TDF_KIND check thinko.
32131 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
32133 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32134 array entries to represent two legal parameterizations of the
32135 overloaded __builtin_cmpb function, as represented by the
32136 P6_OV_BUILTIN_CMPB constant.
32137 (altivec_resolve_overloaded_builtin): Add special case handling
32138 for the __builtin_cmpb function, as represented by the
32139 P6_OV_BUILTIN_CMPB constant.
32140 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
32141 (BU_P6_64BIT_2): New macro.
32142 (BU_P6_OVERLOAD_2): New macro
32143 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
32144 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
32145 (CMPB): Add overload support to represent both 32-bit and 64-bit
32146 compare-bytes function.
32147 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
32148 support for TARGET_CMPB.
32149 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
32150 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32151 documentation of the __builtin_cmpb overloaded built-in function.
32153 2017-05-11 Richard Biener <rguenther@suse.de>
32155 PR tree-optimization/80705
32156 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
32157 bases are not vectorizable.
32159 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32161 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
32162 when counting register pressure.
32164 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32166 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
32167 (iv_ca_more_deps): Renamed to ...
32168 (iv_ca_compare_deps): ... this.
32169 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
32171 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32173 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
32175 (determine_group_iv_costs): ... here.
32176 (find_inv_vars_cb): Record inv var if it's not recorded before.
32178 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32180 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
32181 (get_shiftadd_cost): Ditto.
32183 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32185 * tree-ssa-address.c: Include header file.
32186 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
32188 (add_to_parts): Refactor.
32189 (addr_to_parts): New parameter. Update use of move_hint_to_base.
32190 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
32193 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32195 PR tree-optimization/53090
32196 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
32198 (extract_cond_operands): Detect condition with IV on both sides
32199 and return COMP_IV_EXPR_2.
32200 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
32201 (rewrite_use_compare): Simplify by removing call to function
32202 extract_cond_operands.
32204 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32206 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32207 (extract_cond_operands): Detect condition comparing against non-
32208 invariant bound and return appropriate enum value.
32209 (find_interesting_uses_cond): Update use of extract_cond_operands.
32210 Handle its return value accordingly.
32211 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32213 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32215 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32216 nonlinear iv_use computation in loop invariant sensitive way.
32218 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32220 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32221 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32223 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32225 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32226 (dump_cand): Support iv_cand.inv_exprs.
32227 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32229 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32232 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32234 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32236 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32237 as local function. Include necessary header files.
32238 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32240 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32242 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32244 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32246 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32247 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32248 RSHIFT_EXPR and BIT_NOT_EXPR.
32250 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32252 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32253 (adjust_setup_cost): New parameter supporting round up adjustment.
32254 (struct address_cost_data): Delete.
32255 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32256 (split_address_cost, ptr_difference_cost): Delete.
32257 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32258 (struct ainc_cost_data): New struct.
32259 (get_address_cost_ainc): New function.
32260 (get_address_cost, get_computation_cost): Reimplement.
32261 (determine_group_iv_cost_address): Record inv_expr for all uses of
32263 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32264 (iv_ca_has_deps): Reimplemented to ...
32265 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32267 (iv_ca_extend): Call iv_ca_more_deps.
32269 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32271 * tree-ssa-address.c (struct mem_address): Move to header file.
32272 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32273 * tree-ssa-address.h (struct mem_address): Move from C file.
32274 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32276 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32278 * tree-affine.h (aff_combination_type): New interface.
32279 (aff_combination_zero_p): Remove static.
32280 (aff_combination_const_p): New interface.
32281 (aff_combination_singleton_var_p): New interfaces.
32283 2017-05-11 Richard Biener <rguenther@suse.de>
32285 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32286 Skip unreachable blocks and destinations.
32287 (eliminate): Move stmt removal and fixup ...
32288 (fini_eliminate): ... here. Skip inserted exprs.
32289 (pass_pre::execute): Move fini_pre after fini_eliminate.
32290 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32291 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32292 PRE to get rid of dead code that has invalid SSA form and
32293 split critical edges again.
32295 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32297 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32299 2017-05-11 Richard Biener <rguenther@suse.de>
32301 * passes.c (execute_function_todo): Verify loops if they are
32302 said to be up-to-date.
32303 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32304 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32306 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32309 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32310 handle calling assemble_external ourself.
32313 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32314 modes with zero size. Enhance comment.
32316 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32318 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32319 built-ins for vec_xl and vec_xst with short and char pointer
32322 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32324 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32325 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32326 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32327 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32328 (_mm_maskz_min_round_ss): New intrinsics.
32329 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32330 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32331 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32332 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32333 (__builtin_ia32_minss_mask_round): New builtins.
32334 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32335 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32336 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32338 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32339 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32341 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32344 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32346 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32347 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32348 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32349 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32350 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32351 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32352 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32353 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32354 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32355 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32356 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32357 (__builtin_ia32_mulss_mask_round): New builtins.
32358 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32359 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32360 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32362 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32363 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32365 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32368 2017-05-10 Julia Koval <julia.koval@intel.com>
32370 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32371 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32372 (_mm256_setr_m128i): New intrinsics.
32374 2017-05-10 Julia Koval <julia.koval@intel.com>
32376 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32377 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32378 (_mm_maskz_rcp14_ss): New intrinsics.
32379 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32380 (__builtin_ia32_rcp14ss_mask): New builtins.
32381 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32383 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32385 PR tree-optimization/51513
32386 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32387 (assert_unreachable_fallthru_edge_p): Use it.
32388 (group_case_labels_stmt): Likewise.
32389 * tree-cfg.h: Prototype it.
32390 * stmt.c: Include cfghooks.h and tree-cfg.h.
32391 (emit_case_dispatch_table) <gap_label>: New local variable.
32392 Use it to fill dispatch table gaps.
32393 Test for default_label before updating probabilities.
32394 (expand_case) <default_label>: Remove unneeded initialization.
32395 Test for unreachable default case statement and remove its edge.
32396 Set default_label accordingly.
32397 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32399 2017-05-10 Carl Love <cel@us.ibm.com>
32401 * config/rs6000/rs6000-c: Add support for built-in functions
32402 vector signed char vec_neg (vector signed char)
32403 vector signed short int vec_neg (vector short int)
32404 vector signed int vec_neg (vector signed int)
32405 vector signed long long vec_neg (vector signed long long)
32406 vector float vec_neg (vector float)
32407 vector double vec_neg (vector double)
32408 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32410 * config/rs6000/altivec.h: Add define for vec_neg
32411 * doc/extend.texi: Update the built-in documentation for the
32412 new built-in functions.
32414 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32416 PR tree-optimization/77644
32417 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32419 2017-05-10 Nathan Sidwell <nathan@acm.org>
32421 * dumpfile.h (TDI_lang_all): New.
32422 (TDF_KIND): New. Renumber others
32423 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32425 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32427 (get_dump_file_name): Adjust suffix generation.
32428 (dump_enable_all): Use TDF_KIND.
32429 * doc/invoke.texi (-fdump-lang-all): Document.
32431 * dumpfile.h: Tabify.
32433 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32436 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32437 Move member access before delete.
32439 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32441 * tree-inline.c (expand_call_inline): Split block at stmt
32444 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32447 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32448 are now unused after splitting mov{sf,sd}_hardfloat.
32449 (f32_lr2): Likewise.
32450 (f32_lm): Likewise.
32451 (f32_lm2): Likewise.
32452 (f32_li): Likewise.
32453 (f32_li2): Likewise.
32454 (f32_lv): Likewise.
32455 (f32_sr): Likewise.
32456 (f32_sr2): Likewise.
32457 (f32_sm): Likewise.
32458 (f32_sm2): Likewise.
32459 (f32_si): Likewise.
32460 (f32_si2): Likewise.
32461 (f32_sv): Likewise.
32462 (f32_dm): Likewise.
32463 (f32_vsx): Likewise.
32464 (f32_av): Likewise.
32465 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32466 For movsf, order stores so the VSX stores occur before the GPR
32467 store which encourages the register allocator to use a traditional
32468 FPR instead of a GPR. For movsd, order the stores so that the GPR
32469 store comes before the VSX stores to allow the power6 to work.
32470 This is due to the power6 not having a 32-bit integer store
32471 instruction from a FPR.
32472 (movsf_hardfloat): Likewise.
32473 (movsd_hardfloat): Likewise.
32475 2017-05-09 Martin Sebor <msebor@redhat.com>
32477 PR translation/80280
32478 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32481 PR translation/80280
32482 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32483 data member added in r247778.
32484 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32486 2017-05-09 Nathan Sidwell <nathan@acm.org>
32488 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32490 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32493 2017-05-09 Marek Polacek <polacek@redhat.com>
32495 * doc/invoke.texi: Fix typo.
32497 2017-05-09 Richard Biener <rguenther@suse.de>
32499 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32500 (vrp_val_is_min): Likewise.
32501 (set_value_range_to_value): Likewise.
32502 (set_value_range_to_nonnegative): Likewise.
32503 (gimple_assign_nonzero_p): Likewise.
32504 (gimple_stmt_nonzero_p): Likewise.
32505 (vrp_int_const_binop): Likewise. Remove unreachable case.
32506 (adjust_range_with_scev): Adjust comments.
32507 (compare_range_with_value): Likewise.
32508 (extract_range_from_phi_node): Likewise.
32509 (test_for_singularity): Likewise.
32511 2017-05-09 Richard Biener <rguenther@suse.de>
32513 * tree-vrp.c (get_single_symbol): Add assert that we don't
32514 get overflowed constants as invariant part.
32515 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32516 checks. Use wi::cmp instead of recursing for integer constants.
32517 (compare_values): Just ignore whether we assumed undefined
32518 overflow instead of failing the compare.
32519 (extract_range_for_var_from_comparison_expr): Add comment before the
32520 TREE_NO_WARNING sets.
32521 (test_for_singularity): Likewise.
32522 (extract_range_from_comparison): Do not disable optimization
32523 when we assumed undefined overflow.
32524 (extract_range_basic): Remove init of unused var.
32526 2017-05-09 Richard Biener <rguenther@suse.de>
32528 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32529 (extract_range_from_multiplicative_op_1): Adjust.
32530 (extract_range_from_binary_expr_1): Use int_const_binop.
32532 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32535 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32536 rs6000_store_data_bypass_p in seven define_bypass directives and
32537 in several comments.
32538 * config/rs6000/rs6000-protos.h: Add prototype for
32539 rs6000_store_data_bypass_p function.
32540 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32541 function implements slightly different (rs6000-specific) semantics
32542 than store_data_bypass_p, returning false rather than aborting
32543 with assertion error when arguments do not satisfy the
32544 requirements of store data bypass.
32545 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32546 rs6000_store_data_bypass_p.
32548 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32550 * config/xtensa/xtensa-protos.h
32551 (xtensa_initial_elimination_offset): New declaration.
32552 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32553 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32554 macro definition, add case for FRAME_POINTER_REGNUM when
32555 FRAME_GROWS_DOWNWARD.
32556 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32557 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32558 xtensa_initial_elimination_offset.
32560 2017-05-08 Nathan Sidwell <nathan@acm.org>
32562 * doc/invoke.texi: Alphabetize -fdump options.
32564 2017-05-08 Martin Sebor <msebor@redhat.com>
32566 PR translation/80280
32567 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32569 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32571 * target.def (compute_frame_layout): New optional target hook.
32572 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32573 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32574 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32576 * reload1.c (verify_initial_elim_offsets): Likewise.
32577 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32578 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32579 (arm_get_frame_offsets): Split up into this ...
32580 (arm_compute_frame_layout): ... and this function.
32582 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32584 * config/aarch64/constraints.md (Usa): New constraint.
32585 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32587 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32589 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32590 with_multilib_list after it has been checked.
32592 2017-05-08 Richard Biener <rguenther@suse.de>
32594 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32595 (bitmap_set_subtract_values): Likewise.
32597 2017-05-08 Richard Biener <rguenther@suse.de>
32599 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32600 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32602 (gimple_stmt_nonzero_warnv_p): Rename to ...
32603 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32605 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32606 (extract_range_basic): Adjust, do not disable propagation on
32607 strict overflow sensitive simplification.
32608 (vrp_visit_cond_stmt): Likewise.
32610 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32612 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32613 body size unconditionally.
32615 2017-05-07 Jeff Law <law@redhat.com>
32618 2017-05-06 Jeff Law <law@redhat.com>
32619 PR tree-optimization/78496
32620 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32623 PR tree-optimization/78496
32624 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32625 (simplify_stmt_using_ranges): Call it.
32626 (vrp_dom_walker::before_dom_children): Extract equivalences
32627 from an ASSERT_EXPR with an equality comparison against a
32630 2017-05-06 Jeff Law <law@redhat.com>
32632 PR tree-optimization/78496
32633 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32636 PR tree-optimization/78496
32637 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32638 (simplify_stmt_using_ranges): Call it.
32639 (vrp_dom_walker::before_dom_children): Extract equivalences
32640 from an ASSERT_EXPR with an equality comparison against a
32643 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32645 * lra-constraints.c (lra_copy_reg_equiv): New function.
32646 (split_reg): Use it to copy equivalence information from the
32647 original register to the spill register.
32649 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32651 PR rtl-optimization/75964
32652 * simplify-rtx.c (simplify_const_relational_operation): Remove
32653 invalid handling of comparisons of integer ABS.
32655 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32657 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32658 initialize to zero.
32659 (init_regs): Remove declaration.
32660 (function_arg_advance_32): Initialize error_p as boolean variable.
32662 2017-05-05 Nathan Sidwell <nathan@acm.org>
32664 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32665 lines. Use for (;;).
32667 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32669 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32670 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32671 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32672 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32673 VF=2 that require versioning.
32675 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32677 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32680 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32682 * diagnostic.h (diagnostic_override_option_index): Convert from
32683 macro to inline function.
32685 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32687 * diagnostic.c (last_module_changed_p): New function.
32688 (set_last_module): New function.
32689 (diagnostic_report_current_module): Convert macro usage to
32690 the above functions.
32691 * diagnostic.h (diagnostic_context::last_module): Strengthen
32692 from const line_map * to const line_map_ordinary *.
32693 (diagnostic_last_module_changed): Delete macro.
32694 (diagnostic_set_last_module): Delete macro.
32696 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32698 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32699 with diagnostic_report_diagnostic.
32700 (diagnostic_n_impl_richloc): Likewise.
32701 * diagnostic.h (report_diagnostic): Delete macro.
32702 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32703 with diagnostic_report_diagnostic.
32704 * substring-locations.c (format_warning_va): Likewise.
32706 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32708 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32709 save/restor of format_spec. Move option-printing code to...
32710 (print_option_information): ...this new function, and
32711 reimplement by simply printing to the pretty_printer,
32712 rather than appending to the format string.
32714 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32716 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32717 handling logic into...
32718 (update_effective_level_from_pragmas): ...this new function.
32720 2017-05-04 Andrew Waterman <andrew@sifive.com>
32722 * config/riscv/riscv.opt (mstrict-align): New option.
32723 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32724 (SLOW_UNALIGNED_ACCESS): Define.
32725 (riscv_slow_unaligned_access): Declare.
32726 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32728 (riscv_slow_unaligned_access): New variable.
32729 (rocket_tune_info): Set slow_unaligned_access to true.
32730 (optimize_size_tune_info): Set slow_unaligned_access to false.
32731 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32732 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32733 (riscv_option_override): Set riscv_slow_unaligned_access.
32734 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32736 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32738 * config/riscv/riscv.md: Unify indentation.
32740 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32745 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32747 (extendsi<mode>2): Add support for doing sign extension via
32748 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32749 don't have ISA 3.0 instructions.
32750 (extendsi<mode>2 splitter): Likewise.
32751 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32752 generate the normal insns since SImode can now go in vector
32753 registers. Disallow the special UNSPECs needed for previous
32754 machines to hide SImode being used. Add new insns
32755 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32756 (fix_trunc<mode>si2_stfiwx): Likewise.
32757 (fix_trunc<mode>si2_internal): Likewise.
32758 (fixuns_trunc<mode>si2): Likewise.
32759 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32760 (fctiw<u>z_<mode>_smallint): Likewise.
32761 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32762 of floating point to 32-bit integer from doing a direct move to
32763 the GPR registers to do a store.
32764 (fctiwz_<mode>): Break long line.
32766 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32768 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32769 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32770 (addr_list, addr_offset_valid_p): New.
32771 (split_address_groups): Check offset validity with above function.
32772 (gt-tree-ssa-loop-ivopts.h): Include header file.
32774 2017-05-05 Nathan Sidwell <nathan@acm.org>
32776 * config.gcc (arm*-*-*): Add missing 'fi'.
32778 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32780 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32781 included in -fopt-info does not matter.
32782 * doc/optinfo.texi (-fopt-info): Fix description of default
32783 behavour. Explicitly say order of options included in -fopt-info
32786 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32788 * config.gcc: Allow combinations of aprofile and rmprofile values for
32789 --with-multilib-list.
32790 * config/arm/t-multilib: New file.
32791 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32792 variables. Remove setting of ISA and floating-point ABI in
32793 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32794 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32795 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32796 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32798 * config/arm/t-rmprofile: Likewise except for the matches changes.
32799 * doc/install.texi (--with-multilib-list): Document the combination of
32800 aprofile and rmprofile values and warn about pitfalls in doing that.
32802 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32804 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32805 (movdi_aarch64): Likewise.
32807 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32809 PR tree-optimization/80632
32810 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32812 (build_arrays): Initialize it for virtual phis.
32813 (fix_phi_nodes): Use it for virtual phis.
32815 PR tree-optimization/80558
32816 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32817 [x, y] op z into [x op, y op z] for op & or | if conditions
32820 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32821 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32824 * config/arm/arm.md (use_literal_pool): Remove.
32825 (64-bit immediate split): No longer takes cost into consideration
32826 if arm_disable_literal_pool is enabled.
32827 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32828 used when arm_disable_literal_pool is enabled.
32829 (arm_max_const_double_inline_cost): Remove use of
32830 arm_disable_literal_pool.
32831 (push_minipool_fix): Add assert.
32832 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32833 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32834 (no_literal_pool_sf_immediate): New.
32836 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32838 PR tree-optimization/80613
32839 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32840 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32842 2017-05-05 Richard Biener <rguenther@suse.de>
32844 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32846 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32848 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32849 of this flag from insn conditions due to removal from r247495.
32851 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32853 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32855 (arm_early_store_addr_dep_ptr): Likewise.
32856 * config/arm/aarch-common-protos.h
32857 (arm_early_load_addr_dep_ptr): Add prototype.
32858 (arm_early_store_addr_dep_ptr): Likewise.
32859 * config/arm/cortex-a53.md: Add new bypasses.
32861 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32863 * tree.c (next_type_uid): Change type to unsigned.
32864 (type_hash_canon): Decrement back next_type_uid if
32865 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32866 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32869 2017-05-04 Martin Sebor <msebor@redhat.com>
32871 * builtins.c: Fix a trivial typo in a comment.
32873 PR middle-end/79234
32874 * builtins.c (check_sizes): Adjust to handle reading past the end.
32875 Avoid printing excessive upper bound of ranges. Use %E to print
32876 tree nodes instead of converting them to %wu.
32877 (expand_builtin_memchr): New function.
32878 (compute_dest_size): Rename...
32879 (compute_objsize): ...to this.
32880 (expand_builtin_memcpy): Adjust.
32881 (expand_builtin_mempcpy): Adjust.
32882 (expand_builtin_strcat): Adjust.
32883 (expand_builtin_strcpy): Adjust.
32884 (check_strncat_sizes): Adjust.
32885 (expand_builtin_strncat): Adjust.
32886 (expand_builtin_strncpy): Adjust and simplify.
32887 (expand_builtin_memset): Adjust.
32888 (expand_builtin_bzero): Adjust.
32889 (expand_builtin_memcmp): Adjust.
32890 (expand_builtin): Handle memcmp.
32891 (maybe_emit_chk_warning): Check strncat just once.
32893 2017-05-04 Martin Sebor <msebor@redhat.com>
32895 PR preprocessor/79214
32896 PR middle-end/79222
32897 PR middle-end/79223
32898 * builtins.c (check_sizes): Add inlining context and issue
32899 warnings even when -Wno-system-headers is set.
32900 (check_strncat_sizes): Same.
32901 (expand_builtin_strncat): Same.
32902 (expand_builtin_memmove): New function.
32903 (expand_builtin_stpncpy): Same.
32904 (expand_builtin): Handle memmove and stpncpy.
32906 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32908 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32909 which is not used any more.
32911 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32913 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32915 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32917 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32918 (cortexa53_tunings): Likewise.
32919 (cortexa57_tunings): Likewise.
32920 (cortexa72_tunings): Likewise.
32921 (cortexa73_tunings): Likewise.
32923 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32925 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32926 Set loop alignment to 8.
32928 2017-05-04 Martin Sebor <msebor@redhat.com>
32930 PR translation/80280
32931 * builtins.c (expand_builtin_object_size): Add missing quoting to
32932 %D and like directives.
32933 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32934 (hsa_type_for_tree_type): Same.
32935 (verify_function_arguments): Same.
32936 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32937 * varasm.c (get_section): Same.
32940 2017-05-04 Martin Sebor <msebor@redhat.com>
32942 PR translation/80280
32943 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32945 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32947 * config/aarch64/aarch64.c (generic_addrcost_table):
32948 Change HI/TI mode setting.
32950 2017-05-04 Martin Jambor <mjambor@suse.cz>
32952 PR tree-optimization/80622
32953 * tree-sra.c (comes_initialized_p): New function.
32954 (build_accesses_from_assign): Only set write lazily when
32955 comes_initialized_p is false.
32956 (analyze_access_subtree): Use comes_initialized_p.
32957 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32958 instead of testing for PARM_DECL.
32960 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32962 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32963 constraint on operand 0 to allow more general addressing modes.
32964 Adjust output template.
32965 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32967 * config/aarch64/aarch64-protos.h
32968 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32969 * config/aarch64/constraints.md (Dp): New address constraint.
32970 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32973 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32975 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32976 update use of estimate_ipcp_clone_size_and_time.
32977 (estimate_local_effects): Update use of
32978 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32979 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32980 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32981 Return nonspecialized time.
32983 2017-05-04 Richard Biener <rguenther@suse.de>
32985 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32986 for the last VUSE which def dominates the PHI. Directly call
32988 (get_continuation_for_phi_1): Remove.
32990 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32992 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32993 to explain the use of truncating division. Cap the number of
32994 iterations to the maximum given by nb_iterations_upper_bound,
32997 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32999 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
33000 * configure: Regenerate.
33001 * config.in: Regenerate.
33002 * config/i386/driver-mingw32.c: new file.
33003 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
33004 * config.host: Link driver-mingw32.o on MinGW host.
33005 * doc/install.texi: Document new --enable-mingw-wildcard configure
33008 2017-05-04 Marek Polacek <polacek@redhat.com>
33010 PR tree-optimization/80612
33011 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
33013 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
33014 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
33016 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
33017 (movt splitter): Likewise.
33018 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
33019 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
33020 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
33021 block for Thumb-1 with MOVT.
33022 (thumb2_legitimate_address_p): Move code block ...
33023 (can_avoid_literal_pool_for_label_p): ... into this new function.
33024 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
33026 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
33027 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
33028 "M-profile targets with the MOVT instruction".
33030 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
33032 * config/arm/arm-builtins.c (arm_init_builtins): Rename
33033 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
33034 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
33036 2017-05-04 Martin Liska <mliska@suse.cz>
33038 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
33039 variable cond_code.
33041 2017-05-04 Richard Biener <rguenther@suse.de>
33043 * tree.c (array_at_struct_end_p): Handle arrays at struct
33044 end with flexarrays more conservatively. Refactor and treat
33045 arrays of arrays or aggregates more strict. Fix
33046 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
33047 * tree.c (array_at_struct_end_p): Adjust prototype.
33048 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
33049 * gimple-fold.c (get_range_strlen): Likewise.
33050 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
33052 2017-05-04 Richard Biener <rguenther@suse.de>
33054 PR tree-optimization/31130
33055 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
33057 (supports_overflow_infinity): Likewise.
33058 (is_negative_overflow_infinity): Likewise.
33059 (is_positive_overflow_infinity): Likewise.
33060 (is_overflow_infinity): Likewise.
33061 (stmt_overflow_infinity): Likewise.
33062 (overflow_infinity_range_p): Likewise.
33063 (usable_range_p): Remove as always returning true.
33064 (make_overflow_infinity): Remove.
33065 (negative_overflow_infinity): Likewise.
33066 (positive_overflow_infinity): Likewise.
33067 (avoid_overflow_infinity): Likewise.
33068 (set_value_range): Adjust accordingly.
33069 (set_value_range_to_nonnegative): Likewise, remove now unused
33070 overflow_infinity arg.
33071 (vrp_operand_equal_p): Adjust.
33072 (update_value_range): Likewise.
33073 (range_int_cst_singleton_p): Likewise.
33074 (operand_less_p): Likewise.
33075 (compare_values_warnv): Likewise.
33076 (extract_range_for_var_from_comparison_expr): Likewise.
33077 (vrp_int_const_binop): Likewise.
33078 (zero_nonzero_bits_from_vr): Likewise.
33079 (extract_range_from_multiplicative_op_1): Likewise.
33080 (extract_range_from_binary_expr_1): Likewise.
33081 (extract_range_from_unary_expr): Likewise.
33082 (extract_range_from_comparison): Likewise.
33083 (extract_range_basic): Likewise.
33084 (adjust_range_with_scev): Likewise.
33085 (compare_ranges): Likewise.
33086 (compare_range_with_value): Likewise.
33087 (dump_value_range): Likewise.
33088 (test_for_singularity): Likewise, remove strict_overflow_p parameter
33090 (simplify_cond_using_ranges): Adjust.
33092 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33094 * brig-builtins.def: Added a builtin for class_f64.
33095 * builtin-types.def: Added a builtin type needed by class_f64.
33097 2017-05-03 Jason Merrill <jason@redhat.com>
33099 * timevar.def: Add TV_CONSTEXPR.
33101 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33103 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
33105 2017-05-03 Martin Jambor <mjambor@suse.cz>
33107 * ipa-prop.c (ipa_update_after_lto_read): Removed.
33108 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
33109 * ipa-cp.c (ipcp_propagate_stage): Do not call
33110 ipa_update_after_lto_read.
33111 * ipa-inline.c (ipa_inline): Likewise.
33113 2017-05-03 Martin Jambor <mjambor@suse.cz>
33115 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
33116 tag. Added a default constructor and a destructor.
33117 (ipa_edge_args_sum_t): New class;
33118 (ipa_edge_args_sum): Declare.
33119 (ipa_edge_args_vector): Remove declaration.
33120 (IPA_EDGE_REF): Use ipa_edge_args_sum.
33121 (ipa_free_edge_args_substructures): Remove declaration.
33122 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
33123 (ipa_edge_args_info_available_for_edge_p): Likewise.
33124 * ipa-prop.c (ipa_edge_args_vector): Removed.
33125 (edge_removal_hook_holder): Likewise.
33126 (edge_duplication_hook_holder): Likewise.
33127 (ipa_edge_args_sum): New variable.
33128 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
33129 ipa_edge_args_vector.
33130 (ipa_free_edge_args_substructures): Likewise.
33131 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
33132 ipa_edge_args_vector.
33133 (ipa_edge_removal_hook): Turned into method
33134 ipa_edge_args_sum_t::remove.
33135 (ipa_edge_duplication_hook): Turned into method
33136 ipa_edge_args_sum_t::duplicate.
33137 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
33138 registering edge hooks.
33139 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
33140 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
33141 ipa_edge_args_sum instead of ipa_edge_args_vector.
33142 * ipa-profile.c (ipa_profile): Likewise.
33144 2017-05-03 Martin Jambor <mjambor@suse.cz>
33146 * symbol-summary.h (function_summary): New method exists.
33147 (function_summary::symtab_removal): Deallocate through release.
33148 (call_summary): New class.
33149 (gt_ggc_mx): New overload.
33150 (gt_pch_nx): Likewise.
33151 (gt_pch_nx): Likewise.
33153 2017-05-03 Jeff Law <law@redhat.com>
33155 PR tree-optimization/78496
33156 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
33157 from simplify_cond_using_ranges. Split off code to walk
33158 backwards through casts into ...
33159 (simplify_cond_using_ranges_2): New function.
33160 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
33161 (execute_vrp): After identifying jump threads, call
33162 simplify_cond_using_ranges_2.
33164 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
33167 * ipa-inline.h (inline_summary): Add ctor.
33168 (create_ggc): Do not use ggc_cleared_alloc.
33170 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
33171 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33173 * gcc.c (handle_braces): Support escaping in switch matching
33175 * doc/invoke.texi (Spec Files): Document it.
33176 Remove superfluous @code markup in items.
33178 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33180 * diagnostic-show-locus.c (struct column_range): New struct.
33181 (get_affected_columns): New function.
33182 (get_printed_columns): New function.
33183 (struct correction): New struct.
33184 (correction::ensure_capacity): New function.
33185 (correction::ensure_terminated): New function.
33186 (struct line_corrections): New struct.
33187 (line_corrections::~line_corrections): New dtor.
33188 (line_corrections::add_hint): New function.
33189 (layout::print_trailing_fixits): Reimplement in terms of the new
33191 (selftest::test_overlapped_fixit_printing): New function.
33192 (selftest::diagnostic_show_locus_c_tests): Call it.
33194 2017-05-03 Nathan Sidwell <nathan@acm.org>
33196 Canonicalize canonical type hashing
33197 * tree.h (type_hash_canon_hash): Declare.
33198 * tree.c (type_hash_list, attribute_hash_list): Move into
33199 type_hash_canon_hash.
33200 (build_type_attribute_qual_variant): Break out hash code calc into
33201 type_hash_canon_hash.
33202 (type_hash_canon_hash): New. Generic type hash computation.
33203 (build_range_type_1, build_array_type_1, build_function_type,
33204 build_method_type_directly, build_offset_type, build_complex_type,
33205 make_vector_type): Call it.
33207 2017-05-03 Richard Biener <rguenther@suse.de>
33209 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33210 When all DRs have unknown misaligned do not always peel
33211 when there is a store but apply the same costing model as if
33212 there were only loads.
33214 2017-05-03 Richard Biener <rguenther@suse.de>
33217 PR tree-optimization/80492
33218 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33219 compare_base_decls returning dont-know properly.
33221 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33223 * config/arm/iterators.md (CCSI): New mode iterator.
33224 (arch): New mode attribute.
33225 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33226 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33227 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33228 code iterator for success result mode.
33229 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33230 the corresponding new insn generators.
33232 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33235 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33236 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33238 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33240 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33241 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33242 (DDR_A): Wrap DDR argument in brackets.
33243 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33244 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33245 (DDR_REVERSED_P): Likewise.
33247 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33249 PR tree-optimization/79472
33250 * tree-switch-conversion.c (struct switch_conv_info): Add
33251 contiguous_range and default_case_nonstandard fields.
33252 (collect_switch_conv_info): Compute contiguous_range and
33253 default_case_nonstandard fields, don't clear final_bb if
33254 contiguous_range and only the default case doesn't have the required
33256 (check_all_empty_except_final): Set default_case_nonstandard instead
33257 of failing if contiguous_range and the default case doesn't have empty
33259 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33260 and only the default case doesn't have the required constants. Skip
33262 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33263 if default_case_nonstandard.
33264 (build_constructors): Build constant 1 just once. Assert that default
33265 values aren't inserted in between cases if contiguous_range. Skip
33267 (build_arrays): Skip virtual phis.
33268 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33269 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33270 Handle virtual phis.
33271 (gen_inbound_check): Handle default_case_nonstandard case.
33272 (process_switch): Adjust check_final_bb caller. Call
33273 gather_default_values with the first non-default case instead of
33274 default case if default_case_nonstandard.
33276 2017-05-02 Nathan Sidwell <nathan@acm.org>
33278 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33279 check. Fix formatting.
33281 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33283 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33284 errors when comparing specialized and unspecialized times.
33286 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33288 * diagnostic-show-locus.c
33289 (layout::should_print_annotation_line_p): Make private.
33290 (layout::print_annotation_line): Make private.
33291 (layout::annotation_line_showed_range_p): Make private.
33292 (layout::show_ruler): Make private.
33293 (layout::print_source_line): Make private. Pass in line and
33294 line_width, rather than calling location_get_source_line. Drop
33296 (layout::print_leading_fixits): New method.
33297 (layout::print_any_fixits): Rename to...
33298 (layout::print_trailing_fixits): ...this, and make private.
33299 Don't print newline fixits.
33300 (diagnostic_show_locus): Move logic for printing one row into...
33301 (layout::print_line): ...this new function. Move the
33302 location_get_source_line call and error-handling from
33303 print_source_line to here. Call print_leading_fixits, and rename
33304 print_any_fixits to print_trailing_fixits.
33305 (selftest::test_fixit_insert_containing_newline): Update now that
33306 newlines are partially supported.
33307 (selftest::test_fixit_insert_containing_newline_2): New test.
33308 (selftest::test_fixit_replace_containing_newline): Update comments.
33309 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33310 * edit-context.c (class added_line): New class.
33311 (class edited_line): Describe newline handling in comment.
33312 (edited_line::actually_edited_p): New method.
33313 (edited_line::print_content): Delete redundant decl.
33314 (edited_line::m_predecessors): New field.
33315 (edited_file::print_content): Call edited_line::print_content.
33316 (edited_file::print_diff): Update to support newlines.
33317 (edited_file::print_diff_hunk): Likewise.
33318 (edited_file::print_run_of_changed_lines): New function.
33319 (edited_file::print_diff_line): Convert to...
33320 (print_diff_line): ...this.
33321 (edited_file::get_effective_line_count): New function.
33322 (edited_line::edited_line): Initialize new field m_predecessors.
33323 (edited_line::~edited_line): Clean up m_predecessors.
33324 (edited_line::apply_fixit): Handle newlines.
33325 (edited_line::get_effective_line_count): New function.
33326 (edited_line::print_content): New function.
33327 (edited_line::print_diff_lines): New function.
33328 (selftest::test_applying_fixits_insert_containing_newline): New
33330 (selftest::test_applying_fixits_replace_containing_newline): New
33332 (selftest::insert_line): New function.
33333 (selftest::test_applying_fixits_multiple_lines): Add example of
33335 (selftest::edit_context_c_tests): Call the new tests.
33337 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33339 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33340 parameter cand. Update dump information.
33341 (get_computation_cost): Update uses.
33343 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33345 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33346 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33347 (get_computation_at, rewrite_use_address): Update use of
33348 get_computation_aff.
33350 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33352 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33353 (get_computation): Delete.
33354 (get_computation_cost): Implement like get_computation_cost_at.
33355 Use get_computation_at.
33356 (get_computation_cost_at): Delete.
33357 (rewrite_use_nonlinear_expr): Use get_computation_at.
33358 (rewrite_use_compare, remove_unused_ivs): Ditto.
33360 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33362 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33364 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33366 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33367 (ivopts_global_cost_for_size): Rename parameter and update uses.
33368 (iv_ca_recount_cost): Update uses.
33369 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33370 candidates seperately in n_invs and n_cands.
33371 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33373 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33375 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33376 (find_inv_vars_cb): New.
33377 (find_depends): Renamed to ...
33378 (find_inv_vars): ... this.
33379 (add_candidate_1, force_var_cost): Call find_inv_vars.
33380 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33382 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33384 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33385 inv_vars. Add inv_exprs.
33386 (struct iv_cand): Rename depends_on to inv_vars.
33387 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33388 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33389 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33390 (dump_cand): Dump inv_vars.
33391 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33392 (record_invariant, find_depends, add_candidate_1): Ditto.
33393 (set_group_iv_cost, force_var_cost): Ditto.
33394 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33395 (get_computation_cost_at, get_computation_cost): Ditto.
33396 (determine_group_iv_cost_generic): Ditto.
33397 (determine_group_iv_cost_address): Ditto.
33398 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33399 (determine_group_iv_costs): Ditto.
33400 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33401 (iv_ca_set_remove_invariants): Renamed to ...
33402 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33403 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33404 (iv_ca_set_add_invariants): Renamed to ...
33405 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33406 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33407 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33408 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33409 (create_new_ivs): Remove useless dump.
33411 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33413 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33415 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33416 (iv_ca_set_no_cp, create_new_iv): Ditto.
33418 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33420 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33422 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33424 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33425 function tree_check2.
33427 2017-05-02 Martin Liska <mliska@suse.cz>
33429 * doc/gcov.texi: Add missing preposition.
33430 * gcov.c (function_info::function_info): Properly fill up
33431 all member variables.
33433 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33435 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33437 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33439 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33441 2017-05-02 Martin Liska <mliska@suse.cz>
33444 * lto-streamer-in.c (lto_read_tree_1): Remove
33445 LTO_STREAMER_DEBUG.
33446 * lto-streamer.c (struct tree_hash_entry): Likewise.
33447 (struct tree_entry_hasher): Likewise.
33448 (tree_entry_hasher::hash): Likewise.
33449 (tree_entry_hasher::equal): Likewise.
33450 (lto_streamer_init): Likewise.
33451 (lto_orig_address_map): Likewise.
33452 (lto_orig_address_get): Likewise.
33453 (lto_orig_address_remove): Likewise.
33454 * lto-streamer.h: Likewise.
33455 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33456 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33458 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33460 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33461 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33462 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33463 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33464 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33465 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33466 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33467 (mm_maskz_sub_ss): New intrinsics.
33468 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33469 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33470 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33471 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33472 (__builtin_ia32_subss_mask_round): New builtins.
33473 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33474 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33475 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33477 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33478 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33480 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33483 2017-05-02 Martin Jambor <mjambor@suse.cz>
33485 PR tree-optimization/78687
33486 * tree-sra.c (access): New field parent.
33487 (process_subtree_disqualification): New function.
33488 (disqualify_candidate): Call it.
33489 (build_accesses_from_assign): Reset write flag if creating an
33491 (build_access_subtree): Fill in parent field and also prpagate
33492 down grp_write flag.
33493 (create_artificial_child_access): New parameter set_grp_write, set
33494 grp_write to its value.
33495 (propagate_subaccesses_across_link): Also propagate grp_write flag
33497 (propagate_all_subaccesses): Push the closest parent back to work
33498 queue if add_access_to_work_queue returned true.
33500 2017-05-02 Richard Biener <rguenther@suse.de>
33502 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33503 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33504 -fstrict-overflow documentation.
33505 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33506 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33507 flag_strict_overflow.
33508 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33509 * lto-opts.c (lto_write_options): Do not stream it.
33510 * lto-wrapper.c (merge_and_complain): Do not handle it.
33511 * opts.c (default_options_table): Do not set -fstrict-overflow.
33512 (finish_options): Likewise do not clear it when sanitizing.
33513 * simplify-rtx.c (simplify_const_relational_operation): Do not
33514 test flag_strict_overflow.
33516 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33518 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33519 using enabled attribute.
33520 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33521 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33522 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33523 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33524 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33525 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33526 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33527 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33528 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33529 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33531 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33533 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33535 2017-05-02 Richard Biener <rguenther@suse.de>
33537 PR tree-optimization/80591
33539 2017-04-10 Richard Biener <rguenther@suse.de>
33541 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33544 2017-05-02 Richard Biener <rguenther@suse.de>
33546 PR tree-optimization/80549
33547 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33548 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33549 headers if they do not have a preheader.
33551 2017-05-02 Martin Liska <mliska@suse.cz>
33554 * common.opt: Fix typo.
33555 * doc/invoke.texi: Likewise.
33557 2017-05-01 Jan Beulich <jbeulich@suse.com>
33559 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33560 swapping, add (x,x,m,x,n) alternative.
33562 2017-05-01 Nathan Sidwell <nathan@acm.org>
33564 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33565 unnecessary unadjusted_alignment check.
33567 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33570 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33572 * gimplify.c (gimplify_cilk_detach): New function.
33573 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33574 * tree-core.h: Document EXPR_CILK_SPAWN.
33575 * tree.h (EXPR_CILK_SPAWN): Define.
33577 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33579 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33580 to use new fixit_hint representation, using the "replace" logic.
33581 (get_line_span_for_fixit_hint): Likewise.
33582 (layout::print_any_fixits): Likewise.
33583 (selftest::test_one_liner_many_fixits): Rename to...
33584 (selftest::test_one_liner_many_fixits_1): ...this, and update
33585 comment and expected output to reflect that the multiple fix-it
33586 hints are now consolidated into one insertion.
33587 (selftest::test_one_liner_many_fixits_2): New test.
33588 (selftest::test_diagnostic_show_locus_one_liner): Update for
33590 (selftest::test_fixit_consolidation): Update for fix-it API
33592 * diagnostic.c (print_parseable_fixits): Likewise.
33593 * edit-context.c (edited_line::m_line_events): Convert from
33594 auto_vec <line_event *> to auto_vec <line_event>.
33595 (class line_event): Convert from abstract base class to a concrete
33596 class, taking over the role of replace_event.
33597 (class insert_event): Delete.
33598 (class replace_event): Rename to class line_event. Convert to
33600 (edit_context::add_fixits): Reimplement.
33601 (edit_context::apply_insert): Delete.
33602 (edit_context::apply_replace): Rename to...
33603 (edit_context::apply_fixit): ...this. Convert to half-open range.
33604 (edited_file::apply_insert): Delete.
33605 (edited_file::apply_replace): Rename to...
33606 (edited_file::apply_fixit): ...this.
33607 (edited_line::~edited_line): Drop deletion of events.
33608 (edited_line::apply_insert): Delete.
33609 (edited_line::apply_replace): Rename to...
33610 (edited_line::apply_fixit): ...this. Convert to half-open range.
33611 Update for change to type of m_line_events.
33612 * edit-context.h (edit_context::apply_insert): Delete.
33613 (edit_context::apply_replace): Rename to...
33614 (edit_context::apply_fixit): ...this.
33616 2017-05-01 Martin Sebor <msebor@redhat.com>
33618 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33621 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33624 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33625 __get_cpuid_max returns 0.
33626 (__get_cpuid_count): Ditto.
33628 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33630 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33631 replacement expression is another instance of one of its arguments.
33633 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33636 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33637 check for stack push/pop autoinc.
33638 * config/i386/i386.c (ix86_agi_dependent): Return false
33639 if the only reason why modified_in_p returned true is that
33640 addr is SP based and set_insn is a push or pop.
33642 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33644 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33647 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33650 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33651 (account_size_time): Use two predicates - exec_pred and
33653 (evaluate_conditions_for_known_args): Compute both clause and
33655 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33656 (inline_summary_t::duplicate): Update.
33657 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33659 (compute_inline_parameters): Likewise.
33660 (estimate_edge_size_and_time): Update caluclation of time.
33661 (estimate_node_size_and_time): Compute both time and nonspecialized
33663 (estimate_ipcp_clone_size_and_time): Update.
33664 (inline_merge_summary): Update.
33665 (do_estimate_edge_time): Update.
33666 (do_estimate_edge_size): Update.
33667 (do_estimate_edge_hints): Update.
33668 (inline_read_section, inline_write_summary): Stream both new predicates.
33669 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33671 (compute_inlined_call_time): Cleanup.
33672 (big_speedup_p): Update.
33673 (edge_badness): Update.
33674 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33675 (size_time_entry): Replace predicate by exec_predicate and
33676 nonconst_predicate.
33677 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33678 (estimate_edge_time): Return also nonspec_time.
33679 (reset_edge_growth_cache): Update.
33681 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33683 PR rtl-optimization/80491
33684 * ifcvt.c (noce_process_if_block): When looking for x setter
33685 with missing else_bb, don't check only the insn right before
33686 cond_earliest, but look for the last insn that x is modified in
33687 within the same bb.
33689 PR rtl-optimization/80491
33690 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33692 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33694 PR tree-optimization/80487
33695 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33697 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33699 PR tree-optimization/79697
33700 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33701 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33702 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33704 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33705 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33707 2017-04-28 Martin Sebor <msebor@redhat.com>
33709 PR tree-optimization/80523
33710 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33711 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33713 (maybe_warn, format_directive, parse_directive): Use new functions.
33714 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33716 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33718 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33720 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33722 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33723 target_header_dir): Set correctly.
33724 * configure: Regenerated.
33725 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33726 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33727 instead of SYSTEM_HEADER_DIR.
33729 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33731 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33732 (estimate_local_effects): Likewise.
33733 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33734 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33735 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33736 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33737 do_estimate_edge_time, estimate_edge_time): Likewise.
33738 * ipa-inline-analysis.c (estimate_node_size_and_time,
33739 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33740 (estimate_time_after_inlining): Remove.
33742 2017-04-28 Martin Liska <mliska@suse.cz>
33744 * doc/gcov.texi: Enhance documentation of gcov.
33746 2017-04-28 Martin Liska <mliska@suse.cz>
33748 * doc/gcov.texi: Sort options in alphabetic order.
33749 * doc/gcov-dump.texi: Likewise.
33750 * doc/gcov-tool.texi: Likewise.
33751 * gcov.c (print_usage): Likewise.
33752 * gcov-dump.c (print_usage): Likewise.
33753 * gcov-tool.c (print_merge_usage_message): Likewise.
33754 (print_rewrite_usage_message): Likewise.
33755 (print_overlap_usage_message): Likewise.
33757 2017-04-28 Martin Liska <mliska@suse.cz>
33759 PR gcov-profile/53915
33760 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33762 2017-04-28 Martin Liska <mliska@suse.cz>
33764 PR gcov-profile/79891
33765 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33766 is marked by compiler as living on a line.
33767 (get_cycles_count): Remove usage of the union.
33768 (output_intermediate_file): Likewise.
33769 (find_source): Fix GNU coding style.
33770 (accumulate_line_counts): Remove old non-all block mode.
33771 (output_lines): Remove usage of the union.
33772 * profile.c (output_location): Include all BBs, even if
33773 belonging to a same line (and file) as a previous BB.
33775 2017-04-28 Martin Liska <mliska@suse.cz>
33777 * gcov.c (process_args): Handle new argument 'w'.
33778 (read_graph_file): Assign ID to BBs.
33779 (output_branch_count): Display BB # if verbose flag is set.
33780 (output_lines): Likewise for arcs.
33781 (print_usage): Add '--verbose' option help.
33782 * doc/gcov.texi: Document --verbose (-w) option.
33784 2017-04-28 Martin Liska <mliska@suse.cz>
33786 * gcov.c (struct block_location_info): New struct.
33787 (process_file): Fill up the new structure.
33788 (read_graph_file): Replace usage of encoding by the newly added
33790 (add_line_counts): Likewise.
33791 (accumulate_line_counts): Remove usage of the union.
33792 (function_info::function_info): New function.
33793 (function_info::~function_info): Likewise.
33794 (process_file): Call delete instead of release_function.
33795 (release_function): Release the function.
33796 (release_structures): Call delete instead of release_function.
33797 (solve_flow_graph): Replace usage of num_blocks.
33798 (find_exception_blocks): Likewise.
33799 (output_lines): Fix GNU coding style.
33801 2017-04-28 Martin Liska <mliska@suse.cz>
33804 * coverage.c (coverage_remove_note_file): New function.
33805 * coverage.h: Declare the function.
33806 * toplev.c (finalize): Clean if an error has been seen.
33808 2017-04-28 Martin Liska <mliska@suse.cz>
33810 PR gcov-profile/80031
33811 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33812 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33813 * gcov.c (read_graph_file): Read just number of blocks.
33814 * profile.c (branch_prob): Do not stream 0 flags per a basic
33817 2017-04-28 Martin Liska <mliska@suse.cz>
33819 * gcov-dump.c (tag_*): Add new argument to declarations.
33820 (dump_gcov_file): Likewise.
33821 (tag_blocks): Add and use new argument depth.
33822 (tag_arcs): Likewise.
33823 (tag_lines): Likewise.
33824 (tag_counters): Likewise.
33825 (tag_summary): Likewise.
33826 (dump_working_sets): Use depth to do a proper indentation.
33828 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33831 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33832 * symtab.c (symtab_node::debug_symtab): Move definition here.
33834 2017-04-28 Richard Biener <rguenther@suse.de>
33836 * lto-streamer.h (LTO_major_version): Bump to 7.
33838 2017-04-28 Richard Biener <rguenther@suse.de>
33840 * tree-vrp.c (assert_info): New struct.
33841 (add_assert_info): New helper.
33842 (register_edge_assert_for_2): Refactor to add asserts to a vector
33844 (register_edge_assert_for_1): Likewise.
33845 (register_edge_assert_for): Likewise.
33846 (finish_register_edge_assert_for): New helper actually registering
33847 asserts where live on edge.
33848 (find_conditional_asserts): Adjust.
33849 (find_switch_asserts): Likewise.
33850 (evrp_dom_walker::try_find_new_range): Generalize.
33851 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33853 2017-04-27 Marek Polacek <polacek@redhat.com>
33856 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33857 arg10 and arg11 to itype.
33859 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33861 * doc/extend.texi (Object Size Checking): Improve grammar.
33863 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33866 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33867 that the logic for permitting reciprocal estimates matches that
33870 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33873 * tree.c (type_cache_hasher::equal): Only compare
33874 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33875 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33876 non-aggregate element types.
33877 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33878 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33880 2017-04-27 Richard Biener <rguenther@suse.de>
33882 PR middle-end/80533
33883 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33884 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33885 keeping a reference to a trailing array.
33887 2017-04-27 Richard Biener <rguenther@suse.de>
33889 PR middle-end/80539
33890 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33891 being in loop-closed SSA form conservatively.
33892 (chrec_fold_multiply_poly_poly): Likewise.
33894 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33896 PR middle-end/79665
33897 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33898 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33900 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33903 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33904 (aarch64_function_arg_alignment): Return unsigned int again, but still
33905 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33906 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33907 Don't emit -Wpsabi note.
33908 (aarch64_function_arg_boundary): Likewise.
33909 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33912 2017-04-26 Nathan Sidwell <nathan@acm.org>
33914 * tree.h (crc32_unsigned_n): Declare.
33915 (crc32_unsigned, crc32_unsigned): Make inline.
33916 * tree.c (crc32_unsigned_bits): Replace with ...
33917 (crc32_unsigned_n): ... this.
33918 (crc32_unsigned, crc32_byte): Remove.
33919 (crc32_string): Remove unnecessary braces.
33921 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33923 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33924 * ipa-inline-analysis.c (MAX_TIME): Remove.
33925 (account_size_time): Use sreal for time.
33926 (dump_inline_summary): Update.
33927 (estimate_function_body_sizes): Update.
33928 (estimate_edge_size_and_time): Update.
33929 (estimate_calls_size_and_time): Update.
33930 (estimate_node_size_and_time): Update.
33931 (inline_merge_summary): Update.
33932 (inline_update_overall_summary): Update.
33933 (estimate_time_after_inlining): Update.
33934 (inline_read_section): Update.
33935 (inline_write_summary): Update.
33936 * ipa-inline.c (compute_uninlined_call_time): Update.
33937 (compute_inlined_call_time): Update.
33938 (recursive_inlining): Update.
33939 (inline_small_functions): Update.
33940 (dump_overall_stats): Update.
33941 * ipa-inline.h: Include sreal.h.
33942 (size_time_entry): Turn time to sreal.
33943 (inline_summary): Turn self_time nad time to sreal.
33945 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33947 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33949 (sreal::stream_out, sreal::stream_in): New.
33950 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33952 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33954 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33957 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33960 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33962 (dimode_scalar_chain::compute_convert_gain): Ditto.
33963 (dimode_scalar_chain::make_vector_copies): Ditto.
33964 (dimode_scalar_chain::convert_reg): Ditto.
33965 (dimode_scalar_chain::convert_insn): Ditto.
33966 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33967 (VI248_AVX512BW_1): New mode iterator.
33968 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33969 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33972 2017-04-25 Martin Sebor <msebor@redhat.com>
33974 PR tree-optimization/80497
33975 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33976 constants are representable in HOST_WIDE_INT.
33977 (parse_directive): Ditto.
33979 2017-04-25 Martin Sebor <msebor@redhat.com>
33982 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33983 (new_zero_array): Adjust signature.
33984 (dom_info::dom_init): Used unsigned rather that size_t.
33985 (dom_info::dom_info): Same.
33987 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33988 Jakub Jelinek <jakub@redhat.com>
33991 * config/arm/arm.c: Include gimple.h.
33992 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33993 returns negative, increment ncrn only if it returned positive.
33994 (arm_needs_doubleword_align): Return int instead of bool,
33995 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33996 members, but if there is any such non-FIELD_DECL
33997 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33998 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33999 returns negative, increment nregs only if it returned positive.
34000 (arm_setup_incoming_varargs): Likewise.
34001 (arm_function_arg_boundary): Emit -Wpsabi note if
34002 arm_needs_doubleword_align returns negative, return
34003 DOUBLEWORD_ALIGNMENT only if it returned positive.
34005 2017-04-25 Marek Polacek <polacek@redhat.com>
34008 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34009 first argument to type.
34011 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
34014 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
34015 type checks to test for compatibility instead of equality.
34017 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
34018 Jakub Jelinek <jakub@redhat.com>
34021 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
34023 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
34024 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
34025 the alignment computation, but return their maximum in warn_alignment.
34026 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
34027 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
34029 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
34030 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
34033 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34035 * config/arc/simdext.md (dmpyh): Fix typo.
34037 2017-04-25 Richard Biener <rguenther@suse.de>
34039 PR tree-optimization/80492
34040 * alias.c (compare_base_decls): Handle registers with asm
34041 specification conservatively.
34042 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
34043 compare_base_decls returning dont-know properly.
34045 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34047 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
34048 (legitimate_offset_address_p): New function.
34049 (arc_legitimate_address_p): Use above function.
34051 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34053 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
34055 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34057 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
34058 ACCH registers whenever they are available.
34060 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34062 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
34063 double regs fix when not used.
34065 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34067 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
34069 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
34070 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
34072 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34074 * config/arc/arc.c (arc_output_addsi): Check for h-register class
34075 when emitting short ADD instructions.
34077 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34079 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
34081 (cmpsi_cc_c_insn): Likewise.
34082 (cbranchsi4_scratch): Compute proper instruction length using
34083 compact_hreg_operand.
34084 * config/arc/predicates.md (compact_hreg_operand): New predicate.
34086 2017-04-25 Richard Biener <rguenther@suse.de>
34088 PR middle-end/80509
34089 * passes.c (pass_manager::pass_manager): Initialize
34090 m_name_to_pass_map.
34092 2017-04-25 Richard Biener <rguenther@suse.de>
34094 PR tree-optimization/79201
34095 * tree-ssa-sink.c (statement_sink_location): Handle calls.
34097 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34100 * config/s390/vector.md: Split MEM->GPR vector moves for
34101 non-s_operand addresses.
34103 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34106 * config/s390/predicates.md (reload_const_wide_int_operand): New
34108 * config/s390/s390.md ("movti"): Remove d/P alternative.
34109 ("movti_bigconst"): New pattern definition.
34111 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34114 * s390-protos.h (s390_expand_cs_hqi): Removed.
34115 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
34116 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
34117 modes as well as CCZ1mode and CCZmode.
34118 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
34119 signature of s390_emit_compare_and_swap.
34120 (s390_expand_cs_hqi): Likewise, make static.
34121 (s390_expand_cs_tdsi): Generate an explicit compare before trying
34122 compare-and-swap, in some cases.
34123 (s390_expand_cs): Wrapper function.
34124 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
34126 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
34127 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
34128 patterns for small and large integers. Forbid symref memory operands.
34129 Move expander to s390.c. Require cc register.
34130 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
34131 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
34132 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
34133 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
34134 symref memory operands. Remove CC mode and call s390_match_ccmode
34136 ("atomic_exchange<mode>"): Allow and implement all integer modes.
34138 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34140 * config/s390/s390.md (define_peephole2): New peephole to help
34141 combining the load-and-test pattern with volatile memory.
34143 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34145 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
34146 with CCZmode for TARGET_Z196.
34148 2017-04-25 Jakub Jelinek <jakub@redhat.com>
34150 PR rtl-optimization/80501
34151 * combine.c (make_compound_operation_int): Set subreg_code to SET
34152 even for AND with mask of the sign bit of mode.
34154 PR rtl-optimization/80500
34155 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
34156 sum's initial value.
34158 2017-04-25 Julian Brown <julian@codesourcery.com>
34159 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34161 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
34163 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
34165 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
34167 2017-04-25 Julian Brown <julian@codesourcery.com>
34168 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34170 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
34171 (thunderx2t99_sha): New Reservation.
34173 2017-04-25 Julian Brown <julian@codesourcery.com>
34174 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34176 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
34177 type for 1-element load.
34179 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
34181 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
34183 2017-04-24 Martin Jambor <mjambor@suse.cz>
34185 PR tree-optimization/80293
34186 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
34187 char arrays not totally scalarizable if it is false.
34188 (analyze_all_variable_accesses): Pass correct value in the new
34189 parameter. Add a statistics counter.
34191 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
34193 PR middle-end/79931
34194 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
34196 2017-04-24 Richard Biener <rguenther@suse.de>
34198 PR tree-optimization/80494
34199 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
34200 out for complex types.
34202 2017-04-24 Richard Biener <rguenther@suse.de>
34204 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34205 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34206 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34207 (DFS): Adjust and never fail.
34208 (sccvn_dom_walker::fail): Remove.
34209 (sccvn_dom_walker::before_dom_children): Adjust.
34210 (run_scc_vn): Likewise and never fail.
34211 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34212 (pass_fre::execute): Likewise.
34214 2017-04-24 Richard Biener <rguenther@suse.de>
34216 PR tree-optimization/79725
34217 * tree-ssa-sink.c (statement_sink_location): Return whether
34218 failure reason was zero uses. Move that check later.
34219 (sink_code_in_bb): Deal with zero uses by removing the stmt
34222 2017-04-24 Richard Biener <rguenther@suse.de>
34225 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34226 pointer-based references.
34228 2017-04-24 Richard Biener <rguenther@suse.de>
34231 * pass_manager.h (pass_manager::operator new): Remove.
34232 (pass_manager::operator delete): Likewise.
34233 * passes.c (pass_manager::operator new): Remove.
34234 (pass_manager::operator delete): Likewise.
34235 (pass_manager::pass_manager): Zero individual pass members.
34237 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34240 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34241 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34242 Check "XEXP (src, 1)" operand here.
34243 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34244 Check "XEXP (src, 1)" operand here.
34245 (dimode_scalar_chain::make_vector_copies): Detect count register
34246 of a shift instruction. Zero extend count register from QImode
34247 to DImode to satisfy vector shift pattern count operand predicate.
34248 Substitute vector shift count operand with a DImode copy.
34249 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34252 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34254 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34255 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34256 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34257 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34258 (UNSPEC_NOREX_MEM): Remove definition.
34260 2017-04-21 Richard Biener <rguenther@suse.de>
34262 PR tree-optimization/79547
34263 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34264 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34265 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34266 without any constraints.
34268 2017-04-21 Richard Biener <rguenther@suse.de>
34270 PR tree-optimization/78847
34271 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34273 2017-04-21 Richard Biener <rguenther@suse.de>
34275 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34276 (build_distinct_type_copy): Likewise.
34277 (build_variant_type_copy): Likewise.
34278 * tree.c (build_qualified_type): Pass down mem-stat info.
34279 (build_distinct_type_copy): Likewise.
34280 (build_variant_type_copy): Likewise.
34282 2017-04-21 Richard Biener <rguenther@suse.de>
34284 PR tree-optimization/80237
34285 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34287 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34288 for a simplified result.
34290 2016-04-21 Richard Biener <rguenther@suse.de>
34292 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34293 sth as strict as a simple_iv but a chrec without symbols and an
34294 operand defined in the loop we are peeling (and not some subloop).
34295 (propagate_constants_for_unrolling): Propagate all constants.
34297 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34300 * config/i386/i386.c (print_reg): Remove assert for disalowed
34301 regno values, call output_operand_lossage instead.
34303 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34306 * config/i386/constraints.md (Yc): New register constraint.
34307 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34308 Use Yc constraint for alternative 2 of operand 0. Remove
34309 preferred_for_speed attribute.
34311 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34313 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34314 lastprivate clauses in SIMT case.
34316 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34318 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34320 2017-04-20 Richard Biener <rguenther@suse.de>
34322 PR tree-optimization/57796
34323 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34324 as N scalar stores.
34325 (vect_model_load_cost): Cost gathers as N scalar loads.
34327 2017-04-20 Richard Biener <rguenther@suse.de>
34329 * ggc-page.c (ggc_allocated_p): Rename to ...
34330 (safe_lookup_page_table_entry): ... this and return the lookup
34332 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34334 2017-04-20 Richard Biener <rguenther@suse.de>
34336 PR tree-optimization/80453
34337 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34338 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34339 from the conditions.
34340 (vn_phi_eq): Pass them down.
34341 (vn_phi_lookup): Record them.
34342 (vn_phi_insert): Likewise.
34344 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34346 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34347 uninitialized variable warning to avoid buffer overrun.
34349 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34352 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34353 is suppressed for '{ 0 }' in C.
34355 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34357 * BASE-VER: Set to 8.0.0.
34359 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34361 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34362 priority .init_array and .fini_array section with SECTION_NOTYPE
34365 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34367 PR middle-end/80423
34368 * tree.h (build_array_type): Add typeless_storage default argument.
34369 * tree.c (type_cache_hasher::equal): Also compare
34370 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34371 (build_array_type): Add typeless_storage argument, set
34372 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34374 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34375 (build_array_type): Likewise. Add typeless_storage argument.
34377 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34378 Jakub Jelinek <jakub@redhat.com>
34380 PR tree-optimization/80426
34381 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34382 operation on symbolic operands, also compute the overflow for the
34383 invariant part when the operation degenerates into a negation.
34385 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34388 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34389 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34392 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34394 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34397 * config/avr/avr.c (tree.h): Include it.
34398 (cgraph.h): Include it.
34399 (avr_encode_section_info): Don't warn for uninitialized progmem
34400 variable if it's just an alias.
34402 2017-04-19 Richard Biener <rguenther@suse.de>
34405 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34406 when needed by AutoPGO.
34408 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34411 * doc/lto.texi: Remove an extra 'that'.
34413 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34415 PR rtl-optimization/80429
34416 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34417 are only used in debug insns.
34419 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34420 Vladimir Makarov <vmakarov@redhat.com>
34422 * config/sparc/predicates.md (input_operand): Add comment. Return
34423 true for any memory operand when LRA is in progress.
34424 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34426 2017-04-18 Jeff Law <law@redhat.com>
34429 * mips.md ({return,simple_return}_internal): Do not overwrite
34432 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34434 PR tree-optimization/80443
34435 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34436 instead of adding 1, subtract -1 and similarly instead of subtracting
34439 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34441 PR rtl-optimization/80357
34442 * haifa-sched.c (tmp_bitmap): New variable.
34443 (model_recompute): Handle duplicate use records.
34444 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34445 (free_global_sched_pressure_data): Free it.
34447 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34450 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34451 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34452 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34453 instead of SYSTEM_HEADER_DIR.
34455 2017-04-18 Jeff Law <law@redhat.com>
34457 PR middle-end/80422
34458 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34459 predecessors after walking up the insn chain.
34461 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34464 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34465 sizetype type into debug info.
34467 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34470 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34471 unneeded test for TARGET_UPPER_REGS_SF.
34472 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34474 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34477 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34478 instead of gsi_after_labels.
34480 2017-04-18 Jeff Law <law@redhat.com>
34482 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34486 2017-04-13 Jeff Law <law@redhat.com>
34487 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34488 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34490 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34493 * config/avr/avr.c (intl.h): Include it.
34494 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34496 2017-04-18 Martin Liska <mliska@suse.cz>
34498 PR gcov-profile/78783
34499 * gcov-tool.c (gcov_output_files): Validate that destination
34500 file is either removed by the tool or by a user.
34502 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34503 Guy Benyei <guybe@mellanox.com>
34505 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34506 block, and do not negate it, the stored id is already negative.
34508 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34510 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34512 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34515 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34516 masks of options that should be turned off if the VSX vector
34517 options are turned off.
34518 (OTHER_P8_VECTOR_MASKS): Likewise.
34519 (OTHER_VSX_VECTOR_MASKS): Likewise.
34520 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34521 rs6000_disable_incompatible_switches to validate no type switches
34523 (rs6000_incompatible_switch): New function to disallow turning on
34524 other vector options if -mno-vsx, -mno-power8-vector, or
34525 -mno-power9-vector are specified.
34527 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34529 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34531 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34533 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34534 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34535 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34536 (ARG_POINTER_CFA_OFFSET): Likewise.
34538 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34540 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34541 conditions to take advantage of various optimizations.
34543 2017-04-13 Jeff Law <law@redhat.com>
34545 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34546 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34547 (zero_extendsidi2_dext): Likewise.
34549 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34552 * fold-const.c (fold_ternary_loc): Revert
34553 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34556 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34558 PR rtl-optimization/80343
34559 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34560 new scratch pseudo.
34562 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34565 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34566 to ubsan_encode_value.
34568 2017-04-13 Jeff Law <law@redhat.com>
34570 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34571 appearing in DEBUG_INSNs.
34573 2017-04-13 Martin Liska <mliska@suse.cz>
34575 PR gcov-profile/80413
34576 * gcov-io.c (gcov_write_string): Copy to buffer just when
34577 allocated size is greater than zero.
34579 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34582 * dwarf2out.c (decls_for_scope): Ignore declarations of
34583 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34585 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34588 * ipa-visibility.c (non_local_p): Fix typos.
34589 (localize_node): When localizing symbol in same comdat group,
34590 dissolve the group only when we know external symbols are going
34592 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34594 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34596 PR tree-optimization/79390
34597 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34598 order does not result in usable sequence, retry with reversed operand
34604 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34605 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34606 op0 instead of fold_convert_loc (loc, type, arg0).
34608 2017-04-12 Jeff Law <law@redhat.com>
34610 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34611 has a delay slot in the generated code.
34613 * config/cris/cris.md (cris_preferred_reload_class): Return
34614 GENNONACR_REGS rather than GENERAL_REGS.
34616 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34619 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34620 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34621 signedness of the result type.
34623 2017-04-12 Richard Biener <rguenther@suse.de>
34624 Jeff Law <law@redhat.com>
34626 PR tree-optimization/80359
34627 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34628 trim stores to TARGET_MEM_REFs.
34630 2017-04-12 Richard Biener <rguenther@suse.de>
34632 PR tree-optimization/79390
34633 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34634 threading case even more.
34636 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34639 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34640 for quad_address_p for TImode, instead of just not indexed_address.
34642 2017-04-12 Richard Biener <rguenther@suse.de>
34643 Bernd Edlinger <bernd.edlinger@hotmail.de>
34645 PR middle-end/79671
34646 * alias.c (component_uses_parent_alias_set_from): Handle
34647 TYPE_TYPELESS_STORAGE.
34648 (get_alias_set): Likewise.
34649 * tree-core.h (tree_type_common): Add typeless_storage flag.
34650 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34651 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34652 for types containing members with TYPE_TYPELESS_STORAGE.
34653 (place_field): Likewise.
34654 (layout_type): Likewise for ARRAY_TYPE.
34655 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34656 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34657 TYPE_TYPELESS_STORAGE.
34658 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34660 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34663 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34664 first argument to type.
34666 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34670 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34671 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34672 (rs6000_expand_binop_builtin): Likewise.
34673 (rs6000_expand_ternop_builtin): Likewise; also add missing
34674 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34675 vshasigma built-ins.
34676 * doc/extend.texi: Document that vec_xxpermdi's third argument
34677 must be a constant.
34679 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34681 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34682 Use shift_const cost parameter when calculating gain of STV shifts.
34684 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34686 PR rtl-optimization/70478
34687 * lra-constraints.c (process_alt_operands): Check memory for
34688 disfavoring memory insn operand.
34690 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34692 PR middle-end/80100
34693 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34694 left shift in unsigned HOST_WIDE_INT type.
34696 PR rtl-optimization/80385
34697 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34698 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34701 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34702 if they have any depend clauses.
34704 2017-04-11 Martin Liska <mliska@suse.cz>
34707 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34708 * ipa-split.c (split_function): Create a local comdat symbol
34709 if caller is in a comdat group.
34711 2017-04-11 Martin Liska <mliska@suse.cz>
34714 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34717 2017-04-11 Martin Sebor <msebor@redhat.com>
34719 PR middle-end/80364
34720 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34721 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34723 (directive::set_width, directive::set_precision, format_character):
34725 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34728 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34731 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34732 conflict, set target->arch_name instead of target->cpu_name.
34734 2017-04-11 Richard Biener <rguenther@suse.de>
34736 PR tree-optimization/80374
34737 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34738 build_zero_cst, remove fold_convertible_p check again.
34740 2017-04-11 Martin Liska <mliska@suse.cz>
34743 * ubsan.c (instrument_object_size): Do not instrument register
34746 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34749 * config/i386/i386-builtin-types.def
34750 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34751 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34752 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34753 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34754 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34755 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34756 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34757 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34758 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34759 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34760 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34761 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34762 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34763 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34764 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34765 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34766 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34767 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34768 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34769 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34770 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34771 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34772 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34773 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34774 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34775 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34776 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34777 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34778 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34779 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34780 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34781 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34782 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34783 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34784 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34785 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34786 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34787 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34788 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34789 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34790 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34791 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34792 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34793 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34794 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34795 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34797 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34798 flag to second_arg_count, handle 4 argument function type _COUNT
34799 aliases, handle second_arg_count on second argument rather than last.
34801 2017-04-10 Jeff Law <law@redhat.com>
34803 PR tree-optimization/80374
34804 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34805 record anything if we can not convert integer_zero_node to the
34808 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34811 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34812 Enhance special handling given to the TARGET_P9_MINMAX option in
34813 relation to certain other options.
34815 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34817 PR tree-optimization/80153
34818 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34819 remove POINTER_PLUS_EXPR's base part directly, rather than through
34822 2017-04-10 Richard Biener <rguenther@suse.de>
34823 Bin Cheng <bin.cheng@arm.com>
34825 PR tree-optimization/80153
34826 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34827 the first element of pointer type aff_tree. Build result expr in
34829 (add_elt_to_tree): Convert to type unconditionally. Remove other
34830 fold_convert calls.
34831 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34832 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34834 2017-04-10 Richard Biener <rguenther@suse.de>
34836 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34839 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34841 PR rtl-optimization/70478
34842 * lra-constraints.c (curr_small_class_check): New.
34843 (update_and_check_small_class_inputs): New.
34844 (process_alt_operands): Update curr_small_class_check. Disfavor
34845 alternative insn memory operands. Check available regs for small
34848 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34851 * config/mips/mips.opt (-mvirt): Update description.
34852 * doc/invoke.texi (-mvirt): Likewise.
34854 2017-04-10 Richard Biener <rguenther@suse.de>
34856 PR middle-end/80362
34857 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34858 looking for NEGATE_EXPR in -A / -B to A / B folding.
34860 2017-04-10 Martin Liska <mliska@suse.cz>
34862 PR gcov-profile/80224
34863 * gcov.c (print_usage): Fix usage string.
34864 (get_gcov_intermediate_filename): Remove.
34865 (output_gcov_file): Use both for normal and intermediate format.
34866 (generate_results): Do not initialize special file for
34867 intermediate format.
34869 2017-04-10 Richard Biener <rguenther@suse.de>
34871 PR tree-optimization/80304
34872 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34875 2017-04-10 Nathan Sidwell <nathan@acm.org>
34878 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34879 (rs6000_init_builtins): Use it.
34881 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34883 * config/arm/arm.md (<mrc>): Add mode to SET source.
34884 (<mrrc>): Likewise.
34886 2017-04-10 Richard Biener <rguenther@suse.de>
34888 PR middle-end/80344
34889 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34891 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34894 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34895 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34896 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34897 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34898 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34899 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34900 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34901 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34902 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34903 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34904 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34905 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34906 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34907 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34908 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34909 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34910 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34911 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34912 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34913 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34914 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34915 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34916 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34918 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34920 PR rtl-optimization/70478
34921 * lra-constraints.c: Reverse the last patch.
34923 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34925 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34926 Add comment for WCHAR_T.
34928 2017-04-08 Martin Liska <mliska@suse.cz>
34931 2017-04-07 Martin Liska <mliska@suse.cz>
34934 * ipa-split.c (split_function): Add function part to a same comdat
34937 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34940 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34942 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34944 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34945 (vec_pairing): ...to this.
34946 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34947 (rs6000_sched_init): Adjust for name change.
34948 (struct rs6000_sched_context): Likewise.
34949 (rs6000_init_sched_context): Likewise.
34950 (rs6000_set_sched_context): Likewise.
34952 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34958 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34960 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34961 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34962 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34964 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34966 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34968 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34970 PR rtl-optimization/70703
34971 * ira-color.c (update_conflict_hard_regno_costs): Use
34972 int64_t instead of HOST_WIDE_INT.
34974 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34976 PR rtl-optimization/70478
34977 * lra-constraints.c (process_alt_operands): Disfavor alternative
34978 insn memory operands.
34980 2017-04-07 Jeff Law <law@redhat.com>
34982 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34983 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34985 2017-04-07 Martin Liska <mliska@suse.cz>
34988 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34989 Show error message instead of an ICE.
34991 2017-04-07 Martin Liska <mliska@suse.cz>
34994 * ipa-split.c (split_function): Add function part to a same comdat
34997 2017-04-07 Richard Biener <rguenther@suse.de>
34999 PR middle-end/80341
35000 * tree.c (get_unwidened): Also handle ! for_type case for
35002 * convert.c (do_narrow): Split out from ...
35003 (convert_to_integer_1): ... here. Do not pass final truncation
35004 type to get_unwidened for TRUNC_DIV_EXPR.
35006 2017-04-07 Richard Biener <rguenther@suse.de>
35008 * tree-affine.c (wide_int_ext_for_comb): Take type rather
35010 (aff_combination_const): Adjust.
35011 (aff_combination_scale): Likewise.
35012 (aff_combination_add_elt): Likewise.
35013 (aff_combination_add_cst): Likewise.
35014 (aff_combination_convert): Likewise.
35015 (add_elt_to_tree): Likewise. Remove unused argument.
35016 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
35018 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
35020 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
35022 * config/arm/arm.c (arm_default_short_enums): Use
35023 ARM_DEFAULT_SHORT_ENUMS.
35024 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
35026 2017-04-06 Jakub Jelinek <jakub@redhat.com>
35029 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
35030 members with redundant out-of-class redeclaration.
35032 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
35035 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
35036 * config/i386/i386.md (*zero_extendsidi2):
35037 Add (?*x,*x) and (?*v,*v) alternatives.
35039 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
35042 * config/i386/i386.c (ix86_expand_builtin)
35043 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
35044 mode from insn data. Convert operands to insn operand mode.
35045 Copy operands that don't satisfy insn predicate to a register.
35047 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
35049 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
35052 2017-04-06 Richard Biener <rguenther@suse.de>
35054 PR tree-optimization/80334
35055 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
35056 preserve alignment of accesses.
35058 2017-04-06 Richard Biener <rguenther@suse.de>
35060 PR tree-optimization/80262
35061 * tree-sra.c (build_ref_for_offset): Preserve address-space
35063 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
35064 Drop useless address-space information on MEM_REF offsets.
35066 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
35068 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
35070 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35072 PR rtl-optimization/70703
35073 * ira-color.c (update_conflict_hard_regno_costs): Use
35074 HOST_WIDE_INT instead of long.
35076 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
35079 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
35080 not defined for x86_64 target. Add -mmmx target option when __SSE2__
35082 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
35083 for x86_64 target. Handle -m3dnowa option.
35085 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35087 PR rtl-optimization/70703
35088 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
35089 (update_conflict_hard_regno_costs): Use long instead of unsigned
35090 arithmetic for cost calculation.
35092 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35093 Bernd Edlinger <bernd.edlinger@hotmail.de>
35096 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
35099 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
35102 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
35103 ptr_mode with Pmode throughout.
35104 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
35105 into probe_stack_range and use DImode.
35107 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
35110 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
35111 call_eh_return is true.
35113 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35115 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
35116 Initialize last_match_fntype_index.
35118 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35121 * tree-nvr.c: Include internal-fn.h.
35122 (pass_return_slot::execute): Ignore internal calls without
35125 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35126 Richard Biener <rguenther@suse.de>
35129 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
35130 captures used multiple times, except for the last use.
35131 * generic-match-head.c: Include gimplify.h.
35133 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35135 PR tree-optimization/79390
35136 * target.h (struct noce_if_info): Declare.
35137 * targhooks.h (default_noce_conversion_profitable_p): Declare.
35138 * target.def (noce_conversion_profitable_p): New target hook.
35139 * ifcvt.h (struct noce_if_info): New type, moved from ...
35140 * ifcvt.c (struct noce_if_info): ... here.
35141 (noce_conversion_profitable_p): Renamed to ...
35142 (default_noce_conversion_profitable_p): ... this. No longer
35144 (noce_try_store_flag_constants, noce_try_addcc,
35145 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
35146 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
35147 instead of noce_conversion_profitable_p.
35148 * config/i386/i386.c: Include ifcvt.h.
35149 (ix86_option_override_internal): Don't override
35150 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
35151 (ix86_noce_conversion_profitable_p): New function.
35152 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
35153 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
35154 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
35155 * doc/tm.texi: Regenerated.
35157 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35159 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
35162 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
35165 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
35166 instructions for small multiply cores.
35168 2017-04-04 Jeff Law <law@redhat.com>
35170 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
35172 (mips_expand_vec_perm_const): Initialize elements in orig_perm
35173 that are not set by the loop over the elements.
35175 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35178 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
35179 int mode, convert_modes it to mode as unsigned, otherwise use
35180 lowpart_subreg to mode rather than SImode.
35181 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
35182 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
35183 Use DImode instead of SImode for the shift count operand.
35184 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
35187 2017-04-04 Richard Biener <rguenther@suse.de>
35189 PR middle-end/80281
35190 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
35191 arithmetic done for the negate or the plus. Simplify.
35192 (A - (-B) -> A + B): Likewise.
35193 * fold-const.c (split_tree): Make sure to not negate pointers.
35195 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
35197 PR rtl-optimization/60818
35198 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
35199 a compare of comparisons with the thing compared if this results
35200 in a different machine mode.
35202 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35204 * alias.c (base_alias_check): Fix typo in comment.
35205 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35206 * cgraphunit.c (symbol_table::compile): Likewise.
35207 * collect2.c (maybe_run_lto_and_relink): Likewise.
35208 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35209 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35210 * config/avr/avr.c (avr_map_op_t): Likewise.
35211 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35212 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35213 * config/epiphany/epiphany.md (movcc): Likewise.
35214 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35215 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35217 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35218 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35219 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35220 * config/sh/sh.c (sh_rtx_costs): Likewise.
35221 * fold-const.c (fold_truth_andor): Likewise.
35222 * genautomata.c (collapse_flag): Likewise.
35223 * gengtype.h (struct type::u::s): Likewise.
35224 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35225 * input.c (FORMAT_AMOUNT): Likewise.
35226 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35227 (known_aggs_to_agg_replacement_list): Likewise.
35228 * ipa-inline-analysis.c: Likewise.
35229 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35230 * ipa-polymorphic-call.c
35231 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35232 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35233 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35235 * modulo-sched.c (apply_reg_moves): Likewise.
35236 * omp-expand.c (build_omp_regions_1): Likewise.
35237 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35238 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35239 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35240 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35241 * value-prof.c: Likewise.
35242 * var-tracking.c (val_reset): Likewise.
35244 2017-04-03 Richard Biener <rguenther@suse.de>
35246 PR tree-optimization/80275
35247 * fold-const.c (split_address_to_core_and_offset): Handle
35250 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35252 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35253 descriptors is at least equal to that of functions.
35255 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35257 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35259 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35262 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35263 (mov<IMOD4:mode>): New expander.
35264 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35266 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35268 PR rtl-optimization/79405
35269 * fwprop.c (propagations_left): New variable.
35270 (forward_propagate_into): Decrement it.
35271 (fwprop_init): Initialize it.
35272 (fw_prop): If the variable has reached zero, stop propagating.
35273 (fwprop_addr): Ditto.
35275 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35278 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35279 a FUNCTION_DECL, pass it as decl instead of origin to
35282 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35284 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35287 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35290 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35291 TARGET_VSX_SMALL_INTEGER.
35293 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35295 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35296 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35298 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35300 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35301 extraction from odd-numbered MSA register.
35303 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35305 PR middle-end/80173
35306 * expmed.c (store_bit_field_1): Don't attempt to create
35307 a word subreg out of hard registers wider than word if they
35308 have HARD_REGNO_NREGS of 1 for their mode.
35310 PR middle-end/80163
35311 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35312 conversions to integer types wider than word and pointer.
35315 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35316 (rtx_equal_for_cselib_p): Pass 0 to it.
35317 * cselib.c (cselib_hasher::equal): Likewise.
35318 (rtx_equal_for_cselib_1): Add depth argument. If depth
35319 is 128, don't look up VALUE locs and punt. Increment
35320 depth in recursive calls when walking VALUE locs.
35322 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35324 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35325 (make_gcov_file_name): Use the canonical path name for generating
35327 (read_line): Fix handling of files with ascii null bytes.
35329 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35331 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35332 to initialise a vector register instead
35333 of using a const_int.
35335 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35337 PR translation/80189
35338 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35339 diagnostic messages.
35341 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35344 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35345 (dfp_diex_<mode>): Update mode of operand 1.
35346 * doc/extend.texi (dxex, dxexq): Document change to return type.
35347 (diex, diexq): Document change to argument type.
35349 2017-03-30 Martin Jambor <mjambor@suse.cz>
35352 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35353 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35354 it reflects the signature changes performed at the callee side.
35355 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35356 to cgraph_build_function_type_skip_args.
35357 (build_function_decl_skip_args): Adjust call to the above function.
35359 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35362 * config/i386/sse.md
35363 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35364 register as dest whenever it is a MEM not rtx_equal_p to the
35365 corresponding dup operand, and when forcing into reg move the
35366 reg into the memory afterwards.
35367 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35368 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35369 for the force_reg mode.
35370 (avx512vl_vextractf128<mode>): Use register as dest either
35371 always when a MEM, or when it is a MEM not rtx_equal_p to the
35372 corresponding dup operand, or even not when it is a CONST_VECTOR
35373 depending on the mode and lo vs. hi.
35374 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35376 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35377 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35378 Likewise. Require that operands[2] is even.
35379 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35380 Remove extraneous parens. Require that operands[2] is a multiple
35382 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35383 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35384 disallow memory then.
35386 2017-03-30 Richard Biener <rguenther@suse.de>
35388 PR tree-optimization/77498
35389 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35390 to non-constants over backedges.
35392 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35394 PR rtl-optimization/80233
35395 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35396 as last_combined_insn. Do not test for BARRIER_P separately.
35398 2017-03-29 Andreas Schwab <schwab@suse.de>
35401 * calls.c (prepare_call_address): Convert funexp to Pmode before
35402 copying to temp reg.
35404 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35406 PR tree-optimization/80158
35407 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35408 Handle possible future case of more than one alternate
35410 (replace_rhs_if_not_dup): Likewise.
35411 (replace_one_candidate): Likewise.
35413 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35415 PR rtl-optimization/80193
35416 * ira.c (ira): Do not check allocation for LRA.
35418 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35420 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35421 (nvptx_output_simt_exit): Declare.
35422 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35423 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35424 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35425 (nvptx_declare_function_name): ...here. Emit declaration of local
35426 memory space buffer for omp_simt_enter insn.
35427 (nvptx_output_unisimt_switch): New.
35428 (nvptx_output_softstack_switch): New.
35429 (nvptx_output_simt_enter): New.
35430 (nvptx_output_simt_exit): New.
35431 * config/nvptx/nvptx.h (struct machine_function): New fields
35432 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35433 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35434 (UNSPECV_SIMT_EXIT): Ditto.
35435 (omp_simt_enter_insn): New insn.
35436 (omp_simt_enter): New expansion.
35437 (omp_simt_exit): New insn.
35438 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35440 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35441 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35442 (expand_GOMP_SIMT_EXIT): New.
35443 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35444 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35445 (GOMP_SIMT_EXIT): Ditto.
35446 * target-insns.def (omp_simt_enter): New insn.
35447 (omp_simt_exit): Ditto.
35448 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35450 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35451 (lower_rec_input_clauses): Likewise.
35452 (lower_lastprivate_clauses): Handle SIMT privatization.
35454 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35455 (ompdevlow_adjust_simt_enter): New.
35456 (find_simtpriv_var_op): New.
35457 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35458 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35460 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35461 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35462 (copy_decl_for_dup_finish): Ditto.
35464 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35466 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35469 * config/i386/i386.c (ix86_option_override_internal): Always
35470 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35472 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35474 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35476 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35478 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35479 mark new edge's irreducible flag accordign to it.
35480 (vect_do_peeling): Check loop preheader edge's irreducible flag
35481 and pass it to function slpeel_add_loop_guard.
35483 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35485 PR tree-optimization/80218
35486 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35487 Update block frequencies and counts.
35489 2017-03-28 Richard Biener <rguenther@suse.de>
35491 PR tree-optimization/78644
35492 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35493 of a simplification result we may not use it at all.
35495 2017-03-28 Richard Biener <rguenther@suse.de>
35498 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35499 without arguments, generate default definition of a SSA name.
35501 2017-03-28 Richard Biener <rguenther@suse.de>
35503 PR middle-end/80222
35504 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35505 TYPE_REF_CAN_ALIAS_ALL references.
35506 * fold-const.c (fold_indirect_ref_1): Likewise.
35508 2017-03-28 Martin Liska <mliska@suse.cz>
35511 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35512 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35514 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35515 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35517 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35518 (EXTRA_SPECS): Define.
35519 (SUBTARGET_EXTRA_SPECS): Likewise.
35520 (SUBTARGET_CPP_SPEC): Likewise.
35521 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35522 SUBTARGET_EXTRA_SPECS.
35523 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35525 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35527 * config/arc/simdext.md (vst64_insn): Update pattern.
35528 (vld32wh_insn): Likewise.
35529 (vld32wl_insn): Likewise.
35530 (vld64_insn): Likewise.
35531 (vld32_insn): Likewise.
35533 2017-03-28 Marek Polacek <polacek@redhat.com>
35536 * fold-const.c (fold_comparison): Use protected_set_expr_location
35537 instead of SET_EXPR_LOCATION.
35539 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35541 * tree.c (add_expr): Avoid name lookup warning.
35543 2017-03-27 Jeff Law <law@redhat.com>
35545 PR tree-optimization/80216
35546 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35547 function name. Limit recursion depth.
35548 (record_temporary_equivalences): Corresponding changes.
35550 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35552 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35555 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35558 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35560 * cfgcleanup.c (reg_note_cfa_p): New array.
35561 (insns_have_identical_cfa_notes): New function.
35562 (old_insns_match_p): Don't cross-jump in between /f
35563 and non-/f instructions. If both i1 and i2 are frame related,
35564 verify all CFA notes, their order and content.
35566 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35569 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35570 HImode and SImode with zero extend to DImode to one insn.
35571 (bswap<mode>2_extenddi): Likewise.
35572 (bswapsi2_extenddi): Likewise.
35573 (bswaphi2_extendsi): Likewise.
35574 (bswaphi2): Combine bswap HImode and SImode into one insn.
35575 Separate memory insns from swapping register.
35576 (bswapsi2): Likewise.
35577 (bswap<mode>2): Likewise.
35578 (bswaphi2_internal): Delete, no longer used.
35579 (bswapsi2_internal): Likewise.
35580 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35581 store, and gpr<-gpr swap insns.
35582 (bswap<mode>2_store): Likewise.
35583 (bswaphi2_reg): Register only splitter, combine with the splitter.
35584 (bswaphi2 splitter): Likewise.
35585 (bswapsi2_reg): Likewise.
35586 (bswapsi2 splitter): Likewise.
35587 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35588 the insns into load, store, and register/register insns.
35589 (bswapdi2_ldbrx): Likewise.
35590 (bswapdi2_load): Likewise.
35591 (bswapdi2_store): Likewise.
35592 (bswapdi2_reg): Likewise.
35594 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35596 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35597 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35599 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35602 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35604 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35605 special handling for target option conflicts between dform
35606 options (-mpower9-dform, -mpower9-dform-vector,
35607 -mpower9-dform-scalar) and -mno-direct-move.
35609 2017-03-27 Richard Biener <rguenther@suse.de>
35611 PR tree-optimization/80181
35612 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35614 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35616 * config/arc/predicates.md (move_double_src_operand): Replace the
35617 call to move_double_src_operand with a call to address_operand.
35619 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35621 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35622 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35623 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35625 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35627 * config/arc/predicates.md (long_immediate_loadstore_operand):
35628 Consider scaled addresses cases.
35630 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35632 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35633 restored when in interrupt.
35634 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35635 doesn't have delay slot.
35637 2017-03-27 Richard Biener <rguenther@suse.de>
35640 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35641 inlined thunk clones.
35643 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35646 * asan.c (instrument_derefs): Copy over last operand from
35647 original COMPONENT_REF to the new COMPONENT_REF with
35648 DECL_BIT_FIELD_REPRESENTATIVE.
35649 * ubsan.c (instrument_object_size): Likewise.
35651 2017-03-27 Richard Biener <rguenther@suse.de>
35653 PR tree-optimization/80170
35654 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35655 sure DR/SCEV didnt fold in constants we do not see when looking
35656 at the reference base alignment.
35658 2017-03-27 Richard Biener <rguenther@suse.de>
35660 PR middle-end/80171
35661 * gimple-fold.c (fold_ctor_reference): Properly guard against
35662 NULL return value from canonicalize_constructor_val.
35664 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35667 * config/i386/i386.c (ix86_expand_builtin)
35668 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35669 flags reg setting and flags reg using instructions.
35670 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35671 clobbering instructions to zero extend op2.
35673 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35675 * doc/install.texi (Configuration) <--with-aix-soname>:
35676 Update link to AIX ld.
35678 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35680 PR rtl-optimization/80160
35681 PR rtl-optimization/80159
35682 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35683 reg_alternate_class into account.
35685 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35688 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35689 to consider in curr_insn_transform.
35691 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35693 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35694 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35695 and emit_mode_inner.
35697 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35699 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35700 argument to the overloaded builtin variants. Use the new flag to
35701 deprecate certain builtin variants.
35702 * config/s390/s390-builtin-types.def: Add new builtin types.
35703 * config/s390/s390-builtins.h: Support new flags field for
35704 overloaded builtins.
35705 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35706 (s390_macro_to_expand): Enable vector float data type.
35707 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35708 builtins by incrementing the __VEC__ version number.
35709 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35711 (s390_resolve_overloaded_builtin): Emit error messages depending
35712 on the builtin flags.
35713 * config/s390/s390.c (s390_expand_builtin): Support additional
35714 flags argument. Change error message to match the messages
35715 emitted in s390-c.c.
35716 * config/s390/s390.md: New UNSPEC_* constants.
35717 (op_type): Add new instruction types.
35718 * config/s390/vecintrin.h: Add new builtins and test data class
35720 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35721 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35722 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35723 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35724 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35725 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35727 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35728 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35729 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35730 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35732 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35733 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35734 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35735 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35737 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35738 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35739 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35740 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35741 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35742 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35743 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35745 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35746 ("vec_scatter_element<V_HW_4:mode>_DI")
35747 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35748 ("vec_fpint<mode>", "vflls")
35749 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35750 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35751 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35752 ("*vec_cmphe<mode>_cc"): ... these.
35754 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35755 mode constant instead of magic value.
35757 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35759 * config/s390/s390.c (s390_expand_vec_compare): Support other
35760 vector floating point modes than just V2DF.
35761 (s390_expand_vcond): Likewise.
35762 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35763 (s390_cannot_change_mode_class): Prevent mode changes between TF
35764 and V1TF in vector registers.
35765 * config/s390/s390.md (DF, SF): New mode attributes.
35766 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35767 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35768 SFmode support for VRs.
35769 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35771 (VFT, VF_HW): New mode iterators.
35772 (vw, sdx): New mode attributes.
35773 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35774 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35775 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35776 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35777 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35778 also the new vector floating point modes. Renaming to ...
35780 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35781 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35782 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35783 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35784 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35785 ("vec_unordered<mode>"): ... these.
35787 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35788 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35789 ("*vec_extendv2df"): New insn definitions.
35791 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35793 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35794 ("mulditi3_2", "*muldi3_sign"): New patterns.
35795 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35796 rename the pattern definition.
35798 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35800 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35802 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35804 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35806 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35807 instruction if possible.
35808 * config/s390/vector.md (vec_halfnumelts): New mode
35810 ("*vec_vllezlf<mode>"): New pattern.
35812 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35814 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35815 ("popcountv4si2", "popcountv2di2"): Rename to ...
35816 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35817 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35819 ("popcount<mode>2_vxe"): New pattern.
35821 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35823 * common/config/s390/s390-common.c (processor_flags_table): Add
35825 * config.gcc: Add arch12.
35826 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35827 Default to arch12 for unknown CPU model numbers.
35828 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35829 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35830 PROCESSOR_max sanity check.
35831 * config/s390/s390-opts.h (enum processor_type): Add
35833 * config/s390/s390.c (processor_table): Add arch12.
35834 (s390_expand_builtin): Add check for B_VXE flag.
35835 (s390_issue_rate): Add PROCESSOR_ARCH12.
35836 (s390_get_sched_attrmask): Likewise.
35837 (s390_get_unit_mask): Likewise.
35838 (s390_sched_score): Enable z13 scheduling for arch12.
35839 (s390_sched_reorder): Likewise.
35840 (s390_sched_variable_issue): Likewise.
35841 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35843 (s390_tune_attr): Use z13 scheduling also for arch12.
35844 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35845 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35846 (TARGET_VXE_P): New macros.
35847 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35848 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35849 * config/s390/s390.opt: Add arch12 as processor_type.
35851 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35853 * config/s390/s390.md
35854 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35855 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35856 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35858 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35859 Rename expanders to ...
35861 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35862 ("fixuns_truncdddi2_emu"): ... these.
35864 ("fixuns_trunc<mode>si2_emu"): New expander.
35866 ("*fixuns_truncdfdi2_z13"): Rename to ...
35867 ("*fixuns_truncdfdi2_vx"): ... this.
35869 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35871 * config/s390/2964.md: Remove the single element vector compare
35872 instructions which are no longer used.
35873 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35875 (s390_canonicalize_comparison): Remove handling of DFmode
35877 (s390_expand_vec_compare_scalar): Remove function.
35878 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35879 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35881 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35883 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35885 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35887 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35888 will anyway by matched by mov<mode>_64dfp.
35890 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35892 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35893 vlef/vstef. Add missing operand to vleif.
35895 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35897 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35898 pair for all vector types with 64 bit elements.
35899 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35900 * config/s390/vector.md (V_HW_64): ... here.
35901 (V_128_NOSINGLE): New mode iterator.
35902 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35903 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35904 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35905 ("*vec_load_pairv2di"): Change to ...
35906 ("*vec_load_pair<mode>"): ... this one.
35908 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35910 * config/s390/constraints.md: Add comments.
35911 (jKK): Reject element sizes > 8 bytes.
35912 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35914 * config/s390/s390.md: Add the s_operand checks formerly in
35915 s390_split_ok_p to various splitters where they are still
35917 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35918 for 128 bit vectors. Plus two splitters.
35920 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35922 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35925 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35928 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35929 error if the boundary argument is not constant.
35931 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35933 PR rtl-optimization/80112
35934 * loop-doloop.c (doloop_condition_get): Don't check condition
35935 if cmp isn't SET with IF_THEN_ELSE src.
35937 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35939 PR tree-optimization/80158
35940 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35941 replacing a candidate statement, also replace it for the
35942 candidate's alternate interpretation.
35943 (replace_rhs_if_not_dup): Likewise.
35944 (replace_one_candidate): Likewise.
35946 2017-03-24 Richard Biener <rguenther@suse.de>
35948 PR tree-optimization/80167
35949 * graphite-isl-ast-to-gimple.c
35950 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35952 (translate_isl_ast_to_gimple::get_rename): Likewise.
35954 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35956 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35957 handling of certain combinations of target options, including the
35958 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35959 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35961 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35964 * config/arm/arm.md (*load_multiple): Add reload_completed to
35965 matching condition.
35967 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35968 Richard Biener <rguenth@suse.de>
35970 PR tree-optimization/79908
35971 PR tree-optimization/80136
35972 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35973 been cast away, gimplify_and_add suffices.
35975 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35977 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35979 2017-03-23 Richard Biener <rguenther@suse.de>
35981 PR tree-optimization/80032
35982 * gimplify.c (gimple_push_cleanup): Forced unconditional
35983 cleanups still have to go to the conditional_cleanups
35986 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35988 PR tree-optimization/80072
35989 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35991 (next_operand_entry_id): Change type to unsigned int.
35992 (sort_by_operand_rank): Make sure to return the right return value
35993 even if unsigned fields are bigger than INT_MAX.
35994 (struct oecount): Change cnt and id type to unsigned int.
35995 (oecount_hasher::equal): Formatting fix.
35996 (oecount_cmp): Make sure to return the right return value
35997 even if unsigned fields are bigger than INT_MAX.
35998 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
36001 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
36002 TREE_READONLY on result if writing it more than once.
36005 * doc/invoke.texi (-fsanitize=thread): Document that with
36006 -fnon-call-exceptions atomics are not able to throw
36010 * tsan.c: Include tree-eh.h.
36011 (instrument_builtin_call): Call maybe_clean_eh_stmt or
36012 maybe_clean_or_replace_eh_stmt where needed.
36013 (instrument_memory_accesses): Add cfg_changed argument.
36014 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
36015 if it returned true.
36016 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
36018 PR rtl-optimization/63191
36019 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
36020 wrapper function, moved the whole old content into ...
36021 (ix86_delegitimize_address_1): ... this. New inline function.
36022 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
36023 true as last argument instead of ix86_delegitimize_address.
36025 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
36027 * config/aarch64/aarch64.c (generic_branch_cost): Copy
36028 cortexa57_branch_cost.
36030 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
36032 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
36034 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36037 * doc/md.texi (Constraints): Document wA constraint.
36038 * config/rs6000/constraints.md (wA): New.
36039 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
36040 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
36041 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
36042 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
36044 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
36047 * gimplify.c (is_oacc_declared): New function.
36048 (oacc_default_clause): Use it to set default flags for acc declared
36049 variables inside parallel regions.
36050 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
36051 declared variables.
36052 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
36053 declare attribute to any decl as necessary.
36055 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
36058 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
36059 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
36060 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
36061 (arm_arch_lpae): This.
36062 * config/arm/arm.c (arm_arch7ve): Rename into ...
36063 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
36064 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
36067 2017-03-22 Martin Liska <mliska@suse.cz>
36070 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
36071 error message instead of an ICE.
36073 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36075 * doc/extend.texi (6.11 Additional Floating Types): Revise.
36077 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
36079 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36081 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36084 2017-03-21 Martin Sebor <msebor@redhat.com>
36086 * doc/extend.texi: Use "cannot" instead of "can't."
36087 * doc/hostconfig.texi: Same.
36088 * doc/install.texi: Same.
36089 * doc/invoke.texi: Same.
36090 * doc/loop.texi: Same.
36091 * doc/md.texi: Same.
36092 * doc/objc.texi: Same.
36093 * doc/rtl.texi: Same.
36094 * doc/tm.texi: Same.
36095 * doc/tm.texi.in: Same.
36096 * doc/trouble.texi: Same.
36098 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
36101 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
36102 (collect_checksum_attributes): Set it.
36103 (die_checksum_ordered): Use it.
36105 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36107 PR tree-optimization/79908
36108 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
36109 change: For a VA_ARG whose LHS has been cast away, use
36110 force_gimple_operand to construct the side effects.
36112 2017-03-21 David Malcolm <dmalcolm@redhat.com>
36114 PR translation/80001
36115 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
36116 more amenable to translation.
36117 (oacc_loop_auto_partitions): Likewise.
36119 2017-03-21 Marek Polacek <polacek@redhat.com>
36120 Martin Sebor <msebor@redhat.com>
36122 PR tree-optimization/80109
36123 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
36124 on INTEGRAL_TYPE_P.
36126 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36127 Segher Boessenkool <segher@kernel.crashing.org>
36130 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
36131 check reg_used_between_p between insn and one of succ or succ2
36132 depending on if succ is artificial insn not inserted into insn
36135 2017-03-21 Martin Liska <mliska@suse.cz>
36137 PR gcov-profile/80081
36138 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
36139 * doc/gcc.texi: Include gcov-dump stuff.
36140 * doc/gcov-dump.texi: New file.
36142 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
36144 PR rtl-optimization/79150
36145 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
36146 conditional jump, if the jump is the last insn of the loop.
36148 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36149 Richard Biener <rguenth@suse.de>
36151 PR tree-optimization/79908
36152 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36153 been cast away, use force_gimple_operand to construct the side
36156 2017-03-21 Martin Liska <mliska@suse.cz>
36158 PR libfortran/79956
36159 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
36162 2017-03-21 Brad Spengler <spender@grsecurity.net>
36165 * plugin.c (htab_hash_plugin): New function.
36166 (add_new_plugin): Use it and adjust.
36167 (parse_plugin_arg_opt): Adjust.
36168 (init_one_plugin): Likewise.
36170 2017-03-21 Richard Biener <rguenther@suse.de>
36172 PR tree-optimization/80032
36173 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
36174 if set force the cleanup to happen unconditionally.
36175 (gimplify_target_expr): Push inserted clobbers with force_uncond
36176 to avoid them being removed by control-dependent DCE.
36178 2017-03-21 Richard Biener <rguenther@suse.de>
36180 PR tree-optimization/80122
36181 * tree-inline.c (copy_bb): Do not expans va-arg packs or
36182 va_arg_pack_len when the inlined call stmt requires pack
36184 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
36186 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36189 * tsan.c (instrument_builtin_call): If the memory model argument
36190 is not a constant, assume it is valid.
36193 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
36196 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36198 PR rtl-optimization/79910
36199 * combine.c (can_combine_p): Do not allow combining an I0 or I1
36200 if its dest is used by an insn before I2 (other than the combined
36201 insns themselves, which are properly handled already).
36203 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36206 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36208 * combine.c (record_used_regs): New static function.
36209 (try_combine): Handle situations where there is an additional
36210 instruction between I2 and I3 which needs to have a LOG_LINK
36214 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36216 * combine.c (try_combine): Delete redundant i1 test. Call
36217 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36219 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36222 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36223 alternatives 13/14.
36225 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36227 PR tree-optimization/80054
36228 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36229 the optimization if a PHI or any of its arguments is not dominated
36230 by the candidate's basis. Use gphi* rather than gimple* as
36232 (replace_profitable_candidates): Clean up a gimple* variable that
36233 should be a gphi* variable.
36235 2017-03-20 Martin Sebor <msebor@redhat.com>
36238 * doc/extend.texi (attribute constructor): Document present limitation.
36240 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36243 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36244 __POWER9_VECTOR__ #ifdef control, change template definition to
36245 use Power9-specific built-in function.
36246 (vec_any_eq): Likewise.
36247 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36248 to control outcomes from this test.
36249 (vector_ae_<mode>p): For VEC_F modes, likewise.
36251 2017-03-20 Ian Lance Taylor <iant@google.com>
36253 * config/i386/i386.c (ix86_function_regparm): Save an extra
36254 register for -fsplit-stack with DECL_STATIC_CHAIN.
36256 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36259 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36260 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36262 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36264 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36266 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36269 2017-03-20 Marek Polacek <polacek@redhat.com>
36272 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36274 2017-03-20 Richard Biener <rguenther@suse.de>
36276 PR tree-optimization/80113
36277 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36278 allocate extra SSA name for PHI def.
36279 (add_close_phis_to_outer_loops): Likewise.
36280 (add_close_phis_to_merge_points): Likewise.
36281 (copy_loop_close_phi_args): Likewise.
36282 (copy_cond_phi_nodes): Likewise.
36284 2017-03-20 Martin Liska <mliska@suse.cz>
36286 PR middle-end/79753
36287 * tree-chkp.c (chkp_build_returned_bound): Do not build
36288 returned bounds for a LHS that's not a BOUNDED_P type.
36290 2017-03-20 Martin Liska <mliska@suse.cz>
36294 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36295 COMPLEX_CST and VECTOR_CST.
36297 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36300 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36301 target operand. A new splitter adds the clobber statement in case
36302 the target operand is dead anyway.
36304 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36306 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36307 to age-old versions of binutils and glibc.
36309 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36311 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36313 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36315 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36317 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36319 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36320 requirement for binutils 2.13.
36322 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36324 * combine.c (try_combine): Delete redundant i1 test. Call
36325 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36327 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36329 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36330 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36332 <riscv64-*-elf>: Re-arrange section
36333 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36334 <riscv32-*-linux>: Likewise.
36335 <riscv64-*-elf>: Likewise
36336 <riscv64-*-linux>: Likewise.
36338 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36341 * aarch64.opt(verbose-cost-dump): Fix typo.
36343 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36346 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36347 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36349 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36351 * reload.c (find_reloads): When reloading a nonoffsettable address,
36352 use RELOAD_OTHER for it and its address reloads.
36354 PR rtl-optimization/79910
36355 * combine.c (record_used_regs): New static function.
36356 (try_combine): Handle situations where there is an additional
36357 instruction between I2 and I3 which needs to have a LOG_LINK
36360 2017-03-17 Jeff Law <law@redhat.com>
36362 PR tree-optimization/71437
36363 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36364 conditional in the hash table first.
36365 (vrp_dom_walker::before_dom_children): Extract condition from
36366 ASSERT_EXPR. Record condition, its inverion and any implied
36367 conditions as well.
36369 2017-03-17 Marek Polacek <polacek@redhat.com>
36370 Markus Trippelsdorf <markus@trippelsdorf.de>
36372 PR tree-optimization/80079
36373 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36376 2017-03-17 Richard Biener <rguenther@suse.de>
36378 PR middle-end/80075
36379 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36380 Properly verify the LHS before the RHS possibly claims to be
36382 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36385 2017-03-17 Martin Jambor <mjambor@suse.cz>
36387 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36388 (List of -O2 options): Likewise.
36389 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36392 2017-03-17 Tom de Vries <tom@codesourcery.com>
36394 * gcov-dump.c (print_usage): Print bug_report_url.
36396 2017-03-17 Richard Biener <rguenther@suse.de>
36398 PR middle-end/80050
36399 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36400 (parser::peek): Likewise.
36402 2017-03-17 Richard Biener <rguenther@suse.de>
36404 PR tree-optimization/80048
36405 * sese.c (free_sese_info): Properly release rename_map and
36406 copied_bb_map elements.
36408 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36410 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36411 Add linked-list forward and backlinks. Insert on
36412 construction, remove on destruction.
36413 (class pass_store_merging): Add m_stores_head field.
36414 (pass_store_merging::terminate_and_process_all_chains):
36415 Iterate over m_stores_head list.
36416 (pass_store_merging::terminate_all_aliasing_chains):
36418 (pass_store_merging::execute): Check for debug stmts first.
36419 Push new chains onto the m_stores_head stack.
36421 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36424 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36425 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36426 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36428 2017-03-16 Jeff Law <law@redhat.com>
36430 PR tree-optimization/71437
36431 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36432 member function. Implementation moved into after_dom_children
36433 member function and into the threader's thread_outgoing_edges
36435 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36436 some code into new thread_outgoing_edges.
36437 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36438 definition. Simplify marker handling (do it here). Assume we always
36439 have the available expression and the const/copies tables.
36440 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36442 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36443 * tree-vrp.c (equiv_stack): No longer file scoped.
36444 (vrp_dom_walker): New class.
36445 (vrp_dom_walker::before_dom_children): New member function.
36446 (vrp_dom_walker::after_dom_children): Likewise.
36447 (identify_jump_threads): Setup domwalker. Use it rather than
36448 walking edges in a random order by hand. Simplify setup/finalization.
36449 (finalize_jump_threads): Remove.
36450 (vrp_finalize): Do not call identify_jump_threads here.
36451 (execute_vrp): Do it here instead and call thread_through_all_blocks
36454 PR tree-optimization/71437
36455 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36457 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36459 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36460 (dom_opt_dom_walker::thread_across_edge): Remove
36461 handle_dominating_asserts argument. All callers changed.
36462 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36463 changes. Remove calls to lhs_of_dominating_assert. Other
36464 uses of handle_dominating_asserts turn into unconditional code
36465 (simplify_control_stmt_condition_1): Likewise.
36466 (simplify_control_stmt_condition): Likewise.
36467 (thread_through_normal_block, thread_across_edge): Likewise.
36468 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36469 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36470 object if it is not an SSA_NAME.
36471 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36472 before calling into the VRP specific simplifiers.
36473 (identify_jump_threads): Remove handle_dominating_asserts
36476 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36479 * tree-diagnostic.c (default_tree_printer): No longer static.
36480 * tree-diagnostic.h (default_tree_printer): New prototype.
36482 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36484 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36485 Change ins into fmov.
36487 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36489 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36490 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36491 Use h_con constraint for operand 1.
36492 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36493 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36495 2017-03-15 Jeff Law <law@redhat.com>
36497 PR tree-optimization/71437
36498 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36499 (record_temporary_equivalences): Use it.
36501 PR tree-optimization/71437
36502 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36503 tree-ssa-scopedtables.
36504 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36505 (record_conditions, record_cond, vuse_eq): Likewise.
36506 (record_edge_info): Adjust to API tweak of record_conditions.
36507 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36508 (record_temporary_equivalences, optimize_stmt): Likewise.
36509 (eliminate_redundant_computations): Likewise.
36510 (record_equivalences_from_stmt): Likewise.
36511 * tree-ssa-scopedtables.c: Include options.h and params.h.
36512 (vuse_eq): New function, moved from tree-ssa-dom.c
36513 (build_and_record_new_cond): Likewise.
36514 (record_conditions): Likewise. Accept vector of conditions rather
36515 than edge_equivalence structure for first argument.
36516 for the first argument.
36517 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36518 from tree-ssa-dom.c.
36519 (avail_exprs_stack::record_cond): Likewise.
36520 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36521 from tree-ssa-dom.c.
36522 (avail_exprs_stack): Add new member functions lookup_avail_expr
36524 (record_conditions): Declare.
36526 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36529 * lra-constraints.c (process_alt_operands): Increase reject for
36530 reloading an input/output operand.
36532 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36535 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36536 insns to convert from signed/unsigned char/short to IEEE 128-bit
36538 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36540 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36543 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36544 subreg of inner mode for values already in registers.
36546 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36548 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36549 iteration reg is used after the loop.
36551 2017-03-14 Martin Sebor <msebor@redhat.com>
36553 PR tree-optimization/79800
36554 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36555 precision in negative-positive range.
36556 (format_floating): Call non-const overload with adjusted precision.
36558 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36561 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36564 2017-03-14 Martin Sebor <msebor@redhat.com>
36566 PR middle-end/80020
36567 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36568 * builtins.def (aligned_alloc): Use it.
36571 * Makefile.in (GTFILES): Add calls.c.
36572 * calls.c: Include "gt-calls.h".
36574 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36576 PR rtl-optimization/79728
36577 * regs.h (struct target_regs): New field
36578 x_contains_allocatable_regs_of_mode.
36579 (contains_allocatable_regs_of_mode): New macro.
36580 * reginfo.c (init_reg_sets_1): Initialize it, and change
36581 contains_reg_of_mode so it includes global regs as well.
36582 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36583 rather than contains_regs_of_mode.
36585 2017-03-14 Martin Liska <mliska@suse.cz>
36587 * doc/invoke.texi: Document options that can't be combined with
36588 -fcheck-pointer-bounds.
36590 2017-03-14 Martin Liska <mliska@suse.cz>
36592 PR middle-end/79831
36593 * doc/invoke.texi (-Wchkp): Document the option.
36595 2017-03-14 Martin Liska <mliska@suse.cz>
36597 * Makefile.in: Install gcov-dump.
36599 2017-03-14 Martin Liska <mliska@suse.cz>
36601 * multiple_target.c (expand_target_clones): Bail out for
36602 an invalid attribute.
36604 2017-03-14 Richard Biener <rguenther@suse.de>
36606 * alias.c (struct alias_set_entry): Pack properly.
36607 * cfgloop.h (struct loop): Likewise.
36608 * cse.c (struct set): Likewise.
36609 * ipa-utils.c (struct searchc_env): Likewise.
36610 * loop-invariant.c (struct invariant): Likewise.
36611 * lra-remat.c (struct cand): Likewise.
36612 * recog.c (struct change_t): Likewise.
36613 * rtl.h (struct address_info): Likewise.
36614 * symbol-summary.h (function_summary): Likewise.
36615 * tree-loop-distribution.c (struct partition): Likewise.
36616 * tree-object-size.c (struct object_size_info): Likewise.
36617 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36618 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36619 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36620 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36621 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36622 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36623 (struct _stmt_vec_info): Likewise.
36625 2017-03-14 Martin Liska <mliska@suse.cz>
36628 * multiple_target.c (create_dispatcher_calls): Check that
36629 a target can create a function dispatcher.
36631 2017-03-14 Martin Liska <mliska@suse.cz>
36634 * multiple_target.c (expand_target_clones): Drop local.local
36635 flag for default implementation.
36637 2017-03-14 Richard Biener <rguenther@suse.de>
36639 PR tree-optimization/80030
36640 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36642 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36644 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36645 gcc_fallthrough() instead of __attribute__((fallthrough));
36647 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36649 * doc/gcc.texi: Remove "up" link to (DIR).
36650 * doc/gccint.texi: Ditto.
36652 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36654 * doc/install.texi (Specific) <avr>: Remove reference to
36657 2017-03-13 Jeff Law <law@redhat.com>
36659 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36660 attribute rather than comments.
36662 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36663 match_scratch operand is highest.
36665 2017-03-13 Martin Liska <mliska@suse.cz>
36667 PR middle-end/78339
36668 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36669 is a CHKP clone, use original declaration.
36671 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36673 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36674 (arc_conditional_register_usage): Use a different allocation order
36675 when optimizing for size.
36676 * common/config/arc/arc-common.c (arc_option_optimization_table):
36677 Section anchors default on when optimizing for size.
36679 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36681 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36683 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36685 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36686 * config/arc/arc.md (cpu_facility): Add cd variant.
36687 (*movqi_insn): Add code density variant.
36688 (*movhi_insn): Likewise.
36689 (*movqi_insn): Likewise.
36690 (*addsi3_mixed): Likewise.
36691 (subsi3_insn): Likewise.
36693 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36695 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36697 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36699 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36700 expressions with MINUS and UNARY ops.
36702 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36705 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36707 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36708 between vec_select and vector argument.
36709 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36710 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36711 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36712 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36713 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36714 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36716 2017-03-13 Richard Biener <rguenther@suse.de>
36719 * params.def (vect-max-peeling-for-alignment): Fix typo.
36721 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36723 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36724 issue that only occurred with binutils below 2.18.
36726 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36728 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36729 refer to binutils 2.11/2.12 minimum.
36731 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36733 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36734 ftp.kernel.org and simplify binutils requirement.
36736 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36738 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36740 (Optimize Options): Ditto. Also remove redundancy.
36742 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36744 PR translation/79848
36745 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36747 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36748 to G_ to avoid double translation.
36750 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36752 PR translation/79923
36753 * auto-profile.c (get_combined_location): Convert leading
36754 character of diagnostics to lower case and remove trailing period.
36755 (read_profile): Likewise for various diagnostics.
36756 * config/arm/arm.c (arm_option_override): Remove trailing period
36757 from various diagnostics.
36758 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36759 (msp430_expand_delay_cycles): Likewise.
36761 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36764 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36765 full command-line argument, rather than just "str".
36766 (aarch64_validate_march): Likewise.
36767 (aarch64_validate_mtune): Likewise.
36769 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36771 PR rtl-optimization/78911
36772 * lra-assigns.c (must_not_spill_p): New function.
36773 (spill_for): Use it.
36775 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36777 PR tree-optimization/79981
36778 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36779 ATOMIC_COMPARE_EXCHANGE ifn result.
36780 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36781 IFN_ATOMIC_COMPARE_EXCHANGE.
36783 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36786 * opts.c (parse_sanitizer_options): Add missing question mark to
36787 "did you mean" message.
36789 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36791 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36793 (VMULEUH_UNS): Likewise.
36794 (VMULOUB_UNS): Likewise.
36795 (VMULOUH_UNS): Likewise.
36796 * config/rs6000/rs6000.c (builtin_function_type): Remove
36797 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36799 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36802 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36803 x with result of extra_parsing_for_operand_code_0.
36804 (function_reader::extra_parsing_for_operand_code_0): Convert
36805 return type from void to rtx, returning x. When reading
36806 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36807 larger size containing struct block_symbol.
36809 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36811 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36812 -mfloat128-hardware without -m64.
36814 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36817 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36818 entries to the case statement that marks unsigned arguments to
36819 overloaded functions.
36821 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36823 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36824 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36826 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36829 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36830 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36832 2017-03-10 Martin Liska <mliska@suse.cz>
36836 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36837 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36840 2017-03-10 Olivier Hainque <hainque@adacore.com>
36842 * tree-switch-conversion (array_value_type): Start by resetting
36843 candidate type to it's main variant.
36845 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36847 PR rtl-optimization/79909
36848 * combine.c (try_combine): Use simplify_replace_rtx on individual
36849 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36850 of the whole CALL_INSN_FUNCTION_USAGE.
36852 PR tree-optimization/79972
36853 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36854 get_range_info on SSA_NAMEs. Formatting fixes.
36856 2017-03-10 Richard Biener <rguenther@suse.de>
36857 Jakub Jelinek <jakub@redhat.com>
36859 PR tree-optimization/77975
36860 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36861 edge to be constant.
36862 (get_val_for): For constant x return it. Formatting fix.
36863 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36864 would use the same bases as the current one.
36866 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36868 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36869 instead of vec_select for V1TImode.
36870 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36872 (VSX_LE_128): Add V1TI to this mode iterator.
36873 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36874 (*vsx_le_perm_store_<mode>): Likewise.
36875 (pre-reload splitter for VSX stores): Likewise.
36876 (post-reload splitter for VSX stores): Likewise.
36877 (*vsx_xxpermdi2_le_<mode>): Likewise.
36878 (*vsx_lxvd2x2_le_<mode>): Likewise.
36879 (*vsx_stxvd2x2_le_<mode>): Likewise.
36881 2017-03-09 Michael Eager <eager@eagercon.com>
36883 Correct failures with --enable-checking=yes,rtl.
36885 * config/microblaze/microblaze.c (microblaze_expand_shift):
36886 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36887 test for const0_rtx.
36888 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36889 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36891 2017-03-09 Richard Biener <rguenther@suse.de>
36893 PR tree-optimization/79977
36894 * graphite-scop-detection.c (scop_detection::merge_sese):
36895 Handle the case of extra exits to blocks dominating the entry.
36897 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36899 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36902 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36904 PR rtl-optimization/79949
36905 * lra-constraints.c (process_alt_operands): Check memory when
36906 trying to predict a cycle. Print about the overall increase.
36908 2017-03-09 Richard Biener <rguenther@suse.de>
36910 PR middle-end/79971
36911 * gimple-expr.c (useless_type_conversion_p): Preserve
36912 TYPE_SATURATING for fixed-point types.
36914 2017-03-09 Richard Biener <rguenther@suse.de>
36917 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36918 alignment of BLKmode params.
36920 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36923 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36924 (VALL_NO_V2Q): Likewise.
36926 * config/aarch64/aarch64-simd.md
36927 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36929 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36930 VALL_NO_V2Q mode iterator.
36931 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36933 2017-03-09 Martin Liska <mliska@suse.cz>
36935 PR tree-optimization/79631
36936 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36937 tree_int_cst_sign_bit just for INTEGER constants.
36939 2017-03-09 Martin Liska <mliska@suse.cz>
36943 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36946 2017-03-09 Marek Polacek <polacek@redhat.com>
36949 * tree.c (inchash::add_expr): Handle TREE_VEC.
36951 2017-03-09 Martin Liska <mliska@suse.cz>
36954 (chkp_narrow_size_and_offset): New function.
36955 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36956 (void chkp_parse_bit_field_ref): New function.
36957 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36958 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36960 2017-03-09 Martin Liska <mliska@suse.cz>
36963 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36964 (chkp_find_bounds_1): Remove gcc_unreachable.
36966 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36969 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36970 BUILT_IN_SYNC*, determine the access type from the size suffix and
36971 always build a MEM_REF with that type. Handle forgotten
36972 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36975 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36976 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36977 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36978 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36979 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36980 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36981 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36982 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36983 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36984 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36985 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36986 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36987 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36988 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36989 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36990 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36991 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36992 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36993 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36994 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36995 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36996 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36997 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36998 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36999 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
37000 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
37001 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
37002 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
37003 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
37004 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
37005 definitions outside of __OPTIMIZE__ guarded section.
37008 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
37009 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
37010 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
37011 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
37014 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37016 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
37017 ("vfenez<mode>"): Add missing constraints.
37019 2017-03-08 Martin Sebor <msebor@redhat.com>
37022 * config/nds32/nds32.c (nds32_option_override):
37023 Fix misspelled diagnostic.
37025 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37028 * gimplify.c (gimplify_omp_for): Replace index var in outer
37029 taskloop statement with an artificial variable and add
37030 OMP_CLAUSE_PRIVATE clause for it.
37032 2017-03-08 Richard Biener <rguenther@suse.de>
37034 PR tree-optimization/79955
37035 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
37036 for accesses that are completely outside of the variable.
37038 2017-03-08 Andrew Haley <aph@redhat.com>
37040 PR tree-optimization/79943
37041 * tree-ssa-loop-split.c (compute_new_first_bound): When
37042 calculating the new upper bound, (END-BEG) should be added, not
37045 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37047 * config/avr/avr.md (setmemhi): Make sure match_dup
37048 operand number comes before match_scratch.
37050 2017-03-08 Richard Biener <rguenther@suse.de>
37052 PR tree-optimization/79920
37053 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
37054 with ncopies == 1 to ...
37055 (vect_transform_slp_perm_load): ... here. Properly compute
37056 all element loads by iterating VF times over the group. Do
37057 not handle ncopies (computed in a broken way) in
37058 vect_create_mask_and_perm.
37060 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37063 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
37064 is a uniform vector, use uniform_vector_p return value instead of
37065 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
37067 2017-03-07 Marek Polacek <polacek@redhat.com>
37069 PR middle-end/79809
37070 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
37071 (alloca_call_type): Likewise.
37073 2017-03-07 Martin Liska <mliska@suse.cz>
37075 * gcov.c (process_args): Put comment to correct location.
37077 2017-03-07 Martin Liska <mliska@suse.cz>
37079 PR middle-end/68270
37080 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
37081 Use array_at_struct_end_p instead of DECL_CHAIN (field).
37082 (chkp_narrow_bounds_for_field): Likewise.
37083 (chkp_parse_array_and_component_ref): Pass one more argument to
37086 2017-03-07 Richard Biener <rguenther@suse.de>
37088 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
37091 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
37093 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
37094 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
37096 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37099 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
37100 to end of description.
37101 (PARAM_MAX_STORES_TO_MERGE): Likewise.
37103 2017-03-07 Jakub Jelinek <jakub@redhat.com>
37105 PR rtl-optimization/79901
37106 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
37108 (*avx512f_<code><mode>3<mask_name>): ... this.
37109 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
37110 iterator instead of VI8_AVX2_AVX512BW.
37112 PR rtl-optimization/79901
37113 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
37114 min/max expander, expand it using expand_vec_cond_expr.
37117 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
37120 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37123 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
37124 to void * for PCH reasons.
37125 * dwarf2out.c (output_loc_operands, output_die): Cast
37126 v.val_vec.array to unsigned char *.
37128 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
37131 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
37134 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
37136 PR rtl-optimization/79571
37137 * lra-constraints.c (process_alt_operands): Calculate static
37138 reject and subtract it from overall when only addresses will be
37141 2017-03-06 Julia Koval <julia.koval@intel.com>
37144 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
37145 incoming stack boundary to 128 for 64-bit targets.
37147 2017-03-06 Richard Biener <rguenther@suse.de>
37149 PR tree-optimization/79894
37150 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
37151 to NULL after folding it.
37153 2017-03-06 Richard Biener <rguenther@suse.de>
37155 PR tree-optimization/79824
37156 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
37157 check disabling peeling for gaps.
37159 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
37161 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
37162 attributes): Document gettimeofday.
37164 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37166 * config/s390/s390.c (s390_option_override_internal): Set
37167 PARAM_MIN_VECT_LOOP_BOUND
37169 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37171 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
37172 * config/s390/s390.md: Likewise.
37174 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37177 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
37178 (<avx2_avx512>_perm<mode>): Rename to ...
37179 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
37181 (<avx512>_perm<mode>_mask): Rename to ...
37182 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
37184 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
37185 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
37186 instead of VI8F_256_512.
37187 (avx512f_perm<mode>): New define_expand.
37188 (avx512f_perm<mode>_mask): Likewise.
37189 (avx512f_perm<mode>_1<mask_name>): New define_insn.
37190 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
37192 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37194 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
37195 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
37197 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
37199 2017-03-06 Martin Liska <mliska@suse.cz>
37202 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37203 when having a SSA NAME w/o VAR_DECL assigned to it.
37205 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37207 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37208 msa_dpsub_<su>_d): Fix MODE for vec_select.
37210 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37212 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37214 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37216 2017-03-06 Richard Biener <rguenther@suse.de>
37218 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37219 * plugin.c (register_plugin_info): Likewise.
37220 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37222 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37224 * config/i386/sse.md (sse_storehps, sse_storelps,
37225 avx_<castmode><avxsizesuffix>_<castmode>,
37226 avx512f_<castmode><avxsizesuffix>_<castmode>,
37227 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37228 in condition that at least one operand is not a MEM.
37230 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37232 PR middle-end/79805
37233 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37234 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37236 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37237 gimple_call_nothrow_p flag based on whether original builtin can throw.
37238 If it can, emit following stmts on the fallthrough edge.
37239 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37240 don't create new bb if inserting just debug stmts on the edge, try to
37241 insert them on the fallthru bb or just reset debug stmts.
37243 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37246 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37247 restore recog_data (including the operand rtxes inside it) around
37248 the call to get_insn_template.
37250 2017-03-03 Martin Sebor <msebor@redhat.com>
37252 PR tree-optimization/79699
37253 * context.c (context::~context): Free MPFR caches to avoid
37254 a memory leak on program exit.
37256 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37258 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37259 Use wide_int::ulow () instead of .elt (0).
37261 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37263 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37264 (*pushxf): Limit oF constraint to 32bit targets and add oC
37265 constraint for 64bit targets.
37266 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37267 (*pushdf): Change rmF constraint to rmC.
37269 2017-03-03 Martin Liska <mliska@suse.cz>
37271 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37272 Remove unused variable.
37274 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37277 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37278 is a memory operand, increase num_memory.
37279 (ix86_expand_args_builtin): Likewise.
37281 2017-03-03 Jan Hubicka <jh@suse.cz>
37284 * ipa-devirt.c (maybe_record_node): Properly handle
37285 __cxa_pure_virtual visibility.
37287 2017-03-03 Martin Liska <mliska@suse.cz>
37289 PR tree-optimization/79803
37290 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37292 (pass_loop_prefetch::execute): Disabled optimization if an
37293 assumption about L1 cache size is not met.
37295 2017-03-03 Martin Liska <mliska@suse.cz>
37297 PR rtl-optimization/79574
37298 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37299 (hash_scan_set): Likewise.
37300 (dump_hash_table): Likewise.
37301 (hoist_code): Likewise.
37303 2017-03-03 Richard Biener <rguenther@suse.de>
37305 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37306 in place of uhigh ().
37307 (fixed_convert_from_real): Likewise.
37309 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37312 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37314 2017-03-03 Richard Biener <rguenther@suse.de>
37316 PR middle-end/79818
37317 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37318 TYPE_OVERFLOW_UNDEFINED check.
37320 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37322 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37324 (vector_ae_<mode>_p): Likewise.
37325 (vector_nez_<mode>_p): Likewise.
37326 (vector_ne_v2di_p): Likewise.
37327 (vector_ae_v2di_p): Likewise.
37328 (vector_ne_<mode>_p): Likewise.
37329 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37331 (vsx_tsqrt<mode>2_fe): Likewise.
37333 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37336 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37338 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37340 PR rtl-optimization/79780
37341 * cprop.c (one_cprop_pass): When second and further conditional trap
37342 in a single basic block is turned into an unconditional trap, turn it
37343 into a deleted note to avoid RTL verification failures.
37345 2017-03-02 Richard Biener <rguenther@suse.de>
37347 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37349 2017-03-02 Richard Biener <rguenther@suse.de>
37351 PR tree-optimization/79345
37353 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37354 param and abort the walk, returning -1 if it is hit.
37355 (walk_aliased_vdefs): Take a limit param and pass it on.
37356 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37357 defaulting to 0 and return a signed int.
37358 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37359 (check_defs): New helper.
37360 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37361 about uninitialized memory.
37362 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37363 bogus uninitialized warning.
37364 (fixed_convert_from_real): Likewise.
37366 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37368 PR tree-optimization/66768
37369 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37370 iv_use if base object can't be determined.
37372 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37374 PR tree-optimization/79345
37375 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37376 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37377 (get_pattern_stats): Initialize it.
37378 * genemit.c (gen_expand): Verify match_scratch numbers come after
37379 match_operand/match_dup numbers.
37380 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37381 match_scratch numbers.
37382 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37384 * config/s390/s390.md (trunctdsd2): Likewise.
37386 2017-03-02 Richard Biener <rguenther@suse.de>
37388 * wide-int.h (wide_int_storage::operator=): Implement in terms
37391 2017-03-02 Richard Biener <rguenther@suse.de>
37393 PR tree-optimization/79777
37394 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37395 the to insert expression to sth existing.
37397 2017-03-01 Martin Sebor <msebor@redhat.com>
37399 PR middle-end/79692
37400 * gimple-ssa-sprintf.c
37401 (directive::known_width_and_precision): New function.
37402 (format_integer): Use it.
37403 (get_mpfr_format_length): Consider the full range of precision
37404 when computing %g output with the # flag. Set the likely byte
37405 count to 3 rather than 1 when precision is indeterminate.
37406 (format_floating): Correct the lower bound of precision.
37408 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37410 * doc/invoke.texi: Document default code model for 64-bit Linux.
37412 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37415 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37416 udiv rather than div since input pattern is unsigned.
37418 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37420 * config/i386/i386.c (print_reg): Warn for values of
37421 unsupported size in integer register.
37423 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37426 * config/rs6000/predicates.md (current_file_function_operand): Do
37427 not allow self calls to be local if the function is replaceable.
37429 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37432 * config/rs6000/altivec.h (vec_ctz and others): Change the
37433 preprocessor macro that controls conditional compilation from
37434 _ARCH_PWR9 to __POWER9_VECTOR__.
37435 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37436 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37437 control (instead of _ARCH_PWR9 control) so that template
37438 definition uses power9-specific function.
37439 (vec_any_eq): Likewise.
37440 (vec_all_ne): Change macro definition to use a power9-specific
37441 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37442 _ARCH_PWR9 control).
37443 (vec_any_eq) Likewise.
37444 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37445 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37446 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37447 support for xvcmpnedp instruction.
37448 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37449 macro expansion so that Power9 implementation of vec_all_ne does
37450 not use the AltiVec predicate framework.
37451 (VCMPNEH_P): Likewise.
37452 (VCMPNEW_P): Likewise.
37453 (VCMPNED_P): Likewise.
37454 (VCMPNEFP_P): Likewise.
37455 (VCMPNEDP_P): Likewise.
37456 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37457 implementation of vec_any_eq to not use AltiVec predicate
37459 (VCMPAEH_P): Likewise.
37460 (VCMPAEW_P): Likewise.
37461 (VCMPAED_P): Likewise.
37462 (VCMPAEFP_P): Likewise.
37463 (VCMPAEDP_P): Likewise.
37464 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37465 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37466 not use the AltiVec predicate framework.
37467 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37468 of vec_any_eq to not use AltiVec predicate framework.
37469 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37470 support for predefined __POWER9_VECTOR__ macro to indicate that
37471 Power9 instruction selection is enabled.
37472 (altivec_overloaded_builtins): Remove extraneous
37473 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37474 function argument types RS6000_BTI_bool_V16QI and
37475 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37476 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37477 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37478 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37479 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37480 Power9 for implementations of vec_cmpne. Change the signature for
37481 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37482 (representing vec_all_ne) to remove the previously described first
37483 argument of type RS6000_BTI_INTSI, as this was an artifact of
37484 reliance on the AltiVec predicate framework, which is no longer
37485 used in the implementation of these functions. Add
37486 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37487 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37488 since, unlike the AltiVec predicate framework implementation, we
37489 do not share function descriptors between vec_alle and vec_anyeq.
37490 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37491 set of modes that receive special treatment even when
37492 TARGET_P9_VECTOR is true. The special treatment emits code that
37493 does not depend on Power9 instructions.
37494 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37495 define_expand to not rely on AltiVec predicate framework.
37496 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37498 (vector_ne_v2di_p): Change this define_expand to not rely on
37499 AltiVec predicate framework.
37500 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37502 (vector_ne_<mode>_p): Change this define_expand to not rely on
37503 AltiVec predicate framework.
37504 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37506 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37507 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37508 define_insn pattern.
37509 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37510 define_insn pattern because the xvcmpne<VSs>. instruction is not
37512 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37513 instruction is not supported.
37515 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37517 * config/nvptx/nvptx.c: Include intl.h.
37519 2017-03-01 Martin Jambor <mjambor@suse.cz>
37522 * ipa-prop.h (ipa_bits): Removed field known.
37523 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37524 to pointers. Adjusted their comments to warn about their sharing.
37525 (ipcp_transformation_summary): Change bits to a vector of pointers.
37526 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37527 (ipa_get_ipa_bits_for_value): Declare.
37528 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37529 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37530 (ipa_bits_hash_table): Likewise.
37531 (ipa_vr_ggc_hash_traits): Likewise.
37532 (ipa_vr_hash_table): Likewise.
37533 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37534 being pointers and vr_known being removed.
37535 (ipa_set_jf_unknown): Likewise.
37536 (ipa_get_ipa_bits_for_value): New function.
37537 (ipa_set_jfunc_bits): Likewise.
37538 (ipa_get_value_range): New overloaded functions.
37539 (ipa_set_jfunc_vr): Likewise.
37540 (ipa_compute_jump_functions_for_edge): Use the above functions to
37541 construct bits and vr parts of jump functions.
37542 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37543 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37545 (ipcp_grow_transformations_if_necessary): Also allocate
37546 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37548 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37549 them. Fix too long lines.
37550 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37551 vr_known being removed.
37552 (ipa_read_jump_function): Use new setter functions to construct bits
37553 and vr parts of jump functions or set them to NULL.
37554 (write_ipcp_transformation_info): Adjust for bits being pointers.
37555 (read_ipcp_transformation_info): Likewise.
37556 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37558 Include gt-ipa-prop.h.
37559 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37561 (ipcp_store_bits_results): Likewise.
37562 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37563 Do not write to existing jump functions but use a temporary instead.
37565 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37568 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37569 attempt to use its first operand as BIT_FIELD_REF base.
37571 2017-03-01 Richard Biener <rguenther@suse.de>
37573 PR middle-end/79721
37574 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37575 interpolating formula in wrapping arithmetic.
37576 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37578 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37580 PR tree-optimization/79734
37581 * tree-vect-generic.c (expand_vector_condition): Optimize
37582 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37583 Handle VEC_COND_EXPR where comparison has different inner width from
37584 type's inner width.
37586 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37588 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37589 markup, and similar issues. Remove @opindex entries for things
37590 that aren't options. Add missing -mmpy-option entries.
37592 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37594 PR tree-optimization/79737
37595 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37596 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37597 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37598 instead of byte_size. Formatting fix.
37599 (shift_bytes_in_array_right): Formatting fix.
37601 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37604 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37605 condition on optimize for the leaf function test.
37607 2017-02-28 Martin Liska <mliska@suse.cz>
37610 * read-rtl-function.c (function_reader::handle_unknown_directive):
37611 Bail out when one uses -flto.
37613 2017-02-28 Martin Liska <mliska@suse.cz>
37615 * common.opt: Replace space with tabular for options of <number>
37617 * config/i386/i386.opt: Show <number> value for
37618 -mlarge-data-threshold.
37619 * opts.c (print_filtered_help): Do not display number in hexadecimal
37622 2017-02-28 Martin Liska <mliska@suse.cz>
37624 * common.opt: Fix --help=option -Q for options which are of
37627 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37629 * config/i386/i386.c (print_reg): Error out for values
37630 of 8-bit size in invalid integer register.
37632 2017-02-28 Martin Sebor <msebor@redhat.com>
37634 PR tree-optimization/79691
37635 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37637 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37640 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37641 gcc_unreachable with output_operand_lossage.
37643 2017-02-28 Richard Biener <rguenther@suse.de>
37645 PR tree-optimization/79740
37646 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37648 (visit_nary_op): Insert the nary into the hashtable if we
37649 pattern-matched sth.
37650 * tree-ssa-pre.c (eliminate_insert): Robustify.
37652 2017-02-28 Richard Biener <rguenther@suse.de>
37654 PR middle-end/79731
37655 * fold-const.c (decode_field_reference): Reject out-of-bound
37658 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37660 * config/i386/i386.c: Include intl.h.
37661 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37662 instead of just cond ? "..." : "...".
37663 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37664 * coverage.c (read_counts_file): Likewise.
37665 * omp-offload.c: Include intl.h.
37666 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37667 of just cond ? "..." : "...".
37668 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37669 of just cond ? "..." : "...".
37671 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37674 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37676 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37677 'tune for' CPU name.
37678 * config/arm/arm-cpu-data.h: Regenerated.
37680 2017-02-28 Richard Biener <rguenther@suse.de>
37682 PR tree-optimization/79732
37683 * tree-inline.c (expand_call_inline): Do not shadow var.
37685 2017-02-28 Richard Biener <rguenther@suse.de>
37687 PR tree-optimization/79723
37688 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37689 address-space properly.
37691 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37693 * doc/optinfo.texi (Optimization groups): Fix option used for
37695 * doc/invoke.texi (-fopt-info): Document "omp".
37696 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37697 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37698 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37699 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37700 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37702 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37704 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37705 OPTGROUP_OMP with "omp".
37707 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37710 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37711 for arithmetic shift of unsigned V2DI.
37713 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37715 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37716 arc/linux.h headers.
37717 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37718 (LINK_SPEC): Likewise.
37719 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37720 (EXTRA_SPECS): Likewise.
37721 (STARTFILE_SPEC): Likewise.
37722 (ENDFILE_SPEC): Likewise.
37723 (LIB_SPEC): Likewise.
37724 (TARGET_SDATA_DEFAULT): Likewise.
37725 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37726 (MULTILIB_DEFAULTS): Likewise.
37727 (DWARF2_UNWIND_INFO): Likewise.
37728 * config/arc/big.h: New file.
37729 * config/arc/elf.h: Likewise.
37730 * config/arc/linux.h: Likewise.
37731 * config/arc/t-uClibc: Remove.
37733 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37735 PR tree-optimization/77536
37736 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37737 (tree_transform_and_unroll_loop): Use above function to compute the
37738 estimated niter of unrolled loop and use it when scaling profile.
37739 Also use count info rather than frequency if it's non-zero.
37740 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37741 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37742 (vect_transform_loop): Call above function.
37744 2017-02-27 Richard Biener <rguenther@suse.de>
37746 PR tree-optimization/45397
37747 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37748 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37749 (visit_nary_op): Add pattern matching for CSEing sign-changed
37750 or truncated operations with wider ones.
37752 2017-02-27 Richard Biener <rguenther@suse.de>
37754 PR tree-optimization/79690
37755 * tree-vect-stmts.c (vectorizable_store): Use vector type
37756 built from the DR with address-space.
37758 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37760 * doc/invoke.texi (Optimize Options): Refine the description
37761 of asan-use-after-return.
37763 2017-02-25 Alan Modra <amodra@gmail.com>
37765 PR rtl-optimization/79584
37766 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37767 base, not ad->base_term, the reg within base. Remove assertion
37768 that ad->base == ad->base_term. Replace gen_int_mode using
37769 bogus mode with const0_rtx.
37771 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37773 PR middle-end/79396
37774 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37775 FMA_EXPR like tcc_binary or tcc_unary.
37777 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37780 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37782 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37783 (output_loc_operands): Handle DW_OP_call_ref and
37784 DW_OP_GNU_variable_value.
37785 (struct variable_value_struct): New type.
37786 (struct variable_value_hasher): Likewise.
37787 (variable_value_hash): New variable.
37788 (string_types): Remove.
37789 (copy_loc_descr): New function.
37790 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37791 (prepend_loc_descr_to_each): New function.
37792 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37793 instead of add_loc_descr_to_each if the first argument is single
37794 location list and the second has multiple.
37795 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37796 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37797 when looking for variable value which doesn't have other location info.
37798 (loc_list_from_tree): Formatting fix.
37799 (gen_array_type_die): Simplify DW_AT_string_length handling.
37800 (adjust_string_types): Remove.
37801 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37802 string_types. Call resolve_variable_values.
37803 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37804 (resolve_addr_in_expr): Likewise. Add A argument.
37805 (copy_deref_exprloc): Remove deref argument. Adjust for the
37806 original expression being DW_OP_GNU_variable_value with optionally
37807 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37808 optionally after it.
37809 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37810 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37811 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37812 (variable_value_hasher::hash, variable_value_hasher::equal): New
37814 (resolve_variable_value_in_expr, resolve_variable_value,
37815 resolve_variable_values, note_variable_value_in_expr,
37816 note_variable_value): New functions.
37817 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37820 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37823 * opts.h (handle_generated_option): Add GENERATED_P argument.
37824 * opts-common.c (handle_option): Adjust function comment.
37825 (handle_generated_option): Add GENERATED_P argument, pass it to
37827 (control_warning_option): Pass false to handle_generated_option
37829 * opts.c (maybe_default_option): Pass true to handle_generated_option
37831 * optc-gen.awk: Likewise.
37833 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37835 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37836 a REG, look at the REG it is a SUBREG of.
37837 (splitter for cmpeqsi_t): Ditto.
37839 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37841 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37842 the special USEs with the pattern of the insn, not the insn itself.
37844 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37847 * doc/invoke.texi: Document -mload-store-pairs.
37849 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37850 Sandra Loosemore <sandra@codesourcery.com>
37852 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37853 argument isn't a CONST_INT.
37854 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37855 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37856 (nios2_validate_compare): Bypass alternate compare logic if *op2
37857 is not a CONST_INT.
37858 (ldstwm_operation_p): Return false if first_base is not a REG or
37859 if first_offset is not a CONST_INT.
37861 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37863 * config/cris/cris.md: Use correct operand in a define_peephole2.
37865 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37867 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37869 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37871 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37872 this_insn if it is an INSN or JUMP_INSN.
37873 (force_offsettable): Look at base, not at addr.
37874 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37875 on things that aren't necessarily CONST_INTs.
37877 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37879 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37880 -mfpmath=sse is the default also for x86-32 targets with SSE2
37881 instruction set when @option{-ffast-math} is enabled
37883 2017-02-24 Jeff Law <law@redhat.com>
37885 PR rtl-optimizatoin/79286
37886 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37889 2017-02-24 Richard Biener <rguenther@suse.de>
37891 PR tree-optimization/79389
37892 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37895 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37897 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37898 function comment to reflect reality.
37899 (loop_exits_before_overflow): Fix typo in function description.
37901 2017-02-24 Richard Biener <rguenther@suse.de>
37903 PR tree-optimization/79389
37904 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37905 properly that a threading opportunity exists. Detect conditional
37906 copy/constant propagation opportunities.
37908 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37910 * config/visium/visium.md (type): Add trap.
37911 (b): New mode attribute.
37912 (*btst): Rename into...
37913 (*btst<mode>): ...this and adjust.
37914 (*cbranchsi4_btst_insn): Rename into...
37915 (*cbranch<mode>4_btst_insn): ...this and adjust.
37916 (trap): New define_insn.
37918 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37920 PR tree-optimization/79389
37921 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37922 (noce_reversed_cond_code): New function.
37923 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37924 reversed_comparison_code. Formatting fix.
37925 (noce_try_store_flag): Test rev_cond != NULL in addition to
37926 reversed_comparison_code.
37927 (noce_try_store_flag_constants): Likewise.
37928 (noce_try_store_flag_mask): Likewise.
37929 (noce_try_addcc): Use rev_cond if non-NULL instead of
37930 reversed_comparison_code.
37931 (noce_try_cmove_arith): Likewise. Formatting fixes.
37932 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37933 (noce_find_if_block): Initialize rev_cond.
37934 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37935 instead of false as last argument never attempt to reverse it
37938 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37940 PR tree-optimization/79663
37941 * tree-predcom.c (combine_chains): Process refs in reverse order
37942 only for ZERO length chains, and add explaining comment.
37944 2017-02-23 Jeff Law <law@redhat.com>
37946 PR tree-optimization/79578
37947 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37948 in call to operand_equal_p.
37950 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37953 * config/i386/cpuid.h: Fix another undefined behavior.
37955 2017-02-23 Richard Biener <rguenther@suse.de>
37957 PR tree-optimization/79683
37958 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37959 vector types for data-refs.
37961 2017-02-23 Martin Liska <mliska@suse.cz>
37963 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37965 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37967 PR middle-end/79665
37968 * internal-fn.c (get_range_pos_neg): Moved to ...
37969 * tree.c (get_range_pos_neg): ... here. No longer static.
37970 * tree.h (get_range_pos_neg): New prototype.
37971 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37972 are known to be in between 0 and signed maximum inclusive, try to
37973 expand both unsigned and signed divmod and use the cheaper one from
37976 2017-02-22 Jeff Law <law@redhat.com>
37978 PR tree-optimization/79578
37979 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37980 to compare base operands.
37982 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37985 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37986 gpc_reg_operand instead of fpr_reg_operand.
37988 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37990 * config/mips/mips.c (mips_return_in_memory): Force FP
37991 vector types to be returned in memory for o32 ABI.
37993 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37995 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37996 instead of DW_TAG_member for static data member declarations and don't
37997 set no_linkage_name for static inline data members.
37998 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
38001 2017-02-22 Martin Liska <mliska@suse.cz>
38003 * doc/invoke.texi: Replace inequality signs with square brackets
38006 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
38009 * lra-constraints.c (simplify_operand_subreg): Handle
38010 WORD_REGISTER_OPERATIONS targets.
38012 2017-02-22 Jakub Jelinek <jakub@redhat.com>
38015 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
38016 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
38017 elimination by swapping fld*.
38019 2017-02-22 Richard Biener <rguenther@suse.de>
38021 PR tree-optimization/79673
38022 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
38023 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
38024 irrelevant address-space qualifiers and avoiding a
38025 ADDR_SPACE_CONVERT_EXPR from fold_convert.
38027 2017-02-22 Richard Biener <rguenther@suse.de>
38029 PR tree-optimization/79666
38030 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
38031 to not symbolically negate if that may introduce undefined
38034 2017-02-22 Martin Liska <mliska@suse.cz>
38037 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
38038 * data-streamer-out.c (streamer_write_gcov_count_stream):
38040 * value-prof.c (stream_out_histogram_value): Make assert more
38041 precise based on type of counter.
38043 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
38046 * config/i386/i386.md (standard_x87sse_constant_load splitter):
38047 Use nonimmediate_operand instead of memory_operand for operand 1.
38048 (float-extend standard_x87sse_constant_load splitter): Ditto.
38050 2017-02-21 Jeff Law <law@redhat.com>
38052 PR tree-optimization/79621
38053 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
38054 blocks with edges to themselves.
38056 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38059 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
38060 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
38061 Use gimple_call_builtin_p.
38064 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
38065 on temporarily removed DEBUG_INSNs.
38067 PR tree-optimization/79649
38068 * tree-loop-distribution.c (classify_partition): Give up on
38069 non-generic address space loads/stores.
38071 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
38073 * doc/loop.texi (Loop manipulation): Remove nonexistent
38074 tree_ssa_loop_version from the documentation.
38075 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
38077 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38080 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
38081 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
38082 * config/rs6000/rs6000.c: Include except.h.
38083 (rs6000_expand_split_stack_prologue): Call
38084 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
38086 2017-02-21 Martin Jambor <mjambor@suse.cz>
38089 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
38090 have been analyzed.
38092 2017-02-21 Martin Jambor <mjambor@suse.cz>
38094 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
38095 for backward compatibility only.
38096 * doc/invoke.texi (Option Summary): Remove all references to
38097 -fipa-cp-alignment.
38099 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
38103 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38105 * lra-constraints.c (curr_insn_transform): Handle
38106 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38108 2017-02-21 Martin Liska <mliska@suse.cz>
38110 * config/i386/i386.opt: Replace -masm-dialect with -masm.
38112 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
38114 PR translation/79638
38115 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
38117 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
38120 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
38121 (arm_function_ok_for_sibcall): Return false for an indirect call by
38122 descriptor if all the argument registers are used.
38123 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
38124 alignment of the function.
38126 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38128 PR tree-optimization/61441
38129 * simplify-rtx.c (simplify_const_unary_operation): For
38130 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
38131 the sNaN unmodified.
38133 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
38135 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
38136 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
38137 instead of SYSTEM_HEADER_DIR.
38139 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
38140 Martin Liška <mliska@suse.cz>
38142 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
38143 Fix typos and grammar, use active voice, and clarify.
38145 2017-02-20 Marek Polacek <polacek@redhat.com>
38147 PR middle-end/79537
38148 * gimplify.c (gimplify_expr): Handle unused *&&L;.
38151 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
38153 2017-02-20 Jakub Jelinek <jakub@redhat.com>
38156 * config/i386/i386.c (ix86_expand_builtin): Handle
38157 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
38158 ix86_builtins_isa[fcode].isa as a requirement of those
38159 flags and any other flag in the bitmask.
38160 (ix86_init_mmx_sse_builtins): Use 0 instead of
38161 ~OPTION_MASK_ISA_64BIT as mask.
38162 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
38163 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
38164 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
38165 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
38167 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38170 * lra-constraints.c (split_reg): Check requested split mode
38171 is supported by the register.
38173 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38175 * lra-constraints.c (simplify_operand_subreg): Remove early
38178 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38181 * lra-constraints.c (curr_insn_transform): Tighten condition
38182 for converting SUBREG reloads from OP_OUT to OP_INOUT.
38184 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38187 * lra-constraints.c (curr_insn_transform): Handle
38188 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38190 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
38193 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
38195 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
38197 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
38200 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38203 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38205 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38208 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38210 * final.c (last_columnnum, override_columnnum): New variables.
38211 (final_start_function): Set last_columnnum, pass it to begin_prologue
38212 hook and pass 0 to dwarf2out_begin_prologue.
38213 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38214 to source_line debug hook.
38215 (notice_source_line): Compute last_columnnum and for debug_column_info
38216 return true on column changes.
38217 * debug.h (struct gcc_debug_hooks): Add column argument to
38218 source_line and begin_prologue hooks.
38219 (debug_nothing_int_charstar_int_bool): Remove prototype.
38220 (debug_nothing_int_int_charstar,
38221 debug_nothing_int_int_charstar_int_bool): New prototypes.
38222 (dwarf2out_begin_prologue): Add column argument.
38223 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38224 begin_prologue hooks.
38225 (debug_nothing_int_charstar_int_bool): Remove.
38226 (debug_nothing_int_int_charstar,
38227 debug_nothing_int_int_charstar_int_bool): New functions.
38228 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38229 through to dwarf2out_source_line.
38230 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38231 (dwarf2out_source_line): Add column argument, emit it if requested.
38232 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38234 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38235 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38236 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38237 through to dwarf2out_begin_prologue.
38238 (vmsdbgout_source_line): Add column argument, pass it through to
38239 dwarf2out_source_line.
38240 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38241 dbxout_source_line caller.
38242 (dbxout_source_line): Add column argument.
38244 * common.opt (gno-column-info, gcolumn-info): New options.
38245 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38246 (check_die): Also test for multiple DW_AT_decl_column attributes.
38247 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38248 DW_AT_decl_column if requested.
38249 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38251 (gen_variable_die): Likewise.
38252 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38253 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38256 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38257 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38258 (ix86_handle_option): Handle OPT_m3dnowa.
38259 * doc/invoke.texi (-m3dnowa): Document.
38260 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38261 -m3dnowa instead of -m3dnow -march=athlon.
38264 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38265 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38267 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38270 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38271 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38272 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38273 generator for vsx_xxpermdi_<mode>_be.
38274 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38275 force big-endian semantics.
38276 (vsx_xxpermdi_<mode>_be): New define_expand with same
38277 implementation as previous version of vsx_xxpermdi_<mode>.
38279 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38281 PR tree-optimization/79327
38282 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38283 variable, its initialization and use.
38285 2017-02-17 Julia Koval <julia.koval@intel.com>
38287 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38288 (OPTION_MASK_ISA_PKU_UNSET): New.
38289 (ix86_handle_option): Handle -mrdpid.
38290 * config/i386/cpuid.h (bit_RDPID): New.
38291 * config/i386/driver-i386.c (host_detect_local_cpu):
38292 Detect RDPID feature.
38293 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38294 * config/i386/i386-c.c (ix86_target_macros_internal):
38296 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38297 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38298 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38299 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38300 * config/i386/i386.md (define_insn "rdpid"): New.
38301 * config/i386/i386.opt Add -mrdpid.
38302 * config/i386/immintrin.h (_rdpid_u32): New.
38304 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38306 PR rtl-optimization/79541
38307 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38308 instead of transforming it into USE.
38310 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38312 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38313 If HONOR_SNANS (SFmode) force the input to a register.
38314 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38315 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38316 an frsp or similar insn.
38318 2017-02-17 Martin Liska <mliska@suse.cz>
38320 PR rtl-optimization/79577
38321 * params.def (selsched-max-sched-times): Increase minimum to 1.
38323 2017-02-17 Martin Liska <mliska@suse.cz>
38325 PR rtl-optimization/79574
38326 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38328 2017-02-17 Martin Liska <mliska@suse.cz>
38330 PR tree-optimization/79529
38331 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38332 ssa_defined_default_def_p to handle cases which are implicitly
38334 * tree-ssa.c (ssa_defined_default_def_p): New function.
38335 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38336 which are implicitly defined.
38337 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38339 2017-02-17 Richard Biener <rguenther@suse.de>
38341 PR middle-end/79576
38342 * params.def (max-ssa-name-query-depth): Limit to 10.
38344 2017-02-17 Richard Biener <rguenther@suse.de>
38346 PR tree-optimization/79552
38347 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38350 2017-02-17 Richard Biener <rguenther@suse.de>
38353 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38355 2017-02-17 Marek Polacek <polacek@redhat.com>
38357 PR middle-end/79536
38358 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38359 (fold_negate_expr): New wrapper.
38361 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38363 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38364 Correct terminology and de-emphasize pre-standard behavior.
38366 2017-02-16 Alan Modra <amodra@gmail.com>
38368 PR rtl-optimization/79286
38369 * ira.c (def_dominates_uses): New function.
38370 (update_equiv_regs): Don't create an equivalence for insns that
38371 may trap where the register def does not dominate the use.
38373 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38375 PR rtl-optimization/78127
38376 * lra.c (lra): Call lra_eliminate before finish the loop after
38379 2017-02-16 Richard Biener <rguenther@suse.de>
38381 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38383 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38384 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38385 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38386 (isl_val_int_from_wi): New function.
38387 (extract_affine_gmp): Rename to ...
38388 (extract_affine_wi): ... this, take a widest_int.
38389 (extract_affine_int): Just wrap extract_affine_wi.
38390 (add_param_constraints): Use isl_val_int_from_wi.
38391 (add_loop_constraints): Likewise, and extract_affine_wi.
38393 2017-02-15 Jeff Law <law@redhat.com>
38395 PR middle-end/79521
38396 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38397 ira_init_register_move_cost_if_necessary.
38399 2017-02-15 Martin Sebor <msebor@redhat.com>
38401 PR middle-end/32003
38402 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38403 removed in a prior commit.
38405 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38407 PR tree-optimization/79347
38408 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38409 counters during peeling.
38411 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38413 * Makefile.in (site.exp): Remove "set ISLVER".
38415 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38418 * real.c (real_from_integer): Call real_convert even for decimal.
38420 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38423 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38425 2017-02-14 Andrew Pinski <apinski@cavium.com>
38427 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38428 cores and change the partno/implementer to be correct.
38429 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38430 the 'B" as the implementer.
38431 * config/aarch64/aarch64-tune.md: Regenerate.
38433 2017-02-14 Carl Love <cel@us.ibm.com>
38435 * config/rs6000/rs6000.c: Add case statement entry to make the
38436 xvcvuxdsp built-in argument unsigned.
38437 * config/rs6000/vsx.md: Fix the source and return operand types so they
38438 match the instruction definitions from the ISA document. Fix typo
38439 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38442 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38445 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38446 member early_clobber_alts.
38447 * lra-lives.c (reg_early_clobber_p): New.
38448 (process_bb_lives): Use it.
38449 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38450 (debug_operand_data): Initialize early_clobber_alts.
38451 (setup_operand_alternative): Set up early_clobber_alts.
38452 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38453 alternatives to new_insn_reg.
38454 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38456 (lra_update_insn_regno_info): Pass the new arg.
38458 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38460 PR middle-end/79505
38461 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38462 (new_oacc_loop_raw): Don't clear already cleared fields.
38465 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38466 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38467 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38469 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38472 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38474 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38477 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38478 the extra instruction to the right place to store 128-bit constant
38481 2017-02-14 Martin Sebor <msebor@redhat.com>
38483 PR middle-end/79448
38484 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38485 warning for strings of unknown length.
38487 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38489 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38491 2017-02-14 Jeff Law <law@redhat.com>
38494 * ira-costs.c (scan_one_insn): Initialize register move costs
38495 for pseudos seen in USE/CLOBBER insns.
38497 PR tree-optimization/79095
38498 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38499 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38500 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38501 if the operands are known to be not equal, then the resulting range
38503 (intersect_ranges): If the new range is ~[0,0] and the old range is
38504 wide, then prefer ~[0,0].
38505 * tree-vrp.c (overflow_comparison_p_1): New function.
38506 (overflow_comparison_p): New function.
38507 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38508 if NAME is used in an overflow test.
38509 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38510 overflow check that can be expressed as an equality test, then adjust
38511 ops to be that equality test.
38513 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38515 * config/s390/s390-builtin-types.def: Remove flags argument.
38516 * config/s390/s390.c (s390_init_builtins): Likewise.
38518 2017-02-14 Martin Liska <mliska@suse.cz>
38520 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38521 vector. Fix trailing white spaces.
38523 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38525 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38528 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38530 PR rtl-optimization/68664
38531 * config/arm/arm.c (arm_sched_can_speculate_insn):
38532 New function. Declare prototype.
38533 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38535 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38537 PR rtl-optimization/68664
38538 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38540 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38542 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38544 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38545 max skip bytes for function, loop and jump.
38547 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38549 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38550 ABS_EXPR for gimple dump.
38552 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38555 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38557 PR tree-optimization/79408
38558 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38559 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38560 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38561 also if rhs1 is INTEGER_CST.
38563 2017-02-14 Richard Biener <rguenther@suse.de>
38565 PR middle-end/79432
38566 * tree-into-ssa.c (insert_phi_nodes): When the function can
38567 have abnormal edges rewrite SSA names with broken use-def
38568 dominance out of SSA and register them for PHI insertion.
38570 2017-02-13 Martin Sebor <msebor@redhat.com>
38572 PR middle-end/79496
38573 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38574 clearing info.nowrite flag when snprintf size argument is a range.
38576 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38578 * cprop.c (cprop_jump): Add missing space in string literal.
38579 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38580 (get_constraint_for_component_ref): Likewise.
38581 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38582 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38583 * lra-constraints.c (process_alt_operands): Likewise.
38584 * ipa-inline.c (inline_small_functions): Likewise.
38585 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38586 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38587 * trans-mem.c (diagnose_tm_1_op): Likewise.
38588 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38589 (grid_parallel_clauses_gridifiable): Likewise.
38591 * config/nvptx/mkoffload.c (process): Add space in between
38594 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38595 "MOD4_SSE_REGS" and "ALL_REGS".
38597 * spellcheck.c (test_data): Add , in between "foo" and "food".
38599 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38602 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38603 boundary crossing check and subsequent code generation agree.
38605 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38607 * config/aarch64/aarch64.c (has_memory_op): Delete.
38608 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38611 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38613 PR rtl-optimization/79388
38614 PR rtl-optimization/79450
38615 * combine.c (distribute_notes): When removing TEM_INSN for which
38616 corresponding dest has last value recorded, invalidate that last
38619 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38621 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38622 of explicit '@'. Add missing assembly comment marker on branch costs
38625 2017-02-13 Nathan Sidwell <nathan@acm.org>
38627 * gengtype-lex.l (<in_struct>): Add '/'.
38629 2017-02-13 Martin Liska <mliska@suse.cz>
38632 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38634 2017-02-13 Richard Biener <rguenther@suse.de>
38636 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38638 * configure: Re-generate.
38639 * config.in: Likewise.
38640 * graphite-dependences.c: Simplify as if
38641 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38642 * graphite-isl-ast-to-gimple.c: Likewise.
38643 * graphite-optimize-isl.c: Likewise.
38644 * graphite-poly.c: Likewise.
38645 * graphite-sese-to-poly.c: Likewise.
38646 * graphite.h: Likewise.
38647 * toplev.c: Include isl/version.h and use isl_version () for
38648 printing the ISL version.
38649 * doc/install.texi: Update ISL requirement.
38651 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38653 * doc/standards.texi (Standards): Update reference to
38656 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38658 * doc/extend.texi (Named Address Spaces): sourceware.org now
38660 * doc/install.texi (Binaries): Ditto.
38663 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38665 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38666 standard terminology "stringize"/"stringizing" throughout.
38667 * doc/cppinternals.texi: Likewise.
38669 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38671 * doc/extend.texi: Fix some spelling mistakes and typos.
38672 * doc/invoke.texi: Likewise.
38674 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38677 * params.def (inline-min-speedup) Change from 10 to 8.
38679 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38681 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38684 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38687 * ipa-inline-analysis.c (get_minimal_bb): New function.
38688 (record_modified): Use it.
38689 (remap_edge_change_prob): Handle also ancestor functions.
38691 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38693 * doc/contrib.texi (Contributors): Remove broken link into
38694 the Mauve CVS repository.
38696 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38698 PR middle-end/79454
38699 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38700 result computation whenever lhs doesn't have vector mode, not
38701 just when it has BLKmode.
38703 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38705 * doc/makefile.texi (profiledbootstrap): Refer to the
38706 installation instructions only in textual form.
38708 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38711 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38713 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38715 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38716 (Specific): Update mingw-w64 reference.
38718 (Specific): Remove broken link to Renesas RX processor.
38720 2017-02-10 Richard Biener <rguenther@suse.de>
38722 * toplev.c (process_options): Do not mention obsolete graphite
38723 options when printing sorry message about missing graphite support.
38724 Mention -floop-nest-optimize.
38726 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38728 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38729 (vtst_p16): Likewise.
38730 (vtstq_p8): Likewise.
38731 (vtstq_p16): Likewise.
38733 (vtstq_p64): Likewise.
38734 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38735 (vset_lane_p64): New.
38736 (vsetq_lane_p64): New.
38738 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38740 PR tree-optimization/79411
38741 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38742 stmt operands are SSA_NAMEs used in abnormal phis.
38743 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38746 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38749 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38752 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38754 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38756 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38758 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38762 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38763 not arbitrary TREE_CONSTANT.
38766 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38767 "omp declare target link" attribute unless is_global_var.
38768 * omp-offload.c (find_link_var_op): Likewise.
38770 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38771 Chung-Lin Tang <cltang@codesourcery.com>
38773 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38775 (gimplify_adjust_omp_clauses): Don't delete TILE.
38776 (gimplify_omp_for): Deal with TILE.
38777 * internal-fn.c (expand_GOACC_TILE): New function.
38778 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38780 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38781 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38783 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38784 avoid DIV for outermost collapse var.
38785 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38786 Remove out of date comments, fix whitespace.
38787 * omp-general.c (omp_extract_for_data): Deal with tiling.
38788 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38789 adjust OLF_DIM_BASE value.
38790 (struct omp_for_data): Add tiling field.
38791 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38792 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38793 for auto loops. Remove default auto determining, moved to
38794 oacc_loop_fixed_partitions.
38795 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38796 stmts, add e_mask field.
38797 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38798 (oacc_thread_numbers): Use oacc_dim_call.
38799 (oacc_xform_tile): New.
38800 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38801 (finish_oacc_loop): Adjust for ifns vector.
38802 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38803 add case for GOACC_TILE fns.
38804 (oacc_loop_xform_loop): Delete.
38805 (oacc_loop_process): Iterate over call list directly, and add
38806 handling for GOACC_TILE fns.
38807 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38809 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38810 vector partitioning to outer loops. Assign 2 partitions to loops
38811 when available. Add TILE handling.
38812 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38813 (execite_oacc_device_lower): Process GOACC_TILE fns,
38814 ignore unknown specs.
38815 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38816 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38817 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38819 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38821 * configure.ac (ACX_BUGURL): Update.
38822 * configure: Regenerate.
38824 2017-02-09 Richard Biener <rguenther@suse.de>
38826 PR tree-optimization/69823
38827 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38828 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38830 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38832 * config/arc/arc-c.def: Add __NPS400__ definition.
38833 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38834 (TARGET_NPS400): Define.
38836 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38838 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38840 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38841 pointer, arch_info.
38842 (arc_cpu_types): Fill the arch_info field with a pointer into the
38843 arc_arch_types table.
38844 (arc_selected_cpu): Declare.
38845 * config/arc/arc.c (arc_selected_cpu): Make global.
38846 (arc_selected_arch): Delete.
38847 (arc_base_cpu): Delete.
38848 (arc_override_options): Remove references to deleted variables,
38849 update access to arch information.
38850 (ARC_OPT): Update access to arch information.
38851 (ARC_OPTX): Likewise.
38852 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38853 (TARGET_ARC600): Update access to arch information.
38854 (TARGET_ARC601): Likewise.
38855 (TARGET_ARC700): Likewise.
38856 (TARGET_EM): Likewise.
38857 (TARGET_HS): Likewise.
38858 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38861 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38864 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38865 condition/operands for integer GE/LE/GEU/LEU operations.
38867 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38869 PR translation/79397
38870 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38873 2017-02-08 Martin Jambor <mjambor@suse.cz>
38876 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38877 whether allocation happened.
38878 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38879 nothing was allocated.
38881 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38883 PR tree-optimization/79408
38884 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38885 constant, but SSA_NAME with a known integer range, use the minimum
38886 of that range instead of op1 to determine if modulo can be replaced
38887 with its first operand.
38889 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38891 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38893 2017-02-08 Richard Biener <rguenther@suse.de>
38895 PR tree-optimization/71824
38896 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38897 Check all loops contained in the merged region.
38899 2017-02-07 Andrew Pinski <apinski@cavium.com>
38901 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38903 2017-02-07 Andrew Pinski <apinski@cavium.com>
38905 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38906 (thunderxt88): Likewise.
38907 (thunderxt81): Disable LSE and change v8.1 to v8.
38908 (thunderxt83): Likewise.
38910 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38911 Richard Biener <rguenther@suse.de>
38913 PR middle-end/79399
38914 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38915 type from int to size_t.
38916 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38918 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38920 PR rtl-optimization/79386
38921 * cprop.c (bypass_conditional_jumps): Initialize
38922 bypass_last_basic_block already before splitting bbs after
38923 unconditional traps...
38924 (bypass_conditional_jumps): ... rather than here.
38927 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38928 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38929 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38930 fix -masm=intel patterns.
38932 2017-02-07 Richard Biener <rguenther@suse.de>
38934 PR tree-optimization/79256
38935 PR middle-end/79278
38936 * builtins.c (get_object_alignment_2): Use min_align_of_type
38937 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38938 and ADJUST_FIELD_ALIGN.
38940 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38942 * doc/tm.texi: Regenerate.
38943 * stor-layout.c (layout_decl): Adjust.
38944 (update_alignment_for_field): Likewise.
38945 (place_field): Likewise.
38946 (min_align_of_type): Likewise.
38947 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38948 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38949 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38950 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38951 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38952 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38953 * config/i386/i386.c (x86_field_alignment): Likewise.
38954 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38955 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38956 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38957 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38958 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38959 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38963 2017-01-30 Richard Biener <rguenther@suse.de>
38965 PR tree-optimization/79256
38966 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38967 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38970 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38972 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38973 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38974 builtins to SImode and emit a zero-extend, if necessary.
38976 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38978 * docs/invoke.texi (RISC-V Options): Alphabetize.
38980 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38982 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38985 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38987 * config/riscv/riscv.c: New file.
38988 * common/config/riscv/riscv-common.c: Likewise.
38989 * config.gcc: Likewise.
38990 * config/riscv/constraints.md: Likewise.
38991 * config/riscv/elf.h: Likewise.
38992 * config/riscv/generic.md: Likewise.
38993 * config/riscv/linux.h: Likewise.
38994 * config/riscv/multilib-generator: Likewise.
38995 * config/riscv/peephole.md: Likewise.
38996 * config/riscv/pic.md: Likewise.
38997 * config/riscv/predicates.md: Likewise.
38998 * config/riscv/riscv-builtins.c: Likewise.
38999 * config/riscv/riscv-c.c: Likewise.
39000 * config/riscv/riscv-ftypes.def: Likewise.
39001 * config/riscv/riscv-modes.def: Likewise.
39002 * config/riscv/riscv-opts.h: Likewise.
39003 * config/riscv/riscv-protos.h: Likewise.
39004 * config/riscv/riscv.h: Likewise.
39005 * config/riscv/riscv.md: Likewise.
39006 * config/riscv/riscv.opt: Likewise.
39007 * config/riscv/sync.md: Likewise.
39008 * config/riscv/t-elf-multilib: Likewise.
39009 * config/riscv/t-linux: Likewise.
39010 * config/riscv/t-linux-multilib: Likewise.
39011 * config/riscv/t-riscv: Likewise.
39012 * configure.ac: Likewise.
39013 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
39014 Waterman as RISC-V maintainers.
39015 * doc/install.texi: Add RISC-V entries.
39016 * doc/invoke.texi: Add RISC-V options section.
39017 * doc/md.texi: Add RISC-V constraints section.
39018 * configure: Regenerated.
39020 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
39023 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
39024 false values to be constant vectors with all 0 or all 1 bits set.
39025 (vcondu<mode><mode>): Likewise.
39026 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
39028 (fpmask_comparison_operator): Update comment.
39029 (vecint_comparison_operator): New predicate.
39030 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
39031 vector conditionals when the true and false values are constant
39032 vectors with all 0 bits or all 1 bits set.
39034 2017-02-06 Martin Sebor <msebor@redhat.com>
39036 PR tree-optimization/79376
39037 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
39039 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
39041 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
39042 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
39043 to simplify split condition.
39045 2017-02-06 Jakub Jelinek <jakub@redhat.com>
39047 * omp-expand.c (oxpand_omp_atomic_fetch_op,
39048 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
39051 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
39053 PR rtl-optimization/68664
39054 * target.def (can_speculate_insn): New hook.
39055 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
39056 * doc/tm.texi: Regenerate.
39057 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
39058 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
39059 (rs6000_sched_can_speculate_insn): New function.
39061 2017-02-06 Jakub Jelinek <jakub@redhat.com>
39063 PR tree-optimization/79284
39064 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
39065 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
39066 vectorizable_mask_load_store, vectorizable_operation,
39067 vect_is_simple_cond, get_same_sized_vectype): Use it instead
39068 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
39069 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
39070 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
39071 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
39072 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
39073 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
39074 is_gimple_assign (stmt). Replace another such test with
39075 is_gimple_assign (stmt).
39077 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39080 * config/avr/avr.c (rtl-iter.h): Include it.
39081 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
39082 (avr_legitimate_combined_insn): ...and implementation.
39084 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
39086 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
39087 * config/s390/s390.c (s390_const_operand_ok)
39088 (s390_canonicalize_comparison, s390_extract_part)
39089 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
39090 (s390_contiguous_bitmask_p, s390_rtx_costs)
39091 (legitimize_pic_address): Likewise.
39092 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
39093 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
39094 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
39095 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
39096 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
39098 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39100 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
39101 REGNO($0) == REGNO($1).
39103 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39105 * config/s390/linux.h(SIZE_TYPE): Add comment.
39107 2017-02-06 Julian Brown <julian@codesourcery.com>
39108 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39109 Virendra Pathak <virendra.pathak@broadcom.com>
39111 * config/aarch64/aarch64-cores.def: Change the scheduler
39113 * config/aarch64/aarch64.md: Include thunderx2t99.md.
39114 * config/aarch64/thunderx2t99.md: New file.
39116 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39118 * doc/standards.texi (Go Language): Update link to language
39121 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39123 * tree-eh.c (lower_resx): Sanitize profile.
39124 (cleanup_empty_eh_move_lp): Likewise.
39126 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39129 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
39131 * cfgloopmanip.h (loop_version): Update prototype.
39132 * modulo-sched.c (sms_schedule): Update call of loop_version.
39133 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
39134 * tree-parloops.c (gen_parallel_loop): Likewise.
39135 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
39136 * tree-ssa-loop-split.c (split_loop): Likewise.
39137 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
39138 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
39140 2017-02-05 Martin Liska <mliska@suse.cz>
39143 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
39145 (print_operand_address): Initialize a struct to zero.
39147 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39149 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
39150 garbage collector only in textual form.
39152 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39154 * doc/extend.texi (x86 specific memory model extensions for
39155 transactional memory): Simplify a phrase.
39157 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
39160 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
39161 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
39162 (atomic_storedi_1): Likewise.
39164 2017-02-04 Jakub Jelinek <jakub@redhat.com>
39166 PR tree-optimization/79338
39167 * tree-parloops.c (gather_scalar_reductions): Don't call
39168 vect_analyze_loop_form for loop->inner before destroying loop's
39171 2017-02-03 Martin Sebor <msebor@redhat.com>
39173 PR tree-optimization/79327
39174 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
39175 when precision has resulted in leading zeros.
39176 (format_integer): Adjust the likely counter to assume an unknown
39177 argument that may be zero is non-zero.
39179 2017-02-03 Jason Merrill <jason@redhat.com>
39182 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
39183 avoid copying non-taken branch.
39185 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39187 PR tree-optimization/79340
39188 * tree-vect-loop.c (vectorizable_reduction): Release
39189 vec_defs elements after safe_splicing them into other vectors.
39192 PR tree-optimization/79327
39193 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
39194 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
39196 (format_integer): Use wide_int_to_tree instead of build_int_cst
39197 + to_?hwi. If argmin is NULL, just set argmin and argmax to
39198 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
39199 of shortest and longest sequence.
39201 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39203 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39204 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39206 2017-02-03 Walter Lee <walt@tilera.com>
39209 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39210 after initial stackframe link reg save.
39211 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39213 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39216 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39217 wu for stxssp alternative.
39219 2017-02-03 Martin Sebor <msebor@redhat.com>
39221 PR tree-optimization/79352
39222 * gimple-fold.c (get_range_strlen): Add argument.
39223 (get_range_strlen): Change return type to bool.
39224 (get_maxval_strlen): Pass in a dummy argument.
39225 * gimple-fold.h (get_range_strlen): Change return type to bool.
39226 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39227 * tree.h (array_at_struct_end_p): Add argument.
39228 * tree.c (array_at_struct_end_p): Handle it.
39230 2017-02-03 Martin Liska <mliska@suse.cz>
39233 * multiple_target.c (create_dispatcher_calls): Redirect edge
39234 from a caller of a dispatcher.
39235 (expand_target_clones): Make the clones local.
39236 (ipa_target_clone): Do both target clones and resolvers.
39237 (ipa_dispatcher_calls): Remove the pass.
39238 (pass_dispatcher_calls::gate): Likewise.
39239 (make_pass_dispatcher_calls): Likewise.
39240 * passes.def (pass_target_clone): Put as very first IPA early
39243 2017-02-03 Martin Liska <mliska@suse.cz>
39245 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39246 in case of a function with ifunc attribute.
39248 2017-02-03 Martin Liska <mliska@suse.cz>
39250 * cgraph.c (cgraph_node::dump): Dump function version info.
39251 * symtab.c (symtab_node::dump_base): Add missing new line.
39253 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39255 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39256 (ifcombine_ifandif): Use it.
39258 2017-02-03 Martin Liska <mliska@suse.cz>
39260 * doc/invoke.texi: Document default value for
39261 use-after-scope-direct-emission-threshold.
39263 2017-02-03 Martin Liska <mliska@suse.cz>
39265 PR tree-optimization/79339
39266 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39267 (format_floating): Likewise.
39269 2017-02-03 Martin Liska <mliska@suse.cz>
39272 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39274 (ipa_node_params_t::remove): Likewise.
39275 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39276 initialization from removed ipa_node_params_t::insert.
39277 (ipa_node_params::~ipa_node_params): Move from removed
39278 ipa_node_params_t::release.
39279 * symbol-summary.h (symbol_summary::m_released): New member.
39280 Do not release a summary twice. Do not allow to call finalizer
39281 for types of a summary that live in GGC memory.
39283 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39285 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39288 2017-02-02 Martin Sebor <msebor@redhat.com>
39290 PR middle-end/79275
39291 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39292 (format_string): Tighten up the range of output for non-constant
39293 strings and correct the expected range for wide non-constant strings.
39295 2017-02-02 Martin Sebor <msebor@redhat.com>
39297 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39299 PR middle-end/32003
39300 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39302 (-fdump-tree-@var): Add to index and document how to come up
39303 with pass-specific option and dump file names.
39304 (-fdump-passes): Clarify where to look for output.
39306 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39308 PR middle-end/77445
39309 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39310 statistics of the analyzed path; allow threading for speed when
39311 any of BBs along the path are optimized for speed.
39313 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39315 PR middle-end/78468
39316 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39317 settings of the virtual registers.
39320 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39322 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39323 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39326 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39328 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39329 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39331 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39333 * config/s390/s390.md: Add missing comments with the expanded
39335 * config/s390/vector.md: Likewise.
39336 * config/s390/vx-builtins.md: Likewise.
39338 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39341 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39342 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39343 conditions on a single line.
39345 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39347 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39348 __S390_VX__ to __VX__.
39350 2017-02-01 Andrew Pinski <apinski@cavium.com>
39352 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39353 stmt_info to record_stmt_cost.
39354 (vect_get_known_peeling_cost): Pass stmt_info if known to
39356 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39357 cpu_vector_cost field into
39358 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39359 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39360 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39361 splitting of scalar_stmt_cost and vec_stmt_cost.
39362 (thunderx_vector_cost): Likewise.
39363 (cortexa57_vector_cost): LIkewise.
39364 (exynosm1_vector_cost): Likewise.
39365 (xgene1_vector_cost): Likewise.
39366 (thunderx2t99_vector_cost): Improve after the splitting of the two
39368 (aarch64_builtin_vectorization_cost): Update for the splitting of
39369 scalar_stmt_cost and vec_stmt_cost.
39371 2017-02-01 Torvald Riegel <triegel@redhat.com>
39372 Richard Henderson <rth@redhat.com>
39374 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39375 conditional on existance of a fast atomic load.
39376 * optabs-query.c (can_atomic_load_p): New function.
39377 * optabs-query.h (can_atomic_load_p): Declare it.
39378 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39379 no fast atomic load is available for the particular size of access.
39380 (expand_atomic_compare_and_swap): Likewise.
39381 (expand_atomic_load): Likewise.
39382 (expand_atomic_store): Likewise.
39383 (expand_atomic_fetch_op): Likewise.
39384 * testsuite/lib/target-supports.exp
39385 (check_effective_target_sync_int_128): Remove x86 because it provides
39386 no fast atomic load.
39387 (check_effective_target_sync_int_128_runtime): Likewise.
39389 2017-02-01 Richard Biener <rguenther@suse.de>
39391 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39392 (graphite_transform_loops): Provide opt-info for optimized nests.
39393 * tree-parloop.c (parallelize_loops): Provide opt-info for
39394 parallelized loops.
39396 2017-02-01 Richard Biener <rguenther@suse.de>
39398 PR middle-end/79315
39399 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39400 was not set before.
39402 2017-02-01 Richard Biener <rguenther@suse.de>
39404 PR tree-optimization/71824
39405 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39406 Verify the loops are valid in the merged SESE region.
39407 (scop_detection::can_represent_loop_1): Check analyzing the
39408 evolution of the number of iterations in the region succeeds.
39410 2017-01-31 Ian Lance Taylor <iant@golang.org>
39412 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39413 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39415 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39417 PR preprocessor/79210
39418 * input.c (get_substring_ranges_for_loc): Replace line_width
39419 assertion with error-handling.
39421 2017-01-31 Richard Biener <rguenther@suse.de>
39423 PR tree-optimization/77318
39424 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39425 (create_pw_aff_from_tree): Take loop parameter.
39426 (add_condition_to_pbb): Pass loop of the condition to
39427 create_pw_aff_from_tree.
39429 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39431 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39432 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39434 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39438 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39440 (convert_int_to_float128): Likewise.
39441 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39442 (convert_int_to_float128): Likewise.
39443 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39444 (UNSPEC_IEEE128_CONVERT): Likewise.
39445 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39446 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39447 Use local variables for IBM extended format.
39448 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39449 (fix_trunc<mode>si2_fprs): Likewise.
39450 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39451 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39452 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39453 to know that we can now have integers of all sizes in vector
39455 (fix<uns>_<mode>di2_hw): Likewise.
39456 (float<uns>_<mode>si2_hw): Likewise.
39457 (fix_<mode>si2_hw): Likewise.
39458 (fixuns_<mode>si2_hw): Likewise.
39459 (float<uns>_<mode>di2_hw): Likewise.
39460 (float_<mode>di2_hw): Likewise.
39461 (float_<mode>si2_hw): Likewise.
39462 (floatuns_<mode>di2_hw): Likewise.
39463 (floatuns_<mode>si2_hw): Likewise.
39464 (xscvqp<su>wz_<mode>): Delete, no longer used.
39465 (xscvqp<su>dz_<mode>): Likewise.
39466 (xscv<su>dqp_<mode>): Likewise.
39467 (ieee128_mfvsrd_64bit): Likewise.
39468 (ieee128_mfvsrd_32bit): Likewise.
39469 (ieee128_mfvsrwz): Likewise.
39470 (ieee128_mtvsrw): Likewise.
39471 (ieee128_mtvsrd_64bit): Likewise.
39472 (ieee128_mtvsrd_32bit): Likewise.
39474 2017-01-31 Martin Liska <mliska@suse.cz>
39477 * ipa-prop.c (ipa_free_all_node_params): Call release method
39478 instead of ~sumbol_summary to not to trigger double times
39481 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39483 PR tree-optimization/71691
39484 * bitmap.h (class auto_bitmap): New.
39485 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39486 is_maybe_undefined instead of ssa_undefined_value_p.
39488 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39490 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39491 __S390_ARCH_LEVEL__ to __ARCH__.
39493 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39495 PR tree-optimization/79267
39496 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39497 if should_remove_lhs_p is true.
39499 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39502 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39503 (add_alignment_attribute): New.
39504 (base_type_die): Add alignment attribute.
39505 (subrange_type_die): Likewise.
39506 (modified_type_die): Likewise.
39507 (gen_array_type_die): Likewise.
39508 (gen_descr_array_type_die: Likewise.
39509 (gen_enumeration_type_die): Likewise.
39510 (gen_subprogram_die): Likewise.
39511 (gen_variable_die): Likewise.
39512 (gen_field_die): Likewise.
39513 (gen_ptr_to_mbr_type_die): Likewise.
39514 (gen_struct_or_union_type_die): Likewise.
39515 (gen_subroutine_type_die): Likewise.
39516 (gen_typedef_die): Likewise.
39517 (base_type_cmp): Compare alignment attribute.
39519 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39522 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39523 (setb_unsigned) New pattern for setb with CCUNS.
39524 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39525 subfc./subfe sequence to avoid overflow problems. Generate a
39526 shorter sequence with cmpld/setb for power9.
39527 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39528 for generating subfc. instruction.
39529 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39530 now uses this instruction.
39532 2017-01-30 Ian Lance Taylor <iant@google.com>
39535 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39536 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39538 2017-01-30 Martin Sebor <msebor@redhat.com>
39540 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39541 Move constant to the right of a relational operator.
39542 (get_mpfr_format_length, format_character, format_string): Ditto.
39543 (should_warn_p, maybe_warn): Same.
39545 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39547 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39550 * asan.c (get_translation_unit_decl): Remove function.
39551 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39553 2017-01-30 Martin Liska <mliska@suse.cz>
39555 PR gcov-profile/79259
39556 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39557 -fprofile-generate.
39559 2017-01-30 Martin Liska <mliska@suse.cz>
39562 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39563 Initialize variables with NULL value.
39565 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39568 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39570 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39572 2017-01-30 Richard Biener <rguenther@suse.de>
39574 PR tree-optimization/79276
39575 * tree-vrp.c (process_assert_insertions): Properly adjust common
39576 when removing a duplicate.
39578 2017-01-30 Richard Biener <rguenther@suse.de>
39580 PR tree-optimization/79256
39581 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39582 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39584 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39586 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39589 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39590 ("*r<noxa>sbg_<mode>_sll_bitmask")
39591 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39592 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39593 Use contiguous_bitmask_nowrap_operand.
39595 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39598 * config/rs6000/altivec.h (vec_xl): Revise #define.
39599 (vec_xst): Likewise.
39601 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39603 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39605 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39607 PR rtl-optimization/79194
39608 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39609 traps before call to bypass_conditional_jumps.
39611 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39613 PR tree-optimization/71374
39614 * lra-constraints.c (check_conflict_input_operands): New.
39615 (match_reload): Use it.
39617 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39620 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39621 account to calculate conflict_set.
39623 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39625 PR rtl-optimization/78559
39626 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39627 other_insn in combine.
39629 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39631 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39632 uint16_type_node for BT_UINT16.
39634 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39636 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39637 "RTL Tests" to menu.
39638 (GIMPLE Tests): New node.
39639 (RTL Tests): New node.
39641 2017-01-27 Richard Biener <rguenther@suse.de>
39643 PR tree-optimization/79245
39644 * tree-loop-distribution.c (distribute_loop): Apply cost
39645 modeling also to detected patterns.
39647 2017-01-27 Richard Biener <rguenther@suse.de>
39649 PR tree-optimization/71433
39650 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39651 (compare_assert_loc): New function.
39652 (process_assert_insertions): Sort and optimize assert locations
39653 to remove duplicates and push down identical assertions on
39654 edges to their destination block.
39656 2017-01-27 Richard Biener <rguenther@suse.de>
39658 PR tree-optimization/79244
39659 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39660 out SSA names even if abnormal.
39662 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39664 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39665 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39666 instead of MPFR_RNDN.
39668 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39671 * arm.c (arm_option_override): Don't call build_target_option_node
39672 until after doing all option overrides.
39673 (arm_valid_target_attribute_tree): Likewise.
39675 2017-01-27 Martin Liska <mliska@suse.cz>
39677 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39678 for {cd}tors and C++ {cd}tors.
39680 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39682 * config/s390/s390.md ("*setmem_long_and")
39683 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39685 2017-01-26 Martin Sebor <msebor@redhat.com>
39687 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39690 2017-01-26 Martin Sebor <msebor@redhat.com>
39692 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39693 HAVE_DFmode before using XFmode or DFmode.
39694 (parse_directive): Avoid using the z length modifier to avoid
39695 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39697 PR middle-end/78703
39698 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39699 to accept adjustment as an array.
39700 (get_int_range): New function.
39701 (struct directive): Make width and prec arrays.
39702 (directive::set_width, directive::set_precision): Call get_int_range.
39703 (format_integer, format_floating): Handle width and precision ranges.
39704 (format_string, parse_directive): Same.
39706 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39709 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39710 comdat_type_p set, just clone them, but keep the children in the
39714 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39715 which have direct callers with -fvar-tracking-assignments enabled
39717 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39718 inside of type units.
39720 2017-01-26 Martin Sebor <msebor@redhat.com>
39722 PR middle-end/78703
39723 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39725 (struct format_result): Replace number_chars, number_chars_min,
39726 and number_chars_max with a single member of struct result_range.
39728 (format_result::operator+=): Adjust.
39729 (struct fmtresult): Remove bounded. Handle likely and unlikely
39731 (fmtresult::adjust_for_width_or_precision): New function.
39732 (fmtresult:type_max_digits): New function.
39733 (bytes_remaining): Handle likely and unlikely counters.
39734 (min_bytes_remaining): Remove.
39735 (format_percent): Simplify.
39736 (format_integer, format_floating): Set likely and unlikely counters.
39737 (get_string_length, format_character, format_string): Same.
39738 (format_plain, should_warn_p): New function.
39739 (maybe_warn): Call should_warn_p. Update diagnostic messages
39740 and handle those for all directives, including plain strings.
39741 (format_directive): Handle likely and unlikely counters.
39742 Remove unnecessary quoting from diagnostics. Add an informational
39744 (add_bytes): Remove.
39745 (pass_sprintf_length::compute_format_length): Simplify.
39746 (try_substitute_return_value): Handle likely and unlikely counters.
39748 2017-01-26 Carl Love <cel@us.ibm.com>
39750 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39751 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39753 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39756 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39757 endianess for subregs into account.
39758 * lra-constraints.c (lra_constraints): Do risky transformations
39759 always on the first iteration.
39760 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39761 last_call_used_reg_set.
39762 (process_bb_lives): Define and use last_call_used_reg_set.
39763 * lra.c (lra): Always continue after lra_constraints on the first
39766 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39768 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39770 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39772 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39774 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39775 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39776 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39777 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39778 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39779 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39780 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39781 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39782 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39784 2017-01-26 Marek Polacek <polacek@redhat.com>
39787 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39788 for the third operand.
39790 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39792 PR middle-end/79236
39793 * omp-low.c (struct omp_context): Add simt_stmt field.
39794 (scan_omp_for): Return omp_context *.
39795 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39796 context to the _simt_ SIMD stmt.
39797 (lower_omp_for): For combined SIMD with sibling _simt_
39798 SIMD, make sure to use the same decls in _looptemp_
39799 clauses as in the sibling.
39801 2017-01-26 David Sherwood <david.sherwood@arm.com>
39803 PR middle-end/79212
39804 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39807 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39810 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39811 emit fld b; fld a; if possible.
39813 * brig-builtins.def: Update copyright years.
39814 * config/arm/arm_acle_builtins.def: Update copyright years.
39816 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39819 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39820 constraint instead of o for the stxsd instruction.
39822 2017-01-25 Carl Love <cel@us.ibm.com>
39824 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39825 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39827 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39829 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39831 2017-01-25 Richard Biener <rguenther@suse.de>
39833 PR tree-optimization/69264
39834 * target.def (vector_alignment_reachable): Improve documentation.
39835 * doc/tm.texi: Regenerate.
39836 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39838 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39839 earlier changes with respect to TYPE_USER_ALIGN.
39840 (vector_alignment_reachable_p): Likewise. Improve dumping.
39842 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39845 * config/arm/arm.md (xordi3): Force constant operand into a register
39848 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39850 * doc/invoke.texi (-fstore-merging): Correct default optimization
39851 levels at which it is enabled.
39852 (-O): Move -fstore-merging from list to...
39855 2017-01-25 Richard Biener <rguenther@suse.de>
39858 * omp-expand.c: Include debug.h.
39859 (expand_omp_taskreg): Make sure to generate early debug before
39860 outlining anything from a function.
39861 (expand_omp_target): Likewise.
39862 (grid_expand_target_grid_body): Likewise.
39864 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39867 * asan.c (get_translation_unit_decl): New function.
39868 (asan_add_global): Extract modules file name from globals
39869 TRANSLATION_UNIT_DECL name.
39871 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39874 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39875 for long calls with APCS frame and VFP.
39877 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39879 * cfg.c (original_copy_tables_initialized_p): New function.
39880 * cfg.h (original_copy_tables_initialized_p): New decl.
39881 * cfgrtl.c (relink_block_chain): Guard the call to
39882 free_original_copy_tables with a call to
39883 original_copy_tables_initialized_p.
39884 * cgraph.h (symtab_node::native_rtl_p): New decl.
39885 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39886 (symtab_node::needed_p): Don't assert for early assembly output
39887 for __RTL functions.
39888 (cgraph_node::finalize_function): Set "force_output" for __RTL
39890 (cgraph_node::analyze): Bail out early for __RTL functions.
39891 (analyze_functions): Update assertion to support __RTL functions.
39892 (cgraph_node::expand): Bail out early for __RTL functions.
39893 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39895 * function.h (struct function): Update comment for field
39897 * gimple-expr.c: Include "tree-pass.h".
39898 (gimple_has_body_p): Return false for __RTL functions.
39899 * Makefile.in (OBJS): Add run-rtl-passes.o.
39900 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39902 (gcc::pass_manager::get_clean_slate): New accessor.
39903 * passes.c: Include "insn-addr.h".
39904 (should_skip_pass_p): Add logging. Update logic for running
39905 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39906 property-provider override so it is only done for gimple passes.
39908 (skip_pass): New function.
39909 (execute_one_pass): Call skip_pass when skipping passes.
39910 * read-md.c (md_reader::read_char): Support filtering
39911 the input to a subset of line numbers.
39912 (md_reader::md_reader): Initialize fields
39913 m_first_line and m_last_line.
39914 (md_reader::read_file_fragment): New function.
39915 * read-md.h (md_reader::read_file_fragment): New decl.
39916 (md_reader::m_first_line): New field.
39917 (md_reader::m_last_line): New field.
39918 * read-rtl-function.c (function_reader::create_function): Only
39919 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39920 curr_properties. Set DECL_INITIAL to a dummy block.
39921 (read_rtl_function_body_from_file_range): New function.
39922 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39924 * run-rtl-passes.c: New file.
39925 * run-rtl-passes.h: New file.
39927 2017-01-24 Jeff Law <law@redhat.com>
39929 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39932 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39934 PR tree-optimization/79159
39935 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39936 (record_nonwrapping_iv): Improve boundary using above function if no
39937 value range information.
39939 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39940 Martin Jambor <mjambor@suse.cz>
39942 * brig-builtins.def: New file.
39943 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39944 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39945 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39946 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39947 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39948 * builtin-types.def (BT_INT8): New.
39949 (BT_INT16): Likewise.
39950 (BT_UINT8): Likewise.
39951 (BT_UINT16): Likewise.
39952 (BT_FN_ULONG): Likewise.
39953 (BT_FN_UINT_INT): Likewise.
39954 (BT_FN_UINT_ULONG): Likewise.
39955 (BT_FN_UINT_LONG): Likewise.
39956 (BT_FN_UINT_PTR): Likewise.
39957 (BT_FN_ULONG_PTR): Likewise.
39958 (BT_FN_INT8_FLOAT): Likewise.
39959 (BT_FN_INT16_FLOAT): Likewise.
39960 (BT_FN_UINT32_FLOAT): Likewise.
39961 (BT_FN_UINT16_FLOAT): Likewise.
39962 (BT_FN_UINT8_FLOAT): Likewise.
39963 (BT_FN_UINT64_FLOAT): Likewise.
39964 (BT_FN_UINT16_UINT32): Likewise.
39965 (BT_FN_UINT32_UINT16): Likewise.
39966 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39967 (BT_FN_INT_PTR_INT): Likewise.
39968 (BT_FN_UINT_PTR_UINT): Likewise.
39969 (BT_FN_LONG_PTR_LONG): Likewise.
39970 (BT_FN_ULONG_PTR_ULONG): Likewise.
39971 (BT_FN_VOID_UINT64_UINT64): Likewise.
39972 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39973 (BT_FN_INT8_INT8_INT8): Likewise.
39974 (BT_FN_INT16_INT16_INT16): Likewise.
39975 (BT_FN_INT_INT_INT): Likewise.
39976 (BT_FN_UINT_FLOAT_UINT): Likewise.
39977 (BT_FN_FLOAT_UINT_UINT): Likewise.
39978 (BT_FN_ULONG_UINT_UINT): Likewise.
39979 (BT_FN_ULONG_UINT_PTR): Likewise.
39980 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39981 (BT_FN_UINT_UINT_UINT): Likewise.
39982 (BT_FN_VOID_UINT_PTR): Likewise.
39983 (BT_FN_UINT_UINT_PTR: Likewise.
39984 (BT_FN_UINT32_UINT64_PTR): Likewise.
39985 (BT_FN_INT_INT_UINT_UINT): Likewise.
39986 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39987 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39988 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39989 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39990 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39991 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39992 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39993 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39994 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39995 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39996 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39997 * doc/frontends.texi: List BRIG FE.
39998 * doc/install.texi (Testing): Add BRIG tesring requirements.
39999 * doc/invoke.texi (Overall Options): Mention BRIG.
40000 * doc/standards.texi (Standards): Doucment BRIG HSA version.
40002 2017-01-24 Richard Biener <rguenther@suse.de>
40004 PR translation/79208
40005 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
40007 2017-01-24 Martin Jambor <mjambor@suse.cz>
40010 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
40011 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
40012 and known_contexts.
40014 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
40016 PR middle-end/79123
40017 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
40018 casts from signed to unsigned really don't have a range.
40020 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
40022 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
40023 GMP_RNDx for compatiblity.
40025 2017-01-24 Martin Liska <mliska@suse.cz>
40028 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
40029 that would prevent us to call alloca with -1 as argument.
40031 2017-01-24 Jakub Jelinek <jakub@redhat.com>
40033 * dwarf2out.c (output_compilation_unit_header, output_file_names):
40034 Avoid -Wformat-security warning.
40036 2017-01-23 Andrew Pinski <apinski@cavium.com>
40038 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
40041 2017-01-23 Martin Sebor <msebor@redhat.com>
40043 PR middle-end/78703
40044 * gimple-ssa-sprintf.c (warn_level): New global.
40045 (format_integer): Use it here and throughout the rest of the file.
40046 Use the same switch to compute sign as base.
40047 (maybe_warn): New function.
40048 (format_directive): Factor out warnings into maybe_warn.
40049 Add debugging output. Use warn_level.
40050 (add_bytes): Use warn_level.
40051 (pass_sprintf_length::compute_format_length): Add debugging output.
40052 (try_substitute_return_value): Same.
40053 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
40055 PR middle-end/78703
40056 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
40057 (struct fmtresult, format_integer, format_floating): Adjust.
40058 (fmtresult::fmtresult): Set max correctly in two argument ctor.
40059 (get_string_length, format_string,format_directive): Same.
40060 (pass_sprintf_length::compute_format_length): Same.
40061 (try_substitute_return_value): Simplify slightly.
40063 PR middle-end/78703
40064 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
40065 (fmtresult::operator+=): Outlined.
40066 (struct fmtresult): Add ctors.
40067 (struct conversion_spec): Rename...
40068 (struct directive): ...to this. Add and remove data members.
40069 (directive::set_width, directive::set_precision): New functions.
40070 (format_percent): Use fmtresult ctor.
40071 (get_width_and_precision): Remove.
40072 (format_integer): Make naming changes. Avoid computing width and
40074 (format_floating): Same. Adjust indentation.
40075 (format_character, format_none): New functions.
40076 (format_string): Moved character handling to format_character.
40077 (format_directive): Remove arguments, change return type.
40078 (parse_directive): New function.
40079 (pass_sprintf_length::compute_format_length): Move directive
40080 parsing to parse_directive.
40082 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40084 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
40085 (assign_assembler_name_if_needed): ... this.
40086 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
40087 (assign_assembler_name_if_needed): ... this.
40088 (free_lang_data_in_cgraph): Adjust callers.
40089 * cgraphunit.c (cgraph_node::analyze): Likewise.
40090 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
40093 2017-01-23 Richard Biener <rguenther@suse.de>
40095 PR tree-optimization/79088
40096 PR tree-optimization/79188
40097 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
40098 resetting loop bounds after last path deletion. Reset loop
40099 bounds of the target loop, make code match the comments.
40100 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
40101 Make sure loops need no fixups.
40103 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
40105 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
40106 exponent support with double type for first argument.
40107 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
40108 type returned by __builtin_vec_extract_sig,
40109 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
40110 functions from "vector int" to "vector unsigned int" or from
40111 "vector long long int" to "vector unsigned long long int".
40112 Changed type returned by __builtin_vec_extract_exp,
40113 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
40114 functions from "vector int" to "vector unsigned int" or from
40115 "vector long long int" to "vector unsigned long long int".
40116 Changed return type of __builtin_vec_test_data_class,
40117 __builtin_vec_test_data_class_sp, and
40118 __builtin_vec_test_data_class_dp from "vector int" to
40119 "vector bool int" or from "vector long long int" to "vector bool
40120 long long int" and changed second argument type from "unsigned
40121 int" to "int". Added new overloaded function forms "vector float
40122 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
40123 "vector float __builtin_vec_insert_exp_sp (vector float, vector
40124 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
40125 double, vector unsigned long long int)" and "vector double
40126 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
40127 long int)". Changed return type of
40128 __builtin_scalar_test_data_class and
40129 __builtin_scalar_test_data_class_sp and
40130 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
40131 int" and changed second argument from "unsigned int" to "int".
40132 Changed type returned by __builtin_scalar_test_neg,
40133 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
40134 from "int" to "bool int". Added new overloaded function form
40135 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
40136 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
40137 exponent double-precision with floating point first argument.
40138 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
40139 documentation of scalar_test_data_class, scalar_test_neg,
40140 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
40141 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
40142 vec_test_data_class built-in functions to reflect refinements in
40143 their type signatures.
40145 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
40147 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
40149 (aarch64_elf_asm_destructor): Likewise.
40151 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
40153 PR rtl-optimization/78634
40154 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
40155 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
40156 * ifcvt.c (noce_try_cmove): Add missing cost check.
40158 PR rtl-optimization/71724
40159 * combine.c (if_then_else_cond): Look for situations where it is
40160 beneficial to undo the work of one of the recursive calls.
40162 2017-01-23 Bin Cheng <bin.cheng@arm.com>
40164 PR tree-optimization/70754
40165 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
40166 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
40167 combined stmt before it if not NULL.
40168 (combine_chains): Process refs reversely and compute dominance point
40171 2017-01-23 Martin Liska <mliska@suse.cz>
40173 PR tree-optimization/79196
40174 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
40175 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
40177 (strlen_optimize_stmt): Call the renamed function.
40179 2017-01-23 Michael Matz <matz@suse.de>
40181 PR tree-optimization/78384
40182 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
40184 2017-01-23 Richard Biener <rguenther@suse.de>
40186 PR tree-optimization/79186
40187 * tree-vrp.c (register_new_assert_for): Make sure we've seen
40188 both incoming edges before moving an assert.
40190 2017-01-23 Martin Jambor <mjambor@suse.cz>
40192 * ipa-prop.c (load_from_param_1): Removed.
40193 (load_from_unmodified_param): Bits from load_from_param_1 put back
40195 (load_from_param): Removed.
40196 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
40197 with stmt. Reverted back to use of load_from_unmodified_param.
40199 2017-01-23 Martin Jambor <mjambor@suse.cz>
40202 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40203 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40204 field a pointer to garbage collected vector, mark lattices and
40205 ipcp_orig_node with GTY((skip)).
40206 (ipa_get_param_count): Adjust to descriptors being a pointer.
40207 (ipa_get_param): Likewise.
40208 (ipa_get_type): Likewise.
40209 (ipa_get_param_move_cost): Likewise.
40210 (ipa_set_param_used): Likewise.
40211 (ipa_get_controlled_uses): Likewise.
40212 (ipa_set_controlled_uses): Likewise.
40213 (ipa_is_param_used): Likewise.
40214 (ipa_node_params_t): Move into garbage collector. New methods insert
40216 (ipa_node_params_sum): Annotate wth GTY(()).
40217 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40219 (ipa_load_from_parm_agg): Adjust declaration.
40220 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40221 * ipa-profile.c (ipa_profile): Likewise.
40222 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40223 (ipa_populate_param_decls): Make descriptors parameter garbage
40225 (ipa_dump_param): Adjust to descriptors being a pointer.
40226 (ipa_alloc_node_params): Likewise.
40227 (ipa_initialize_node_params): Likewise.
40228 (load_from_param_1): Make descriptors parameter garbage collected.
40229 (load_from_unmodified_param): Likewise.
40230 (load_from_param): Likewise.
40231 (ipa_load_from_parm_agg): Likewise.
40232 (ipa_node_params::~ipa_node_params): Removed.
40233 (ipa_free_all_node_params): Remove call to delete operator.
40234 (ipa_node_params_t::insert): New.
40235 (ipa_node_params_t::remove): Likewise.
40236 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40237 copy known_csts and known_contexts vectors.
40238 (ipa_read_node_info): Adjust to descriptors being a pointer.
40239 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40241 (ipcp_transform_function): Make descriptors variable garbage
40244 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40246 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40247 * config/i386/avx512dqintrin.h: Ditto.
40248 * config/i386/avx512fintrin.h: Ditto.
40249 * config/i386/i386.c: Handle new builtins.
40250 * config/i386/i386-builtin.def: Add new builtins.
40251 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40252 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40254 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40255 Martin Liska <mliska@suse.cz>
40257 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40258 * asan.c (asan_expand_poison_ifn): Support stores and use
40259 appropriate ASAN report function.
40260 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40261 * internal-fn.def (ASAN_POISON_USE): Declare.
40262 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40263 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40264 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40265 ASAN_POISON calls w/o LHS.
40266 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40267 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40268 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40269 * gimplify.c (asan_poison_variables): Add attribute
40270 use_after_scope_memory to variables that really needs to live
40272 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40273 having the attribute.
40275 2017-01-23 Martin Liska <mliska@suse.cz>
40277 * asan.c (create_asan_shadow_var): New function.
40278 (asan_expand_poison_ifn): Likewise.
40279 * asan.h (asan_expand_poison_ifn): New declaration.
40280 * internal-fn.c (expand_ASAN_POISON): Likewise.
40281 * internal-fn.def (ASAN_POISON): New builtin.
40282 * sanopt.c (pass_sanopt::execute): Expand
40283 asan_expand_poison_ifn.
40284 * tree-inline.c (copy_decl_for_dup_finish): Make function
40286 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40287 * tree-ssa.c (is_asan_mark_p): New function.
40288 (execute_update_addresses_taken): Rewrite local variables
40289 (identified just by use-after-scope as addressable) into SSA.
40291 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40293 * doc/install.texi (Specific): opensource.apple.com uses https
40294 now. Remove trailing slash.
40296 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40298 * README.Portability: Remove note on an Irix compatibility issue.
40300 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40302 * gcov.c (INCLUDE_ALGORITHM): Define.
40303 (INCLUDE_VECTOR): Define.
40304 No longer include <vector> and <algorithm> directly.
40306 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40308 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40310 * doc/invoke.texi (Code Gen Options): Ditto.
40312 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40315 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40317 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40319 rtl-optimization/79125
40320 * cprop.c (local_cprop_pass): Handle cases where we make an
40321 unconditional trap.
40323 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40327 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40328 read from, for big endian.
40330 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40332 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40333 register pauth builtins for LP64 only.
40335 2017-01-20 Marek Polacek <polacek@redhat.com>
40338 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40341 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40343 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40345 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40346 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40347 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40349 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40352 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40353 in big-endian mode when they are not a single duplicated value.
40355 2017-01-20 Richard Biener <rguenther@suse.de>
40357 * BASE-VER: Bump to 7.0.1.
40359 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40361 * omp-low.c (omplow_simd_context): New struct. Use it...
40362 (lower_rec_simd_input_clauses): ...here and...
40363 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40364 references to idx, lane, max_vf, is_simt.
40366 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40368 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40371 2017-01-20 Martin Jambor <mjambor@suse.cz>
40373 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40374 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40376 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40377 (GTFILES): Rename hsa.c to hsa-common.c.
40378 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40379 * hsa-dump.c: Likewise.
40380 * hsa-gen.c: Likewise.
40381 * hsa-regalloc.c: Likewise.
40382 * ipa-hsa.c: Likewise.
40383 * omp-expand.c: Likewise.
40384 * omp-low.c: Likewise.
40385 * toplev.c: Likewise.
40387 2017-01-20 Marek Polacek <polacek@redhat.com>
40390 * doc/invoke.texi: Document -Wduplicated-branches.
40391 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40392 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40393 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40394 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40395 return 0 only when not OEP_LEXICOGRAPHIC.
40396 (fold_build_cleanup_point_expr): Use the expression
40397 location when building CLEANUP_POINT_EXPR.
40398 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40399 * tree.c (add_expr): Handle error_mark_node.
40401 2017-01-20 Martin Liska <mliska@suse.cz>
40404 * tree-profile.c (init_ic_make_global_vars): Do not call
40406 (gimple_init_gcov_profiler): Likewise.
40408 2017-01-20 Martin Liska <mliska@suse.cz>
40411 * cgraph.h (maybe_create_reference): Remove argument and
40413 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40415 * ipa-cp.c (create_specialized_node): Likewise.
40416 * symtab.c (symtab_node::maybe_create_reference): Handle
40417 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40419 2017-01-20 Martin Liska <mliska@suse.cz>
40421 * read-rtl-function.c (function_reader::create_function): Use
40422 build_decl instread of build_decl_stat.
40424 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40426 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40427 * config/i386/avx512dqintrin.h: Ditto.
40428 * config/i386/avx512fintrin.h: Ditto.
40429 * config/i386/i386-builtin-types.def: Add new types.
40430 * config/i386/i386.c: Handle new types.
40431 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40432 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40433 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40434 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40435 (__builtin_ia32_kshiftridi): New.
40436 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40438 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40442 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40443 define to rs6000_init_stack_protect_guard.
40444 (rs6000_init_stack_protect_guard): New function.
40446 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40447 Yunqiang Su <yunqiang.su@imgtec.com>
40449 * config.gcc (supported_defaults): Add madd4.
40450 (with_madd4): Add validation.
40451 (all_defaults): Add madd4.
40452 * config/mips/mips.opt (mmadd4): New option.
40453 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40455 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40457 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40458 (ISA_HAS_FUSED_MADD4): Likewise.
40459 * doc/invoke.texi (-mmadd4): Document the new option.
40460 * doc/install.texi (--with-madd4): Document the new option.
40462 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40464 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40465 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40466 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40467 (aarch64_init_pauth_hint_builtins): New.
40468 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40469 (aarch64_expand_builtin): Expand new builtins.
40471 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40473 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40474 * combine-stack-adj.c (no_unhandled_cfa): Handle
40475 REG_CFA_TOGGLE_RA_MANGLE.
40476 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40477 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40478 info for return address signing.
40479 (aarch64_expand_epilogue): Likewise.
40481 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40483 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40484 * config/aarch64/aarch64-protos.h
40485 (aarch64_return_address_signing_enabled): New declaration.
40486 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40488 (aarch64_expand_prologue): Sign return address before it's pushed onto
40490 (aarch64_expand_epilogue): Authenticate return address fetched from
40492 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40493 (aarch64_attributes): New function attributes for "sign-return-address".
40494 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40495 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40496 ("*do_return"): Generate combined instructions according to key index.
40497 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40498 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40500 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40501 * config/aarch64/aarch64.opt (msign-return-address=): New.
40502 * doc/extend.texi (AArch64 Function Attributes): Documents
40503 "sign-return-address=".
40504 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40506 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40508 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40509 overall option summary.
40511 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40513 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40514 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40515 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40516 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40518 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40520 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40521 -mpower9-minmax by default for -mcpu=power9.
40522 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40523 128-bit floating point.
40525 2017-01-20 Alan Modra <amodra@gmail.com>
40527 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40528 optimizing for size.
40530 2017-01-20 Alan Modra <amodra@gmail.com>
40533 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40534 for strcmp and strncmp from corresponding builtin decl.
40536 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40538 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40539 instead of i386/rtems-64.h.
40540 * config/i386/rtems-64.h: Remove.
40542 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40546 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40548 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40550 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40552 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40553 Change int to HOST_WIDE_INT.
40554 * config/aarch64/aarch64-protos.h
40555 (aarch64_simd_gen_const_vector_dup): Likewise.
40556 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40558 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40560 * langhooks-def.h (lhd_type_for_size): New decl.
40561 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40562 * langhooks.c (lhd_type_for_size): New function, taken from
40565 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40567 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40568 define_bypass for CR latency.
40569 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40570 (power9-alu2): Add define_bypass for CR latency.
40572 (power9-mul): Update insn latency.
40573 (power9-mul-compare): Update insn latency, bypass latency and remove
40576 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40578 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40580 * config/aarch64/aarch64.md
40581 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40582 aarch64_nopcrelative_literal_loads.
40583 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40585 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40587 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40588 TARGET_LOONGSON_3A.
40589 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40591 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40594 * config.gcc (supported_defaults): Add lxc1-sxc1.
40595 (with_lxc1_sxc1): Add validation.
40596 (all_defaults): Add lxc1-sxc1.
40597 * config/mips/mips.opt (mlxc1-sxc1): New option.
40598 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40600 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40601 __mips_no_lxc1_sxc1.
40602 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40603 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40604 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40606 2017-01-19 Richard Biener <rguenther@suse.de>
40608 PR tree-optimization/72488
40609 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40610 sure to restore SSA info.
40611 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40613 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40615 PR rtl-optimization/79121
40616 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40617 of the inner type when shifting an extended value.
40619 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40622 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40623 interposable aliases.
40625 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40628 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40629 Use the evmergelohi instruction.
40630 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40631 (mov_sitf_e500_subreg8_2_be): Likewise.
40632 (mov_sitf_e500_subreg12_2_le): Likewise.
40633 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40634 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40635 (mov_sitf_e500_subreg8_2_le): Likewise.
40636 (mov_sitf_e500_subreg12_2_be): Likewise.
40638 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40640 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40641 attribute from vecsimple to vecperm.
40642 (altivec_vbpermq2): Likewise.
40644 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40647 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40649 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40650 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40651 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40652 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40653 case where N arg is SIZE_MAX.
40654 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40655 (cmpstrsi): Add pattern.
40657 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40659 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40660 __builtin_vec_revb builtins.
40661 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40662 built-in functions to support generation of the ISA 3.0 XXBR<x>
40663 vector byte reverse instructions.
40664 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40665 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40666 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40667 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40668 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40669 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40670 (P9V_BUILTIN_VEC_REVB): Likewise.
40671 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40672 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40673 (p9_xxbrq_v16qi): Likewise.
40674 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40675 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40676 (p9_xxbrh_v8hi): Likewise.
40677 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40678 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40679 vec_revb built-in functions.
40681 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40683 PR rtl-optimization/78952
40684 * config/i386/i386.md (any_extract): New code iterator.
40685 (*insvqi_2): Use any_extract for source operand.
40686 (*insvqi_3): Use any_shiftrt for source operand.
40688 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40690 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40692 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40694 2017-01-18 Matthias Klose <doko@ubuntu.com>
40696 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40698 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40700 * config/rs6000/altivec.h (vec_bperm): Change #define.
40701 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40702 (altivec_vbpermq2): New define_insn.
40703 (altivec_vbpermd): Likewise.
40704 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40705 function interface.
40706 (VBPERMD): Likewise.
40707 (VBPERM): New polymorphic function interface.
40708 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40709 Add entries for P9V_BUILTIN_VEC_VBPERM.
40710 * doc/extend.texi: Add interfaces for vec_bperm.
40712 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40714 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40715 first letter of error messages.
40716 (s390_resolve_overloaded_builtin): Likewise.
40717 * config/s390/s390.c (s390_expand_builtin): Likewise.
40718 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40719 (s390_valid_target_attribute_inner_p): Likewise.
40720 * config/s390/s390.md ("tabort"): Likewise.
40722 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40724 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40725 (ISA_AVOID_DIV_HILO): New macro.
40726 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40727 (ISA_HAS_DDIV): Likewise.
40729 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40731 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40733 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40735 * doc/invoke.texi (fabi-version): Spelling fix.
40737 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40740 * doc/invoke.texi (fabi-version): Mention mangling fix for
40743 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40746 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40748 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40751 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40752 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40754 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40756 (stack_protect_test): Ditto.
40757 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40758 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40760 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40761 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40762 -mstack-protector-guard-offset=.
40763 (RS/6000 and PowerPC Options): Ditto.
40765 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40767 * config/i386/i386.h (MASK_CLASS_P): New define.
40768 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40769 there are no registers from different register sets also when
40770 mask registers are used. Update function comment.
40771 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40772 to (*k/*r) and (*k/*km) alternatives.
40774 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40776 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40777 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40778 (EH_RETURN_HANDLER_RTX): New define.
40779 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40780 Force frame pointer in EH return functions.
40781 (aarch64_expand_epilogue): Add barrier for eh_return.
40782 (aarch64_final_eh_return_addr): Remove.
40783 (aarch64_eh_return_handler_rtx): New function.
40784 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40786 (aarch64_eh_return_handler_rtx): New prototype.
40788 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40790 * config/rs6000/altivec.h (vec_rlmi): New #define.
40791 (vec_vrlnm): Likewise.
40792 (vec_rlnm): Likewise.
40793 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40794 (UNSPEC_VRLNM): Likewise.
40795 (VIlong): New mode iterator.
40796 (altivec_vrl<VI_char>mi): New define_insn.
40797 (altivec_vrl<VI_char>nm): Likewise.
40798 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40800 (VRLDNM): Likewise.
40801 (RLNM): New polymorphic function entry.
40802 (VRLWMI): New monomorphic function entry.
40803 (VRLDMI): Likewise.
40804 (RLMI): New polymorphic function entry.
40805 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40806 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40807 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40810 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40813 * dwarf2out.c (field_byte_offset): Restore the
40814 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40815 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40816 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40819 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40822 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40824 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40827 * dwarf2out.c (add_data_member_location_attribute): For constant
40828 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40829 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40830 DW_AT_byte_size attributes.
40832 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40834 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40835 after forcing to constant memory when the code model is medium.
40837 2017-01-17 Julia Koval <julia.koval@intel.com>
40840 * config/i386/avx512fintrin.h
40841 (_mm512_i32gather_ps): Change __addr type to void const*.
40842 (_mm512_mask_i32gather_ps): Ditto.
40843 (_mm512_i32gather_pd): Ditto.
40844 (_mm512_mask_i32gather_pd): Ditto.
40845 (_mm512_i64gather_ps): Ditto.
40846 (_mm512_mask_i64gather_ps): Ditto.
40847 (_mm512_i64gather_pd): Ditto.
40848 (_mm512_mask_i64gather_pd): Ditto.
40849 (_mm512_i32gather_epi32): Ditto.
40850 (_mm512_mask_i32gather_epi32): Ditto.
40851 (_mm512_i32gather_epi64): Ditto.
40852 (_mm512_mask_i32gather_epi64): Ditto.
40853 (_mm512_i64gather_epi32): Ditto.
40854 (_mm512_mask_i64gather_epi32): Ditto.
40855 (_mm512_i64gather_epi64): Ditto.
40856 (_mm512_mask_i64gather_epi64): Ditto.
40857 (_mm512_i32scatter_ps): Change __addr type to void*.
40858 (_mm512_mask_i32scatter_ps): Ditto.
40859 (_mm512_i32scatter_pd): Ditto.
40860 (_mm512_mask_i32scatter_pd): Ditto.
40861 (_mm512_i64scatter_ps): Ditto.
40862 (_mm512_mask_i64scatter_ps): Ditto.
40863 (_mm512_i64scatter_pd): Ditto.
40864 (_mm512_mask_i64scatter_pd): Ditto.
40865 (_mm512_i32scatter_epi32): Ditto.
40866 (_mm512_mask_i32scatter_epi32): Ditto.
40867 (_mm512_i32scatter_epi64): Ditto.
40868 (_mm512_mask_i32scatter_epi64): Ditto.
40869 (_mm512_i64scatter_epi32): Ditto.
40870 (_mm512_mask_i64scatter_epi32): Ditto.
40871 (_mm512_i64scatter_epi64): Ditto.
40872 (_mm512_mask_i64scatter_epi64): Ditto.
40873 * config/i386/avx512pfintrin.h
40874 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40875 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40876 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40877 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40878 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40879 (_mm512_prefetch_i32scatter_ps): Ditto.
40880 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40881 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40882 (_mm512_prefetch_i64scatter_pd): Ditto.
40883 (_mm512_prefetch_i64scatter_ps): Ditto.
40884 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40885 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40886 * config/i386/avx512vlintrin.h
40887 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40888 (_mm_mmask_i32gather_ps): Ditto.
40889 (_mm256_mmask_i32gather_pd): Ditto.
40890 (_mm_mmask_i32gather_pd): Ditto.
40891 (_mm256_mmask_i64gather_ps): Ditto.
40892 (_mm_mmask_i64gather_ps): Ditto.
40893 (_mm256_mmask_i64gather_pd): Ditto.
40894 (_mm_mmask_i64gather_pd): Ditto.
40895 (_mm256_mmask_i32gather_epi32): Ditto.
40896 (_mm_mmask_i32gather_epi32): Ditto.
40897 (_mm256_mmask_i32gather_epi64): Ditto.
40898 (_mm_mmask_i32gather_epi64): Ditto.
40899 (_mm256_mmask_i64gather_epi32): Ditto.
40900 (_mm_mmask_i64gather_epi32): Ditto.
40901 (_mm256_mmask_i64gather_epi64): Ditto.
40902 (_mm_mmask_i64gather_epi64): Ditto.
40903 (_mm256_i32scatter_ps): Change __addr type to void*.
40904 (_mm256_mask_i32scatter_ps): Ditto.
40905 (_mm_i32scatter_ps): Ditto.
40906 (_mm_mask_i32scatter_ps): Ditto.
40907 (_mm256_i32scatter_pd): Ditto.
40908 (_mm256_mask_i32scatter_pd): Ditto.
40909 (_mm_i32scatter_pd): Ditto.
40910 (_mm_mask_i32scatter_pd): Ditto.
40911 (_mm256_i64scatter_ps): Ditto.
40912 (_mm256_mask_i64scatter_ps): Ditto.
40913 (_mm_i64scatter_ps): Ditto.
40914 (_mm_mask_i64scatter_ps): Ditto.
40915 (_mm256_i64scatter_pd): Ditto.
40916 (_mm256_mask_i64scatter_pd): Ditto.
40917 (_mm_i64scatter_pd): Ditto.
40918 (_mm_mask_i64scatter_pd): Ditto.
40919 (_mm256_i32scatter_epi32): Ditto.
40920 (_mm256_mask_i32scatter_epi32): Ditto.
40921 (_mm_i32scatter_epi32): Ditto.
40922 (_mm_mask_i32scatter_epi32): Ditto.
40923 (_mm256_i32scatter_epi64): Ditto.
40924 (_mm256_mask_i32scatter_epi64): Ditto.
40925 (_mm_i32scatter_epi64): Ditto.
40926 (_mm_mask_i32scatter_epi64): Ditto.
40927 (_mm256_i64scatter_epi32): Ditto.
40928 (_mm256_mask_i64scatter_epi32): Ditto.
40929 (_mm_i64scatter_epi32): Ditto.
40930 (_mm_mask_i64scatter_epi32): Ditto.
40931 (_mm256_i64scatter_epi64): Ditto.
40932 (_mm256_mask_i64scatter_epi64): Ditto.
40933 (_mm_i64scatter_epi64): Ditto.
40934 (_mm_mask_i64scatter_epi64): Ditto.
40935 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40936 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40937 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40938 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40939 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40940 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40941 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40942 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40943 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40944 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40945 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40946 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40947 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40948 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40949 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40950 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40951 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40952 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40953 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40954 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40955 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40956 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40957 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40958 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40959 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40960 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40961 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40962 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40963 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40964 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40965 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40966 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40967 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40968 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40969 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40970 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40971 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40972 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40973 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40974 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40975 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40976 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40977 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40978 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40979 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40980 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40981 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40982 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40983 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40984 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40985 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40986 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40987 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40988 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40989 definitions accordingly.
40991 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40992 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40995 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40998 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
41001 * ira-conflicts.c (ira_build_conflicts): Update total conflict
41002 hard regs for inner regno.
41004 2017-01-17 Martin Liska <mliska@suse.cz>
41007 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
41008 assumption and add comment.
41010 2017-01-17 Nathan Sidwell <nathan@acm.org>
41012 * ipa-visibility.c (localize_node): New function, broken out of ...
41013 (function_and_variable_visibility): ... here. Call it.
41015 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
41017 PR middle-end/77445
41018 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
41019 correctly set frequency of oudgoing edge.
41020 (duplicate_thread_path): Fix profile updating.
41022 2017-01-17 Jakub Jelinek <jakub@redhat.com>
41025 * configure.ac: Add GCC_BASE_VER.
41026 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
41027 version from BASE-VER file.
41028 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
41029 (gcc.o): Depend on $(BASEVER).
41030 * common.opt (dumpfullversion): New option.
41031 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
41032 * doc/invoke.texi: Document -dumpfullversion.
41033 * doc/install.texi: Document --with-gcc-major-version-only.
41034 * configure: Regenerated.
41036 2017-01-17 Richard Biener <rguenther@suse.de>
41038 PR tree-optimization/71433
41039 * tree-vrp.c (register_new_assert_for): Merge same asserts
41040 on all incoming edges.
41041 (process_assert_insertions_for): Handle insertions at the
41044 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
41046 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
41047 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
41049 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
41052 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
41055 2017-01-17 Alan Modra <amodra@gmail.com>
41058 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
41059 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
41060 symbolic stack limit when pic.
41062 2017-01-16 Martin Sebor <msebor@redhat.com>
41064 PR tree-optimization/78608
41065 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
41067 2017-01-16 Jeff Law <law@redhat.com>
41070 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
41071 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
41072 for several include directories that may be relative to sysroot.
41073 * config/i386/x-mingw32 (gplus_includedir): Define.
41074 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
41075 (native_system_includedir): Likewise.
41076 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
41077 override if TARGET_SYSTEM_ROOT is defined.
41078 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
41080 PR tree-optimization/79090
41081 PR tree-optimization/33562
41082 PR tree-optimization/61912
41083 PR tree-optimization/77485
41084 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
41085 and computed trims into the dump file.
41087 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
41089 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
41091 2017-01-16 Jakub Jelinek <jakub@redhat.com>
41094 * gimplify.c (gimplify_init_constructor): If want_value and
41095 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
41099 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
41100 sequence. Formatting fixes.
41101 (doloop_optimize): Formatting fixes.
41104 * gcc.c (debug_level_greater_than_spec_func): New function.
41105 (static_spec_functions): Add debug-level-gt spec function.
41106 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
41108 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41109 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41110 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
41111 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
41112 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
41113 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
41115 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
41117 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
41118 QImode fixups to general and mask registers only.
41120 2017-01-16 Carl Love <cel@us.ibm.com>
41122 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
41123 for built-in functions
41124 vector signed char vec_nabs (vector signed char)
41125 vector signed short vec_nabs (vector signed short)
41126 vector signed int vec_nabs (vector signed int)
41127 vector signed long long vec_nabs (vector signed long long)
41128 vector float vec_nabs (vector float)
41129 vector double vec_nabs (vector double)
41130 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
41132 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
41133 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
41134 * doc/extend.texi: Update the documentation file for the new built-in
41137 2017-01-16 Martin Sebor <msebor@redhat.com>
41139 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
41142 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41144 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
41145 UNSPEC_VSX__XXSPLTD to require special splat handling.
41147 2017-01-16 David Malcolm <dmalcolm@redhat.com>
41150 * system.h: Poison strndup.
41152 2017-01-16 Alan Modra <amodra@gmail.com>
41155 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
41158 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
41160 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
41162 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
41164 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
41165 call recog here. Assert that INSN_CODE (insn) is non-negative.
41167 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41170 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
41172 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
41173 in the currently scheduled RTL fragment.
41175 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41177 PR rtl-optimization/78751
41178 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
41181 2017-01-14 Jeff Law <law@redhat.com>
41183 PR tree-optimization/79090
41184 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
41185 variable length stores.
41186 (compute_trims): Delete dead assignment to *trim_tail.
41187 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
41190 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
41192 PR rtl-optimization/78626
41193 PR rtl-optimization/78727
41194 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
41195 of a block, and split such blocks after everything else is finished.
41197 2017-01-14 Alan Modra <amodra@gmail.com>
41200 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
41201 target legitimate_combined_insn.
41202 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41203 (rs6000_legitimate_combined_insn): New function.
41204 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41206 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41207 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41208 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41210 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41212 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41214 2017-01-13 Jeff Law <law@redhat.com>
41216 PR tree-optimization/33562
41217 PR tree-optimization/61912
41218 PR tree-optimization/77485
41219 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41221 (delete_dead_assignment): Likewise.
41222 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41223 statement to delete_dead_call and delete_dead_assignment.
41225 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41228 * substring-locations.c (format_warning_va): Strengthen case 1 so
41229 that both endpoints of the substring must be within the format
41230 range for just the substring to be printed.
41232 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41234 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41235 * config/i386/i386.c (ix86_target_string): Add missing options
41236 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41237 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41238 flags_other and ix86_target_other to flags2_other. Display unknown
41240 (ix86_valid_target_attribute_inner_p): Add missing options and
41241 reorder options by implied ISAs, as in ix86_target_string.
41243 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41245 * hash-table.h (hash_table::too_empty_p): New function.
41246 (hash_table::expand): Use it.
41247 (hash_table::traverse): Likewise.
41248 (hash_table::empty_slot): Use sizeof (value_type) instead of
41249 sizeof (PTR) to convert bytes to elements. Shrink the table
41250 if the current size is excessive for the current number of
41253 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41255 * ira-costs.c (record_reg_classes): Break from the inner loop
41256 early once alt_fail is known to be true. Update outer loop
41257 handling accordingly.
41259 2017-01-13 Jeff Law <law@redhat.com>
41261 * tree-ssa-dse.c (decrement_count): New function.
41262 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41263 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41264 when we know the partially dead statement is a mem* function.
41266 PR tree-optimization/61912
41267 PR tree-optimization/77485
41268 * tree-ssa-dse.c: Include expr.h.
41269 (maybe_trim_constructor_store): New function.
41270 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41272 PR tree-optimization/33562
41273 PR tree-optimization/61912
41274 PR tree-optimization/77485
41275 * doc/invoke.texi: Document new dse-max-object-size param.
41276 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41277 * tree-ssa-dse.c: Include params.h.
41278 (dse_store_status): New enum.
41279 (initialize_ao_ref_for_dse): New, partially extracted from
41281 (valid_ao_ref_for_dse, normalize_ref): New.
41282 (setup_live_bytes_from_ref, compute_trims): Likewise.
41283 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41284 (maybe_trim_partially_dead_store): Likewise.
41285 (maybe_trim_complex_store): Likewise.
41286 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41287 Track what bytes live from the original store. Return tri-state
41288 for dead, partially dead or live.
41289 (dse_dom_walker): Add constructor, destructor and new private members.
41290 (delete_dead_call, delete_dead_assignment): New extracted from
41292 (dse_optimize_stmt): Make a member of dse_dom_walker.
41293 Use initialize_ao_ref_for_dse.
41295 PR tree-optimization/33562
41296 PR tree-optimization/61912
41297 PR tree-optimization/77485
41298 * sbitmap.h (bitmap_count_bits): Prototype.
41299 (bitmap_clear_range, bitmap_set_range): Likewise.
41300 * sbitmap.c (bitmap_clear_range): New function.
41301 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41303 2017-01-13 Martin Liska <mliska@suse.cz>
41306 * function.c (set_cfun): Add new argument force.
41307 * function.h (set_cfun): Likewise.
41308 * ipa-inline-transform.c (inline_call): Use the function when
41309 strict alising from is dropped for function we inline to.
41311 2017-01-13 Richard Biener <rguenther@suse.de>
41313 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41314 for dumping GIMPLE INTEGER_CSTs.
41316 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41318 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41319 to 201112L since C++17.
41321 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41324 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41325 if -fsanitize=kernel-address is present.
41327 2017-01-13 Richard Biener <rguenther@suse.de>
41329 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41330 as _Literal ( type ) number in case usual suffixes do not
41331 preserve all information.
41333 2017-01-13 Richard Biener <rguenther@suse.de>
41335 PR tree-optimization/77283
41336 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41337 and ssa-iterators.h.
41338 (is_feasible_trace): Implement a cost model based on joiner
41341 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41344 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41345 char or short to __float128/_Float128 directly.
41347 2017-01-12 Martin Sebor <msebor@redhat.com>
41349 to -Wformat-overflow.
41350 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41351 (min_bytes_remaining): Same.
41352 (get_string_length): Same.
41353 (format_string): Same.
41354 (format_directive): Same.
41356 (pass_sprintf_length::handle_gimple_call): Same.
41358 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41360 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41361 info.nowrite calls with no lhs that can't throw. Return bool
41362 whether gsi_remove has been called or not.
41363 (pass_sprintf_length::handle_gimple_call): Return bool whether
41364 try_substitute_return_value called gsi_remove. Formatting fix.
41365 (pass_sprintf_length::execute): Don't use gsi_remove if
41366 handle_gimple_call returned true.
41369 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41370 be removed due to side-effects, don't remove following barrier nor
41371 turn the successor edge into fallthru edge.
41373 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41376 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41377 element-reversing loads and stores as not swappable.
41379 2017-01-12 Nathan Sidwell <nathan@acm.org>
41380 Nicolai Stange <nicstange@gmail.com>
41382 * combine.c (try_combine): Don't ignore result of overlap checking
41383 loop. Combine overlap & asm check into single loop.
41385 2017-01-12 Richard Biener <rguenther@suse.de>
41387 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41388 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41390 2017-01-12 Richard Biener <rguenther@suse.de>
41392 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41393 and TS_TARGET_OPTION directly derive from TS_BASE.
41394 * tree-core.h (tree_optimization_option): Derive from tree_base.
41395 (tree_target_option): Likewise.
41397 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41399 * config/i386/i386.c (memory_address_length): Increase len
41400 only when rip_relative_addr_p returns false.
41402 2017-01-11 Julia Koval <julia.koval@intel.com>
41404 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41405 (OPTION_MASK_ISA_SGX_SET): New.
41406 (ix86_handle_option): Handle OPT_msgx.
41407 * config.gcc: Added sgxintrin.h.
41408 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41409 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41410 * config/i386/i386.c (ix86_target_string): Add -msgx.
41412 (ix86_option_override_internal): Handle new options.
41413 (ix86_valid_target_attribute_inner_p): Add sgx.
41414 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41415 * config/i386/i386.opt: Add msgx.
41416 * config/i386/sgxintrin.h: New file.
41417 * config/i386/x86intrin.h: Add sgxintrin.h.
41419 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41422 * fold-const.c (maybe_nonzero_address): Return 1 for function
41424 (tree_single_nonzero_warnv_p): Don't handle function local objects
41428 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41431 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41434 * opts.c: Include "spellcheck.h"
41435 (struct string_fragment): New struct.
41436 (struct edit_distance_traits<const string_fragment &>): New
41438 (get_closest_sanitizer_option): New function.
41439 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41441 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41443 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41445 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41446 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41447 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41448 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41449 for initial die_offset if dwarf_split_debug_info.
41450 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41451 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41453 (output_skeleton_debug_sections): Formatting fix. Use
41454 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41455 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41457 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41459 * config/arm/cortex-a53.md: Add bypasses for
41460 cortex_a53_r2f_cvt.
41461 (cortex_a53_r2f): Only use for transfers.
41462 (cortex_a53_f2r): Likewise.
41463 (cortex_a53_r2f_cvt): Add reservation for conversions.
41464 (cortex_a53_f2r_cvt): Likewise.
41466 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41468 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41469 to all inlined functions, change static to extern.
41471 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41474 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41476 (arm_assemble_integer): Likewise.
41478 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41480 * config.gcc: Use new awk script to check CPU, FPU and architecture
41481 parameters for --with-... options.
41482 * config/arm/parsecpu.awk: New file
41483 * config/arm/arm-cpus.in: New file.
41484 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41486 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41488 * config/arm/t-arm: Update dependency rules.
41489 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41490 of processing .def files.
41491 * config/arm/genopt.sh: Deleted.
41492 * config/arm/gentune.sh: Deleted.
41493 * config/arm/arm-cores.def: Deleted.
41494 * config/arm/arm-arches.def: Deleted.
41495 * config/arm/arm-fpus.def: Deleted.
41496 * config/arm/arm-tune.md: Regenerated.
41497 * config/arm/arm-tables.opt: Regenerated.
41498 * config/arm/arm-cpu.h: New generated file.
41499 * config/arm/arm-cpu-data.h: New generated file.
41500 * config/arm/arm-cpu-cdata.h: New generated file.
41502 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41505 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41507 (input_varpool_node): Unpack dynamically_initialized bit.
41509 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41511 PR rtl-optimization/79032
41512 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41513 the alignment of the adjusted memory reference against that of MODE,
41514 instead of the alignment of the original memory reference.
41516 2017-01-11 Martin Jambor <mjambor@suse.cz>
41518 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41520 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41521 decorated functions.
41523 2017-01-11 Richard Biener <rguenther@suse.de>
41525 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41526 set range/nonnull info for PHI results. Do not set it on
41527 stmts marked for removal.
41529 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41531 * expr.c (store_field): In the bitfield case, fetch the return value
41532 from the registers before applying a single big-endian adjustment.
41533 Always do a final load for a BLKmode value not larger than a word.
41535 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41538 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41539 that we correctly handle column numbers greater than
41540 LINE_MAP_MAX_COLUMN_NUMBER.
41542 2017-01-10 Martin Sebor <msebor@redhat.com>
41544 PR middle-end/78245
41545 * gimple-ssa-sprintf.c (get_destination_size): Call
41546 {init,fini}object_sizes.
41547 * tree-object-size.c (addr_object_size): Adjust.
41548 (pass_through_call): Adjust.
41549 (pass_object_sizes::execute): Adjust.
41550 * tree-object-size.h (fini_object_sizes): Declare.
41552 2017-01-10 Martin Sebor <msebor@redhat.com>
41554 PR tree-optimization/78775
41555 * builtins.c (get_size_range): Move...
41556 * calls.c: ...to here.
41557 (alloc_max_size): Accept zero argument.
41558 (operand_signed_p): Remove.
41559 (maybe_warn_alloc_args_overflow): Call get_size_range.
41560 * calls.h (get_size_range): Declare.
41562 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41564 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41565 from TI's devices.csv file as of September 2016.
41566 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41568 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41570 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41571 * doc/invoke.texi: Likewise.
41572 * doc/md.texi: Likewise.
41573 * doc/objc.texi: Likewise.
41575 2017-01-10 Joshua Conner <joshconner@google.com>
41577 * config/arm/fuchsia-elf.h: New file.
41578 * config/fuchsia.h: New file.
41579 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41580 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41582 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41584 2016-01-10 Richard Biener <rguenther@suse.de>
41586 PR tree-optimization/79034
41587 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41588 Propagate out degenerate PHIs in the joiner.
41590 2017-01-10 Martin Liska <mliska@suse.cz>
41592 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41593 (sort_congruence_classes_by_decl_uid): Likewise.
41594 (sort_congruence_class_groups_by_decl_uid): Likewise.
41595 (sem_item_optimizer::merge_classes): Sort class, groups in these
41596 classes and members in the groups by DECL_UID of declarations.
41597 This would make merge operations stable.
41599 2017-01-10 Martin Liska <mliska@suse.cz>
41601 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41602 usage of m_classes_vec.
41603 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41604 (sem_item_optimizer::get_group_by_hash): Likewise.
41605 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41606 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41607 (sem_item_optimizer::verify_classes): Likewise.
41608 (sem_item_optimizer::process_cong_reduction): Likewise.
41609 (sem_item_optimizer::dump_cong_classes): Likewise.
41610 (sem_item_optimizer::merge_classes): Likewise.
41611 * ipa-icf.h (congruence_class_hash): Rename from
41612 congruence_class_group_hash. Remove declaration of m_classes_vec.
41614 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41616 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41617 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41618 * config.gcc: Add avx512vpopcntdqintrin.h.
41619 * config/i386/avx512vpopcntdqintrin.h: New.
41620 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41621 * config/i386/i386-builtin-types.def: Add new types.
41622 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41623 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41624 __builtin_ia32_vpopcountq_v8di_mask): New.
41625 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41626 __AVX512VPOPCNTDQ__.
41627 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41628 (PTA_AVX512VPOPCNTDQ): Define.
41629 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41630 TARGET_AVX512VPOPCNTDQ_P): Define.
41631 * config/i386/i386.opt: Add mavx512vpopcntdq.
41632 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41633 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41635 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41637 PR middle-end/77484
41638 * predict.def (PRED_CALL): Set to 67.
41640 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41642 * expr.c (store_field): In the bitfield case, if the value comes from
41643 a function call and is of an aggregate type returned in registers, do
41644 not modify the field mode; extract the value in all cases if the mode
41645 is BLKmode and the size is not larger than a word.
41647 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41650 * config/i386/cpuid.h: Fix undefined behavior.
41652 2017-01-04 Jeff Law <law@redhat.com>
41654 PR tree-optimization/79007
41655 PR tree-optimization/67955
41656 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41657 conservative for pt.null when flag_non_call_exceptions is on.
41659 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41661 PR translation/79019
41662 PR translation/79020
41663 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41664 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41665 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41667 * config/avr/avr.opt (maccumulate-args): Likewise.
41668 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41669 * common.opt (freport-bug): Likewise.
41670 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41671 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41672 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41673 translatable string.
41674 * config/i386/i386.c (function_value_32): Likewise.
41675 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41676 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41678 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41679 * common/config/msp430/msp430-common.c (msp430_handle_option):
41681 * symtab.c (symtab_node::verify_base): Likewise.
41682 * opts.c (set_debug_level): Likewise.
41683 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41684 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41685 missing whitespace to translatable strings.
41686 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41687 * config/sh/superh.h: Likewise.
41688 * config/i386/xopintrin.h: Likewise.
41689 * config/i386/znver1.md: Likewise.
41690 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41691 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41692 * double-int.h (struct double_int): Likewise.
41693 * double-int.c (div_and_round_double): Likewise.
41694 * wide-int.cc: Likewise.
41695 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41696 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41697 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41698 (crossjumps_occurred): ... this.
41699 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41702 PR tree-optimization/78899
41703 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41704 returning bool return struct loop *, NULL for failure and the new
41706 (versionable_outer_loop_p): Don't version outer loop if it has
41707 dont_vectorized bit set.
41708 (tree_if_conversion): When versioning outer loop, ensure
41709 tree_if_conversion is performed also on the inner loop of the
41710 non-vectorizable outer loop copy.
41711 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41712 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41713 prevent vectorization of it.
41714 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41715 the outer loop vectorization of the non-scalar version is attempted
41716 before vectorization of the inner loop in scalar version. If
41717 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41718 vectorization of its inner loop.
41719 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41720 has 2 inner loops, rename also on edges from bb whose single pred
41721 is outer_loop->header. Fix typo in function comment.
41723 2017-01-09 Martin Sebor <msebor@redhat.com>
41726 * asan.c (asan_emit_stack_protection): Increase local buffer size
41727 to avoid snprintf truncation warning.
41729 2017-01-09 Andrew Pinski <apinski@cavium.com>
41731 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41732 to reference thunderx2t99 for the tuning structure
41733 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41735 (thunderx2t99_extra_costs): This.
41736 * config/aarch64/aarch64-tune.md: Regenerate.
41737 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41738 (vulcan_addrcost_table): This.
41739 (vulcan_regmove_cost): Rename to ...
41740 (thunderx2t99_regmove_cost): This.
41741 (vulcan_vector_cost): Rename to ...
41742 (thunderx2t99_vector_cost): this.
41743 (vulcan_branch_cost): Rename to ...
41744 (thunderx2t99_branch_cost): This.
41745 (vulcan_tunings): Rename to ...
41746 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41747 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41749 2017-01-09 Martin Jambor <mjambor@suse.cz>
41753 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41754 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41755 (propagate_vr_accross_jump_function): Use the above function for all
41756 value range computations for pass-through jump functions and type
41757 converasion from explicit value range values.
41758 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41759 parameters from TYPE_ARG_TYPES.
41760 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41761 (ipa_write_node_info): Stream type of the actual argument.
41762 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41764 2017-01-09 Martin Liska <mliska@suse.cz>
41767 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41768 (lookup_compiler): Do not show error message with have_E.
41770 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41772 PR tree-optimization/78938
41773 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41774 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41775 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41776 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41779 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41781 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41784 2017-01-09 Richard Biener <rguenther@suse.de>
41786 PR tree-optimization/78997
41787 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41788 name condition properly.
41790 2017-01-09 Richard Biener <rguenther@suse.de>
41793 * dwarf2out.c (is_cxx): New overload with context.
41794 (is_naming_typedef_decl): Use it.
41796 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41798 * invoke.texi (Option Summary): Correct spacing in option lists
41799 and add line breaks to fix over-long lines.
41801 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41803 PR middle-end/17660
41805 * extend.texi (Common Variable Attributes): Add xref to GCC
41806 Internals manual to explain mode attribute keywords.
41808 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41811 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41812 and Preprocessor Options.
41813 (Options for Linking): Document -pthread here....
41814 (RS/6000 and PowerPC Options): ...not here.
41815 (Solaris 2 Options): ...or here.
41816 * doc/cppopts.texi: Document -pthread.
41818 2017-01-08 Martin Sebor <msebor@redhat.com>
41820 PR middle-end/77708
41821 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41822 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41823 New member functions.
41824 (format_directive): Used them.
41826 (pass_sprintf_length::handle_gimple_call): Same.
41827 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41828 to avoid truncation for any argument.
41829 (extract_affine_mul): Same.
41830 * tree.c (get_file_function_name): Same.
41832 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41834 PR middle-end/77484
41835 * predict.def (PRED_INDIR_CALL): Set to 86.
41837 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41839 PR preprocessor/54124
41840 * doc/cppopts.texi: Reformat -d subtable to list the full name
41841 of the options. Add cross-reference to the docs for the general
41842 compiler -d options.
41843 * doc/invoke.texi (Developer Options): Add cross-reference to the
41844 preprocessor-specific -d option documentation.
41846 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41848 PR preprocessor/13498
41849 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41850 redudant material, and reflect new command-line options.
41851 (System Headers): Likewise.
41853 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41855 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41856 -isystem, and -idirafter. Copy-edit.
41857 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41858 default for -ftrack-macro-expansion. Delete obsolete and
41859 badly-formatted implementation details about -fdebug-cpp output.
41860 * doc/cppwarnopts.texi: Copy-edit.
41862 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41865 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41866 that the transition from a max line width >= 1<<10 to narrower
41867 lines works correctly.
41869 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41871 * doc/options.texi (PerFunction): New.
41872 * opt-functions.awk (switch_flags): Map both Optimization and
41873 PerFunction to CL_OPTIMIZATION.
41874 * opth-gen.awk: Test for PerFunction flag along with
41876 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41877 it only when the latter is present. Skip those that don't in
41878 the hash function generator.
41879 * common.opt (fvar-tracking): Mark as PerFunction instead of
41881 (fvar-tracking-assignments): Likewise.
41882 (fvar-tracking-assignments-toggle): Likewise.
41883 (fvar-tracking-uninit): Likewise.
41885 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41887 PR translation/79018
41888 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41891 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41894 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41895 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41896 TARGET_LONG_JUMP_TABLE_OFFSETS.
41897 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41898 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41899 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41900 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41901 * config/m68k/m68k.md (tablejump expander): Likewise.
41902 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41903 TARGET_LONG_JUMP_TABLE_OFFSETS.
41904 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41905 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41907 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41908 David Holsgrove <david.holsgrove@xilinx.com>
41910 * common/config/microblaze/microblaze-common.c
41911 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41912 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41914 * config/microblaze/microblaze.c (microblaze_must_save_register)
41915 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41917 (microblaze_eh_return): New function.
41918 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41919 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41920 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41921 * config/microblaze/microblaze.md (eh_return): New pattern.
41923 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41925 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41926 GCC_DIAGNOSTIC_STRINGIFY): Define.
41928 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41930 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41932 * config/arm/arm.md (<mcrr>): New.
41934 * config/arm/arm.c (arm_arch5te): New.
41935 (arm_option_override): Set arm_arch5te.
41936 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41938 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41939 (arm_mcrr_qualifiers): ... this. New.
41940 (MRRC_QUALIFIERS): Define to...
41941 (arm_mrrc_qualifiers): ... this. New.
41942 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41944 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41945 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41946 (MRRCI, mrrc, MRRC): New.
41947 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41948 VUNSPEC_MRRC2): New.
41950 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41952 * config/arm/arm.md (<mcr>): New.
41954 * config/arm/arm.c (arm_coproc_builtin_available): Add
41955 support for mcr, mrc, mcr2 and mrc2.
41956 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41957 (arm_mcr_qualifiers): ... this. New.
41958 (MRC_QUALIFIERS): Define to ...
41959 (arm_mrc_qualifiers): ... this. New.
41960 (MCR_QUALIFIERS): Define to ...
41961 (arm_mcr_qualifiers): ... this. New.
41962 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41964 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41965 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41966 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41967 VUNSPEC_MRC2): New.
41969 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41971 * config/arm/arm.md (*ldc): New.
41975 * config/arm/arm.c (arm_coproc_builtin_available): Add
41976 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41977 (arm_coproc_ldc_stc_legitimate_address): New.
41978 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41979 'qualifier_const_pointer'.
41980 (LDC_QUALIFIERS): Define to...
41981 (arm_ldc_qualifiers): ... this. New.
41982 (STC_QUALIFIERS): Define to...
41983 (arm_stc_qualifiers): ... this. New.
41984 * config/arm/arm-protos.h
41985 (arm_coproc_ldc_stc_legitimate_address): New.
41986 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41987 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41988 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41989 stc2, stcl, stc2l): New.
41990 * config/arm/constraints.md (Uz): New.
41991 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41992 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41993 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41994 VUNSPEC_STC2L): New.
41996 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41998 * config/arm/arm.md (<cdp>): New.
41999 * config/arm/arm.c (neon_const_bounds): Rename this ...
42000 (arm_const_bounds): ... this.
42001 (arm_coproc_builtin_available): New.
42002 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
42003 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
42004 (CDP_QUALIFIERS): Define to...
42005 (arm_cdp_qualifiers): ... this. New.
42007 (arm_expand_builtin_args): Add case for 6 arguments.
42008 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
42009 (arm_const_bounds): ... this.
42010 (arm_coproc_builtin_available): New.
42011 * config/arm/arm_acle.h (__arm_cdp): New.
42013 * config/arm/arm_acle_builtins.def (cdp): New.
42015 * config/arm/iterators.md (CDPI,CDP,cdp): New.
42016 * config/arm/neon.md: Rename all 'neon_const_bounds' to
42017 'arm_const_bounds'.
42018 * config/arm/types.md (coproc): New.
42019 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
42020 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
42021 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
42022 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
42024 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42026 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
42027 (UBINOP_QUALIFIERS): New.
42029 (acle_builtin_data): New. Change comment.
42030 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
42031 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
42032 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
42033 arm_acle_builtins.def.
42034 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
42035 (arm_init_acle_builtins): New.
42036 (CRC32_BUILTIN): Remove.
42037 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
42038 crc32cb, crc32ch and crc32cw.
42039 (arm_init_crc32_builtins): Remove.
42040 (arm_init_builtins): Use arm_init_acle_builtins rather
42041 than arm_init_crc32_builtins.
42042 (arm_expand_acle_builtin): New.
42043 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
42044 * config/arm/arm_acle_builtins.def: New.
42046 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42048 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
42049 (arm_builtin_datum): ... this.
42050 (arm_init_neon_builtin): Rename to ...
42051 (arm_init_builtin): ... this. Add a new parameters PREFIX
42052 and USE_SIG_IN_NAME.
42053 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
42054 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
42055 'arm_builtin_datum'.
42056 (arm_init_vfp_builtins): Likewise.
42057 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
42058 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
42059 (arm_expand_neon_args): Rename to ...
42060 (arm_expand_builtin_args): ... this. Rename builtin_arg
42061 enum values and differentiate between ARG_BUILTIN_MEMORY
42062 and ARG_BUILTIN_NEON_MEMORY.
42063 (arm_expand_neon_builtin_1): Rename to ...
42064 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
42065 values, arm_expand_builtin_args and add bool parameter NEON.
42066 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
42067 (arm_expand_vfp_builtin): Likewise.
42068 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
42070 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42072 PR middle-end/77484
42073 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
42074 * predict.c (tree_estimate_probability_bb): Reverse direction of
42075 polymorphic call predictor.
42077 2017-01-06 David Malcolm <dmalcolm@redhat.com>
42079 * passes.c (execute_one_pass): Split out pass-skipping logic into...
42080 (determine_pass_name_match): ...this new function and...
42081 (should_skip_pass_p): ...this new function.
42083 2017-01-06 Nathan Sidwell <nathan@acm.org>
42085 * ipa-visibility.c (function_and_variable_visibility): Reformat
42086 comments and long lines. Remove extrneous if.
42087 * symtab.c (symtab_node::make_decl_local): Fix code format.
42088 (symtab_node::set_section_for_node): Fix comment typo.
42090 2017-01-06 Martin Liska <mliska@suse.cz>
42093 * lra-constraints.c: Rename invariant to lra_invariant.
42094 * predict.c (set_even_probabilities): Initialize e to NULL.
42096 2017-01-05 Martin Sebor <msebor@redhat.com>
42098 PR tree-optimization/78910
42099 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
42100 (format_integer): Correct off-by-one error in the handling
42101 of precision with negative numbers in signed conversions..
42103 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
42105 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
42107 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42109 PR tree-optimization/71016
42110 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
42111 factor_out_conditional_conversion. Formatting fix.
42112 (factor_out_conditional_conversion): Add cond_stmt argument.
42113 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
42114 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
42117 2017-01-05 David Malcolm <dmalcolm@redhat.com>
42119 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
42120 read-rtl-function.o, and selftest-rtl.o.
42121 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
42122 (selftest::aarch64_test_loading_full_dump): New function.
42123 (selftest::aarch64_run_selftests): New function.
42124 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
42125 selftest::aarch64_run_selftests.
42126 * config/i386/i386.c
42127 (selftest::ix86_test_loading_dump_fragment_1): New function.
42128 (selftest::ix86_test_loading_call_insn): New function.
42129 (selftest::ix86_test_loading_full_dump): New function.
42130 (selftest::ix86_test_loading_unspec): New function.
42131 (selftest::ix86_run_selftests): Call the new functions.
42132 * emit-rtl.c (maybe_set_max_label_num): New function.
42133 * emit-rtl.h (maybe_set_max_label_num): New decl.
42134 * function.c (instantiate_decls): Guard call to
42135 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
42136 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
42138 * gensupport.c (gen_reader::gen_reader): Pass "false"
42139 for new "compact" param of rtx_reader.
42140 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
42141 rather than an empty string for NULL strings.
42142 * read-md.c: Potentially include config.h rather than bconfig.h.
42143 Wrap include of errors.h with #ifdef GENERATOR_FILE.
42144 (have_error): New global, copied from errors.c.
42145 (md_reader::read_name): Rename to...
42146 (md_reader::read_name_1): ...this, adding "out_loc" param,
42147 and converting "missing name or number" to returning false, rather
42149 (md_reader::read_name): Reimplement in terms of read_name_1.
42150 (md_reader::read_name_or_nil): New function.
42151 (md_reader::read_string): Handle "(nil)" by returning NULL.
42152 (md_reader::md_reader): Add new param "compact".
42153 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
42154 (md_reader::read_file): New method.
42155 * read-md.h (md_reader::md_reader): Add new param "compact".
42156 (md_reader::read_file): New method.
42157 (md_reader::is_compact): New accessor.
42158 (md_reader::read_name): Convert return type from void to file_location.
42159 (md_reader::read_name_or_nil): New decl.
42160 (md_reader::read_name_1): New decl.
42161 (md_reader::m_compact): New field.
42162 (noop_reader::noop_reader): Pass "false" for new "compact" param
42164 (rtx_reader::rtx_reader): Add new "compact" param.
42165 (rtx_reader::read_rtx_operand): Make virtual and convert return
42166 type from void to rtx.
42167 (rtx_reader::read_until): New decl.
42168 (rtx_reader::handle_any_trailing_information): New virtual function.
42169 (rtx_reader::postprocess): New virtual function.
42170 (rtx_reader::finalize_string): New virtual function.
42171 (rtx_reader::m_in_call_function_usage): New field.
42172 (rtx_reader::m_reuse_rtx_by_id): New field.
42173 * read-rtl-function.c: New file.
42174 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
42175 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
42176 (selftest::verify_three_block_rtl_cfg): New decl.
42177 * read-rtl-function.h: New file.
42178 * read-rtl.c: Potentially include config.h rather than bconfig.h.
42179 For host, include function.h, memmodel.h, and emit-rtl.h.
42180 (one_time_initialization): New function.
42181 (struct compact_insn_name): New struct.
42182 (compact_insn_names): New array.
42183 (find_code): Handle insn codes in compact dumps.
42184 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
42185 (bind_subst_iter_and_attr): Likewise.
42186 (add_condition_to_string): Likewise.
42187 (add_condition_to_rtx): Likewise.
42188 (apply_attribute_uses): Likewise.
42189 (add_current_iterators): Likewise.
42190 (apply_iterators): Likewise.
42191 (initialize_iterators): Guard usage of apply_subst_iterator with
42192 #ifdef GENERATOR_FILE.
42193 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
42194 (md_reader::read_mapping): Likewise.
42195 (add_define_attr_for_define_subst): Likewise.
42196 (add_define_subst_attr): Likewise.
42197 (read_subst_mapping): Likewise.
42198 (check_code_iterator): Likewise.
42199 (rtx_reader::read_rtx): Likewise. Move one-time initialization
42201 (one_time_initialization): New function.
42202 (rtx_reader::read_until): New method.
42203 (read_flags): New function.
42204 (parse_reg_note_name): New function.
42205 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42206 Handle reuse_rtx ids.
42207 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42208 Add parsing support for RTL dumps, mirroring the special-cases in
42209 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42210 values, and calling handle_any_trailing_information.
42211 (rtx_reader::read_rtx_operand): Convert return type from void
42212 to rtx, returning return_rtx. Handle case 'e'. Call
42213 finalize_string on XSTR and XTMPL fields.
42214 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42215 "(nil)" values were omitted. Call the postprocess vfunc on the
42217 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42218 class ctor. Initialize m_in_call_function_usage. Call
42219 one_time_initialization.
42220 * rtl-tests.c (selftest::test_uncond_jump): Call
42221 set_new_first_and_last_insn.
42222 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42223 * selftest-rtl.c: New file.
42224 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42225 (selftest::get_insn_by_uid): New decl.
42226 * selftest-run-tests.c (selftest::run_tests): Call
42227 read_rtl_function_c_tests.
42228 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42229 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42232 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42234 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42235 operands in a special way. Assert that pos+len <= mode precision.
42237 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42239 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42240 3 argument Alias with unlimited for the negative form.
42241 (fno-vect-cost-model): Removed.
42243 2017-01-05 Martin Liska <mliska@suse.cz>
42245 * hsa-gen.c (gen_hsa_divmod): New function.
42246 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42248 2017-01-05 Martin Liska <mliska@suse.cz>
42251 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42254 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42256 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42257 small constant length operands.
42259 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42261 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42262 between loop iterations.
42264 2017-01-05 Martin Liska <mliska@suse.cz>
42267 * gimplify.c (gimplify_decl_expr): Compare to
42268 asan_poisoned_variables instread of checking flags.
42269 (gimplify_target_expr): Likewise.
42270 (gimplify_expr): Likewise.
42271 (gimplify_function_tree): Conditionally initialize
42272 asan_poisoned_variables.
42274 2017-01-04 Jeff Law <law@redhat.com>
42276 PR tree-optimizatin/78812
42277 * rtl.h (contains_mem_rtx_p): Prototype.
42278 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42279 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42280 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42281 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42282 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42284 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42286 * input.c (assert_char_at_range): Default-initialize actual_range.
42288 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42290 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42291 to match the caller.
42293 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42295 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42296 insns after final jump in test to emit dummy move.
42298 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42300 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42301 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42303 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42305 * multiple_target.c (create_dispatcher_calls): Init e_next.
42306 * tree-ssa-loop-split.c (split_loop): Init border.
42307 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42310 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42315 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42316 (altivec_register_operand): Do not return true if the operand
42317 contains a SUBREG mixing SImode and SFmode.
42318 (vsx_register_operand): Likewise.
42319 (vsx_reg_sfsubreg_ok): New predicate.
42320 (vfloat_operand): Do not return true if the operand contains a
42321 SUBREG mixing SImode and SFmode.
42322 (vint_operand): Likewise.
42323 (vlogical_operand): Likewise.
42324 (gpc_reg_operand): Likewise.
42325 (int_reg_operand): Likewise.
42326 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42327 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42328 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42330 (rs6000_emit_move_si_sf_subreg): New helper function.
42331 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42332 fixup SUBREGs involving SImode and SFmode.
42333 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42334 numbers for the new peephole2 optimization.
42335 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42336 the GLIBC math library that do AND/IOR/XOR operations on single
42337 precision floating point.
42338 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42339 target macros to say whether we need to avoid SUBREGs mixing
42341 (TARGET_ALLOW_SF_SUBREG): Likewise.
42342 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42343 (UNSPEC_SI_FROM_SF): Likewise.
42344 (iorxor): Change spacing.
42345 (and_ior_xor): New iterator for AND, IOR, and XOR.
42346 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42347 (movdi_from_sf_zero_ext): Likewise.
42348 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42349 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42350 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42351 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42352 (fms<mode>4): Likewise.
42353 (fnma<mode>4): Likewise.
42354 (fnms<mode>4): Likewise.
42355 (nfma<mode>4): Likewise.
42356 (nfms<mode>4): Likewise.
42358 2017-01-04 Marek Polacek <polacek@redhat.com>
42361 * doc/invoke.texi: Document -Wpointer-compare.
42363 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42365 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42368 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42369 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42372 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42375 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42376 documentation of the powerpc_popcntb_ok attribute.
42377 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42378 code to issue warning messages if a requested CPU configuration is
42379 not supported by the binary (assembler and loader) toolchain.
42380 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42381 made to define a built-in function that has been disabled.
42382 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42383 made to define a built-in function that has been disabled.
42384 (altivec_init_builtins): Add comment explaining why definition
42385 of the DST built-in functions is not preceded by an assertion
42386 check. Add assertions to prevent ICE if attempts are made to
42387 define an altivec predicate or an abs* built-in function that has
42389 (htm_init_builtins): Add comment explaining why definition of the
42390 htm built-in functions is not preceded by an assertion check.
42392 2017-01-04 Jeff Law <law@redhat.com>
42394 PR tree-optimizatin/67955
42395 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42396 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42397 the points-to solution does not include pt_null. Use DECL_PT_UID
42400 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42402 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42403 Use gen_int_mode instead of gen_lopwart for const_int operands.
42405 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42407 PR tree-optimization/71563
42408 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42409 out of range value - has low bits known to be zero.
42411 2017-01-04 Alan Modra <amodra@gmail.com>
42413 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42414 * configure: Regenerate.
42415 * config.in: Regenerate.
42417 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42420 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42421 a substring of the message, but strcmp with the whole message. Ifdef
42422 ENABLE_NLS, translate the message first using dgettext.
42424 2017-01-03 Jeff Law <law@redhat.com>
42426 PR tree-optimizatin/78856
42427 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42428 (mark_threaded_blocks): Remove code to truncate thread paths that
42429 cross multiple loop headers. Instead invalidate the cached loop
42430 iteration information and handle case of a thread path walking
42431 into an irreducible region.
42433 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42436 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42437 assertions. Add support for doing the signbit if the IEEE 128-bit
42438 floating point value is in a GPR.
42439 * config/rs6000/rs6000.md (Fsignbit): Delete.
42440 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42441 Update the length attribute if the value is in a GPR.
42442 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42443 the sign or zero extension instruction, since the value is always 0/1.
42444 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42447 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42448 extracting SImode to a GPR register so that we can generate a
42449 store, limit the vector to be in a traditional Altivec register
42450 for the vextuwrx instruction.
42452 2017-01-03 Ian Lance Taylor <iant@google.com>
42454 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42456 2017-01-03 Martin Sebor <msebor@redhat.com>
42458 PR tree-optimization/78696
42459 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42460 precision. Use MPFR for %f for greater fidelity. Correct handling
42462 (pass_sprintf_length::compute_format_length): Set width and precision
42463 specified by asrerisk to void_node for vararg functions.
42464 (try_substitute_return_value): Adjust dump output.
42466 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42468 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42470 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42472 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42473 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42474 -mlra/-mno-lra was passed to the compiler.
42476 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42478 PR rtl-optimization/65618
42479 * emit-rtl.c (try_split): Move initialization of "before" and
42480 "after" to just before the call to emit_insn_after_setloc.
42482 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42484 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42486 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42488 * dwarf2out.c (gen_enumeration_type_die): When
42489 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42491 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42493 PR tree-optimization/78965
42494 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42495 Change first argument from const call_info & to call_info &. For %n
42496 set info.nowrite to false.
42498 PR middle-end/78901
42499 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42500 possibly throwing calls.
42502 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42503 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42504 and fns handling, rather than in a separate case SSA_NAME.
42506 2017-01-02 Jeff Law <law@redhat.com>
42508 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42509 fixes for first_period and second_period variables.
42511 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42514 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42515 (*insvqi_1): New insn pattern.
42516 (*insvqi_1_mem_rex64): Ditto.
42517 (*insvqi_2): Ditto.
42518 (*insvqi_3): Rename from *insvqi.
42520 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42522 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42524 * doc/cfg.texi (Edges): Remove reference to Java.
42525 (Maintaining the CFG): Ditto.
42527 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42529 PR middle-end/77674
42530 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42531 transparent aliases.
42533 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42535 PR middle-end/77484
42536 * predict.def (PRED_CALL): Update hitrate.
42537 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42538 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42539 into direct/indirect/polymorphic variants.
42541 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42543 Update copyright years.
42545 * gcc.c (process_command): Update copyright notice dates.
42546 * gcov-dump.c (print_version): Ditto.
42547 * gcov.c (print_version): Ditto.
42548 * gcov-tool.c (print_version): Ditto.
42549 * gengtype.c (create_file): Ditto.
42550 * doc/cpp.texi: Bump @copying's copyright year.
42551 * doc/cppinternals.texi: Ditto.
42552 * doc/gcc.texi: Ditto.
42553 * doc/gccint.texi: Ditto.
42554 * doc/gcov.texi: Ditto.
42555 * doc/install.texi: Ditto.
42556 * doc/invoke.texi: Ditto.
42558 Copyright (C) 2017 Free Software Foundation, Inc.
42560 Copying and distribution of this file, with or without modification,
42561 are permitted in any medium without royalty provided the copyright
42562 notice and this notice are preserved.