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 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
7 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
8 Alan Hayward <alan.hayward@arm.com>
9 David Sherwood <david.sherwood@arm.com>
11 * combine.c (can_change_dest_mode): Handle polynomial
13 * expmed.c (store_bit_field_1): Likewise.
14 * expr.c (store_constructor): Likewise.
15 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
16 and polynomial REGMODE_NATURAL_SIZE.
17 (gen_lowpart_common): Likewise.
18 * reginfo.c (record_subregs_of_mode): Likewise.
19 * rtlanal.c (read_modify_subreg_p): Likewise.
21 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22 Alan Hayward <alan.hayward@arm.com>
23 David Sherwood <david.sherwood@arm.com>
25 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
28 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29 Alan Hayward <alan.hayward@arm.com>
30 David Sherwood <david.sherwood@arm.com>
32 * match.pd: Cope with polynomial numbers of vector elements.
34 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
35 Alan Hayward <alan.hayward@arm.com>
36 David Sherwood <david.sherwood@arm.com>
38 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
39 in a POINTER_PLUS_EXPR.
41 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
42 Alan Hayward <alan.hayward@arm.com>
43 David Sherwood <david.sherwood@arm.com>
45 * omp-simd-clone.c (simd_clone_subparts): New function.
46 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
47 (ipa_simd_modify_function_body): Likewise.
49 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
50 Alan Hayward <alan.hayward@arm.com>
51 David Sherwood <david.sherwood@arm.com>
53 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
54 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
55 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
56 (expand_vector_condition, vector_element): Likewise.
57 (subparts_gt): New function.
58 (get_compute_type): Use subparts_gt.
59 (count_type_subparts): Delete.
60 (expand_vector_operations_1): Use subparts_gt instead of
63 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
64 Alan Hayward <alan.hayward@arm.com>
65 David Sherwood <david.sherwood@arm.com>
67 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
68 (vect_compile_time_alias): ...this new function. Do the calculation
69 on poly_ints rather than trees.
70 (vect_prune_runtime_alias_test_list): Update call accordingly.
72 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
73 Alan Hayward <alan.hayward@arm.com>
74 David Sherwood <david.sherwood@arm.com>
76 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
78 (vect_schedule_slp_instance): Likewise.
80 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
81 Alan Hayward <alan.hayward@arm.com>
82 David Sherwood <david.sherwood@arm.com>
84 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
85 constant and extern definitions for variable-length vectors.
86 (vect_get_constant_vectors): Note that the number of units
87 is known to be constant.
89 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
90 Alan Hayward <alan.hayward@arm.com>
91 David Sherwood <david.sherwood@arm.com>
93 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
94 of units as polynomial. Choose between WIDE and NARROW based
97 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
98 Alan Hayward <alan.hayward@arm.com>
99 David Sherwood <david.sherwood@arm.com>
101 * tree-vect-stmts.c (simd_clone_subparts): New function.
102 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
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-stmts.c (vectorizable_call): Treat the number of
109 vectors as polynomial. Use build_index_vector for
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-stmts.c (get_load_store_type): Treat the number of
117 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
118 for variable-length vectors.
119 (vectorizable_mask_load_store): Treat the number of units as
120 polynomial, asserting that it is constant if the condition has
121 already been enforced.
122 (vectorizable_store, vectorizable_load): Likewise.
124 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
125 Alan Hayward <alan.hayward@arm.com>
126 David Sherwood <david.sherwood@arm.com>
128 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
129 of units as polynomial. Punt if we can't tell at compile time
130 which vector contains the final result.
132 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
133 Alan Hayward <alan.hayward@arm.com>
134 David Sherwood <david.sherwood@arm.com>
136 * tree-vect-loop.c (vectorizable_induction): Treat the number
137 of units as polynomial. Punt on SLP inductions. Use an integer
138 VEC_SERIES_EXPR for variable-length integer reductions. Use a
139 cast of such a series for variable-length floating-point
142 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
143 Alan Hayward <alan.hayward@arm.com>
144 David Sherwood <david.sherwood@arm.com>
146 * tree.h (build_index_vector): Declare.
147 * tree.c (build_index_vector): New function.
148 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
149 of units as polynomial, forcibly converting it to a constant if
150 vectorizable_reduction has already enforced the condition.
151 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
152 to create a {1,2,3,...} vector.
153 (vectorizable_reduction): Treat the number of units as polynomial.
154 Choose vectype_in based on the largest scalar element size rather
155 than the smallest number of units. Enforce the restrictions
158 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
159 Alan Hayward <alan.hayward@arm.com>
160 David Sherwood <david.sherwood@arm.com>
162 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
163 number of units as polynomial.
165 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
166 Alan Hayward <alan.hayward@arm.com>
167 David Sherwood <david.sherwood@arm.com>
169 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
170 * target.def (autovectorize_vector_sizes): Return the vector sizes
171 by pointer, using vector_sizes rather than a bitmask.
172 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
173 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
174 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
176 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
177 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
178 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
179 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
180 * omp-general.c (omp_max_vf): Likewise.
181 * omp-low.c (omp_clause_aligned_alignment): Likewise.
182 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
183 * tree-vect-loop.c (vect_analyze_loop): Likewise.
184 * tree-vect-slp.c (vect_slp_bb): Likewise.
185 * doc/tm.texi: Regenerate.
186 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
188 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
189 the vector size as a poly_uint64 rather than an unsigned int.
190 (current_vector_size): Change from an unsigned int to a poly_uint64.
191 (get_vectype_for_scalar_type): Update accordingly.
192 * tree.h (build_truth_vector_type): Take the size and number of
193 units as a poly_uint64 rather than an unsigned int.
194 (build_vector_type): Add a temporary overload that takes
195 the number of units as a poly_uint64 rather than an unsigned int.
196 * tree.c (make_vector_type): Likewise.
197 (build_truth_vector_type): Take the number of units as a poly_uint64
198 rather than an unsigned int.
200 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
201 Alan Hayward <alan.hayward@arm.com>
202 David Sherwood <david.sherwood@arm.com>
204 * target.def (get_mask_mode): Take the number of units and length
205 as poly_uint64s rather than unsigned ints.
206 * targhooks.h (default_get_mask_mode): Update accordingly.
207 * targhooks.c (default_get_mask_mode): Likewise.
208 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
209 * doc/tm.texi: Regenerate.
211 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
212 Alan Hayward <alan.hayward@arm.com>
213 David Sherwood <david.sherwood@arm.com>
215 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
216 * omp-general.c (omp_max_vf): Likewise.
217 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
218 (expand_omp_simd): Handle polynomial safelen.
219 * omp-low.c (omplow_simd_context): Add a default constructor.
220 (omplow_simd_context::max_vf): Change from int to poly_uint64.
221 (lower_rec_simd_input_clauses): Update accordingly.
222 (lower_rec_input_clauses): Likewise.
224 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
225 Alan Hayward <alan.hayward@arm.com>
226 David Sherwood <david.sherwood@arm.com>
228 * tree-vectorizer.h (vect_nunits_for_cost): New function.
229 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
230 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
231 (vect_analyze_slp_cost): Likewise.
232 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
233 (vect_model_load_cost): Likewise.
235 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
236 Alan Hayward <alan.hayward@arm.com>
237 David Sherwood <david.sherwood@arm.com>
239 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
240 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
241 from an unsigned int * to a poly_uint64_pod *.
242 (calculate_unrolling_factor): New function.
243 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
245 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
246 Alan Hayward <alan.hayward@arm.com>
247 David Sherwood <david.sherwood@arm.com>
249 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
250 from an unsigned int to a poly_uint64.
251 (_loop_vec_info::slp_unrolling_factor): Likewise.
252 (_loop_vec_info::vectorization_factor): Change from an int
254 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
255 (vect_get_num_vectors): New function.
256 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
257 (vect_get_num_copies): Use vect_get_num_vectors.
258 (vect_analyze_data_ref_dependences): Change max_vf from an int *
259 to an unsigned int *.
260 (vect_analyze_data_refs): Change min_vf from an int * to a
262 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
263 than an unsigned HOST_WIDE_INT.
264 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
265 (vect_analyze_data_ref_dependence): Change max_vf from an int *
266 to an unsigned int *.
267 (vect_analyze_data_ref_dependences): Likewise.
268 (vect_compute_data_ref_alignment): Handle polynomial vf.
269 (vect_enhance_data_refs_alignment): Likewise.
270 (vect_prune_runtime_alias_test_list): Likewise.
271 (vect_shift_permute_load_chain): Likewise.
272 (vect_supportable_dr_alignment): Likewise.
273 (dependence_distance_ge_vf): Take the vectorization factor as a
274 poly_uint64 rather than an unsigned HOST_WIDE_INT.
275 (vect_analyze_data_refs): Change min_vf from an int * to a
277 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
278 vfm1 as a poly_uint64 rather than an int. Make the same change
279 for the returned bound_scalar.
280 (vect_gen_vector_loop_niters): Handle polynomial vf.
281 (vect_do_peeling): Likewise. Update call to
282 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
283 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
285 * tree-vect-loop.c (vect_determine_vectorization_factor)
286 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
287 (vect_get_known_peeling_cost): Likewise.
288 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
289 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
290 (vect_transform_loop): Likewise. Use the lowest possible VF when
291 updating the upper bounds of the loop.
292 (vect_min_worthwhile_factor): Make static. Return an unsigned int
294 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
295 polynomial unroll factors.
296 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
297 (vect_make_slp_decision): Likewise.
298 (vect_supported_load_permutation_p): Likewise, and polynomial
300 (vect_analyze_slp_cost): Handle polynomial vf.
301 (vect_slp_analyze_node_operations): Likewise.
302 (vect_slp_analyze_bb_1): Likewise.
303 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
304 than an unsigned HOST_WIDE_INT.
305 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
306 (vectorizable_load): Handle polynomial vf.
307 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
309 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
311 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
312 Alan Hayward <alan.hayward@arm.com>
313 David Sherwood <david.sherwood@arm.com>
315 * match.pd: Handle bit operations involving three constants
316 and try to fold one pair.
318 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
320 * tree-vect-loop-manip.c: Include gimple-fold.h.
321 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
322 niters_maybe_zero parameters. Handle other cases besides a step of 1.
323 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
324 Add a path that uses a step of VF instead of 1, but disable it
326 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
327 and niters_no_overflow parameters. Update calls to
328 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
329 Create a new SSA name if the latter choses to use a ste other
330 than zero, and return it via niters_vector_mult_vf_var.
331 * tree-vect-loop.c (vect_transform_loop): Update calls to
332 vect_do_peeling, vect_gen_vector_loop_niters and
333 slpeel_make_loop_iterate_ntimes.
334 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
335 (vect_gen_vector_loop_niters): Update declarations after above changes.
337 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
339 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
340 128-bit round to integer instructions.
341 (ceil<mode>2): Likewise.
342 (btrunc<mode>2): Likewise.
343 (round<mode>2): Likewise.
345 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
347 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
348 unaligned VSX load/store on P8/P9.
349 (expand_block_clear): Allow the use of unaligned VSX
352 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
354 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
356 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
357 swap associated with both a load and a store.
359 2018-01-02 Andrew Waterman <andrew@sifive.com>
361 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
362 * config/riscv/riscv.md (clear_cache): Use it.
364 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
366 * web.c: Remove out-of-date comment.
368 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
370 * expr.c (fixup_args_size_notes): Check that any existing
371 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
372 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
373 (emit_single_push_insn): ...here.
375 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
377 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
378 (const_vector_encoded_nelts): New function.
379 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
380 (const_vector_int_elt, const_vector_elt): Declare.
381 * emit-rtl.c (const_vector_int_elt_1): New function.
382 (const_vector_elt): Likewise.
383 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
386 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
388 * expr.c: Include rtx-vector-builder.h.
389 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
390 directly on the tree encoding.
391 (const_vector_from_tree): Likewise.
392 * optabs.c: Include rtx-vector-builder.h.
393 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
394 sequence of "u" values.
395 * vec-perm-indices.c: Include rtx-vector-builder.h.
396 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
397 directly on the vec_perm_indices encoding.
399 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
401 * doc/rtl.texi (const_vector): Describe new encoding scheme.
402 * Makefile.in (OBJS): Add rtx-vector-builder.o.
403 * rtx-vector-builder.h: New file.
404 * rtx-vector-builder.c: Likewise.
405 * rtl.h (rtx_def::u2): Add a const_vector field.
406 (CONST_VECTOR_NPATTERNS): New macro.
407 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
408 (CONST_VECTOR_DUPLICATE_P): Likewise.
409 (CONST_VECTOR_STEPPED_P): Likewise.
410 (CONST_VECTOR_ENCODED_ELT): Likewise.
411 (const_vec_duplicate_p): Check for a duplicated vector encoding.
412 (unwrap_const_vec_duplicate): Likewise.
413 (const_vec_series_p): Check for a non-duplicated vector encoding.
414 Say that the function only returns true for integer vectors.
415 * emit-rtl.c: Include rtx-vector-builder.h.
416 (gen_const_vec_duplicate_1): Delete.
417 (gen_const_vector): Call gen_const_vec_duplicate instead of
418 gen_const_vec_duplicate_1.
419 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
420 (gen_const_vec_duplicate): Use rtx_vector_builder.
421 (gen_const_vec_series): Likewise.
422 (gen_rtx_CONST_VECTOR): Likewise.
423 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
424 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
425 Build a new vector rather than modifying a CONST_VECTOR in-place.
426 (handle_special_swappables): Update call accordingly.
427 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
428 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
429 Build a new vector rather than modifying a CONST_VECTOR in-place.
430 (handle_special_swappables): Update call accordingly.
432 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
434 * simplify-rtx.c (simplify_const_binary_operation): Use
435 CONST_VECTOR_ELT instead of XVECEXP.
437 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
439 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
440 the selector elements to be different from the data elements
441 if the selector is a VECTOR_CST.
442 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
443 ssizetype for the selector.
445 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
447 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
448 before testing each element individually.
449 * tree-vect-generic.c (lower_vec_perm): Likewise.
451 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
453 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
454 * selftest-run-tests.c (selftest::run_tests): Call it.
455 * vector-builder.h (vector_builder::operator ==): New function.
456 (vector_builder::operator !=): Likewise.
457 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
458 (vec_perm_indices::all_from_input_p): New function.
459 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
460 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
461 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
462 instead of reading the VECTOR_CST directly. Detect whether both
463 vector inputs are the same before constructing the vec_perm_indices,
464 and update the number of inputs argument accordingly. Use the
465 utility functions added above. Only construct sel2 if we need to.
467 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
469 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
470 the broadcast of the low byte.
471 (expand_mult_highpart): Use an explicit encoding for the permutes.
472 * optabs-query.c (can_mult_highpart_p): Likewise.
473 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
474 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
475 (vectorizable_bswap): Likewise.
476 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
477 explicit encoding for the power-of-2 permutes.
478 (vect_permute_store_chain): Likewise.
479 (vect_grouped_load_supported): Likewise.
480 (vect_permute_load_chain): Likewise.
482 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
484 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
485 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
486 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
487 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
488 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
489 (vect_gen_perm_mask_any): Likewise.
491 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
493 * int-vector-builder.h: New file.
494 * vec-perm-indices.h: Include int-vector-builder.h.
495 (vec_perm_indices): Redefine as an int_vector_builder.
496 (auto_vec_perm_indices): Delete.
497 (vec_perm_builder): Redefine as a stand-alone class.
498 (vec_perm_indices::vec_perm_indices): New function.
499 (vec_perm_indices::clamp): Likewise.
500 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
501 (vec_perm_indices::new_vector): New function.
502 (vec_perm_indices::new_expanded_vector): Update for new
503 vec_perm_indices class.
504 (vec_perm_indices::rotate_inputs): New function.
505 (vec_perm_indices::all_in_range_p): Operate directly on the
506 encoded form, without computing elided elements.
507 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
508 encoding. Update for new vec_perm_indices class.
509 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
510 the given vec_perm_builder.
511 (expand_vec_perm_var): Update vec_perm_builder constructor.
512 (expand_mult_highpart): Use vec_perm_builder instead of
513 auto_vec_perm_indices.
514 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
515 vec_perm_indices instead of auto_vec_perm_indices. Use a single
516 or double series encoding as appropriate.
517 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
518 vec_perm_indices instead of auto_vec_perm_indices.
519 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
520 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
521 (vect_permute_store_chain): Likewise.
522 (vect_grouped_load_supported): Likewise.
523 (vect_permute_load_chain): Likewise.
524 (vect_shift_permute_load_chain): Likewise.
525 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
526 (vect_transform_slp_perm_load): Likewise.
527 (vect_schedule_slp_instance): Likewise.
528 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
529 (vectorizable_mask_load_store): Likewise.
530 (vectorizable_bswap): Likewise.
531 (vectorizable_store): Likewise.
532 (vectorizable_load): Likewise.
533 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
534 vec_perm_indices instead of auto_vec_perm_indices. Use
535 tree_to_vec_perm_builder to read the vector from a tree.
536 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
537 vec_perm_builder instead of a vec_perm_indices.
538 (have_whole_vector_shift): Use vec_perm_builder and
539 vec_perm_indices instead of auto_vec_perm_indices. Leave the
540 truncation to calc_vec_perm_mask_for_shift.
541 (vect_create_epilog_for_reduction): Likewise.
542 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
543 from auto_vec_perm_indices to vec_perm_indices.
544 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
545 instead of changing individual elements.
546 (aarch64_vectorize_vec_perm_const): Use new_vector to install
547 the vector in d.perm.
548 * config/arm/arm.c (expand_vec_perm_d::perm): Change
549 from auto_vec_perm_indices to vec_perm_indices.
550 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
551 instead of changing individual elements.
552 (arm_vectorize_vec_perm_const): Use new_vector to install
553 the vector in d.perm.
554 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
555 Update vec_perm_builder constructor.
556 (rs6000_expand_interleave): Likewise.
557 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
558 (rs6000_expand_interleave): Likewise.
560 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
562 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
563 to qimode could truncate the indices.
564 * optabs.c (expand_vec_perm_var): Likewise.
566 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
568 * Makefile.in (OBJS): Add vec-perm-indices.o.
569 * vec-perm-indices.h: New file.
570 * vec-perm-indices.c: Likewise.
571 * target.h (vec_perm_indices): Replace with a forward class
573 (auto_vec_perm_indices): Move to vec-perm-indices.h.
574 * optabs.h: Include vec-perm-indices.h.
575 (expand_vec_perm): Delete.
576 (selector_fits_mode_p, expand_vec_perm_var): Declare.
577 (expand_vec_perm_const): Declare.
578 * target.def (vec_perm_const_ok): Replace with...
579 (vec_perm_const): ...this new hook.
580 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
581 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
582 * doc/tm.texi: Regenerate.
583 * optabs.def (vec_perm_const): Delete.
584 * doc/md.texi (vec_perm_const): Likewise.
585 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
586 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
587 expand_vec_perm for constant permutation vectors. Assert that
588 the mode of variable permutation vectors is the integer equivalent
589 of the mode that is being permuted.
590 * optabs-query.h (selector_fits_mode_p): Declare.
591 * optabs-query.c: Include vec-perm-indices.h.
592 (selector_fits_mode_p): New function.
593 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
594 is defined, instead of checking whether the vec_perm_const_optab
595 exists. Use targetm.vectorize.vec_perm_const instead of
596 targetm.vectorize.vec_perm_const_ok. Check whether the indices
597 fit in the vector mode before using a variable permute.
598 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
599 vec_perm_indices instead of an rtx.
600 (expand_vec_perm): Replace with...
601 (expand_vec_perm_const): ...this new function. Take the selector
602 as a vec_perm_indices rather than an rtx. Also take the mode of
603 the selector. Update call to shift_amt_for_vec_perm_mask.
604 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
605 Use vec_perm_indices::new_expanded_vector to expand the original
606 selector into bytes. Check whether the indices fit in the vector
607 mode before using a variable permute.
608 (expand_vec_perm_var): Make global.
609 (expand_mult_highpart): Use expand_vec_perm_const.
610 * fold-const.c: Includes vec-perm-indices.h.
611 * tree-ssa-forwprop.c: Likewise.
612 * tree-vect-data-refs.c: Likewise.
613 * tree-vect-generic.c: Likewise.
614 * tree-vect-loop.c: Likewise.
615 * tree-vect-slp.c: Likewise.
616 * tree-vect-stmts.c: Likewise.
617 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
619 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
620 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
621 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
622 (aarch64_vectorize_vec_perm_const): ...this new function.
623 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
624 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
625 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
626 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
627 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
628 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
629 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
631 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
632 check for NEON modes.
633 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
634 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
635 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
636 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
638 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
639 the old VEC_PERM_CONST conditions.
640 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
641 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
642 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
643 (ia64_vectorize_vec_perm_const_ok): Merge into...
644 (ia64_vectorize_vec_perm_const): ...this new function.
645 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
646 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
647 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
648 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
649 * config/mips/mips.c (mips_expand_vec_perm_const)
650 (mips_vectorize_vec_perm_const_ok): Merge into...
651 (mips_vectorize_vec_perm_const): ...this new function.
652 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
653 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
654 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
655 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
656 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
657 (rs6000_expand_vec_perm_const): Delete.
658 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
660 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
661 (altivec_expand_vec_perm_const_le): Take each operand individually.
662 Operate on constant selectors rather than rtxes.
663 (altivec_expand_vec_perm_const): Likewise. Update call to
664 altivec_expand_vec_perm_const_le.
665 (rs6000_expand_vec_perm_const): Delete.
666 (rs6000_vectorize_vec_perm_const_ok): Delete.
667 (rs6000_vectorize_vec_perm_const): New function.
668 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
669 an element count and rtx array.
670 (rs6000_expand_extract_even): Update call accordingly.
671 (rs6000_expand_interleave): Likewise.
672 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
673 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
674 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
675 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
676 (rs6000_expand_vec_perm_const): Delete.
677 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
678 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
679 (altivec_expand_vec_perm_const_le): Take each operand individually.
680 Operate on constant selectors rather than rtxes.
681 (altivec_expand_vec_perm_const): Likewise. Update call to
682 altivec_expand_vec_perm_const_le.
683 (rs6000_expand_vec_perm_const): Delete.
684 (rs6000_vectorize_vec_perm_const_ok): Delete.
685 (rs6000_vectorize_vec_perm_const): New function. Remove stray
686 reference to the SPE evmerge intructions.
687 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
688 an element count and rtx array.
689 (rs6000_expand_extract_even): Update call accordingly.
690 (rs6000_expand_interleave): Likewise.
691 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
692 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
694 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
696 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
698 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
699 vector mode and that that mode matches the mode of the data
701 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
702 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
703 directly using expand_vec_perm_1 when forcing selectors into
705 (expand_vec_perm_var): New function, split out from expand_vec_perm.
707 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
709 * optabs-query.h (can_vec_perm_p): Delete.
710 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
711 * optabs-query.c (can_vec_perm_p): Split into...
712 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
713 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
714 particular selector is valid.
715 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
716 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
717 (vect_grouped_load_supported): Likewise.
718 (vect_shift_permute_load_chain): Likewise.
719 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
720 (vect_transform_slp_perm_load): Likewise.
721 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
722 (vectorizable_bswap): Likewise.
723 (vect_gen_perm_mask_checked): Likewise.
724 * fold-const.c (fold_ternary_loc): Likewise. Don't take
725 implementations of variable permutation vectors into account
726 when deciding which selector to use.
727 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
728 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
729 with a false third argument.
730 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
731 to test whether the constant selector is valid and can_vec_perm_var_p
732 to test whether a variable selector is valid.
734 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
736 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
737 * optabs-query.c (can_vec_perm_p): Likewise.
738 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
739 instead of vec_perm_indices.
740 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
741 (vect_gen_perm_mask_checked): Likewise,
742 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
743 (vect_gen_perm_mask_checked): Likewise,
745 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
747 * optabs-query.h (qimode_for_vec_perm): Declare.
748 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
749 (qimode_for_vec_perm): ...this new function.
750 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
752 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
754 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
755 does not have a conditional at the top.
757 2018-01-02 Richard Biener <rguenther@suse.de>
759 * ipa-inline.c (big_speedup_p): Fix expression.
761 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
764 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
767 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
771 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
772 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
773 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
774 cond_taken_branch_cost 3->4.
776 2018-01-01 Jakub Jelinek <jakub@redhat.com>
778 PR tree-optimization/83581
779 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
780 TODO_cleanup_cfg if any changes have been made.
783 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
784 convert_modes if target mode has the right side, but different mode
788 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
789 last argument when extracting from CONCAT. If either from_real or
790 from_imag is NULL, use expansion through memory. If result is not
791 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
792 the parts directly to inner mode, if even that fails, use expansion
796 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
797 check for bswap in mode rather than HImode and use that in expand_unop
800 2017-12-31 Jakub Jelinek <jakub@redhat.com>
803 * config/i386/i386.c (ix86_attribute_table): Remove excess
804 initializer for "shared" attribute.
806 2017-12-30 Tom de Vries <tom@codesourcery.com>
809 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
811 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
812 offload_funcs entries that no longer have a corresponding cgraph_node.
813 Mark the remaining ones as DECL_PRESERVE_P.
814 (output_lto): Call prune_offload_funcs.
816 2017-12-30 Jakub Jelinek <jakub@redhat.com>
818 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
819 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
820 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
821 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
822 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
823 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
824 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
825 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
826 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
827 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
828 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
829 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
830 vaesenc_<mode>, vpclmulqdq_<mode>,
831 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
833 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
835 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
836 _Float<N> and _Float<N>X variants for rounding built-in
838 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
839 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
840 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
841 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
842 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
843 * builtins.c (mathfn_built_in_2): Likewise.
844 * internal-fn.def (CEIL): Likewise.
846 (NEARBYINT): Likewise.
850 * convert.c (convert_to_integer_1): Likewise.
851 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
852 (integer_valued_real_call_p): Likewise.
853 * fold-const-call.c (fold_const_call_ss): Likewise.
854 * gencfn-macros.c (print_case_cfn): Change CFN and operator
855 printers to take a const char * suffix instead of a bool.
856 (print_define_operator_list): Likewise.
857 (fltall_suffixes): New list of suffixes, that include the
858 traditional suffixes as well as all of the _Float<N> and
860 (main): For _Float<N> and _Float<N>X functions, emit both
861 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
862 has the _Float<N> and _Float<N>X case names or operators. The
863 <name>_ALL variant has both the traditional and the
864 _Float<N>/_Float<N>X case names or operators.
865 * match.pd (COPYSIGN optimizations): Provide optimizations for
866 _Float<N> and _Float<N>X types where possible.
867 (MIN/MAX optimizations): Likewise.
868 (sqrt optimizations): Likewise.
869 (rounding optimizations): Likewise.
871 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
873 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
874 gen_int_mode rather than GEN_INT.
875 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
877 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
879 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
880 instead of CONSTANT_P.
881 (gen_vec_series): Likewise.
882 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
884 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
886 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
889 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
891 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
892 (aarch64_ld1x2<VDC:mode>): Likewise.
893 (aarch64_simd_ld1<mode>_x2): Likewise.
894 (aarch64_simd_ld1<mode>_x2): Likewise.
895 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
896 (vld1_s8_x2): Likewise.
897 (vld1_u16_x2): Likewise.
898 (vld1_s16_x2): Likewise.
899 (vld1_u32_x2): Likewise.
900 (vld1_s32_x2): Likewise.
901 (vld1_u64_x2): Likewise.
902 (vld1_s64_x2): Likewise.
903 (vld1_f16_x2): Likewise.
904 (vld1_f32_x2): Likewise.
905 (vld1_f64_x2): Likewise.
906 (vld1_p8_x2): Likewise.
907 (vld1_p16_x2): Likewise.
908 (vld1_p64_x2): Likewise.
909 (vld1q_u8_x2): Likewise.
910 (vld1q_s8_x2): Likewise.
911 (vld1q_u16_x2): Likewise.
912 (vld1q_s16_x2): Likewise.
913 (vld1q_u32_x2): Likewise.
914 (vld1q_s32_x2): Likewise.
915 (vld1q_u64_x2): Likewise.
916 (vld1q_s64_x2): Likewise.
917 (vld1q_f16_x2): Likewise.
918 (vld1q_f32_x2): Likewise.
919 (vld1q_f64_x2): Likewise.
920 (vld1q_p8_x2): Likewise.
921 (vld1q_p16_x2): Likewise.
922 (vld1q_p64_x2): Likewise.
924 2017-12-27 Martin Liska <mliska@suse.cz>
926 PR tree-optimization/83552
927 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
928 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
930 2017-12-27 Tom de Vries <tom@codesourcery.com>
932 * config/nvptx/nvptx.c (nvptx_option_override): Disable
933 -gstatement-frontiers.
935 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
937 PR rtl-optimization/83513
938 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
939 before priority comparison.
941 2017-12-25 Jakub Jelinek <jakub@redhat.com>
944 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
945 ix86_isa_flags2 to ix86_isa_flags.
946 * config/i386/i386-c.c (ix86_target_macros_internal): Test
947 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
948 isa_flags rather than isa_flags2.
949 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
950 and -mavx512bitalg from isa2_opts to isa_opts.
951 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
952 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
953 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
955 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
956 OPTION_MASK_ISA_AVX512F ored with another option similarly to
957 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
958 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
959 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
960 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
961 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
962 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
963 builtins from bdesc_args2 to bdesc_args section.
964 (__builtin_ia32_compressstoreuqi512_mask,
965 __builtin_ia32_compressstoreuhi512_mask,
966 __builtin_ia32_compressstoreuqi256_mask,
967 __builtin_ia32_expandloadqi512_mask,
968 __builtin_ia32_expandloadqi512_maskz,
969 __builtin_ia32_expandloadhi512_mask,
970 __builtin_ia32_expandloadhi512_maskz,
971 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
972 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
973 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
974 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
975 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
976 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
977 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
978 __builtin_ia32_vpshldv_v32hi_maskz,
979 __builtin_ia32_vpopcountb_v64qi_mask,
980 __builtin_ia32_vpopcountw_v32hi_mask,
981 __builtin_ia32_vpshufbitqmb512_mask,
982 __builtin_ia32_vpshufbitqmb256_mask): Add
983 " | OPTION_MASK_ISA_AVX512BW".
984 (__builtin_ia32_expandloadqi256_mask,
985 __builtin_ia32_expandloadqi256_maskz,
986 __builtin_ia32_vpopcountb_v32qi_mask): Add
987 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
988 (__builtin_ia32_expandloadhi256_mask,
989 __builtin_ia32_expandloadhi256_maskz,
990 __builtin_ia32_expandloadqi128_mask,
991 __builtin_ia32_expandloadqi128_maskz,
992 __builtin_ia32_expandloadhi128_mask,
993 __builtin_ia32_expandloadhi128_maskz,
994 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
995 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
996 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
997 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
998 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
999 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
1000 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
1001 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
1002 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
1003 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
1004 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
1005 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
1006 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
1007 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
1008 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
1009 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
1010 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
1011 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
1012 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
1013 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
1014 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
1015 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1016 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1017 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1018 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
1019 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
1020 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
1021 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1022 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1023 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
1024 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
1025 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
1026 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1027 __builtin_ia32_vpopcountw_v8hi_mask): Add
1028 " | OPTION_MASK_ISA_AVX512VL".
1029 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1030 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
1031 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
1032 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
1033 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
1034 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
1035 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1036 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
1037 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
1038 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
1039 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
1040 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
1041 _mm_mask_bitshuffle_epi64_mask): Likewise.
1042 * common/config/i386/i386-common.c
1043 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
1044 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
1045 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1046 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
1047 OPTION_MASK_ISA_AVX512BITALG_UNSET.
1048 (OPTION_MASK_ISA2_AVX512F_UNSET,
1049 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
1050 (ix86_handle_option): For -mno-general-regs-only, clear from
1051 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
1052 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
1053 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
1054 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
1055 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
1056 for moving from ix86_isa_flags2 to ix86_isa_flags.
1058 2017-12-23 Jakub Jelinek <jakub@redhat.com>
1061 * fold-const.c (struct contains_label_data): New type.
1062 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
1063 inside of a SWITCH_BODY seen during the walk.
1064 (contains_label_p): Use walk_tree instead of
1065 walk_tree_without_duplicates, prepare data for contains_label_1 and
1068 2017-12-22 Jakub Jelinek <jakub@redhat.com>
1071 * tree-iterator.c (alloc_stmt_list): Start with cleared
1072 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
1076 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
1077 enable avx512f explicitly in #pragma GCC target.
1078 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
1079 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
1080 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
1081 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
1082 __builtin_ia32_vpdpbusds_v8si_mask,
1083 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
1084 __builtin_ia32_vpdpbusds_v4si_mask,
1085 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
1086 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
1087 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
1088 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
1089 __builtin_ia32_vpdpwssds_v8si_mask,
1090 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
1091 __builtin_ia32_vpdpwssds_v4si_mask,
1092 __builtin_ia32_vpdpwssds_v4si_maskz): Use
1093 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
1094 just OPTION_MASK_ISA_AVX512VNNI.
1096 2017-12-22 Martin Jambor <mjambor@suse.cz>
1099 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
1102 2017-12-22 Julia Koval <julia.koval@intel.com>
1103 Sebastian Peryt <sebastian.peryt@intel.com>
1105 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
1106 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
1107 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
1108 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
1109 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
1110 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
1111 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
1112 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
1113 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
1114 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
1115 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
1116 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
1117 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
1118 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
1119 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
1120 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
1121 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
1122 _mm256_maskz_popcnt_epi64): New intrinsics.
1123 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
1124 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
1125 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
1126 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
1127 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
1128 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
1129 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
1130 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
1131 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
1132 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
1133 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
1134 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
1135 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
1136 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
1137 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1138 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
1139 __builtin_ia32_vpshufbitqmb256_mask,
1140 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
1141 * config/i386/i386-c.c (__AVX512BITALG__): New.
1142 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
1143 (ix86_valid_target_attribute_inner_p): Ditto.
1144 (ix86_expand_args_builtin): Handle new types.
1145 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
1146 * config/i386/i386.opt: Add -mavx512bitalg.
1147 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
1148 avx512bitalgintrin.h.
1149 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
1150 (vpopcount<mode><mask_name>): Add more types.
1151 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
1152 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
1154 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1156 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
1157 Or in OPTION_MASK_ISA_AVX512F_SET.
1158 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1159 OPTION_MASK_ISA_AVX512VNNI_UNSET.
1160 (ix86_handle_option): Adjust for
1161 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
1162 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
1164 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
1165 isa_flag instead of isa_flag2.
1166 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
1167 isa_opts2 to isa_opts.
1168 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
1171 2017-12-22 Mike Stump <mikestump@comcast.net>
1172 Eric Botcazou <ebotcazou@adacore.com>
1174 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
1176 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1179 * config/i386/i386.c (ix86_function_arg_boundary): Return
1180 PARM_BOUNDARY for TYPE_EMPTY_P types.
1183 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
1184 if navail is >= dir.len.
1186 2017-12-21 Steve Ellcey <sellcey@cavium.com>
1188 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
1191 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1193 PR rtl-optimization/80747
1194 PR rtl-optimization/83512
1195 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
1196 succ edge from ENTRY, copy partition from e->dest to the newly
1198 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
1199 ENTRY, use BB_PARTITION of its successor block as current_partition.
1200 Don't copy partition when splitting succ edge from ENTRY.
1202 PR tree-optimization/83523
1203 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
1204 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
1205 (convert_mult_to_fma): Likewise.
1207 PR tree-optimization/83521
1208 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
1209 gimple_build_assign without code on result of
1210 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
1211 a VIEW_CONVERT_EXPR.
1213 2017-12-21 Andrew Pinski <apinski@cavium.com>
1214 Steve Ellcey <sellcey@cavium.com>
1216 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
1217 multi-arch for ilp32.
1219 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
1222 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1223 constraints to operand 2.
1224 (*ashl<mode>3_mask_1): Ditto.
1225 (*<shift_insn><mode>3_mask): Ditto.
1226 (*<shift_insn><mode>3_mask_1): Ditto.
1227 (*<rotate_insn><mode>3_mask): Ditto.
1228 (*<rotate_insn><mode>3_mask_1): Ditto.
1230 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1232 * reorg.c (make_return_insns): Reemit each insn with its own location.
1234 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1236 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1237 generation for cases where splatting a value is not useful.
1238 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1239 across a vec_duplicate and a paradoxical subreg forming a vector
1240 mode to a vec_concat.
1242 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1244 * combine.c (simplify_set): Do not transform subregs to zero_extends
1245 if the destination is not a scalar int mode.
1247 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1250 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1251 shared zero if the input has overflowed.
1253 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1255 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1256 support for Cortex-A55 and Cortex-A75.
1258 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1260 * common/config/arm/arm-common.c (compare_opt_names): Add function
1261 comment. Use strcmp instead of manual loop.
1263 2017-12-21 Martin Liska <mliska@suse.cz>
1265 PR gcov-profile/83509
1266 * gcov-dump.c (dump_gcov_file): Do not read info about
1267 support_unexecuted_blocks for gcda files.
1269 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1271 PR rtl-optimization/82973
1272 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1273 (valid_for_const_vector_p): ... this.
1274 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1275 (valid_for_const_vector_p): ... this. Adjust function comment.
1276 (gen_vec_duplicate): Adjust caller.
1277 * optabs.c (expand_vector_broadcast): Likewise.
1278 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1279 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1281 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1282 instead of GET_CODE == CONST_FIXED.
1283 (simplify_subreg): Use CONST_FIXED_P macro instead of
1284 GET_CODE == CONST_FIXED.
1287 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1288 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1289 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1290 (ix86_option_override_internal): Adjust for
1291 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1292 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1293 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1294 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1295 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1296 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1297 def_builtin2 for CET builtins.
1298 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1299 up formatting in IX86_BUILTIN_RDPID code.
1300 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1301 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1302 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1303 ix86_isa_flags2 to ix86_isa_flags.
1304 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1306 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1307 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1308 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1310 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1311 Or in OPTION_MASK_ISA_AVX512F_SET.
1312 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1313 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1314 (ix86_handle_option): Adjust for
1315 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1316 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
1318 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1319 Alan Hayward <alan.hayward@arm.com>
1320 David Sherwood <david.sherwood@arm.com>
1322 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
1323 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1324 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1325 Track polynomial offsets.
1327 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1328 Alan Hayward <alan.hayward@arm.com>
1329 David Sherwood <david.sherwood@arm.com>
1331 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1332 Treat drb->init as a poly_int. Fail if its misalignment wrt
1333 vector_alignment isn't known.
1335 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1336 Alan Hayward <alan.hayward@arm.com>
1337 David Sherwood <david.sherwood@arm.com>
1339 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1341 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1342 (vect_loop_versioning): Take the loop versioning threshold as a
1344 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1345 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1346 versioning_threshold.
1347 (vect_analyze_loop_2): Compute the loop versioning threshold
1348 whenever loop versioning is needed, and store it in the new
1349 field rather than combining it with the cost model threshold.
1350 (vect_transform_loop): Update call to vect_loop_versioning.
1351 Try to combine the loop versioning and cost thresholds here.
1353 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1354 Alan Hayward <alan.hayward@arm.com>
1355 David Sherwood <david.sherwood@arm.com>
1357 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1358 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1359 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1360 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1361 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1362 (group_compare_offset): Update accordingly.
1363 (split_small_address_groups_p): Likewise.
1364 (record_use): Take addr_offset as a poly_uint64 rather than
1365 an unsigned HOST_WIDE_INT.
1366 (strip_offset): Return the offset as a poly_uint64 rather than
1367 an unsigned HOST_WIDE_INT.
1368 (record_group_use, split_address_groups): Track polynomial offsets.
1369 (add_iv_candidate_for_use): Likewise.
1370 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1371 than a HOST_WIDE_INT.
1372 (strip_offset_1): Return the offset as a poly_int64 rather than
1375 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1376 Alan Hayward <alan.hayward@arm.com>
1377 David Sherwood <david.sherwood@arm.com>
1379 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1380 rather than a HOST_WIDE_INT.
1381 * tree.c (get_binfo_at_offset): Likewise.
1383 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1384 Alan Hayward <alan.hayward@arm.com>
1385 David Sherwood <david.sherwood@arm.com>
1387 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1388 rather than a HOST_WIDE_INT.
1389 * tree-sra.c (build_ref_for_offset): Likewise.
1391 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1392 Alan Hayward <alan.hayward@arm.com>
1393 David Sherwood <david.sherwood@arm.com>
1395 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1397 * tree.c (mem_ref_offset): Likewise.
1398 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1399 * builtins.c (get_object_alignment_2): Likewise.
1400 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1401 * gimple-fold.c (get_base_constructor): Likewise.
1402 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1403 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1405 * ipa-polymorphic-call.c
1406 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1407 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1408 (get_ancestor_addr_info): Likewise.
1409 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1410 * match.pd: Likewise.
1411 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1412 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1413 * tree-eh.c (tree_could_trap_p): Likewise.
1414 * tree-object-size.c (addr_object_size): Likewise.
1415 * tree-ssa-address.c (copy_ref_info): Likewise.
1416 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1417 (indirect_refs_may_alias_p): Likewise.
1418 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1419 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1420 (non_rewritable_mem_ref_base): Likewise.
1421 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1422 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1423 * varasm.c (decode_addr_const): Likewise.
1425 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1426 Alan Hayward <alan.hayward@arm.com>
1427 David Sherwood <david.sherwood@arm.com>
1429 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1430 offsets for MEM_REFs.
1432 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1433 Alan Hayward <alan.hayward@arm.com>
1434 David Sherwood <david.sherwood@arm.com>
1436 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1437 as a poly_uint64 rather than an unsigned int.
1438 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1440 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1441 Alan Hayward <alan.hayward@arm.com>
1442 David Sherwood <david.sherwood@arm.com>
1444 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1446 (decode_addr_const): Update accordingly.
1448 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1449 Alan Hayward <alan.hayward@arm.com>
1450 David Sherwood <david.sherwood@arm.com>
1452 * tree.h (bit_field_size, bit_field_offset): New functions.
1453 * hsa-gen.c (gen_hsa_addr): Use them.
1454 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1455 (simplify_vector_constructor): Likewise.
1456 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1457 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1458 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1459 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
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 * expr.h (emit_group_load, emit_group_load_into_temps)
1467 (emit_group_store): Take the size as a poly_int64 rather than an int.
1468 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1469 (emit_group_load_into_temp, emit_group_store): Likewise.
1471 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1472 Alan Hayward <alan.hayward@arm.com>
1473 David Sherwood <david.sherwood@arm.com>
1475 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1476 an unsigned int to a poly_uint64.
1477 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1478 sizes as poly_uint64s rather than unsigned ints.
1479 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1482 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1483 Alan Hayward <alan.hayward@arm.com>
1484 David Sherwood <david.sherwood@arm.com>
1486 * lra-constraints.c (emit_inc): Change inc_amount from an int
1489 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1490 Alan Hayward <alan.hayward@arm.com>
1491 David Sherwood <david.sherwood@arm.com>
1493 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1495 (add_stack_var, stack_var_cmp, partition_stack_vars)
1496 (dump_stack_var_partition): Update accordingly.
1497 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1498 than a HOST_WIDE_INT.
1499 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1500 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1501 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1502 rather than a HOST_WIDE_INT.
1504 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1505 Alan Hayward <alan.hayward@arm.com>
1506 David Sherwood <david.sherwood@arm.com>
1508 * target.def (return_pops_args): Treat both the input and output
1509 sizes as poly_int64s rather than HOST_WIDE_INTS.
1510 * targhooks.h (default_return_pops_args): Update accordingly.
1511 * targhooks.c (default_return_pops_args): Likewise.
1512 * doc/tm.texi: Regenerate.
1513 * emit-rtl.h (incoming_args): Change pops_args, size and
1514 outgoing_args_size from int to poly_int64_pod.
1515 * function.h (expr_status): Change x_pending_stack_adjust and
1516 x_stack_pointer_delta from int to poly_int64.
1517 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1518 (ARGS_SIZE_RTX): Update accordingly.
1519 * calls.c (highest_outgoing_arg_in_use): Change from int to
1521 (stack_usage_watermark, stored_args_watermark): New variables.
1522 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1523 (emit_call_1): Change the stack_size and rounded_stack_size
1524 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1526 (save_fixed_argument_area): Check stack_usage_watermark.
1527 (initialize_argument_information): Change old_pending_adj from
1528 a HOST_WIDE_INT * to a poly_int64_pod *.
1529 (compute_argument_block_size): Return the size as a poly_int64
1531 (finalize_must_preallocate): Track polynomial argument sizes.
1532 (compute_argument_addresses): Likewise.
1533 (internal_arg_pointer_based_exp): Track polynomial offsets.
1534 (mem_overlaps_already_clobbered_arg_p): Rename to...
1535 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1536 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1537 Check stored_args_used_watermark.
1538 (load_register_parameters): Update accordingly.
1539 (check_sibcall_argument_overlap_1): Likewise.
1540 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1541 args size as a poly_int64 rather than an int. Return a bool
1542 indicating whether the optimization was possible and return
1543 the new adjustment by reference.
1544 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1545 Update stored_args_watermark.
1546 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1547 (expand_call): Likewise. Maintain stack_usage_watermark and
1548 stored_args_watermark. Update calls to
1549 combine_pending_stack_adjustment_and_call.
1550 (emit_library_call_value_1): Handle polynomial argument sizes.
1551 Call stack_region_maybe_used_p and mark_stack_region_used.
1552 Maintain stack_usage_watermark.
1553 (store_one_arg): Likewise. Update call to
1554 mem_overlaps_already_clobbered_arg_p.
1555 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1557 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1558 * config/microblaze/microblaze.c (microblaze_function_prologue):
1560 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1561 TARGET_RETURN_POPS_ARGS interface.
1562 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1564 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1565 * config/i386/i386.c (ix86_return_pops_args): Update for new
1566 TARGET_RETURN_POPS_ARGS interface.
1567 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1568 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1569 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1570 TARGET_RETURN_POPS_ARGS interface.
1571 * config/vax/vax.c (vax_return_pops_args): Likewise.
1572 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1573 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1574 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1575 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1576 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1577 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1578 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1579 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1580 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1581 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1582 and x_stack_pointer_delta from int to poly_int64.
1583 * dojump.c (do_pending_stack_adjust): Update accordingly.
1584 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1585 stack_pointer_deltas.
1586 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1587 (pad_to_arg_alignment): Track polynomial offsets.
1588 (assign_parm_find_stack_rtl): Likewise.
1589 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1590 * toplev.c (output_stack_usage): Update reference to
1591 current_function_pushed_stack_size.
1593 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1594 Alan Hayward <alan.hayward@arm.com>
1595 David Sherwood <david.sherwood@arm.com>
1597 * function.c (in_arg_offset, var_offset, dynamic_offset)
1598 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1599 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1600 rather than a HOST_WIDE_INT.
1601 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1602 (instantiate_virtual_regs_in_insn): Likewise.
1604 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1605 Alan Hayward <alan.hayward@arm.com>
1606 David Sherwood <david.sherwood@arm.com>
1608 * rtl.h (get_args_size, add_args_size_note): New functions.
1609 (find_args_size_adjust): Return a poly_int64 rather than a
1611 (fixup_args_size_notes): Likewise. Make the same change to the
1612 end_args_size parameter.
1613 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1614 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1615 * calls.c (emit_call_1): Likewise.
1616 * explow.c (adjust_stack_1): Likewise.
1617 * cfgcleanup.c (old_insns_match_p): Update use of
1618 find_args_size_adjust.
1619 * combine.c (distribute_notes): Track polynomial arg sizes.
1620 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1621 end_true_args_size, beg_delay_args_size and end_delay_args_size
1622 from HOST_WIDE_INT to poly_int64.
1623 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1624 than a HOST_WIDE_INT.
1625 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1626 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1627 polynomial arg sizes.
1628 * emit-rtl.c (try_split): Use get_args_size.
1629 * recog.c (peep2_attempt): Likewise.
1630 * reload1.c (reload_as_needed): Likewise.
1631 * expr.c (find_args_size_adjust): Return the adjustment as a
1632 poly_int64 rather than a HOST_WIDE_INT.
1633 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1634 to a poly_int64 and change the return type in the same way.
1635 (emit_single_push_insn): Track polynomial arg sizes.
1637 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1638 Alan Hayward <alan.hayward@arm.com>
1639 David Sherwood <david.sherwood@arm.com>
1641 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1642 from HOST_WIDE_INT to poly_int64.
1643 * expr.c (push_block, emit_push_insn): Likewise.
1645 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1646 Alan Hayward <alan.hayward@arm.com>
1647 David Sherwood <david.sherwood@arm.com>
1649 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1651 (get_frame_size): Return the size as a poly_int64 rather than a
1653 (frame_offset_overflow): Take the offset as a poly_int64 rather
1654 than a HOST_WIDE_INT.
1655 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1656 (assign_stack_temp): Likewise for the size.
1657 * function.c (get_frame_size): Return a poly_int64 rather than
1659 (frame_offset_overflow): Take the offset as a poly_int64 rather
1660 than a HOST_WIDE_INT.
1661 (try_fit_stack_local): Take the start, length and size as poly_int64s
1662 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1663 rather than a HOST_WIDE_INT.
1664 (add_frame_space): Take the start and end as poly_int64s rather than
1666 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1667 (assign_stack_temp): Likewise for the size.
1668 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1670 (find_temp_slot_from_address): Handle polynomial offsets.
1671 (combine_temp_slots): Likewise.
1672 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1674 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1675 poly_int64 rather than a HOST_WIDE_INT.
1676 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1677 than a HOST_WIDE_INT.
1678 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1679 polynomial frame offsets.
1680 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1681 as a poly_int64 rather than an int.
1682 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1683 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1684 * config/v850/v850.c (compute_frame_size): Likewise.
1685 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1686 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1687 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1688 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1689 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1690 rather than a HOST_WIDE_INT.
1691 * explow.c (get_dynamic_stack_base): Likewise.
1692 * final.c (final_start_function): Use the constant lower bound
1693 of the frame size for -Wframe-larger-than.
1694 * ira.c (do_reload): Adjust for new get_frame_size return type.
1695 * lra.c (lra): Likewise.
1696 * reload1.c (reload): Likewise.
1697 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1698 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1699 * rtlanal.c (get_initial_register_offset): Return the offset as
1700 a poly_int64 rather than a HOST_WIDE_INT.
1702 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1703 Alan Hayward <alan.hayward@arm.com>
1704 David Sherwood <david.sherwood@arm.com>
1706 * reload1.c (elim_table): Change initial_offset, offset and
1707 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1708 (offsets_at): Change the target array's element type from
1709 HOST_WIDE_INT to poly_int64_pod.
1710 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1711 (elimination_costs_in_insn, update_eliminable_offsets)
1712 (verify_initial_elim_offsets, set_offsets_for_label)
1713 (init_eliminable_invariants): Update after above changes.
1715 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1716 Alan Hayward <alan.hayward@arm.com>
1717 David Sherwood <david.sherwood@arm.com>
1719 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1720 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1721 (decomposition): Change start and end from HOST_WIDE_INT
1723 (decompose, immune_p): Update accordingly.
1724 (find_inc_amount): Return a poly_int64 rather than an int.
1725 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1728 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1729 Alan Hayward <alan.hayward@arm.com>
1730 David Sherwood <david.sherwood@arm.com>
1732 * tree.h (get_inner_reference): Return the bitsize and bitpos
1733 as poly_int64_pods rather than HOST_WIDE_INT.
1734 * fold-const.h (ptr_difference_const): Return the pointer difference
1735 as a poly_int64_pod rather than a HOST_WIDE_INT.
1736 * expr.c (get_inner_reference): Return the bitsize and bitpos
1737 as poly_int64_pods rather than HOST_WIDE_INT.
1738 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1740 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1741 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1742 (optimize_bit_field_compare): Update call to get_inner_reference.
1743 (decode_field_reference): Likewise.
1744 (fold_unary_loc): Track polynomial offsets and sizes.
1745 (split_address_to_core_and_offset): Return the bitpos as a
1746 poly_int64_pod rather than a HOST_WIDE_INT.
1747 (ptr_difference_const): Likewise for the pointer difference.
1748 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1749 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1750 * dbxout.c (dbxout_expand_expr): Likewise.
1751 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1752 (loc_list_from_tree_1, fortran_common): Likewise.
1753 * gimple-laddress.c (pass_laddress::execute): Likewise.
1754 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1755 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1756 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1757 * tree-affine.c (tree_to_aff_combination): Likewise.
1758 (get_inner_reference_aff): Likewise.
1759 * tree-data-ref.c (split_constant_offset_1): Likewise.
1760 (dr_analyze_innermost): Likewise.
1761 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1762 * tree-sra.c (ipa_sra_check_caller): Likewise.
1763 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1764 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1765 (instrument_bool_enum_load, instrument_object_size): Likewise.
1766 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1767 to get_inner_reference.
1768 * hsa-gen.c (gen_hsa_addr): Likewise.
1769 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1770 * tsan.c (instrument_expr): Likewise.
1771 * match.pd: Update call to ptr_difference_const.
1773 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1774 Alan Hayward <alan.hayward@arm.com>
1775 David Sherwood <david.sherwood@arm.com>
1777 * fold-const.c (fold_comparison): Track sizes and offsets as
1778 poly_int64s rather than HOST_WIDE_INTs when folding address
1781 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1782 Alan Hayward <alan.hayward@arm.com>
1783 David Sherwood <david.sherwood@arm.com>
1785 * expr.h (get_bit_range): Return the bitstart and bitend as
1786 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1787 as a poly_int64 rather than a HOST_WIDE_INT.
1788 * expr.c (get_bit_range): Likewise.
1789 (expand_assignment): Update call accordingly.
1790 * fold-const.c (optimize_bit_field_compare): Likewise.
1792 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1793 Alan Hayward <alan.hayward@arm.com>
1794 David Sherwood <david.sherwood@arm.com>
1796 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1797 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1798 as an optional final parameter.
1799 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1800 (aarch64_legitimate_address_p): Likewise.
1801 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1802 instead of an rtx code.
1803 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1804 (aarch64_legitimate_address_hook_p): Likewise.
1805 (aarch64_print_ldpstp_address): Likewise.
1806 (aarch64_print_operand_address): Likewise.
1807 (aarch64_address_cost): Likewise.
1808 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1809 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1810 (aarch64_mem_pair_lanes_operand): Likewise.
1812 2017-12-20 Richard Biener <rguenther@suse.de>
1814 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1815 update_stmt_if_modified.
1817 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1819 PR tree-optimization/83491
1820 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1821 before walking uses. Improve coding style and comments.
1823 2017-12-20 Tom de Vries <tom@codesourcery.com>
1825 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1827 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1830 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1832 (pass_ipa_free_fn_summary): Add small_p private data member,
1833 initialize to false in the ctor.
1834 (pass_ipa_free_fn_summary::clone,
1835 pass_ipa_free_fn_summary::set_pass_param,
1836 pass_ipa_free_fn_summary::gate): New methods.
1837 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1838 | TODO_dump_symtab if small_p.
1839 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1840 entry and add another instance of the pass with false parm after
1842 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1843 ipa_free_fn_summary here.
1845 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1847 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1849 2017-12-20 Martin Sebor <msebor@redhat.com>
1852 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1853 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1854 (expand_builtin_strncmp): Same.
1856 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1859 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1860 allowed debug stmts before labels.
1861 (expand_gimple_basic_block): Likewise.
1862 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1863 * gimple-iterator.h (gsi_after_labels): Likewise.
1864 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1865 rename reused variable, and simplify using gsi_move_before.
1866 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1867 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1868 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1869 (gimple_verify_flow_info, gimple_block_label): Likewise.
1870 (make_blocks): Move debug markers after adjacent labels.
1871 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1872 allowed debug insns outside blocks.
1873 * df-scan.c (df_insn_delete): Likewise.
1874 * lra-constraints.c (update_ebb_live_info): Likewise.
1875 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1876 (vt_initialize, delete_vta_debug_insns): Likewise.
1877 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1879 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1880 Alan Hayward <alan.hayward@arm.com>
1881 David Sherwood <david.sherwood@arm.com>
1883 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1884 (round_up_to_byte_boundary): Likewise.
1885 * expr.h (get_bit_range): Add temporary shim.
1886 * gimple-ssa-store-merging.c (store_operand_info): Change the
1887 bitsize, bitpos, bitregion_start and bitregion_end fields from
1888 unsigned HOST_WIDE_INT to poly_uint64.
1889 (merged_store_group): Likewise load_align_base.
1890 (compatible_load_p, compatible_load_p): Update accordingly.
1891 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1892 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1893 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1894 bitregion_start and bitregion_end as poly_uint64s rather than
1895 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1896 (handled_load): Take the bitsize, bitpos,
1897 bitregion_start and bitregion_end as poly_uint64s rather than
1898 unsigned HOST_WIDE_INTs.
1899 (pass_store_merging::process_store): Update call to
1900 mem_valid_for_store_merging.
1902 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1903 Alan Hayward <alan.hayward@arm.com>
1904 David Sherwood <david.sherwood@arm.com>
1906 * builtins.c (get_object_alignment_2): Track polynomial offsets
1907 and sizes. Update the alignment handling.
1909 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1910 Alan Hayward <alan.hayward@arm.com>
1911 David Sherwood <david.sherwood@arm.com>
1913 * tree.h (get_inner_reference): Add a version that returns the
1914 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1915 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1916 the case in which bitpos is not associated with the first byte.
1918 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1919 Alan Hayward <alan.hayward@arm.com>
1920 David Sherwood <david.sherwood@arm.com>
1922 * tree-affine.h (get_inner_reference_aff): Return the size as a
1924 * tree-affine.c (get_inner_reference_aff): Likewise.
1925 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1926 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1928 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1929 Alan Hayward <alan.hayward@arm.com>
1930 David Sherwood <david.sherwood@arm.com>
1932 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1933 HOST_WIDE_INT rather than a poly_int64.
1935 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1936 Alan Hayward <alan.hayward@arm.com>
1937 David Sherwood <david.sherwood@arm.com>
1939 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1940 HOST_WIDE_INT to poly_int64_pod.
1941 (perform_symbolic_merge): Update accordingly.
1942 (bswap_replace): Likewise.
1944 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1945 Alan Hayward <alan.hayward@arm.com>
1946 David Sherwood <david.sherwood@arm.com>
1948 * tree-affine.h (aff_tree::offset): Change from widest_int
1950 (wide_int_ext_for_comb): Delete.
1951 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1952 constants as poly_widest_int rather than widest_int.
1953 (aff_combination_constant_multiple_p): Return the multiplier
1954 as a poly_widest_int.
1955 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1957 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1958 version static and add an overload for poly_widest_int.
1959 (aff_combination_const, aff_combination_add_cst)
1960 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1961 the constants as poly_widest_int rather than widest_int.
1962 (tree_to_aff_combination): Generalize INTEGER_CST case to
1964 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1965 (aff_combination_add_product, aff_combination_mult): Handle
1967 (aff_combination_constant_multiple_p): Return the multiplier
1968 as a poly_widest_int.
1969 * tree-predcom.c (determine_offset): Return the offset as a
1971 (split_data_refs_to_components, suitable_component_p): Update
1973 (valid_initializer_p): Update call to
1974 aff_combination_constant_multiple_p.
1975 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1976 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1977 as a poly_int64 rather than a HOST_WIDE_INT.
1978 (get_address_cost): Handle polynomial offsets.
1979 (iv_elimination_compare_lt): Likewise.
1980 (rewrite_use_nonlinear_expr): Likewise.
1982 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1983 Alan Hayward <alan.hayward@arm.com>
1984 David Sherwood <david.sherwood@arm.com>
1986 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1987 as a poly_int64_pod rather than a HOST_WIDE_INT.
1988 (get_addr_base_and_unit_offset): Likewise.
1989 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1990 (get_addr_base_and_unit_offset): Likewise.
1991 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1992 to poly_int64 in example.
1993 * fold-const.c (fold_binary_loc): Update call to
1994 get_addr_base_and_unit_offset.
1995 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1996 (maybe_canonicalize_mem_ref_addr): Likewise.
1997 (gimple_fold_stmt_to_constant_1): Likewise.
1998 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2000 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
2001 * match.pd: Likewise.
2002 * omp-low.c (lower_omp_target): Likewise.
2003 * tree-sra.c (build_ref_for_offset): Likewise.
2004 (build_debug_ref_for_model): Likewise.
2005 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
2006 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
2007 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
2008 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
2009 (constant_pointer_difference): Likewise.
2010 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
2011 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
2012 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
2013 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2014 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
2015 (set_ssa_val_to): Likewise.
2016 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
2017 (maybe_diag_stxncpy_trunc): Likewise.
2018 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
2019 * tree.c (build_simple_mem_ref_loc): Likewise.
2020 (array_at_struct_end_p): Likewise.
2022 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2023 Alan Hayward <alan.hayward@arm.com>
2024 David Sherwood <david.sherwood@arm.com>
2026 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
2027 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
2028 (get_ref_base_and_extent_hwi): Declare.
2029 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
2030 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
2031 (get_ref_base_and_extent_hwi): New function.
2032 * cfgexpand.c (expand_debug_expr): Update call to
2033 get_ref_base_and_extent.
2034 * dwarf2out.c (add_var_loc_to_decl): Likewise.
2035 * gimple-fold.c (get_base_constructor): Return the offset as a
2036 poly_int64_pod rather than a HOST_WIDE_INT.
2037 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
2038 * ipa-polymorphic-call.c
2039 (ipa_polymorphic_call_context::set_by_invariant)
2040 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
2041 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
2042 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
2043 rather than get_ref_base_and_extent.
2044 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
2045 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
2046 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
2048 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
2049 call to get_ref_base_and_extent.
2050 * tree-sra.c (create_access, get_access_for_expr): Likewise.
2051 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
2052 (stmt_kills_ref_p): Likewise.
2053 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2054 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
2056 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
2057 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
2058 when calling native_encode_expr.
2059 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
2060 call to get_ref_base_and_extent.
2061 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
2062 get_ref_base_and_extent.
2063 * var-tracking.c (track_expr_p): Likewise.
2065 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2066 Alan Hayward <alan.hayward@arm.com>
2067 David Sherwood <david.sherwood@arm.com>
2069 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
2070 HOST_WIDE_INT to poly_int64_pod.
2071 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
2072 polynomail parameter offsets.
2074 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2075 Alan Hayward <alan.hayward@arm.com>
2076 David Sherwood <david.sherwood@arm.com>
2078 * gengtype.c (main): Handle poly_int64_pod.
2079 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
2080 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
2081 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
2082 from HOST_WIDE_INT to poly_int64_pod.
2083 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
2084 (copy_cfa): New function.
2085 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
2086 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
2087 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
2088 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
2089 (initial_return_save): Treat offsets as poly_ints.
2090 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
2091 offset is nonconstant.
2092 (reg_save): Take the offset as a poly_int64. Fall back to
2093 DW_CFA_expression for nonconstant offsets.
2094 (queue_reg_save): Take the offset as a poly_int64.
2095 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
2097 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2098 Alan Hayward <alan.hayward@arm.com>
2099 David Sherwood <david.sherwood@arm.com>
2101 * rtl.h (operand_subword, operand_subword_force): Take the offset
2102 as a poly_uint64 an unsigned int.
2103 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
2105 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2106 Alan Hayward <alan.hayward@arm.com>
2107 David Sherwood <david.sherwood@arm.com>
2109 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
2110 'p' format code. Use INT_LIST rather than SUBREG as the example of
2111 a code with an XINT and an XEXP. Remove the implication that
2112 accessing an rtx field using XINT is expected to work.
2113 * rtl.def (SUBREG): Change format from "ei" to "ep".
2114 * rtl.h (rtunion::rt_subreg): New field.
2115 (XCSUBREG): New macro.
2116 (SUBREG_BYTE): Use it.
2117 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
2118 Update constructor accordingly.
2119 (subreg_shape::operator ==): Update accordingly.
2120 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
2121 than an unsigned int.
2122 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
2123 a poly_uint64 rather than an unsigned int.
2124 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
2125 than an unsigned int.
2126 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
2127 (subreg_size_highpart_offset): Return a poly_uint64 rather than
2128 an unsigned int. Take the sizes as poly_uint64s.
2129 (subreg_offset_from_lsb): Return a poly_uint64 rather than
2130 an unsigned int. Take the shift as a poly_uint64 rather than
2132 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
2133 as a poly_uint64 rather than an unsigned int.
2134 (simplify_subreg_regno): Likewise.
2135 (byte_lowpart_offset): Return the memory offset as a poly_int64
2137 (subreg_memory_offset): Likewise. Take the subreg offset as a
2138 poly_uint64 rather than an unsigned int.
2139 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
2140 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
2141 poly_uint64 rather than an unsigned int.
2142 * rtl.c (rtx_format): Describe 'p' in comment.
2143 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
2144 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
2145 offset as a poly_uint64 rather than an unsigned int.
2146 (byte_lowpart_offset): Return the memory offset as a poly_int64
2148 (subreg_memory_offset): Likewise. Take the subreg offset as a
2149 poly_uint64 rather than an unsigned int.
2150 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
2151 mode sizes as poly_uint64s rather than unsigned ints. Return a
2152 poly_uint64 rather than an unsigned int.
2153 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
2154 (copy_insn_1): Handle 'p'.
2155 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
2156 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
2157 an unsigned int. Return the shift in the same way.
2158 (subreg_lsb): Return the shift as a poly_uint64 rather than an
2160 (subreg_size_offset_from_lsb): Take the sizes and shift as
2161 poly_uint64s rather than unsigned ints. Return the offset as
2163 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
2164 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
2166 * rtlhash.c (add_rtx): Handle 'p'.
2167 * genemit.c (gen_exp): Likewise.
2168 * gengenrtl.c (type_from_format, gendef): Likewise.
2169 * gensupport.c (subst_pattern_match, get_alternatives_number)
2170 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
2171 (subst_dup): Likewise.
2172 * gengtype.c (adjust_field_rtx_def): Likewise.
2173 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
2174 (match_pattern_2): Likewise.
2175 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
2176 (rtx_test::subreg_field): New function.
2177 (operator ==, safe_to_hoist_p, transition_parameter_type)
2178 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
2179 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
2180 * genpeep.c (match_rtx): Likewise.
2181 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
2182 (rtx_writer::print_rtx_operand): Handle 'p'.
2183 (print_value): Handle SUBREG.
2184 * read-rtl.c (apply_int_iterator): Likewise.
2185 (rtx_reader::read_rtx_operand): Handle 'p'.
2186 * alias.c (rtx_equal_for_memref_p): Likewise.
2187 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
2188 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
2190 * calls.c (expand_call): Likewise.
2191 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
2192 (make_extraction, gen_lowpart_for_combine): Likewise.
2193 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
2195 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
2196 rather than an unsigned int. Treat subreg offsets as poly_ints.
2197 (exp_equiv_p): Handle 'p'.
2198 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
2199 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
2200 * dse.c (find_shift_sequence): Likewise.
2201 * dwarf2out.c (rtl_for_decl_location): Likewise.
2202 * expmed.c (extract_low_bits): Likewise.
2203 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
2204 (expand_expr_real_2): Likewise.
2205 * final.c (alter_subreg): Likewise.
2206 (leaf_renumber_regs_insn): Handle 'p'.
2207 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
2208 Treat subreg offsets as poly_ints.
2209 * fwprop.c (forward_propagate_and_simplify): Likewise.
2210 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
2211 * ira.c (get_subreg_tracking_sizes): Likewise.
2212 * ira-conflicts.c (go_through_subreg): Likewise.
2213 * ira-lives.c (process_single_reg_class_operands): Likewise.
2214 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
2215 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
2216 as a poly_uint64 rather than an unsigned int.
2217 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
2218 subreg offsets as poly_ints.
2219 * lra-constraints.c (operands_match_p): Handle 'p'.
2220 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
2221 * lra-spills.c (assign_mem_slot): Likewise.
2222 * postreload.c (move2add_valid_value_p): Likewise.
2223 * recog.c (general_operand, indirect_operand): Likewise.
2224 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2225 (copyprop_hardreg_forward_1): Likewise.
2226 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2227 (record_subregs_of_mode): Likewise.
2228 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2229 * reload.c (operands_match_p): Handle 'p'.
2230 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2231 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2232 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2233 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2234 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2235 (simplify_const_poly_int_tests<N>::run): Likewise.
2236 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2237 a poly_uint64 rather than an unsigned int.
2238 * valtrack.c (debug_lowpart_subreg): Likewise.
2239 * var-tracking.c (var_lowpart): Likewise.
2240 (loc_cmp): Handle 'p'.
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 * ira.c (get_subreg_tracking_sizes): New function.
2247 (init_live_subregs): Take an integer size rather than a register.
2248 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2249 to init_live_subregs.
2251 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2252 Alan Hayward <alan.hayward@arm.com>
2253 David Sherwood <david.sherwood@arm.com>
2255 * expr.c (store_constructor_field): Change bitsize from a
2256 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2257 HOST_WIDE_INT to a poly_int64.
2258 (store_constructor): Change size from a HOST_WIDE_INT to
2260 (store_field): Likewise bitsize and bitpos.
2262 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2263 Alan Hayward <alan.hayward@arm.com>
2264 David Sherwood <david.sherwood@arm.com>
2266 * expmed.h (store_bit_field): Change bitregion_start and
2267 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2268 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2269 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2270 (store_fixed_bit_field, store_split_bit_field): Likewise.
2271 * expr.c (store_constructor_field, store_field): Likewise.
2272 (optimize_bitfield_assignment_op): Likewise. Make the same change
2273 to bitsize and bitpos.
2274 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2275 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2276 same change in the constructor arguments.
2277 (get_best_mode): Change bitregion_start and bitregion_end from
2278 unsigned HOST_WIDE_INT to poly_uint64.
2279 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2280 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2282 (bit_field_mode_iterator::next_mode): Update for new types
2283 of m_bitregion_start and m_bitregion_end.
2284 (get_best_mode): Change bitregion_start and bitregion_end from
2285 unsigned HOST_WIDE_INT to poly_uint64.
2287 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2288 Alan Hayward <alan.hayward@arm.com>
2289 David Sherwood <david.sherwood@arm.com>
2291 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2292 accepts poly_uint64 offsets.
2293 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2294 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2295 * expmed.c (lowpart_bit_field_p): Likewise.
2296 (extract_bit_field_as_subreg): New function, split out from...
2297 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2298 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2299 extractions, check that BITSIZE matches the size of the extracted
2300 value and that BITNUM is an exact multiple of that size.
2301 If all else fails, try forcing the value into memory if
2302 BITNUM is variable, and adjusting the address so that the
2303 offset is constant. Split the part that can only handle constant
2304 bitsize and bitnum out into...
2305 (extract_integral_bit_field): ...this new function.
2306 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2307 rather than unsigned HOST_WIDE_INTs.
2309 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2310 Alan Hayward <alan.hayward@arm.com>
2311 David Sherwood <david.sherwood@arm.com>
2313 * expmed.h (store_bit_field): Take bitsize and bitnum as
2314 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2315 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2316 that returns the byte size.
2317 (store_bit_field_1): Take bitsize and bitnum as
2318 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2319 to simple_mem_bitfield_p. Split the part that can only handle
2320 constant bitsize and bitnum out into...
2321 (store_integral_bit_field): ...this new function.
2322 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2323 than unsigned HOST_WIDE_INTs.
2324 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2326 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2327 Alan Hayward <alan.hayward@arm.com>
2328 David Sherwood <david.sherwood@arm.com>
2330 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2331 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2333 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2334 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2335 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2336 offset as a poly_int64 rather than an int.
2337 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2338 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2339 * lra-constraints.c (equiv_address_substitution): Track offsets
2341 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2342 (curr_insn_transform): Handle the new form of sp_offset.
2343 * lra-eliminations.c (lra_elim_table): Change previous_offset
2344 and offset from HOST_WIDE_INT to poly_int64.
2345 (print_elim_table, update_reg_eliminate): Update accordingly.
2346 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2347 (get_elimination): Update accordingly.
2348 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2349 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2350 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2351 poly_int64 offsets generally.
2352 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2353 (mark_not_eliminable, init_elimination): Update accordingly.
2354 (remove_reg_equal_offset_note): Return a bool and pass the new
2355 offset back by pointer as a poly_int64.
2356 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2357 rather than a HOST_WIDE_INT.
2358 (do_remat): Track offsets poly_int64s.
2359 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2361 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2362 Alan Hayward <alan.hayward@arm.com>
2363 David Sherwood <david.sherwood@arm.com>
2365 * rtl.h (mem_attrs): Add a default constructor. Change size and
2366 offset from HOST_WIDE_INT to poly_int64.
2367 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2368 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2369 (widen_memory_access): Take the sizes and offsets as poly_int64s
2370 rather than HOST_WIDE_INTs.
2371 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2372 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2374 (adjust_offset_for_component_ref): Change the offset from a
2375 HOST_WIDE_INT to a poly_int64.
2376 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2377 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2378 * dce.c (find_call_stack_args): Likewise.
2379 * dse.c (record_store): Likewise.
2380 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2381 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2382 * read-rtl-function.c (test_loading_mem): Likewise.
2383 * rtlanal.c (may_trap_p_1): Likewise.
2384 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2385 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2386 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2387 (mem_attrs::mem_attrs): New function.
2388 (set_mem_attributes_minus_bitpos): Change bitpos from a
2389 HOST_WIDE_INT to poly_int64.
2390 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2391 (clear_mem_offset, clear_mem_size, change_address)
2392 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2393 initialize mem_attrs.
2394 (set_mem_offset, set_mem_size, adjust_address_1)
2395 (adjust_automodify_address_1, offset_address, widen_memory_access):
2396 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2398 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2399 Alan Hayward <alan.hayward@arm.com>
2400 David Sherwood <david.sherwood@arm.com>
2402 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2403 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2404 rather than 0 to represent an unknown size. Assert that the size
2405 is known when the mode isn't BLKmode.
2406 (may_trap_p_1): Use -1 for unknown sizes.
2407 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2409 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2410 Alan Hayward <alan.hayward@arm.com>
2411 David Sherwood <david.sherwood@arm.com>
2413 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2414 to poly_int64. Update commentary for positions_needed.large.
2415 (read_info_type): Change offset and width from HOST_WIDE_INT
2417 (set_usage_bits): Likewise.
2418 (canon_address): Return the offset as a poly_int64 rather than
2419 a HOST_WIDE_INT. Use strip_offset_and_add.
2420 (set_all_positions_unneeded, any_positions_needed_p): Use
2421 positions_needed.large to track stores with non-constant widths.
2422 (all_positions_needed_p): Likewise. Take the offset and width
2423 as poly_int64s rather than ints. Assert that rhs is nonnull.
2424 (record_store): Cope with non-constant offsets and widths.
2425 Nullify the rhs of an earlier store if we can't tell which bytes
2427 (find_shift_sequence): Take the access_size and shift as poly_int64s
2429 (get_stored_val): Take the read_offset and read_width as poly_int64s
2430 rather than HOST_WIDE_INTs.
2431 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2432 non-constant offsets and widths.
2434 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2435 Alan Hayward <alan.hayward@arm.com>
2436 David Sherwood <david.sherwood@arm.com>
2438 * inchash.h (inchash::hash::add_poly_int): New function.
2439 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2440 Use poly_int64 rather than HOST_WIDE_INT.
2441 (ao_ref::max_size_known_p): New function.
2442 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2443 rather than HOST_WIDE_INT.
2444 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2445 to temporaries until its interface is adjusted to match.
2446 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2447 (aliasing_component_refs_p, decl_refs_may_alias_p)
2448 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2449 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2450 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2452 * alias.c (ao_ref_from_mem): Likewise.
2453 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2454 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2455 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2456 (maybe_trim_complex_store, maybe_trim_constructor_store)
2457 (live_bytes_read, dse_classify_store): Likewise.
2458 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2459 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2460 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2461 (vn_reference_lookup_3): Likewise.
2462 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2464 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2465 Alan Hayward <alan.hayward@arm.com>
2466 David Sherwood <david.sherwood@arm.com>
2468 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2469 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2470 instead of ranges_overlap_p.
2472 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2473 Alan Hayward <alan.hayward@arm.com>
2474 David Sherwood <david.sherwood@arm.com>
2476 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2477 sizes as poly_int64s rather than HOST_WIDE_INTs.
2479 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2480 Alan Hayward <alan.hayward@arm.com>
2481 David Sherwood <david.sherwood@arm.com>
2483 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2484 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2485 * gimple-fold.c (fold_ctor_reference): Likewise.
2487 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2488 Alan Hayward <alan.hayward@arm.com>
2489 David Sherwood <david.sherwood@arm.com>
2491 * target.def (dwarf_poly_indeterminate_value): New hook.
2492 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2493 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2494 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2495 * doc/tm.texi: Regenerate.
2496 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2497 offset as a poly_int64.
2498 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2499 offset as a poly_int64.
2500 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2501 Take the offset as a poly_int64.
2502 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2503 (frame_pointer_fb_offset): Change to a poly_int64.
2504 (int_loc_descriptor): Take the offset as a poly_int64. Use
2505 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2506 (based_loc_descr): Take the offset as a poly_int64.
2507 Use strip_offset_and_add to handle (plus X (const)).
2508 Use new_reg_loc_descr instead of an open-coded version of the
2509 previous implementation.
2510 (mem_loc_descriptor): Handle CONST_POLY_INT.
2511 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2512 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2514 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2515 Alan Hayward <alan.hayward@arm.com>
2516 David Sherwood <david.sherwood@arm.com>
2518 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2519 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2520 * inchash.h (inchash::hash::add_poly_hwi): New function.
2521 * gengtype.c (main): Register poly_int64.
2522 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2523 offset as a poly_int.
2524 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2525 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2526 offset as a poly_int64.
2527 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2528 * print-rtl.c (print_poly_int): New function.
2529 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2531 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2533 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2534 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2535 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2536 Enforce track_offset_p for parts of a PARALLEL.
2537 (vt_add_function_parameter): Use const_offset for the final
2538 offset to track. Use get_tracked_reg_offset for the parts
2541 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2542 Alan Hayward <alan.hayward@arm.com>
2543 David Sherwood <david.sherwood@arm.com>
2545 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2546 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2547 * doc/tm.texi: Regenerate.
2548 * hooks.h (hook_bool_uint_uint_true): Delete.
2549 (hook_bool_puint64_puint64_true): Declare.
2550 * hooks.c (hook_bool_uint_uint_true): Delete.
2551 (hook_bool_puint64_puint64_true): New function.
2552 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2553 instead of unsigned ints.
2554 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2555 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2557 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2558 Alan Hayward <alan.hayward@arm.com>
2559 David Sherwood <david.sherwood@arm.com>
2561 * optabs.h (expand_operand): Add an int_value field.
2562 (create_expand_operand): Add an int_value parameter and use it
2563 to initialize the new expand_operand field.
2564 (create_integer_operand): Replace with a declaration of a function
2565 that accepts poly_int64s. Move the implementation to...
2566 * optabs.c (create_integer_operand): ...here.
2567 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2568 the mode preserves the value of int_value, instead of calling
2569 const_int_operand on the rtx. Use gen_int_mode to generate
2572 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2573 Alan Hayward <alan.hayward@arm.com>
2574 David Sherwood <david.sherwood@arm.com>
2576 * dumpfile.h (dump_dec): Declare.
2577 * dumpfile.c (dump_dec): New function.
2578 * pretty-print.h (pp_wide_integer): Turn into a function and
2579 declare a poly_int version.
2580 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2582 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2583 Alan Hayward <alan.hayward@arm.com>
2584 David Sherwood <david.sherwood@arm.com>
2586 * doc/generic.texi (POLY_INT_CST): Document.
2587 * tree.def (POLY_INT_CST): New tree code.
2588 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2589 * tree-core.h (tree_poly_int_cst): New struct.
2590 (tree_node): Add a poly_int_cst field.
2591 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2592 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2593 instead of a wide_int_ref.
2594 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2596 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2597 instead of an unsigned HOST_WIDE_INT.
2598 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2599 (ptrdiff_tree_p): Declare.
2600 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2601 extern inline implementations if the target doesn't use POLY_INT_CST.
2602 (poly_int_tree_p): New function.
2603 (wi::unextended_tree): New class.
2604 (wi::int_traits <unextended_tree>): New override.
2605 (wi::extended_tree): Add a default constructor.
2606 (wi::extended_tree::get_tree): New function.
2607 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2608 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2609 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2610 (wi::tree_to_poly_wide_ref): New typedefs.
2611 (wi::ints_for): Provide overloads for extended_tree and
2613 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2614 (wi::to_wide): New functions.
2615 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2616 * tree.c (poly_int_cst_hasher): New struct.
2617 (poly_int_cst_hash_table): New variable.
2618 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2619 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2621 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2622 (init_ttree): Initialize poly_int_cst_hash_table.
2623 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2624 a poly_int64 instead of a HOST_WIDE_INT.
2625 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2626 instead of an unsigned HOST_WIDE_INT.
2627 (wide_int_to_tree): Rename to...
2628 (wide_int_to_tree_1): ...this.
2629 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2630 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2631 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2632 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2633 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2634 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2636 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2637 (streamer_read_tree_body): Likewise.
2638 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2639 (streamer_write_tree_body): Likewise.
2640 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2641 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2642 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2643 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2644 * gimple-expr.h (is_gimple_constant): Likewise.
2645 * gimplify.c (maybe_with_size_expr): Likewise.
2646 * print-tree.c (print_node): Likewise.
2647 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2648 * tree-pretty-print.c (dump_generic_node): Likewise.
2649 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2650 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2651 * tree-vrp.c (compare_values_warnv): Likewise.
2652 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2653 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2654 (force_expr_to_var_cost): Likewise.
2655 * tree-ssa-loop.c (for_each_index): Likewise.
2656 * fold-const.h (build_invariant_address, size_int_kind): Take a
2657 poly_int64 instead of a HOST_WIDE_INT.
2658 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2659 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2661 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2662 (int_const_binop_2): New function, split out from...
2663 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2664 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2665 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2666 * gimple-ssa-strength-reduction.c (slsr_process_add)
2667 (slsr_process_mul): Check for INTEGER_CSTs before using them
2669 * stor-layout.c (bits_from_bytes): New function.
2670 (bit_from_pos): Use it.
2671 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2672 by BITS_PER_UNIT to get the TYPE_SIZE.
2673 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2674 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2676 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2677 Alan Hayward <alan.hayward@arm.com>
2678 David Sherwood <david.sherwood@arm.com>
2680 * doc/rtl.texi (const_poly_int): Document. Also document the
2681 rtl sharing behavior.
2682 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2683 * rtl.h (const_poly_int_def): New struct.
2684 (rtx_def::u): Add a cpi field.
2685 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2686 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2687 (wi::rtx_to_poly_wide_ref): New typedef
2688 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2689 (poly_int_rtx_p): New functions.
2690 (trunc_int_for_mode): Declare a poly_int64 version.
2691 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2692 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2694 (strip_offset): Declare.
2695 (strip_offset_and_add): New function.
2696 * rtl.def (CONST_POLY_INT): New rtx code.
2697 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2698 (shared_const_p): Use poly_int_rtx_p.
2699 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2701 (gen_int_shift_amount): Likewise.
2702 * emit-rtl.c (const_poly_int_hasher): New class.
2703 (const_poly_int_htab): New variable.
2704 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2705 (const_poly_int_hasher::hash): New function.
2706 (const_poly_int_hasher::equal): Likewise.
2707 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2708 (immed_wide_int_const): Rename to...
2709 (immed_wide_int_const_1): ...this and make static.
2710 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2711 instead of a wide_int_ref.
2712 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2713 (gen_lowpart_common): Handle CONST_POLY_INT.
2714 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2715 * cselib.c (cselib_hash_rtx): Likewise.
2716 * dwarf2out.c (const_ok_for_output_1): Likewise.
2717 * expr.c (convert_modes): Likewise.
2718 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2719 * rtlhash.c (add_rtx): Likewise.
2720 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2721 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2722 Handle existing CONST_POLY_INT rtxes.
2723 * expmed.h (expand_shift): Take a poly_int64 instead of a
2725 * expmed.c (expand_shift): Likewise.
2726 * rtlanal.c (strip_offset): New function.
2727 (commutative_operand_precedence): Give CONST_POLY_INT the same
2728 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2730 * rtl-tests.c (const_poly_int_tests): New struct.
2731 (rtl_tests_c_tests): Use it.
2732 * simplify-rtx.c (simplify_const_unary_operation): Handle
2734 (simplify_const_binary_operation): Likewise.
2735 (simplify_binary_operation_1): Fold additions of symbolic constants
2736 and CONST_POLY_INTs.
2737 (simplify_subreg): Handle extensions and truncations of
2739 (simplify_const_poly_int_tests): New struct.
2740 (simplify_rtx_c_tests): Use it.
2741 * wide-int.h (storage_ref): Add default constructor.
2742 (wide_int_ref_storage): Likewise.
2743 (trailing_wide_ints): Use GTY((user)).
2744 (trailing_wide_ints::operator[]): Add a const version.
2745 (trailing_wide_ints::get_precision): New function.
2746 (trailing_wide_ints::extra_size): Likewise.
2748 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2749 Alan Hayward <alan.hayward@arm.com>
2750 David Sherwood <david.sherwood@arm.com>
2752 * emit-rtl.h (gen_int_shift_amount): Declare.
2753 * emit-rtl.c (gen_int_shift_amount): New function.
2754 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2756 * calls.c (shift_return_value): Likewise.
2757 * cse.c (fold_rtx): Likewise.
2758 * dse.c (find_shift_sequence): Likewise.
2759 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2760 (expand_shift, expand_smod_pow2): Likewise.
2761 * lower-subreg.c (shift_cost): Likewise.
2762 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2763 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2764 (expand_vec_perm_var): Likewise.
2765 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2766 (simplify_binary_operation_1): Likewise.
2767 * combine.c (try_combine, find_split_point, force_int_to_mode)
2768 (simplify_shift_const_1, simplify_shift_const): Likewise.
2769 (change_zero_ext): Likewise. Use simplify_gen_binary.
2771 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2773 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2775 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2777 * doc/invoke.texi (ARM Options): Document accepted extension options
2778 for -march=armv8.3-a.
2780 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2783 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2784 or --with-float=softfp, set the default CPU to arm10e.
2786 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2788 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2789 * config/visium/predicates.md (const_shift_operand): Likewise.
2790 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2791 (visium_legitimize_reload_address): Likewise.
2793 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2796 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2798 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2801 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2803 2017-12-20 Julia Koval <julia.koval@intel.com>
2805 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2806 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2807 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2808 * config.gcc: Include vpclmulqdqintrin.h.
2809 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2810 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2812 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2813 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2814 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2815 * config/i386/i386.c (isa2_opts): Add -mcx16.
2816 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2817 (ix86_option_override_internal): Move mcx16 to flags2.
2818 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2819 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2820 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2821 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2822 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2823 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2824 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2825 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2826 * doc/invoke.texi: Add -mvpclmulqdq.
2828 2017-12-20 Tom de Vries <tom@codesourcery.com>
2831 * config/i386/i386.c (ix86_static_chain): Move
2832 DECL_STATIC_CHAIN test ...
2833 * calls.c (rtx_for_static_chain): ... here. New function.
2834 * calls.h (rtx_for_static_chain): Declare.
2835 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2836 instead of targetm.calls.static_chain.
2837 * df-scan.c (df_get_entry_block_def_set): Same.
2839 2017-12-19 Tom de Vries <tom@codesourcery.com>
2841 PR tree-optimization/83493
2842 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2845 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2847 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2848 inform with hardcoded english plural handling.
2850 2017-12-18 Jeff Law <law@redhat.com>
2852 PR tree-optimization/83477
2853 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2854 a non-virtual PHI, always push a new range.
2856 2017-12-19 Martin Sebor <msebor@redhat.com>
2859 * builtins.c (compute_objsize): Handle non-constant offsets.
2861 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2863 PR tree-optimization/83444
2864 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2865 character load case, if get_stridx on MEM_REF's operand doesn't
2866 look usable, retry with get_addr_stridx.
2868 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2871 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2872 (variable_tracking_main_1): Keep markers even when VTA fails.
2875 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2876 even if there are markers before it.
2877 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2879 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2881 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2882 typical order conditions.
2883 * sel-sched.c (extract_new_fences_from): Likewise.
2884 * config/visium/constraints.md (J, K, L): Likewise.
2885 * config/visium/predicates.md (const_shift_operand): Likewise.
2886 * config/visium/visium.c (visium_legitimize_address,
2887 visium_legitimize_reload_address): Likewise.
2888 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2889 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2890 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2891 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2892 SET_DUMP_DETAIL): Likewise.
2893 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2894 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2895 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2896 avr_set_core_architecture, avr_set_current_function,
2897 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2898 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2899 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2900 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2901 avr_map_decompose, avr_fold_builtin): Likewise.
2902 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2903 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2904 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2905 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2906 * config/m32c/m32c.c (m32c_conditional_register_usage,
2907 m32c_address_cost): Likewise.
2908 * config/m32c/predicates.md (shiftcount_operand,
2909 longshiftcount_operand): Likewise.
2910 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2911 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2912 can_use_cdx_ldstw): Likewise.
2913 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2914 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2916 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2917 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2918 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2919 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2920 * config/vax/vax.c (adjacent_operands_p): Likewise.
2921 * config/ft32/constraints.md (L, b, KA): Likewise.
2922 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2924 * cfgexpand.c (expand_stack_alignment): Likewise.
2925 * gcse.c (insert_expr_in_table): Likewise.
2926 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2927 * cgraphunit.c (cgraph_node::expand): Likewise.
2928 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2929 * emit-rtl.c (add_insn): Likewise.
2930 * input.c (dump_location_info): Likewise.
2931 * passes.c (NEXT_PASS): Likewise.
2932 * read-rtl-function.c (parse_note_insn_name,
2933 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2935 * sched-rgn.c (sched_rgn_init): Likewise.
2936 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2937 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2938 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2940 * explow.c (eliminate_constant_term): Likewise.
2941 * final.c (leaf_renumber_regs_insn): Likewise.
2942 * cfgrtl.c (print_rtl_with_bb): Likewise.
2943 * genhooks.c (emit_init_macros): Likewise.
2944 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2945 * tree-data-ref.c (conflict_fn): Likewise.
2946 * selftest.c (assert_streq): Likewise.
2947 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2948 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2949 fold_binary_loc, multiple_of_p): Likewise.
2950 * reload.c (push_reload, find_equiv_reg): Likewise.
2951 * et-forest.c (et_nca, et_below): Likewise.
2952 * dbxout.c (dbxout_symbol_location): Likewise.
2953 * reorg.c (relax_delay_slots): Likewise.
2954 * dojump.c (do_compare_rtx_and_jump): Likewise.
2955 * gengtype-parse.c (type): Likewise.
2956 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2957 simplify_const_relational_operation): Likewise.
2958 * reload1.c (do_output_reload): Likewise.
2959 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2960 * gengtype.c (type_for_name): Likewise.
2961 * gimple-ssa-sprintf.c (format_directive): Likewise.
2963 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2966 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2967 accessing it. Adjust comment.
2969 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2972 * predict.c (zero_one_minusone): New function.
2973 (apply_return_prediction): Avoid return prediction for functions
2974 returning only -1, 0 and 1 values, unless they only return -1 and 0
2977 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2979 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2980 fall-through warning.
2981 (arc_compute_frame_size): Remove unused variables.
2982 (arc_print_operand): Fix fprintif format.
2983 (arc_can_follow_jump): Clean fall-through warning.
2985 2017-12-19 Marek Polacek <polacek@redhat.com>
2988 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2991 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2993 * config/arc/arc.c (overriderregs): New variable.
2994 (arc_override_options): Track fixed/call saved/call options.
2995 (arc_conditional_register_usage): Check against overrideregs
2996 variable whenever we change register properties.
2998 2017-12-19 Nathan Sidwell <nathan@acm.org>
3000 * opts.c (finish_options): Don't prefix dump_base_name if it
3001 already contains directories.
3003 2017-12-19 Martin Liska <mliska@suse.cz>
3005 PR rtl-optimization/82675
3006 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
3007 more element in sbitmap.
3009 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
3011 * gimplify.c (gimplify_expr): Use error_operand_p.
3013 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
3016 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
3019 2017-12-19 Daniel Cederman <cederman@gaisler.com>
3021 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
3022 the jump is to a label.
3024 2017-12-19 Jakub Jelinek <jakub@redhat.com>
3026 PR tree-optimization/83444
3027 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
3032 * ipa-inline.c (flatten_remove_node_hook): New function.
3033 (ipa_inline): Keep only nodes with flatten attribute at the end of
3034 the array in the order from ipa_reverse_postorder, only walk that
3035 portion of array for flattening, if there is more than one such
3036 node, temporarily register a removal hook and ignore removed nodes.
3038 PR tree-optimization/80631
3039 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
3040 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
3041 IFN_REDUC_MAX or IFN_REDUC_MIN.
3043 2017-12-18 Jeff Law <law@redhat.com>
3045 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
3046 record symbolic equivalences from backedges in the CFG.
3049 2017-11-19 Jeff Law <law@redhat.com>
3051 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3052 of degenerates resulting from ignoring an edge.
3054 2017-12-18 Martin Sebor <msebor@redhat.com>
3057 PR tree-optimization/78450
3058 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
3059 (handle_builtin_strlen): Call it.
3061 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
3063 PR rtl-optimization/83424
3064 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
3066 2017-12-18 Marek Polacek <polacek@redhat.com>
3069 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
3070 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
3073 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
3075 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
3076 (umaddsidi4, umaddsidi_split): Likewise.
3078 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
3080 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
3083 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3086 * config/s390/s390.c (s390_option_override): Avoid strncpy.
3088 2017-12-18 Richard Biener <rguenther@suse.de>
3090 PR tree-optimization/81877
3091 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
3092 (outermost_indep_loop): Adjust.
3093 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
3094 (can_sm_ref_p): Adjust.
3096 2017-12-18 Richard Biener <rguenther@suse.de>
3099 * tree.c (array_at_struct_end_p): Return true if the underlying
3100 object has space for at least one element in excess of what
3101 the array domain specifies.
3103 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3105 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
3106 example to avoid overfull hbox.
3107 * doc/invoke.texi (Option Summary): Add missing @gol.
3108 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
3111 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3112 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3114 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
3115 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
3116 (AArch64 Options): Document -mverbose-cost-dump.
3117 (ARM Options): Likewise, plus -mflip-thumb.
3119 2017-12-17 Martin Sebor <msebor@redhat.com>
3122 * gimple-ssa-warn-restrict.c
3123 (builtin_memref::offset_out_of_bounds): Correct the handling of
3126 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3128 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
3130 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
3132 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
3134 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
3136 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
3139 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
3141 * doc/invoke.texi: Fix some typos.
3143 2017-12-16 Martin Sebor <msebor@redhat.com>
3145 PR tree-optimization/78918
3146 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
3147 * builtins.c (check_sizes): Rename...
3148 (check_access): ...to this. Rename function arguments for clarity.
3149 (check_memop_sizes): Adjust names.
3150 (expand_builtin_memchr, expand_builtin_memcpy): Same.
3151 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
3152 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
3153 (check_strncat_sizes, expand_builtin_strncat): Same.
3154 (expand_builtin_strncpy, expand_builtin_memset): Same.
3155 (expand_builtin_bzero, expand_builtin_memcmp): Same.
3156 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
3157 (maybe_emit_sprintf_chk_warning): Same.
3158 (expand_builtin_strcpy): Adjust.
3159 (expand_builtin_stpcpy): Same.
3160 (expand_builtin_with_bounds): Detect out-of-bounds accesses
3161 in pointer-checking forms of memcpy, memmove, and mempcpy.
3162 (gcall_to_tree_minimal, max_object_size): Define new functions.
3163 * builtins.h (max_object_size): Declare.
3164 * calls.c (alloc_max_size): Call max_object_size instead of
3165 hardcoding ssizetype limit.
3166 (get_size_range): Handle new argument.
3167 * calls.h (get_size_range): Add a new argument.
3168 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
3169 * doc/invoke.texi (-Wrestrict): Adjust, add example.
3170 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
3172 (gimple_fold_builtin_memory_chk): Same.
3173 (gimple_fold_builtin_stxcpy_chk): New function.
3174 * gimple-ssa-warn-restrict.c: New source.
3175 * gimple-ssa-warn-restrict.h: New header.
3176 * gimple.c (gimple_build_call_from_tree): Propagate location.
3177 * passes.def (pass_warn_restrict): Add new pass.
3178 * tree-pass.h (make_pass_warn_restrict): Declare.
3179 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
3181 (handle_builtin_strcat): Same.
3182 (strlen_optimize_stmt): Rename...
3183 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
3184 stpncpy, strncpy, and their checking forms.
3186 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
3188 PR rtl-optimization/82849
3189 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
3190 and get_max_loop_iterations_int.
3192 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3193 Alan Hayward <alan.hayward@arm.com>
3194 David Sherwood <david.sherwood@arm.com>
3196 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
3197 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
3198 size as a poly_uint64.
3199 (mode_for_vector, mode_for_int_vector): Take the number of vector
3200 elements as a poly_uint64.
3201 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
3202 size as a poly_uint64.
3203 (mode_for_vector, mode_for_int_vector): Take the number of vector
3204 elements as a poly_uint64.
3206 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3207 Alan Hayward <alan.hayward@arm.com>
3208 David Sherwood <david.sherwood@arm.com>
3210 * machmode.h (MACRO_MODE): New macro.
3211 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
3212 * targhooks.c (default_libcall_value, default_secondary_reload)
3213 (default_memory_move_cost, default_register_move_cost)
3214 (default_class_max_nregs): Likewise.
3216 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3217 Alan Hayward <alan.hayward@arm.com>
3218 David Sherwood <david.sherwood@arm.com>
3220 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
3221 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3223 * genautomata.c (main): Likewise.
3224 * genconditions.c (write_header): Likewise.
3225 * genemit.c (main): Likewise.
3226 * genextract.c (print_header): Likewise.
3227 * genopinit.c (main): Likewise.
3228 * genoutput.c (output_prologue): Likewise.
3229 * genpeep.c (main): Likewise.
3230 * genpreds.c (write_insn_preds_c): Likewise.
3231 * genrecog.c (writer_header): Likewise.
3232 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3233 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3234 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3235 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3236 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3237 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3238 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3239 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3240 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3241 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3242 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3243 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3244 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3245 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3246 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3247 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3248 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3249 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3250 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3251 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3252 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3253 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3254 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3255 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3256 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3257 * config/darwin.c (IN_TARGET_CODE): Likewise.
3258 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3259 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3260 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3261 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3262 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3263 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3264 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3265 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3266 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3267 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3268 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3269 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3270 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3271 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3272 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3273 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3274 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3275 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3276 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3277 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3278 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3279 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3280 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3281 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3282 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3283 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3284 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3285 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3286 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3287 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3288 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3289 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3290 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3291 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3292 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3293 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3294 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3295 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3296 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3297 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3298 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3299 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3300 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3301 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3302 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3303 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3304 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3305 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3306 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3307 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3308 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3309 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3310 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3311 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3312 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3313 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3314 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3315 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3316 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3317 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3318 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3319 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3320 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3321 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3322 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3323 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3324 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3325 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3326 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3327 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3328 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3329 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3330 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3331 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3332 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3333 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3334 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3335 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3336 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3337 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3338 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3339 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3340 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3341 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3342 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3343 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3344 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3345 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3346 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3347 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3348 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3349 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3350 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3351 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3352 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3353 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3354 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3355 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3356 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3357 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3358 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3359 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3360 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3361 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3362 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3363 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3365 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3366 Alan Hayward <alan.hayward@arm.com>
3367 David Sherwood <david.sherwood@arm.com>
3369 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3372 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3373 Alan Hayward <alan.hayward@arm.com>
3374 David Sherwood <david.sherwood@arm.com>
3376 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3377 * doc/md.texi (vec_series@var{m}): Document.
3378 * tree.def (VEC_SERIES_EXPR): New tree code.
3379 * tree.h (build_vec_series): Declare.
3380 * tree.c (build_vec_series): New function.
3381 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3382 * tree-pretty-print.c (dump_generic_node): Likewise.
3383 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3384 * tree-inline.c (estimate_operator_cost): Likewise.
3385 * expr.c (expand_expr_real_2): Likewise.
3386 * optabs-tree.c (optab_for_tree_code): Likewise.
3387 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3388 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3389 * expmed.c (make_tree): Handle VEC_SERIES.
3390 * optabs.def (vec_series_optab): New optab.
3391 * optabs.h (expand_vec_series_expr): Declare.
3392 * optabs.c (expand_vec_series_expr): New function.
3393 * tree-vect-generic.c (expand_vector_operations_1): Check that
3394 the operands also have vector type.
3396 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3397 Alan Hayward <alan.hawyard@arm.com>
3398 David Sherwood <david.sherwood@arm.com>
3400 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3401 (VEC_COND_EXPR): Add missing @tindex.
3402 * doc/md.texi (vec_duplicate@var{m}): Document.
3403 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3404 * tree.c (build_vector_from_val): Add stubbed-out handling of
3405 variable-length vectors, using VEC_DUPLICATE_EXPR.
3406 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3407 * cfgexpand.c (expand_debug_expr): Likewise.
3408 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3409 * tree-inline.c (estimate_operator_cost): Likewise.
3410 * tree-pretty-print.c (dump_generic_node): Likewise.
3411 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3412 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3413 (test_vec_duplicate_folding): New function.
3414 (fold_const_c_tests): Call it.
3415 * optabs.def (vec_duplicate_optab): New optab.
3416 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3417 * optabs.h (expand_vector_broadcast): Declare.
3418 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3419 vec_duplicate_optab.
3420 * expr.c (store_constructor): Try using vec_duplicate_optab for
3422 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3424 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3427 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3428 div/mod latencies a bit.
3430 2017-12-15 Jeff Law <law@redhat.com>
3432 PR tree-optimization/36550
3433 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3434 (mark_threaded_blocks): Rewrite code to avoid block copying when
3435 optimizing for size. Don't pessimize blocks which will be
3436 copied, but all the statements will be dead.
3438 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3440 PR tree-optimization/81165
3441 * tree-ssa-threadupdate.c (uses_in_bb): New.
3442 (estimate_threading_killed_stmts): New.
3443 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3444 * tree-ssa-threadedge.c
3445 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3448 2017-12-15 Jeff Law <law@redhat.com>
3450 PR tree-optimization/83410
3451 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3452 threads when parallelizing loops.
3454 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3456 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3458 * config/alpha/alpha.c (vms_attribute_table): Swap
3459 affects_type_identity and handler fields, adjust comments.
3460 * config/mips/mips.c (mips_attribute_table): Likewise.
3461 * config/visium/visium.c (visium_attribute_table): Likewise.
3462 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3463 * config/microblaze/microblaze.c (microblaze_attribute_table):
3465 * config/spu/spu.c (spu_attribute_table): Likewise.
3466 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3467 * config/arc/arc.c (arc_attribute_table): Likewise.
3468 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3469 * config/v850/v850.c (v850_handle_interrupt_attribute,
3470 v850_handle_data_area_attribute): Formatting fixes.
3471 (v850_attribute_table): Swap affects_type_identity and handler
3472 fields, adjust comments.
3473 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3474 * config/arm/arm.c (arm_attribute_table): Likewise.
3475 * config/avr/avr.c (avr_attribute_table): Likewise.
3476 * config/s390/s390.c (s390_attribute_table): Likewise.
3477 * config/sh/sh.c (sh_attribute_table): Likewise.
3478 * config/i386/i386.c (ix86_handle_cconv_attribute,
3479 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3480 (ix86_attribute_table): Swap affects_type_identity and handler
3481 fields, adjust comments.
3482 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3483 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3484 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3485 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3486 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3487 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3488 * config/rx/rx.c (rx_attribute_table): Likewise.
3489 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3490 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3491 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3492 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3493 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3494 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3495 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3496 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3497 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3498 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3499 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3501 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3503 (register_scoped_attributes, decl_attributes): Formatting fixes.
3505 PR tree-optimization/83269
3506 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3507 subtraction in arg0's type if type is signed and arg0 is unsigned.
3511 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3512 convert? on @0 instead of convert. Check type of @1, not @0.
3513 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3514 convert? on @0 instead of convert. Check type of @1, not @0.
3515 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3516 only optimize if either both @1 and @2 types are narrower
3517 precision, or both are wider or equal precision, and in the former
3518 case only if both have undefined overflow.
3520 2017-12-15 Richard Biener <rguenther@suse.de>
3523 * internal-fn.def (IFN_NOP): Add.
3524 * internal-fn.c (expand_NOP): Do nothing.
3525 * lto-streamer-in.c (input_function): Instead of removing
3526 sanitizer calls replace them with IFN_NOP calls.
3528 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3529 Alan Hayward <alan.hayward@arm.com>
3530 David Sherwood <david.sherwood@arm.com>
3532 * dse.c (store_info, read_info_type): Replace begin and end with
3534 (print_range): New function.
3535 (set_all_positions_unneeded, any_positions_needed_p)
3536 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3538 (record_store): Likewise. Optimize the case in which all positions
3540 (get_stored_val): Replace read_begin and read_end with read_offset
3542 (replace_read): Update call accordingly.
3544 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3546 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3547 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3548 (loop_cand::loop_cand): Initialize above members.
3549 (loop_cand::supported_operations): Delete.
3550 (loop_cand::can_interchange_p): Inline above function.
3551 (loop_cand::classify_simple_reduction): Record number of constant
3552 initialized simple reductions.
3553 (should_interchange_loops): New parameters. Check stmt cost of loops
3555 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3556 Update call to should_interchange_loops.
3557 (should_interchange_loop_nest): Update call to
3558 should_interchange_loops.
3560 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3563 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3564 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3565 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3567 2017-12-15 Julia Koval <julia.koval@intel.com>
3569 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3570 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3572 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3573 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3574 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3576 2017-12-15 Julia Koval <julia.koval@intel.com>
3578 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3579 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3580 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3581 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3582 _mm_aesenc_epi128): New intrinsics.
3584 2017-12-15 Julia Koval <julia.koval@intel.com>
3586 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3587 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3589 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3590 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3591 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3593 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3595 * gimple-ssa-strength-reduction.c (analyze_increments):
3596 Distinguish replacement costs for constant strides from those for
3599 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3601 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3603 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3605 * doc/invoke.texi: Document -Wcast-function-type.
3606 * recog.h (stored_funcptr): Change signature.
3607 * tree-dump.c (dump_node): Avoid warning.
3608 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3610 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3613 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3616 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3618 PR tree-optimization/83312
3619 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3620 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3621 "val" param, and to cope with arbitrary basic blocks.
3622 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3623 handle NULL_TREE for "val", dropping "bb" param.
3624 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3625 drop "bb" param. Handle NULL_TREE for "val".
3626 (find_case_label_for_value): Make "switch_stmt" param const.
3627 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3629 (vrp_prop::check_all_array_refs): Reimplement as...
3630 (check_array_bounds_dom_walker::before_dom_children): ...this new
3631 vfunc. Replace linear search through BB block list, excluding
3632 those with non-executable in-edges via dominator walk.
3634 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3636 * config/arm/arm.opt (mverbose-cost-dump): New option.
3637 * config/arm/arm.c (arm_rtx_costs): Use it.
3639 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3642 * reload1.c (emit_input_reload_insns): Skip debug markers.
3644 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3646 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3650 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3651 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3652 * config/sh/sh.c: Skip debug insns besides notes.
3653 * config/sh/sh.md: Likewise.
3654 * config/sh/sh_treg_combine.cc: Likewise.
3655 * config/sh/sync.md: Likewise.
3657 2017-12-14 Tom de Vries <tom@codesourcery.com>
3659 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3660 Add item for weak_undefined.
3662 2017-12-14 Richard Biener <rguenther@suse.de>
3664 PR tree-optimization/67842
3665 * sese.h (bb_in_region): Remove #if 0'ed code.
3667 2017-12-14 Richard Biener <rguenther@suse.de>
3669 PR tree-optimization/83326
3670 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3671 may_be_zero parameter and handle it by not marking the first
3672 peeled copy as not exiting the loop.
3673 (try_peel_loop): Likewise.
3674 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3675 to handle the case of constant or zero iterations and perform
3676 loop header copying on-the-fly.
3678 2017-12-14 Richard Biener <rguenther@suse.de>
3680 PR tree-optimization/83418
3681 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3682 Instead of asserting we don't get unfolded comparisons deal with them.
3684 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3687 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3688 basic blocks. Assert debug bind insns don't appear outside of bbs,
3689 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3692 PR tree-optimization/83198
3693 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3694 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3695 value if arg is a REAL_CST with incompatible type.
3697 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3698 Bin Cheng <bin.cheng@arm.com>
3701 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3703 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3706 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3708 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3709 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3711 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3712 Alan Hayward <alan.hayward@arm.com>
3713 David Sherwood <david.sherwood@arm.com>
3715 * poly-int.h: New file.
3716 * poly-int-types.h: Likewise.
3717 * coretypes.h: Include them.
3718 (POLY_INT_CONVERSION): Define.
3719 * target.def (estimated_poly_value): New hook.
3720 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3721 * doc/tm.texi: Regenerate.
3722 * doc/poly-int.texi: New file.
3723 * doc/gccint.texi: Include it.
3724 * doc/rtl.texi: Describe restrictions on subreg modes.
3725 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3726 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3727 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3728 * targhooks.h (default_estimated_poly_value): Declare.
3729 * targhooks.c (default_estimated_poly_value): New function.
3730 * target.h (estimated_poly_value): Likewise.
3731 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3732 (wi::unary_traits): Delete.
3733 (wi::binary_traits::signed_shift_result_type): Define for
3734 offset_int << HOST_WIDE_INT, etc.
3735 (generic_wide_int::operator <<=): Define for all types and use
3736 wi::lshift instead of <<.
3737 (wi::hwi_with_prec): Add a default constructor.
3738 (wi::ints_for): New class.
3739 (operator <<): Define for all wide-int types.
3740 (operator /): New function.
3741 (operator %): Likewise.
3742 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3743 (ASSERT_MAYBE_NE_AT): New macros.
3745 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3746 Dominik Vogt <vogt@linux.vnet.ibm.com>
3749 * emit-rtl.c (init_emit): Remove ??? comment.
3750 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3751 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3752 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3753 alignment of 3 virtual registers to BITS_PER_WORD.
3755 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3757 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3759 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3760 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3762 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3765 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3767 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3769 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3770 with the exception of debug begin stmt markers appear before
3774 * final.c (rest_of_handle_final): Call variable_tracking_main only
3775 if !flag_var_tracking.
3777 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3778 Jakub Jelinek <jakub@redhat.com>
3782 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3783 labels debug stmts that can only appear after labels.
3785 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3787 PR rtl-optimization/82398
3788 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3789 EXPR_USEFULNESS in priority comparison.
3791 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3793 PR rtl-optimization/83393
3794 * combine.c (move_deaths): If reg_stat points to a too new insn in
3795 last_death, do not use it: find the proper insn instead.
3797 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3800 * config/rs6000/rs6000-string.c (expand_block_compare,
3801 expand_strn_compare): Fix set_mem_size() calls.
3803 2017-12-12 Jeff Law <law@redhat.com>
3805 PR tree-optimization/83298
3806 PR tree-optimization/83362
3807 PR tree-optimization/83383
3808 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3809 push_value_range a public interface. Add new argument to
3810 record_ranges_from_stmt.
3811 * gimple-ssa-evrp-analyze.c
3812 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3813 Update comments. Handle recording temporary equivalences.
3814 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3815 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3816 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3817 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3818 gimple-ssa-evrp-analyze.h.
3819 (record_temporary_equivalences_from_phis): Add new argument. When
3820 the PHI arg is an SSA_NAME, set the result's range to the range
3822 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3823 from statements too.
3824 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3825 Pass it down to children as needed.
3826 (thread_outgoing_edges): Likewise.
3827 (thread_across_edge): Likewise. Push/pop range state as needed.
3828 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3830 2017-12-12 Julia Koval <julia.koval@intel.com>
3832 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3833 (PTA_CANNONLAKE): Remove PTA_CLWB.
3835 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3838 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3839 (vcondv2div2df): Likewise.
3840 (vconduv2dfv2di): Likewise.
3842 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3844 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3845 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3846 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3847 ATTR_MATHFN_FPROUNDING_ERRNO.
3849 2017-12-12 Richard Biener <rguenther@suse.de>
3851 PR tree-optimization/83385
3852 * graphite-scop-detection.c (get_order, order): Remove.
3853 (bb_to_rpo): New global.
3855 (build_scops): Sort pbbs in RPO order.
3857 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3859 * combine.c (simplify_set): Do not transform subregs to zero_extends
3860 if the destination mode is a vector mode.
3862 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3864 PR tree-optimization/83359
3865 * tree-cfg.h (fold_loop_internal_call): Declare.
3866 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3867 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3868 (find_loop_dist_alias): New function.
3869 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3870 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3873 PR tree-optimization/80631
3874 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3875 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3876 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3877 hardcoding zero as the value if COND_EXPR is never true. For
3878 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3879 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3880 hardcoding MAX_EXPR as the reduction operation.
3881 (is_nonwrapping_integer_induction): Allow negative step.
3882 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3883 vect_create_epilog_for_reduction, if no value is suitable, don't
3884 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3886 2017-12-12 Richard Biener <rguenther@suse.de>
3888 PR tree-optimization/81889
3889 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3890 range info from the non-wrapping IV instead of just the range
3893 2017-12-12 Julia Koval <julia.koval@intel.com>
3895 * config.gcc: Add vaesintrin.h.
3896 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3898 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3899 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3901 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3902 * config/i386/immintrin.h: Include vaesintrin.h.
3903 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3904 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3905 _mm_aesdec_epi128): New intrinsics.
3907 2017-12-12 Julia Koval <julia.koval@intel.com>
3909 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3910 OPTION_MASK_ISA_VAES_UNSET): New.
3911 (ix86_handle_option): Handle -mvaes.
3912 * config/i386/cpuid.h: Define bit_VAES.
3913 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3914 * config/i386/i386-c.c (__VAES__): New.
3915 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3916 (ix86_valid_target_attribute_inner_p): Ditto.
3917 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3918 * config/i386/i386.opt: Add -mvaes.
3919 * doc/invoke.texi: Ditto.
3921 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3923 * debug.h (gcc_debug_hooks): Add inline_entry.
3924 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3925 * debug.c (do_nothing_debug_hooks): Likewise.
3926 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3927 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3928 (dwarf2_lineno_debug_hooks): Likewise.
3930 * common.opt (gstatement-frontiers): New, setting
3931 debug_nonbind_markers_p.
3932 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3933 * toplev.c (process_options): Autodetect value for debug statement
3935 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3936 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3938 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3939 markers. Integrate source bind into debug stmt expand loop.
3940 (pass_expand::execute): Check debug marker limit. Avoid deep
3941 TER and expand debug locations for debug bind insns only.
3942 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3944 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3945 * final.c (reemit_insn_block_notes): Take current block from
3946 nonbind markers. Declare note where it's first set.
3947 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3948 begin stmt markers if enabled.
3949 (notice_source_line): Handle nonbind markers. Fail if their
3950 location is unknown or that of builtins.
3951 (rest_of_handle_final): Convert begin stmt markers to notes if
3952 var-tracking didn't run.
3953 (rest_of_clean_state): Skip begin stmt markers.
3954 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3956 * function.c (allocate_struct_function): Set begin_stmt_markers.
3957 * function.h (struct function): Add debug_marker_count counter
3958 and debug_nonbind_markers flag.
3959 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3960 * gimple-low.c (lower_function_body): Adjust
3961 debug_nonbind_markers.
3962 (lower_stmt): Drop or skip gimple debug stmts.
3963 (lower_try_catch): Skip debug stmts.
3964 * gimple.c (gimple_build_debug_begin_stmt): New.
3965 (gimple_copy): Increment debug_marker_count if copying one.
3966 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3967 * gimplify.c (rexpr_location): New.
3968 (rexpr_has_location): New.
3969 (warn_switch_unreachable_r): Handle gimple debug stmts.
3970 (shortcut_cond_r): Call expr_location.
3972 (find_goto_label): New.
3973 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3975 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3977 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3978 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3979 (LANG_HOOKS_INITIALIZER): ... this.
3980 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3981 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3983 (update_ebb_live_info): Skip debug insn markers.
3984 * lra.c (debug_insn_static_data): Rename to...
3985 (debug_bind_static_data): ... this.
3986 (debug_marker_static_data): New.
3987 (lra_set_insn_recog_data): Select one of the above depending
3989 (lra_update_isn_regno_info): Don't assume debug insns have
3991 (push_insns): Skip debug insns.
3992 * lto-streamer-in.c (input_function): Drop debug stmts
3993 depending on active options. Adjust debug_nonbind_markers.
3994 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3995 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3996 begin stmt marker notes.
3997 (print_insn): Likewise.
3998 * recog.c (extract_insn): Recognize rtl for debug markers.
3999 * rtl.def (DEBUG_MARKER): New.
4000 * tree-inline.c: Include params.h.
4001 (remap_gimple_stmt): Handle nonbind markers.
4002 (maybe_move_debug_stmts_to_successors): Likewise.
4003 (copy_debug_stmt): Likewise.
4004 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
4005 markers regardless of no side effects.
4006 (tsi_link_before): Don't update container's side effects when adding
4007 a begin stmt marker.
4008 (tsi_link_after): Likewise.
4009 (expr_first): Skip begin stmt markers.
4010 (expr_last): Likewise.
4011 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
4012 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4013 Disregard nonbind markers.
4014 * tree.c (make_node_stat): Don't set side effects for begin stmt
4016 (build1_stat): Likewise.
4017 * tree.def (DEBUG_BEGIN_STMT): New.
4018 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
4019 * var-tracking.c (delete_debug_insns): Renamed to...
4020 (delete_vta_debug_insns): ... this.
4021 (reemit_marker_as_note): New.
4022 (vt_initialize): Reemit markers.
4023 (delete_vta_debug_insns): Likewise.
4024 (vt_debug_insns_local): Reemit or delete markers.
4025 (variable_tracking_main_1): Likewise.
4026 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
4027 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
4028 (gimple_debug_nonbind_marker_p): New.
4029 (gimple_build_debug_bind): Adjust.
4030 (gimple_build_debug_begin_stmt): New.
4031 * doc/invoke.texi (max-debug-marker-count): New param.
4032 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
4033 (debug_parameter_ref, debug_marker): New.
4034 (NOTE_INSN_BEGIN_STMT): New.
4035 (DEBUG_INSN): Describe begin stmt markers.
4037 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
4038 without debug insns, we wouldn't, but clean up debug insns
4039 after a control flow insn nevertheless.
4041 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
4042 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
4043 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
4044 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
4045 (rtl_tidy_fallthru_edge): Likewise.
4046 (rtl_verify_fallthru): Likewise.
4047 (rtl_verify_bb_layout): Likewise.
4048 (skip_insns_after_block): Likewise.
4049 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
4050 * dwarf2out.c: Include print-rtl.h.
4051 (dwarf2out_next_real_insn): New.
4052 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
4053 Dump debug binds in asm comments.
4054 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
4055 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
4056 callers to use gsi_start_nondebug_bb instead.
4057 (gsi_after_labels): Skip gimple debug stmts.
4058 (gsi_start_nondebug): New.
4059 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
4060 (proper_loop_form_for_interchange): Adjust.
4061 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
4062 * gimple.h (gimple_seq_last_nondebug_stmt): New.
4063 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
4064 (collect_fallthrough_labels): Likewise.
4065 (should_warn_for_implicit_fallthrough): Likewise.
4066 (warn_implicit_fallthrough_r): Likewise.
4067 (expand_FALLTHROUGH_r): Likewise.
4068 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
4069 (graphite_copy_stmts_from_block): Skip nonbind markers.
4070 * haifa-sched.c (sched_extend_bb): Skip debug insns.
4071 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
4072 * jump.c (clean_barriers): Skip debug insns.
4073 * omp-expand.c (expand_parallel_call): Skip debug insns.
4074 (expand_task_call): Likewise.
4075 (remove_exit_barrier): Likewise.
4076 (expand_omp_taskreg): Likewise.
4077 (expand_omp_for_init_counts): Likewise.
4078 (expand_omp_for_generic): Likewise.
4079 (expand_omp_for_static_nochunk): Likewise.
4080 (expand_omp_for_static_chunk): Likewise.
4081 (expand_omp_simd): Likewise.
4082 (expand_omp_taskloop_for_outer): Likewise.
4083 (expand_omp_taskloop_for_inner): Likewise.
4084 (expand_oacc_for): Likewise.
4085 (expand_omp_sections): Likewise.
4086 (expand_omp_single): Likewise.
4087 (expand_omp_synch): Likewise.
4088 (expand_omp_atomic_load): Likewise.
4089 (expand_omp_atomic_store): Likewise.
4090 (expand_omp_atomic_fetch_op): Likewise.
4091 (expand_omp_atomic_pipeline): Likewise.
4092 (expand_omp_atomic_mutex): Likewise.
4093 (expand_omp_target): Likewise.
4094 (grid_expand_omp_for_loop): Likewise.
4095 (grid_expand_target_grid_body): Likewise.
4096 (build_omp_regions_1): Likewise.
4097 * omp-low.c (check_combined_parallel): Skip debug stmts.
4098 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
4099 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
4100 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
4102 * tree-cfg.c (make_blobs_1): Skip debug stmts.
4103 (make_edges): Likewise.
4104 (cleanup_dead_labels): Likewise.
4105 (gimple_can_merge_blocks_p): Likewise.
4106 (stmt_starts_bb_p): Likewise.
4107 (gimple_block_label): Likewise.
4108 (gimple_redirect_edge_and_branch): Likewise.
4109 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
4111 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
4113 * tree-pretty-print (print_declaration): Omit initializer in slim
4115 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
4117 (eliminate_unnecessary_stmts): Stabilize block removal order.
4118 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
4119 * var-tracking.c (get_first_insn): New.
4120 (vt_emit_notes): Call it.
4121 (vt_initialize): Walk any insns before the first BB.
4122 (delete_debug_insns): Likewise.
4124 * gimple.h (enum gimple_debug_subcode): Add
4125 GIMPLE_DEBUG_BEGIN_STMT.
4126 (gimple_debug_begin_stmt_p): New.
4127 (gimple_debug_nonbind_marker_p): New.
4128 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
4129 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
4130 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
4131 * insn-notes.def (BEGIN_STMT): New.
4132 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
4133 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
4134 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
4135 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
4136 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
4137 (INSN_DEBUG_MARKER_KIND): New.
4138 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
4139 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
4140 (INSN_VAR_LOCATION_PTR): New.
4141 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
4143 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
4144 deps only if debug bind insns are enabled.
4145 (pass_expand::execute): Avoid deep TER and expand
4146 debug locations for debug bind insns only.
4147 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
4148 debug stmts special handling down to debug bind stmts.
4149 * combine.c (try_combine): Narrow debug insns special handling
4150 down to debug bind insns.
4151 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
4152 Narrow debug insns preexisting special handling down to debug
4154 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
4155 handling down to debug bind insns.
4156 * function.c (instantiate_virtual_regs): Skip debug markers,
4157 adjust handling of debug binds.
4158 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
4159 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
4160 * haifa-sched.c (schedule_insn): Narrow special handling of debug
4161 insns to debug bind insns.
4162 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
4163 special handling of debug stmts to debug bind stmts.
4164 * ipa-split.c (split_function): Likewise.
4165 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
4166 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
4168 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
4169 * regrename.c (build_def_use): Likewise.
4170 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
4171 (pass_cprop_hardreg): Narrow special casing of debug insns to
4173 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
4174 * reload1.c (reload): Likewise.
4175 * sese.c (sese_insert_phis_for_liveouts): Narrow special
4176 casing of debug stmts to debug bind stmts.
4177 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
4178 * ssa-iterators.h (num_imm_uses): Likewise.
4179 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
4180 debug stmts to debug bind stmts.
4181 * tree-inline.c (tree_function_versioning): Narrow special casing
4182 of debug stmts to debug bind stmts.
4183 * tree-loop-distribution.c (generate_loops_for_partition):
4184 Narrow special casing of debug stmts to debug bind stmts.
4185 * tree-sra.c (analyze_access_subtree): Narrow special casing
4186 of debug stmts to debug bind stmts.
4187 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
4188 stmts to debug bind stmts.
4189 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
4190 casing of debug stmts to debug bind stmts.
4191 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
4192 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
4193 casing of debug stmts to debug bind stmts.
4194 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4196 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
4197 debug stmts to debug bind stmts.
4198 (gimple_replace_ssa_lhs): Likewise.
4199 (insert_debug_temp_for_var_def): Likewise.
4200 (insert_debug_temps_for_defs): Likewise.
4201 (reset_debug_uses): Likewise.
4202 * tree-ssanames.c (release_ssa_name_fn): Likewise.
4203 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
4204 (adjust_debug_stmts): Likewise.
4205 (adjust_phi_and_debug_stmts): Likewise.
4206 (vect_do_peeling): Likewise.
4207 * tree-vect-loop.c (vect_transform_loop): Likewise.
4208 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
4209 * var-tracking.c (adjust_mems): Narrow special casing of debug
4210 insns to debug bind insns.
4211 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
4212 (compute_bb_dataflow, vt_find_locations): Likewise.
4213 (vt_expand_loc, emit_notes_for_changes): Likewise.
4214 (vt_init_cfa_base): Likewise.
4215 (vt_emit_notes): Likewise.
4216 (vt_initialize): Likewise.
4217 (vt_finalize): Likewise.
4219 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
4220 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
4221 (next_nonnote_nondebug_insn_bb): New.
4222 (prev_nonnote_nondebug_insn_bb): New.
4223 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4224 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4225 (prev_nonnote_nondebug_insn_bb): Declare.
4226 (next_nonnote_nondebug_insn_bb): Declare.
4227 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4228 * cfgrtl.c (get_last_bb_insn): Likewise.
4229 * lra.c (push_insns): Likewise.
4231 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4234 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4236 * selftest.c (selftest::test_runner::test_runner): New ctor.
4237 (selftest::test_runner::~test_runner): New dtor.
4238 * selftest.h (class selftest::test_runner): New class.
4240 2017-12-11 Carl Love <cel@us.ibm.com>
4242 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4243 vec_extract_fp32_from_shortl]): Add #defines.
4244 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4245 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4246 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4247 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4248 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4249 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4250 * doc/extend.texi: Add documentation for the added builtins.
4252 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4254 PR rtl-optimization/80693
4255 PR rtl-optimization/81019
4256 PR rtl-optimization/81020
4257 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4258 are not mentioned in i3. Place the REG_UNUSED note on i2,
4259 possibly modified to REG_DEAD, if it did not originate in i3.
4261 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4263 * recog.c (store_data_bypass_p_1): New function.
4264 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4265 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4267 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4269 PR rtl-optimization/83361
4270 * ifcvt.c (if_convert): Call fixup_partitions.
4272 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4274 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4275 early folding of splat_u{8,16,32}.
4277 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4279 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4280 output_operand_lossage first argument with capital letter.
4281 (aarch64_override_options): Don't start error and sorry first argument
4282 with capital letter.
4284 2017-12-11 Andi Kleen <ak@linux.intel.com>
4286 PR gcov-profile/83355
4287 * auto-profile.c (string_table::get_index_by_decl): Don't
4288 recurse when abstract origin points to itself.
4290 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4292 PR tree-optimization/83320
4293 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4294 (prune_datarefs_not_in_loop): Ditto.
4296 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4298 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4300 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4302 PR tree-optimization/83337
4303 * gimple-loop-interchange.cc (compute_access_stride): Handle
4304 bitfield DRs properly.
4306 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4308 PR tree-optimization/83338
4309 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4310 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4311 vector of unsigned integers to vector of signed integers.
4313 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4315 PR rtl-optimization/83317
4316 * lra-constraints.c (process_address_1): Add insn code check.
4318 2017-12-08 Michael Matz <matz@suse.de>
4320 Fix PR tree-optimization/83323
4321 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4322 head-controlled loops and loop BBs.
4323 * common.opt (funroll-and-jam): Remove, instead ...
4324 (floop-unroll-and-jam): ... reuse this option.
4325 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4326 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4327 (-floop-unroll-and-jam): ... this option.
4329 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4331 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4332 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4333 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4335 2017-12-08 Richard Biener <rguenther@suse.de>
4338 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4339 handle accesses outside of zero-sized vars.
4341 2017-12-08 Martin Jambor <mjambor@suse.cz>
4343 PR tree-optimization/83141
4344 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4345 test for MEM_REFs implicitely changing types with padding. Remove
4347 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4349 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4351 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4353 * config/avr/avr.c (avr_attribute_table): Likewise.
4354 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4355 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4356 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4357 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4358 of struct attribute_spec.
4359 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4361 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4364 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4366 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4369 * profile-count.c (profile_count::from_gcov_type): Move from
4370 profile-count.h; handle overflow.
4371 * profile-count.h (profile_count::from_gcov_type): Move offline.
4373 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4375 PR rtl-optimization/83304
4376 * combine.c (move_deaths): If we do not know where a register died,
4379 2017-12-08 Richard Biener <rguenther@suse.de>
4381 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4382 Provide -fopt-info-loop feedback when we interchange in a nest.
4384 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4386 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4387 for armv6 ARM CPU IDs.
4389 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4391 * common/config/arm/arm-common.c: Include <algorithm>.
4392 (INCLUDE_VECTOR): Define.
4393 (compare_opt_names): New function.
4394 (arm_rewrite_selected_arch): Only strip out extensions that can be
4395 expressed through -mfpu. Sort the remaining extensions
4398 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4400 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4401 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4402 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4403 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4405 2017-06-08 Tristan Gingold <gindold@adacore.com>
4408 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4409 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4411 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4413 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4414 of struct attribute_spec.
4416 2017-12-08 Julia Koval <julia.koval@intel.com>
4418 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4419 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4420 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4421 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4422 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4423 _mm_maskz_dpwssds_epi32): Ditto.
4425 2017-12-08 Richard Biener <rguenther@suse.de>
4427 PR tree-optimization/81303
4428 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4429 conditions try to create a comparison vector type matching
4430 the data vector type.
4431 (vectorizable_condition): Adjust.
4432 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4433 Leave invariant conditions alone in case we can vectorize those.
4435 2017-12-08 Julia Koval <julia.koval@intel.com>
4437 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4438 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4439 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4440 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4441 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4443 2017-12-08 Julia Koval <julia.koval@intel.com>
4445 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4446 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4447 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4448 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4449 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4450 _mm_maskz_dpbusds_epi32): New intrinsics.
4452 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4454 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4456 (ld<bh>io_signed): Likewise.
4457 (st<bhw>io): Likewise.
4458 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4459 SMALL_INT12 constant integer operand.
4461 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4463 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4465 (nios2_large_constant_allowed): ...this. Adjust uses.
4466 (nios2_plus_symbolic_constant_p): Rename to...
4467 (nios2_plus_large_constant_p): ...this. Adjust uses.
4468 (nios2_legitimate_address_p): Correct CONST_INT handling.
4469 (nios2_symbolic_memory_operand_p): Rename to...
4470 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4471 (nios2_large_constant_p): Check for large constant integers too.
4472 (nios2_split_large_constant): Handle constant integers.
4473 (nios2_split_symbolic_memory_operand): Rename to...
4474 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4475 (nios2_legitimize_constant_address): Handle constant integers.
4476 (r0rel_constant_p): Handle small constant integers.
4477 (nios2_print_operand_address): Handle r0-relative integer addresses.
4478 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4479 * config/nios2/nios2.md: Adjust for renamed functions.
4481 2017-12-07 Andrew Waterman <andrew@sifive.com>
4483 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4484 (TARGET_HAVE_SRODATA_SECTION): New define.
4485 (riscv_select_section): New function.
4487 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4488 Alexander Monakov <amonakov@ispras.ru>
4489 Jakub Jelinek <jakub@redhat.com>
4492 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4494 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4496 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4497 a bool success value. Don't call output_operand_lossage here.
4498 (aarch64_print_ldpstp_address): Return a bool success value.
4499 (aarch64_print_operand_address): Call output_addr_const if
4500 aarch64_print_address_internal fails.
4501 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4502 'y'; call output_operand_lossage instead. Call output_operand_lossage
4503 if aarch64_print_ldpstp_address fails.
4505 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4507 * tree-vector-builder.h
4508 (tree_vector_builder::binary_encoded_nelts): Declare.
4509 * tree-vector-builder.c
4510 (tree_vector_builder::binary_encoded_nelts): New function.
4511 * fold-const.c (negate_expr_p): Likewise.
4512 (operand_equal_p, fold_checksum_tree): Likewise.
4513 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4514 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4515 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4516 (uniform_vector_p): Likewise.
4517 * varasm.c (const_hash_1, compare_constant): Likewise.
4518 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4519 (valid_lattice_transition): Operate directly on the VECTOR_CST
4521 * ipa-icf.c: Include tree-vector-builder.h.
4522 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4523 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4525 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4527 * tree.c (build_vector): Delete.
4528 * tree.h (build_vector): Make static and move into the self-testing
4531 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4533 * vector-builder.h (vector_builder::derived): New const overload.
4534 (vector_builder::elt): New function.
4535 * tree-vector-builder.h (tree_vector_builder::type): New function.
4536 (tree_vector_builder::apply_step): Declare.
4537 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4539 * gimple-fold.h (tree_vector_builder): Declare.
4540 (gimple_build_vector): Take a tree_vector_builder instead of a
4541 type and vector of elements.
4542 * gimple-fold.c (gimple_build_vector): Likewise.
4543 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4545 (get_initial_defs_for_reduction): Likewise.
4546 (vectorizable_induction): Likewise.
4548 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4550 * tree-vector-builder.h
4551 (tree_vector_builder::new_binary_operation): Declare.
4552 * tree-vector-builder.c
4553 (tree_vector_builder::new_binary_operation): New function.
4554 * fold-const.c (fold_relational_const): Use it.
4555 (const_binop): Likewise. Check that both input vectors have
4556 the same number of elements, thus excluding things like WIDEN_SUM.
4557 Check whether it is possible to operate directly on the encodings
4560 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4562 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4563 new_unary_operation, operating only on the encoded elements.
4564 (const_unop): Likewise.
4565 (exact_inverse): Likewise.
4566 (distributes_over_addition_p): New function.
4567 (const_binop): Use tree_vector_builder and new_unary_operation
4568 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4569 on the encoded elements unless the encoding is strided and the
4570 operation does not distribute over addition.
4571 (fold_convert_const): Use tree_vector_builder and
4572 new_unary_operation. Operate only on the encoded elements
4573 for truncating integer conversions, or for non-stepped encodings.
4575 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4577 * config/sparc/sparc.c: Include tree-vector-builder.h.
4578 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4579 * expmed.c: Include tree-vector-builder.h.
4580 (make_tree): Use tree_vector_builder instead of build_vector.
4581 * fold-const.c: Include tree-vector-builder.h.
4582 (const_binop): Use tree_vector_builder instead of build_vector.
4583 (const_unop): Likewise.
4584 (native_interpret_vector): Likewise.
4585 (fold_vec_perm): Likewise.
4586 (fold_ternary_loc): Likewise.
4587 * gimple-fold.c: Include tree-vector-builder.h.
4588 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4590 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4591 (simplify_vector_constructor): Use tree_vector_builder instead
4593 * tree-vect-generic.c: Include tree-vector-builder.h.
4594 (add_rshift): Use tree_vector_builder instead of build_vector.
4595 (expand_vector_divmod): Likewise.
4596 (optimize_vector_constructor): Likewise.
4597 * tree-vect-loop.c: Include tree-vector-builder.h.
4598 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4599 of build_vector. Explicitly use a stepped encoding for
4601 * tree-vect-slp.c: Include tree-vector-builder.h.
4602 (vect_get_constant_vectors): Use tree_vector_builder instead
4604 (vect_transform_slp_perm_load): Likewise.
4605 (vect_schedule_slp_instance): Likewise.
4606 * tree-vect-stmts.c: Include tree-vector-builder.h.
4607 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4608 (vect_gen_perm_mask_any): Likewise.
4609 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4610 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4612 (build_vector_from_val): Likewise. Explicitly use a duplicate
4615 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4617 * doc/generic.texi (VECTOR_CST): Describe new representation of
4619 * vector-builder.h: New file.
4620 * tree-vector-builder.h: Likewise.
4621 * tree-vector-builder.c: Likewise.
4622 * Makefile.in (OBJS): Add tree-vector-builder.o.
4623 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4624 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4625 (tree_vector): Change the number of elements to
4626 vector_cst_encoded_nelts.
4627 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4628 (VECTOR_CST_ELTS): Delete.
4629 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4630 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4631 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4632 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4633 (VECTOR_CST_ENCODED_ELT): Likewise.
4634 (vector_cst_encoded_nelts): New function.
4635 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4636 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4637 (vector_cst_int_elt, vector_cst_elt): Declare.
4638 * tree.c: Include tree-vector-builder.h.
4639 (tree_code_size): Abort if passed VECTOR_CST.
4640 (tree_size): Update for new VECTOR_CST layout.
4641 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4642 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4643 (build_vector): Use tree_vector_builder.
4644 (vector_cst_int_elt, vector_cst_elt): New functions.
4645 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4646 encoded elements and then create the vector in the canonical form.
4647 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4648 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4649 (tree_c_tests): Call test_vector_cst_patterns.
4650 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4652 (hash_tree): Likewise.
4653 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4654 (streamer_write_tree_header): Likewise.
4655 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4656 (streamer_alloc_tree): Likewise. Update call to make_vector.
4657 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4659 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4661 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4662 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4663 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4665 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4666 Richard Biener <rguenther@suse.de>
4668 PR tree-optimization/81303
4669 * Makefile.in (gimple-loop-interchange.o): New object file.
4670 * common.opt (floop-interchange): Reuse the option from graphite.
4671 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4672 -floop-interchange and mention it for -O3.
4673 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4674 * gimple-loop-interchange.cc: New file.
4675 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4676 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4677 * passes.def (pass_linterchange): New pass.
4678 * timevar.def (TV_LINTERCHANGE): New time var.
4679 * tree-pass.h (make_pass_linterchange): New declaration.
4680 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4681 interchange. Record IV before/after increment in new parameters.
4682 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4683 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4685 (check_reduction_path): ...New function here.
4686 * tree-vectorizer.h (check_reduction_path): New declaration.
4688 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4691 PR rtl-optimization/80818
4692 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4693 always early clobbered.
4694 * lra-lives.c (process_bb_lives): Check input hard regs for early
4695 clobbered non-operand hard reg.
4697 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4700 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4701 types_compatible_p, just that TYPE_MODE is the same.
4703 2017-12-07 Martin Sebor <msebor@redhat.com>
4706 * attribs.c (empty_attribute_table): Initialize new member of
4707 struct attribute_spec.
4708 (decl_attributes): Add argument. Handle mutually exclusive
4709 combinations of attributes.
4710 (selftests::test_attribute_exclusions): New function.
4711 (selftests::attribute_c_tests): Ditto.
4712 * attribs.h (decl_attributes): Add default argument.
4713 * selftest.h (attribute_c_tests): Declare.
4714 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4715 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4717 * doc/extend.texi (Common Function Attributes): Update const and pure.
4718 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4719 of struct attribute_spec.
4720 * config/arc/arc.c (arc_attribute_table): Same.
4721 * config/arm/arm.c (arm_attribute_table): Same.
4722 * config/avr/avr.c ( avr_attribute_table): Same.
4723 * config/bfin/bfin.c (bfin_attribute_table): Same.
4724 * config/cr16/cr16.c (cr16_attribute_table): Same.
4725 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4726 * config/h8300/h8300.c (h8300_attribute_table): Same.
4727 * config/i386/i386.c (ix86_attribute_table): Same.
4728 * config/ia64/ia64.c (ia64_attribute_table): Same.
4729 * config/m32c/m32c.c (m32c_attribute_table): Same.
4730 * config/m32r/m32r.c (m32r_attribute_table): Same.
4731 * config/m68k/m68k.c (m68k_attribute_table): Same.
4732 * config/mcore/mcore.c (mcore_attribute_table): Same.
4733 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4734 * config/mips/mips.c (mips_attribute_table): Same.
4735 * config/msp430/msp430.c (msp430_attribute_table): Same.
4736 * config/nds32/nds32.c (nds32_attribute_table): Same.
4737 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4738 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4739 * config/rl78/rl78.c (rl78__attribute_table): Same.
4740 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4741 * onfig/rx/rx.c (rx_attribute_table): Same.
4742 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4743 * config/sh/sh.c (sh_attribute_table): Same.
4744 * config/sparc/sparc.c (sparc_attribute_table): Same.
4745 * config/spu/spu.c (spu_attribute_table): Same.
4746 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4747 * config/v850/v850.c (v850_attribute_table): Same.
4748 * config/visium/visium.c (visium_attribute_table): Same.
4750 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4753 * config/arm/arm.c (INCLUDE_STRING): Define.
4754 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4755 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4758 2017-12-07 Michael Matz <matz@suse.de>
4760 Add unroll and jam pass
4762 * gimple-loop-jam.c: New file.
4763 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4764 * common.opt (funroll-and-jam): New option.
4765 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4766 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4767 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4768 * passes.def: Add pass_loop_jam.
4769 * timevar.def (TV_LOOP_JAM): Add.
4770 * tree-pass.h (make_pass_loop_jam): Declare.
4771 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4772 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4773 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4774 to flow_loop_tree_node_add.
4775 (duplicate_subloops, copy_loops_to): Append to sibling list.
4776 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4777 * doc/invoke.texi (-funroll-and-jam): Document new option.
4778 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4780 2017-12-07 Richard Biener <rguenther@suse.de>
4782 PR tree-optimization/83296
4783 PR tree-optimization/67769
4784 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4785 flow sensitive info in an unrelated BB.
4786 (value_replacement): Use reset_flow_sensitive_info.
4787 (minmax_replacement): Reset flow sensitive info on the def
4788 we move. Do not reset flow sensitive info in the whole BB
4789 we move the stmt to.
4790 (abs_replacement): Likewise.
4792 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4795 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4796 rs6000_cpu to the given -mcpu=, or to the default processor.
4798 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4800 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4801 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4802 instead of rs6000_cpu_attr.
4803 (rs6000_adjust_cost): Ditto.
4804 (is_microcoded_insn): Ditto.
4805 (rs6000_adjust_priority): Ditto.
4806 (rs6000_issue_rate): Ditto.
4807 (rs6000_use_sched_lookahead): Ditto.
4808 (rs6000_use_sched_lookahead_guard): Ditto.
4809 (rs6000_sched_reorder): Ditto.
4810 (force_new_group): Ditto.
4811 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4812 (group_ending_nop): Ditto.
4814 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4816 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4817 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4818 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4819 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4821 (rs6000_reassociation_width): Ditto.
4822 (rs6000_emit_epilogue): Ditto.
4823 (rs6000_adjust_cost): Ditto.
4824 (is_microcoded_insn): Ditto.
4825 (is_cracked_insn): Ditto.
4826 (rs6000_adjust_priority): Ditto.
4827 (rs6000_sched_reorder): Ditto.
4828 (rs6000_sched_reorder2): Ditto.
4829 (insn_must_be_first_in_group): Ditto.
4830 (insn_must_be_last_in_group): Ditto.
4831 (rs6000_register_move_cost): Ditto.
4832 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4835 2017-12-07 Julia Koval <julia.koval@intel.com>
4837 * config.gcc: Add vaesintrin.h.
4838 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4840 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4841 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4843 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4844 * config/i386/immintrin.h: Include vaesintrin.h.
4845 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4846 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4847 _mm_aesdec_epi128): New intrinsics.
4849 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4851 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4852 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4853 c-family/c-spellcheck.cc.
4854 (best_macro_match::best_macro_match): Likewise.
4856 (struct edit_distance_traits<cpp_hashnode *>): Move to
4857 c-family/c-spellcheck.h.
4858 (class best_macro_match): Likewise.
4860 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4862 PR tree-optimization/83293
4863 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4864 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4865 might insert into empty bb.
4868 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4869 simplify for plus with :c added, and pointer_plus without that.
4870 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4871 with undefined overflow and the conversion is not widening,
4872 perform negation in utype and only convert to type afterwards.
4873 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4874 simplify for plus with :c added, and pointer_plus without that.
4875 If type is integral with undefined overflow and the conversion is
4876 not widening, perform minus in utype and only convert to type
4877 afterwards. Move the last pointer_diff_expr simplify into the
4880 2017-12-06 Martin Sebor <msebor@redhat.com>
4882 PR tree-optimization/82646
4883 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4884 strncpy, not maxlen.
4886 2017-12-06 Martin Sebor <msebor@redhat.com>
4888 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4891 PR tree-optimization/83075
4892 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4893 strncat/strncpy don't change length of source string.
4895 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4898 2017-11-29 Martin Aberg <maberg@gaisler.com>
4900 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4901 to prevent b2bst errata sequence.
4902 (sqrtdf2_fix): Likewise.
4904 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4906 PR tree-optimization/81945
4907 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4908 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4909 to dest_cfun has orig_loop_num set, either remap it to the new
4910 loop number if the loop got moved too, or clear it.
4912 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4914 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4916 (thunderx2t99_nothing): New insn reservation.
4917 (thunderx2t99_mrs): New insn reservation.
4918 (thunderx2t99_multiple): New insn reservation.
4919 (thunderx2t99_alu_basi): Add bfx to reservation.
4920 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4922 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4925 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4927 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4929 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4930 tree-ssa-pre.c::remove_dead_inserted_code.
4931 * tree-ssa-dce.h: New file.
4932 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4933 (remove_dead_inserted_code): Move and rename to function
4934 tree-ssa-dce.c::simple_dce_from_worklist.
4935 (pass_pre::execute): Update use.
4937 2017-12-05 Richard Biener <rguenther@suse.de>
4939 PR tree-optimization/83277
4940 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4941 to code-gen liveout vars.
4943 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4945 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4946 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4947 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4950 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4952 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4954 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4955 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4958 2017-12-05 Richard Biener <rguenther@suse.de>
4960 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4961 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4962 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4963 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4965 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4967 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4969 2017-12-05 Martin Liska <mliska@suse.cz>
4970 Jakub Jelinek <jakub@redhat.com>
4972 * doc/invoke.texi: Document the options.
4973 * flag-types.h (enum sanitize_code): Add
4974 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4975 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4976 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4977 * opts.c: Define new sanitizer options.
4978 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4979 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4981 2017-12-05 Julia Koval <julia.koval@intel.com>
4983 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4984 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4985 (ix86_handle_option): Handle -mavx512vnni.
4986 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4987 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4988 * config/i386/i386-c (__AVX512VNNI__): New.
4989 * config/i386/i386.c (ix86_target_string): Handle new option.
4990 (ix86_valid_target_attribute_inner_p): Handle new option.
4991 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4992 * config/i386/i386.opt (mavx512vnni): New option.
4994 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4997 * athlon.md: Disable for generic.
4998 * haswell.md: Enable for generic.
4999 * i386.c (ix86_sched_init_global): Add core hooks for generic.
5000 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
5002 (ix86_adjust_cost): Move generic to haswell path.
5004 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
5006 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
5007 instead of MEM_P in a couple more places. Fix formatting issues.
5009 2017-12-04 Jim Wilson <jimw@sifive.com>
5011 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
5012 instead of GP_REG_LAST-1.
5013 (riscv_adjust_libcall_cfi_prologue): Likewise.
5014 (riscv_adjust_libcall_cri_epilogue): Likewise.
5015 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
5018 2017-12-04 Luis Machado <luis.machado@linaro.org>
5020 * ipa-pure-const.c (check_decl): Add missing newline.
5021 (state_from_flags): Likewise.
5023 2017-12-04 Jeff Law <law@redhat.com>
5025 PR tree-optimizatin/78496
5026 * gimple-ssa-evrp-analyze.h
5027 (evrp_range_analyzer::get_vr_values): Simplify.
5028 * gimple-ssa-evrp-analyze.c: Corresponding changes.
5029 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
5030 and gimple-ssa-evrp-analyze.h.
5031 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
5032 (simplify_stmt_for_jump_threading): Copy a blob of code from
5033 tree-vrp.c to use ranges to simplify statements.
5034 (dom_opt_dom_walker::before_dom_children): Call
5035 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
5036 (dom_opt_dom_walker::after_dom_children): Similarly for
5037 evrp_range_analyzer::leave.
5038 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
5041 * gimple-ssa-evrp-analyze.c
5042 (evrp_range_analyzer::extract_range_from_stmt): Always use
5043 vr_values::update_value_range so preexisting range info is
5044 medged with new range info, even if the new range is VR_VARYING.
5046 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5048 * combine.c: Adjust comment.
5049 (use_crosses_set_p): Delete.
5050 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
5051 (try_combine): Ditto.
5053 2017-12-04 Richard Biener <rguenther@suse.de>
5055 PR tree-optimization/83255
5056 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
5057 Re-add zero-iteration check.
5059 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5061 PR rtl-optimization/83245
5062 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
5063 hard registers as earlyclobber, also if not in an asm.
5065 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5069 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5072 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5073 rs6000_cpu based on cpu_index, not tune_index.
5075 2017-12-04 Richard Biener <rguenther@suse.de>
5077 PR tree-optimization/83238
5078 * graphite-scop-detection.c (scop_detection::merge_sese): Make
5079 code match comment, rejecting invalid SESE regions.
5081 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
5083 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
5084 require base operand is a REG_POINTER prior to reload on targets
5085 with non-equivalent space registers.
5087 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5089 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
5090 (update_specialized_profile): Fix updating of counts.
5091 (perhaps_add_new_callers): Likewise.
5093 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5096 * x86-tune.def: Remove obsolette FIXMEs.
5097 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
5098 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
5099 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
5101 (X86_TUNE_PAD_RETURNS): Disable for generic.
5103 2017-12-02 Jakub Jelinek <jakub@redhat.com>
5105 PR tree-optimization/83170
5106 PR tree-optimization/83241
5107 * gimple-ssa-store-merging.c
5108 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
5109 gimple_vuse (ins_stmt) in case it has changed.
5110 (imm_store_chain_info::output_merged_store): Likewise.
5112 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
5116 * tree-cfg.c (pass_warn_function_return::execute): Handle
5117 __builtin_ubsan_handle_missing_return like __builtin_unreachable
5118 with BUILTINS_LOCATION.
5122 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
5123 is BLKmode for vector field with vector raw mode, use TYPE_MODE
5124 instead of DECL_MODE.
5126 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
5127 last argument to rtx pointer.
5128 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
5129 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
5130 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
5131 depending on the chosen ISAs.
5132 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
5133 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
5134 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
5136 * config/i386/sse.md (mov<mode>_internal): Likewise.
5137 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
5139 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5141 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
5143 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5145 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
5146 parameters from prototype.
5147 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
5148 parameters. Don't print a comment.
5149 (emit_fusion_gpr_load): Adjust.
5150 (emit_fusion_load_store): Adjust.
5151 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
5152 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
5153 comment on the second line.
5155 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5158 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5159 rs6000_cpu based on cpu_index, not tune_index.
5161 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5163 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
5164 which_alternative instead of which_alternative + 1.
5165 (output_asm_insn): Print an extra tab if the template is short.
5167 2017-12-01 Jim Wilson <jimw@sifive.com>
5169 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
5171 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
5172 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
5174 * doc/tm.texi: Regenerate.
5176 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
5179 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
5180 whether we can allocate pseudos before trying to fix an address.
5181 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
5182 memory address is indexed or indirect.
5183 (floatuns_<mode>si2_hw2): Likewise.
5185 2017-12-01 Jason Merrill <jason@redhat.com>
5187 * Makefile.in (TAGS): Add c-family/*.cc.
5189 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
5191 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
5192 (call_value_insn): Likewise.
5193 (sibcall_insn): Likewise.
5194 (sibcall_value_insn): Likewise.
5195 (movsi_aarch64): Likewise.
5196 (movdi_aarch64): Likewise.
5197 (add_losym_): Likewise.
5198 (ldr_got_small_): Likewise.
5199 (ldr_got_small_sidi): Likewise.
5200 (ldr_got_small_28k_): Likewise.
5201 (ldr_got_small_28k_sidi): Likewise.
5202 * config/aarch64/aarch64.c (aarch64_print_address_internal):
5203 Move output_addr_const to symbolic case. Add error check.
5205 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5207 * config/s390/predicates.md (plus16_Q_operand): New predicate.
5208 * config/s390/s390.md: Disable MVC merging peephole if it would
5209 disable operand forwarding.
5210 (new peephole2): Split MVCs if it would turn them into up to 2
5213 2017-12-01 Richard Biener <rguenther@suse.de>
5215 PR tree-optimization/83232
5216 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
5217 detection of same access. Instead of breaking the group here
5218 do not consider the duplicate. Add comment explaining real fix.
5220 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
5222 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5224 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5226 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5228 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5230 * function.h (struct function): Remove cilk_frame_decl,
5231 is_cilk_function and calls_cilk_spawn fields.
5232 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5234 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5235 cilk_elemental field.
5236 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5237 * target.def: Adjust comment.
5238 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5239 Don't test cilk_elemental.
5241 PR tree-optimization/83233
5242 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5243 bswap_stat name for the struct.
5246 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5247 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5248 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5249 SWITCH_BREAK_LABEL_P set on the label.
5250 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5251 added for default case if it was missing and not all cases covered.
5252 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5253 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5255 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5256 to avoid -Wimplicit-fallthrough warning.
5257 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5258 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5260 PR tree-optimization/83221
5261 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5263 (init_reassoc): Formatting fix.
5266 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5267 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5271 * tree-cfg.c: Include asan.h.
5272 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5274 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5276 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5278 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5280 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5281 vpcmpeqd instruction.
5283 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5285 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5286 registers type for 128bit mode.
5288 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5290 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5292 * gengtype-state.c (read_state_pair): Likewise.
5293 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5294 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5295 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5296 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5297 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5298 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5299 * builtins.c (fold_builtin_3): Likewise.
5300 * graphite-scop-detection.c
5301 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5302 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5305 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5308 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5309 parameter. Return true if ignore_decl_rtl_set_p is true and other
5310 conditions are satisfied.
5311 * asan.h (asan_protect_global): Add new parameter.
5312 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5313 to asan_protect_global calls.
5315 2017-11-30 Jim Wilson <jimw@sifive.com>
5317 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5318 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5319 -mstrict-align, add info on default value. Delete redundant lines for
5320 -mabi. Add missing -mexplicit-relocs docs.
5322 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5324 * config/arc/arc.md (trap): New pattern.
5326 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5328 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5329 instruction to end into a delay slot.
5330 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5331 can be placed into a delay slot against reg_note.
5333 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5335 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5336 labels number of usages.
5338 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5340 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5342 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5344 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5347 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5348 multiplication by power of 2 constant into two shifts + comparison.
5350 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5353 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5355 2017-11-30 Richard Biener <rguenther@suse.de>
5357 PR tree-optimization/83202
5358 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5359 allow_peel argument and guard peeling.
5360 (canonicalize_loop_induction_variables): Likewise.
5361 (canonicalize_induction_variables): Pass false.
5362 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5363 peeling from cunrolli.
5365 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5367 * combine.c (try_combine): Print a message to dump file whenever
5368 I0, I1, or I2 cannot be combined into I3.
5370 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5372 PR rtl-optimization/83156
5373 PR rtl-optimization/82621
5374 * combine.c (try_combine): Don't split an I2 if one of the dests is
5375 set again before I3. Allow unused dests.
5377 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5379 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5381 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5383 PR rtl-optimization/80818
5384 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5385 recursively. Use insn code for clobber.
5386 (lra_set_insn_recog_data): Pass the new arg to
5387 collect_non_operand_hard_regs.
5388 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5390 (lra_update_insn_regno_info): Pass insn to
5391 add_regs_to_insn_regno_info.
5393 2017-11-29 Jim Wilson <jimw@sifive.com>
5394 Andrew Waterman <andrew@sifive.com>
5396 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5397 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5398 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5399 use SINGLE_SHIFT_COST.
5400 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5401 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5403 2017-11-29 Julia Koval <julia.koval@intel.com>
5405 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5406 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5407 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5408 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5409 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5410 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5411 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5412 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5413 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5414 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5415 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5416 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5417 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5418 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5419 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5420 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5421 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5422 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5423 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5424 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5425 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5426 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5427 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5428 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5429 __builtin_ia32_vpshldv_v2di_mask,
5430 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5431 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5432 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5434 2017-11-29 Julia Koval <julia.koval@intel.com>
5436 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5437 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5438 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5439 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5440 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5441 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5442 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5443 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5444 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5445 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5446 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5447 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5448 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5449 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5450 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5451 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5452 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5453 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5454 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5455 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5456 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5458 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5460 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5461 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5464 2017-11-29 Martin Aberg <maberg@gaisler.com>
5466 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5467 to prevent b2bst errata sequence.
5468 (sqrtdf2_fix): Likewise.
5470 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5472 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5473 (div_sqrt_insn_p): New function.
5474 (sparc_do_work_around_errata): Insert NOP instructions to
5475 prevent sequences that could trigger the TN-0013 errata for
5476 certain LEON3 processors.
5477 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5478 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5479 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5480 (in_branch_delay): Prevent div and sqrt in delay slot if
5482 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5484 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5486 * config/sparc/sparc.c (atomic_insn_p): New function.
5487 (sparc_do_work_around_errata): Insert NOP instructions to
5488 prevent sequences that could trigger the TN-0010 errata for
5490 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5491 instruction referable in atomic_insns_p.
5493 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5495 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5496 (atomic_compare_and_swap_leon3_1): Likewise.
5499 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5501 * config/sparc/sparc.c (fpop_insn_p): New function.
5502 (sparc_do_work_around_errata): Insert NOP instructions to
5503 prevent sequences that could trigger the TN-0012 errata for
5505 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5506 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5507 (in_branch_annul_delay): Prevent floating-point instructions
5508 in delay slot of annulled integer branch.
5510 2017-11-29 Richard Biener <rguenther@suse.de>
5512 PR tree-optimization/83202
5513 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5515 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5516 nodes vectorized to the same stmts multiple times.
5517 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5518 (vect_analyze_slp_instance): Adjust.
5519 (scalar_stmts_to_slp_tree_map_t): New typedef.
5520 (vect_schedule_slp_instance): Add a map recording the SLP node
5521 representing the vectorized stmts for a set of scalar stmts.
5522 Avoid code-generating redundancies.
5523 (vect_schedule_slp): Allocate map and pass it down.
5525 2017-11-29 Nathan Sidwell <nathan@acm.org>
5528 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5531 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5533 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5534 cases for printing LDP/STP memory addresses.
5535 (aarch64_print_address_internal): Renamed from
5536 aarch64_print_operand_address, added parameter, add Pmode check.
5537 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5538 (aarch64_print_operand_address): Indirect to
5539 aarch64_print_address_internal.
5540 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5541 'y' operand output specifier.
5543 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5546 * tree.c (build_simple_mem_ref_loc): Handle
5547 get_addr_base_and_unit_offset returning a MEM_REF.
5550 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5553 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5556 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5558 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5560 (layout::move_to_column): Likewise.
5562 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5565 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5566 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5568 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5569 Martin Jambor <mjambor@suse.cz>
5572 * tree.h (expr_type_first_operand_type_p): Declare
5573 * tree.c (expr_type_first_operand_type_p): New function.
5574 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5575 (ipa_value_from_jfunc): Adjust declaration.
5576 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5577 Use it as result type for arithmetics, unless it is NULL in which case
5578 be more conservative.
5579 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5580 ipa_get_jf_pass_through_result.
5581 (propagate_vals_across_pass_through): Likewise.
5582 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5583 is to propagate_vals_across_pass_through.
5584 (propagate_constants_across_call): Pass PARM_TYPE to
5585 propagate_scalar_across_jump_function.
5586 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5587 ipa_value_from_jfunc.
5588 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5589 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5590 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5591 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5592 target_type, pass it to ipa_value_from_jfunc.
5593 (update_indirect_edges_after_inlining): Pass parameter type to
5594 try_make_edge_direct_simple_call.
5596 2017-11-28 Jeff Law <law@redhat.com>
5598 * gimple-ssa-evrp-analyze.c
5599 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5600 refine ranges if scev_initialized_p returns true.
5601 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5603 2017-11-28 Julia Koval <julia.koval@intel.com>
5605 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5606 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5607 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5608 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5609 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5610 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5611 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5612 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5613 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5614 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5615 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5616 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5617 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5618 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5619 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5620 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5621 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5622 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5623 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5624 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5625 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5626 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5628 2017-11-28 Julia Koval <julia.koval@intel.com>
5630 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5631 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5632 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5633 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5634 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5635 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5636 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5637 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5638 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5639 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5640 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5641 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5642 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5643 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5644 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5645 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5646 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5647 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5648 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5649 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5650 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5651 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5652 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5653 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5654 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5655 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5656 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5657 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5658 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5659 __builtin_ia32_vpshld_v2di_mask): New builtins.
5660 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5661 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5663 2017-11-28 Richard Biener <rguenther@suse.de>
5665 PR tree-optimization/80776
5666 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5668 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5670 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5671 If the incoming edge is an effective fallthru because the other
5672 edge only reaches a __builtin_unreachable () then record ranges
5673 derived from the controlling condition in SSA info.
5674 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5675 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5677 2017-11-28 Olivier Hainque <hainque@adacore.com>
5679 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5680 on mingw build hosts.
5682 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5685 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5687 2017-11-28 Richard Biener <rguenther@suse.de>
5690 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5691 copies generated from memcpy use a character array as reference
5694 2017-11-28 Julia Koval <julia.koval@intel.com>
5695 Sebastian Peryt <sebastian.peryt@intel.com>
5697 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5698 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5699 cilk-common.o, cilk.h, cilk-common.c): Remove.
5701 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5702 * builtins.c (is_builtin_name): Remove cilkplus condition.
5703 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5704 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5705 cilk-builtins.def, cilkplus.def): Remove.
5706 * cif-code.def (CILK_SPAWN): Remove.
5707 * cilk-builtins.def: Delete.
5708 * cilk-common.c: Ditto.
5710 * cilkplus.def: Ditto.
5711 * config/darwin.h (fcilkplus): Delete.
5712 * cppbuiltin.c: Ditto.
5713 * doc/extend.texi: Remove cilkplus doc.
5714 * doc/generic.texi: Ditto.
5715 * doc/invoke.texi: Ditto.
5716 * doc/passes.texi: Ditto.
5717 * gcc.c (fcilkplus): Remove.
5718 * gengtype.c (cilk.h): Remove.
5719 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5721 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5723 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5725 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5726 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5727 cilkplus conditions.
5728 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5729 inline_read_section): Ditto.
5730 * ipa-inline-analysis.c (cilk.h): Remove.
5731 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5732 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5733 append_linker_options): Remove condition for fcilkplus.
5734 * lto/lto-lang.c (cilk.h): Remove.
5735 (lto_init): Remove condition for fcilkplus.
5736 * omp-expand.c (expand_cilk_for_call): Delete.
5737 (expand_omp_taskreg, expand_omp_for_static_chunk,
5738 expand_omp_for): Remove cilkplus
5740 (expand_cilk_for): Delete.
5741 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5742 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5743 execute_lower_omp, diagnose_sb_0): Ditto.
5744 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5745 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5746 * tree-nested.c: Ditto.
5747 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5748 (dump_generic_node): Ditto.
5749 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5750 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5752 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5754 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5756 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5757 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5758 for vector mode and !TARGET_SIMD.
5760 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5762 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5764 * tree.h (SWITCH_LABELS): Remove.
5765 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5766 assert SWITCH_BODY is non-NULL.
5767 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5769 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5771 PR tree-optimization/80788
5772 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5773 has TREE_OVERFLOW set, call drop_tree_overflow.
5775 2017-11-28 Richard Biener <rguenther@suse.de>
5777 PR tree-optimization/83158
5778 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5780 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5783 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5784 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5786 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5788 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5789 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5791 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5794 * varasm.c (compare_constant): Take the mode of the constants into
5795 account when comparing floating point constants.
5797 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5799 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5800 from explicit instantiation of debug_helper.
5801 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5803 2017-11-27 Richard Biener <rguenther@suse.de>
5805 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5808 2017-11-27 Richard Biener <rguenther@suse.de>
5810 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5811 (type_hash_canon): Also clear int_cst_hash_table entry for
5813 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5815 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5817 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5818 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5819 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5821 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5823 * hash-map.h (gt_cleare_cache): Avoid UB.
5825 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5827 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5828 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5829 (decide_unroll_runtime_iterations): Likewise.
5830 (decide_unroll_stupid): Likewise.
5832 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5835 * config/i386/i386.md: Add a loop with incssp.
5837 2017-11-27 Martin Jambor <mjambor@suse.cz>
5839 PR tree-optimization/81248
5840 * tree-sra.c (splice_param_accesses): Remove size check.
5841 (decide_one_param_reduction): Fix size check.
5842 * gimple-pretty-print.c (dump_profile): Silence warning.
5843 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5845 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5848 * dbxout.c (lastlineno): New variable.
5849 (dbx_debug_hooks): Use dbxout_switch_text_section as
5850 switch_text_section debug hook.
5851 (dbxout_function_end): Switch to current_function_section
5852 rather than function_section. If crtl->has_bb_partition,
5853 output just one N_FUN, depending on in_cold_section_p.
5854 (dbxout_source_line): Remember last lineno in lastlineno.
5855 (dbxout_switch_text_section): New function.
5856 (dbxout_function_decl): Adjust dbxout_block caller.
5857 (dbx_block_with_cold_children): New function.
5858 (dbxout_block): Return true if any LBRAC/RBRAC have been
5859 emitted. Use dbx_block_with_cold_children at depth == 0
5860 in second partition. Add PARENT_BLOCKNUM argument, pass
5861 it optionally adjusted to children. Output LBRAC/RBRAC
5862 around recursive call only if the block is in the current
5863 partition, if not and anything was output, emit empty
5865 * final.c (final_scan_insn): Compute cold_function_name
5866 before calling switch_text_section debug hook. Call
5867 that hook even if dwarf2out_do_frame if not emitting
5871 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5872 TREE_READONLY decls.
5874 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5876 PR rtl-optimization/82488
5877 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5879 2017-11-26 Julia Koval <julia.koval@intel.com>
5881 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5883 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5886 2017-11-26 Julia Koval <julia.koval@intel.com>
5888 * config/i386/driver-i386.c (host_detect_local_cpu):
5889 Detect skylake-avx512.
5891 2017-11-26 Julia Koval <julia.koval@intel.com>
5893 * config.gcc: Add -march=cannonlake.
5894 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5895 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5896 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5897 (PTA_CANNONLAKE): New.
5898 (processor_target_table): Add cannonlake.
5899 (ix86_option_override_internal): Ditto.
5900 (fold_builtin_cpu): Ditto.
5901 (get_builtin_code_for_version): Handle cannonlake.
5902 (M_INTEL_COREI7_CANNONLAKE): New.
5903 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5904 * doc/invoke.texi: Add -march=cannonlake.
5906 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5908 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5909 (try_init_one_plugin): Alternative implementation for MinGW.
5910 * Makefile.in (plugin_implib): New.
5911 (gengtype-lex.c): Fix broken AIX workaround.
5912 * configure: Regenerate.
5913 * doc/plugins.texi: Document support for MinGW.
5915 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5917 PR rtl-optimization/81553
5918 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5919 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5920 is a shift where C1 has different mode than the whole shift, use C1's
5921 mode for MULT rather than the shift's mode.
5924 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5925 builtins not enabled in the currently selected ISA.
5927 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5929 PR tree-optimization/71026
5930 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5931 (insert_reciprocals): Change to insert reciprocals before a division
5932 by a square and to insert the square of a reciprocal.
5933 (execute_cse_reciprocals_1): Change to consider division by a square.
5934 (register_division_in): Add importance parameter.
5936 2017-11-24 Richard Biener <rguenther@suse.de>
5938 PR tree-optimization/82402
5939 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5940 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5942 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5944 * match.pd (0-ptr): New transformation.
5946 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5949 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5950 when skipping checking.
5952 2017-11-24 Richard Biener <rguenther@suse.de>
5954 PR tree-optimization/83128
5955 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5956 (vn_reference_lookup_3): Likewise.
5958 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5961 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5962 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5965 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5966 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5967 check for the latter with gimple_call_builtin_p. Do not handle
5968 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5970 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5972 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5975 2017-11-23 Julia Koval <julia.koval@intel.com>
5977 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5978 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5979 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5980 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5981 _mm512_maskz_expandloadu_epi16): New intrinsics.
5982 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5983 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5984 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5985 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5986 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5987 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5988 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5989 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5990 _mm256_maskz_expandloadu_epi8): New intrinsics.
5991 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5992 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5993 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5994 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5995 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5996 * config/i386/sse.md (VI248_VLBW): New iterator.
5997 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5999 2017-11-23 Julia Koval <julia.koval@intel.com>
6001 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
6002 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
6003 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
6004 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
6005 _mm512_mask_compressstoreu_epi16): New.
6006 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
6007 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
6008 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
6009 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
6010 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
6011 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
6012 _mm256_mask_compressstoreu_epi8): New.
6013 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
6014 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
6015 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
6016 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
6017 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
6018 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
6019 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
6020 __builtin_ia32_compresshi128_mask,
6021 __builtin_ia32_compressstoreuqi512_mask,
6022 __builtin_ia32_compressstoreuhi512_mask,
6023 __builtin_ia32_compressstoreuqi256_mask,
6024 __builtin_ia32_compressstoreuqi128_mask,
6025 __builtin_ia32_compressstoreuhi256_mask,
6026 __builtin_ia32_compressstoreuhi128_mask): New builtins.
6027 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
6029 (ix86_expand_special_args_builtin): Handle new types.
6030 (s4fma_expand): Handle new builtin array.
6031 * config/i386/immintrin.h: Include new headers.
6032 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
6033 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
6035 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6037 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
6038 fixes. Declare temp and g variables at the top in order to avoid
6039 {} in most of the cases.
6041 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
6043 * match.pd (ptr-0): New transformation.
6045 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
6047 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
6048 (struct addr_mode_cost_table): New.
6049 (struct tune_params): Add field addr_mode_costs.
6050 * config/arm/arm.c (generic_addr_mode_costs): New.
6051 (arm_slowmul_tune): Initialise addr_mode_costs field.
6052 (arm_fastmul_tune): Likewise.
6053 (arm_strongarm_tune): Likewise.
6054 (arm_xscale_tune): Likewise.
6055 (arm_9e_tune): Likewise.
6056 (arm_marvell_pj4_tune): Likewise.
6057 (arm_v6t2_tune): Likewise.
6058 (arm_cortex_tune): Likewise.
6059 (arm_cortex_a8_tune): Likewise.
6060 (arm_cortex_a7_tune): Likewise.
6061 (arm_cortex_a15_tune): Likewise.
6062 (arm_cortex_a35_tune): Likewise.
6063 (arm_cortex_a53_tune): Likewise.
6064 (arm_cortex_a57_tune): Likewise.
6065 (arm_exynosm1_tune): Likewise.
6066 (arm_xgene1_tune): Likewise.
6067 (arm_cortex_a5_tune): Likewise.
6068 (arm_cortex_a9_tune): Likewise.
6069 (arm_cortex_a12_tune): Likewise.
6070 (arm_cortex_a73_tune): Likewise.
6071 (arm_v7m_tune): Likewise.
6072 (arm_cortex_m7_tune): Likewise.
6073 (arm_v6m_tune): Likewise.
6074 (arm_fa726te_tune): Likewise.
6075 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
6077 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
6079 * config/arm/arm.c (arm_mem_costs): New function.
6080 (arm_rtx_costs_internal): Use arm_mem_costs.
6082 2017-11-23 Mark Wielaard <mark@klomp.org>
6084 * dwarf2out.c (init_sections_and_labels): Use generation to create
6085 unique ranges_section_label and ranges_base_label. Return generation.
6086 (output_rnglists): Add generation argument. Use generation to create
6087 unique ranges labels.
6088 (dwarf2out_finish): Get generation from init_sections_and_labels
6089 and pass generation to output_rnglists.
6091 2017-11-23 Mike Stump <mikestump@comcast.net>
6092 Eric Botcazou <ebotcazou@adacore.com>
6094 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
6095 * cfgloop.h (struct loop): Add unroll field.
6096 * function.h (struct function): Add has_unroll bitfield.
6097 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
6098 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
6099 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
6100 (pass_rtl_unroll_loops::gate): Likewise.
6101 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
6102 for which loop->unroll==1.
6103 (decide_unroll_constant_iterations): Use note for consistency and
6104 take loop->unroll into account. Return early if loop->unroll is set.
6105 Fix thinko in existing test.
6106 (decide_unroll_runtime_iterations): Use note for consistency and
6107 take loop->unroll into account.
6108 (decide_unroll_stupid): Likewise.
6109 * lto-streamer-in.c (input_cfg): Read loop->unroll.
6110 * lto-streamer-out.c (output_cfg): Write loop->unroll.
6111 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
6113 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
6114 (print_loop): Print loop->unroll if set.
6115 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
6116 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
6117 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
6119 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
6120 loop->unroll is set and smaller than the trip count. Otherwise bypass
6121 entirely the heuristics if loop->unroll is set. Remove dead note.
6122 Fix off-by-one bug in other note.
6123 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
6124 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
6126 (tree_unroll_loops_completely): Make static.
6127 (pass_complete_unroll::execute): Use correct type for variable.
6128 (pass_complete_unrolli::execute): Fix formatting.
6129 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
6131 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
6133 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
6134 enable when TARGET_PREFER_AVX128 is set.
6136 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6138 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
6140 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
6141 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
6143 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6145 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
6146 frequencies. Fix estimation of aggregate parameters.
6148 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6150 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
6153 2017-11-23 Tom de Vries <tom@codesourcery.com>
6155 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
6157 2017-11-23 Tom de Vries <tom@codesourcery.com>
6159 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
6162 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
6165 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
6166 sibcall_value_pcrel_fdpic): Use local variable instead of
6168 (calli_tbr_rel): Add missing operand 2.
6169 (call_valuei_tbr_rel): Add missing operand 3.
6171 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6174 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
6175 bitpos/bitsize covering the whole destination, use store_expr only if
6176 the complex mode is the same. Otherwise, use expand_normal and if
6177 it returns CONCAT, subreg each part separately instead of trying to
6178 subreg the whole result.
6180 2017-11-23 Richard Biener <rguenther@suse.de>
6182 PR tree-optimization/23094
6183 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
6184 come here from walking over backedges in the first iteration.
6185 (vn_reference_lookup_3): Skip clobbers that store the same value.
6187 2017-11-23 Richard Biener <rguenther@suse.de>
6189 PR tree-optimization/81403
6190 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
6191 a block we need a leader relative to.
6192 (phi_translate_1): For nary processing require a leader from
6193 get_representative_for given we run expression simplification
6194 using match-and-simplify. Remove previous fix.
6196 2017-11-22 Jeff Law <law@redhat.com>
6198 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
6199 Use new method allocate_value_range rather than accessing the
6200 vrp_value_range_pool data member directly.
6201 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
6202 to use extract_range_from_stmt method to avoid need for
6203 extract_range_from_assignment method.
6204 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
6205 method rather than setting values_propgated data member directly.
6206 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
6207 and values propagated data members and extract_range_from_assignment
6208 method. Reorder private data members to conform to standards.
6209 Add new methods set_lattice_propagation_complete and
6210 allocate_value_range.
6212 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
6214 PR rtl-optimization/83030
6215 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
6216 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
6217 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
6218 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
6219 CROSSING_JUMP_P flag is already set before setting it.
6220 * cfgrtl.c (fixup_partition_crossing): Likewise.
6221 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6224 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6226 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6227 Handle the case where both arguments are using gen_const_vec_series.
6229 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6232 * pretty-print.c (pp_format): Move quoting implementation to
6233 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6234 to pass address of "quote" local.
6235 (pp_begin_quote): New function.
6236 (pp_end_quote): New function.
6237 * pretty-print.h (printer_fn): Convert penultimate param from bool
6239 (pp_begin_quote): New decl.
6240 (pp_end_quote): New decl.
6241 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6242 param from bool to bool *.
6243 * tree-diagnostic.h (default_tree_printer): Likewise.
6245 2017-11-22 Jeff Law <law@redhat.com>
6247 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6248 Initialize vr_values.
6249 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6250 vr_values via vr_values class instance rather than delegators.
6251 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6252 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6253 (evrp_range_analyzer::push_value_range): Likewise.
6254 (evrp_range_analyzer::pop_value_range): Likewise.
6255 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6256 most delegators. Those remaining are exposed as public interfaces.
6257 Make vr_values a pointer and private.
6258 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6260 (evrp_range_analyzer::get_vr_value): New method.
6261 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6262 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6263 (class evrp_dom_walker): Attach evrp_folder class, initialize
6264 it in the ctor. Remove temporary delegators.
6265 (evrp_dom_walker::before_dom_children): Call methods in attached
6266 evrp_range_analyzer class via class instance pointer. Use
6267 free value_range_constant_singleton to remove need for
6268 op_with_constant_singleton_value delegator method. Do not
6269 create a vrp_prop class instance for every call! Narrow
6270 scope of a couple variables.
6271 (evrp_dom_walker::cleanup): Call methods in attached
6272 evrp_range_analyzer class via class instance pointer.
6273 * vr-values.h (class vr_values): Privatize many methods and
6276 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6278 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6279 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6280 * expr.c (expand_expr_real_2): Likewise.
6281 * fold-const.c (const_unop): Likewise.
6282 * optabs-tree.c (optab_for_tree_code): Likewise.
6283 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6284 * tree-inline.c (estimate_operator_cost): Likewise.
6285 * tree-pretty-print.c (dump_generic_node): Likewise.
6286 (op_code_prio): Likewise.
6287 (op_symbol_code): Likewise.
6288 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6289 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6290 * internal-fn.c (direct_internal_fn_optab): New function.
6291 (direct_internal_fn_array, direct_internal_fn_supported_p
6292 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6293 * fold-const-call.c (fold_const_reduction): New function.
6294 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6296 * tree-vect-loop.c: Include internal-fn.h.
6297 (reduction_code_for_scalar_code): Rename to...
6298 (reduction_fn_for_scalar_code): ...this and return an internal
6300 (vect_model_reduction_cost): Take an internal_fn rather than
6302 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6304 (vectorizable_reduction): Use internal functions rather than tree
6305 codes for the reduction operation. Update calls to the functions
6307 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6308 Use calls to internal functions rather than REDUC tree codes.
6309 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6311 2017-11-22 Olivier Hainque <hainque@adacore.com>
6313 * config/vxworks.c (vxworks_override_options): Pick default
6314 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6315 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6316 DWARF_GNAT_ENCODINGS_DEFAULT.
6317 * config/vxworksae.h: Likewise.
6319 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6321 PR tree-optimization/83104
6322 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6325 2017-11-22 Marek Polacek <polacek@redhat.com>
6326 H.J. Lu <hongjiu.lu@intel.com>
6327 Jason Merrill <jason@redhat.com>
6332 * calls.c (initialize_argument_information): Call
6333 warn_parameter_passing_abi target hook.
6334 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6335 argument onto stack.
6336 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6337 * common.opt: Update -fabi-version description.
6338 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6339 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6341 (ix86_is_empty_record): New function.
6342 (ix86_warn_parameter_passing_abi): New function.
6343 (TARGET_EMPTY_RECORD_P): Redefine.
6344 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6345 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6346 * doc/tm.texi: Regenerated.
6347 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6348 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6349 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6350 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6351 instead of int_size_in_bytes.
6352 * expr.c (copy_blkmode_to_reg): Likewise.
6353 * function.c (aggregate_value_p): Return 0 for empty types.
6354 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6356 (locate_and_pad_parm): Call arg size_in_bytes instead
6358 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6359 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6360 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6362 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6363 (std_gimplify_va_arg_expr): Skip empty records. Call
6364 arg_size_in_bytes instead size_in_bytes.
6365 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6366 * tree-core.h (tree_type_common): Add empty_flag.
6367 (tree_decl_common): Update comments.
6368 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6370 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6371 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6373 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6374 * tree.c (default_is_empty_type): New function.
6375 (default_is_empty_record): New function.
6376 (arg_int_size_in_bytes): New function.
6377 (arg_size_in_bytes): New function.
6378 (get_ultimate_context): New function.
6379 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6380 TRANSLATION_UNIT_WARN_EMPTY_P.
6381 (default_is_empty_record, arg_int_size_in_bytes,
6382 arg_size_in_bytes, get_ultimate_context): Declare.
6384 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6386 * config/arm/arm.c (cmse_clear_registers): New function.
6387 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6388 code by call to cmse_clear_registers.
6389 (cmse_nonsecure_entry_clear_before_return): Likewise.
6391 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6393 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6394 (vdot_s32, vdotq_s32): New.
6395 (vdot_lane_u32, vdotq_lane_u32): New.
6396 (vdot_lane_s32, vdotq_lane_s32): New.
6399 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6402 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6403 for unsigned values with fewer HWIs than the precision.
6404 (test_overflow): New function.
6405 (wide_int_cc_tests): Call it.
6407 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6409 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6410 has the same mode class as Pmode.
6412 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6414 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6415 padding_bits_to_clear_ptr.
6416 (cmse_nonsecure_entry_clear_before_return): Likewise.
6418 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6420 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6421 auto_sbitap instead of integer bitfield to control register needing
6424 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6426 PR tree-optimization/83044
6427 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6428 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6429 that need the upper bound. Subtract offset from
6430 get_addr_base_and_unit_offset only if positive and subtract it
6431 before division by eltsize rather than after it.
6434 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6435 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6436 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6440 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6441 expand_binop with *mul_widen_optab, make sure at least one of the
6442 operands doesn't have VOIDmode.
6445 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6447 PR rtl-optimization/82044
6448 PR tree-optimization/82042
6449 * dse.c (record_store): Check for overflow.
6450 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6451 add_wild_read instead of clear_rhs_from_active_local_stores on
6452 overflow and log it into dump_file.
6454 2017-11-22 Richard Biener <rguenther@suse.de>
6456 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6457 fake edges to exit when looking for a place to insert.
6458 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6460 (insert_into_preds_of_block): Commit edge insertion immediately,
6461 assert that doesn't require new BBs.
6462 (fini_pre): Release expressions.
6463 (pass_pre::execute): Shuffle things around a bit, if the fn
6464 is too large do not compute AVAIL either as this is really the
6467 2017-11-22 Richard Biener <rguenther@suse.de>
6469 PR tree-optimization/83089
6470 * tree-if-conv.c (pass_if_conversion::execute): If anything
6471 changed reset SCEV and free the number of iteration estimates.
6473 2017-11-21 Martin Sebor <msebor@redhat.com>
6475 PR tree-optimization/82945
6476 * calls.h (warn_nonstring_bound): Remove unused function.
6478 2017-11-21 Martin Sebor <msebor@redhat.com>
6480 PR tree-optimization/82945
6481 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6482 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6483 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6485 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6486 * calls.h (get_attr_nonstring_decl): Declare new function.
6487 * doc/extend.texi (attribute nonstring): Update.
6488 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6489 get_attr_nonstring_decl and handle it.
6490 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6491 detection of nul-termination.
6492 (strlen_to_stridx): Change to a pointer.
6493 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6494 (pass_strlen::execute): Same.
6496 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6498 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6499 for the new option -mprefer-vector-width=[none|128|256|512].
6500 * config/i386/i386.c (ix86_target_string): remove old style options
6501 -mprefer-avx256 and make -mprefer-avx128 as alias.
6502 (ix86_option_override_internal): Apply defaults for the
6503 -mprefer-vector-width=[128|256] option.
6504 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6505 Implement macros to work with -mprefer-vector-width=.
6506 * config/i386/i386.opt: Implemented option
6507 -mprefer-vector-width=[none|128|256|512].
6508 * doc/invoke.texi: Documentation for
6509 -mprefer-vector-width=[none|128|256|512].
6511 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6513 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6515 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6517 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6519 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6521 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6523 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6524 * expr.c (expand_expr_real_2): Likewise.
6525 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6526 fold_binary_loc): Likewise.
6527 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6528 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6529 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6530 MINUS_EXPR transformations.
6531 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6532 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6533 * tree-inline.c (estimate_operator_cost): Likewise.
6534 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6535 op_symbol_code): Likewise.
6536 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6537 * vr-values.c (extract_range_from_binary_expr): Likewise.
6538 * varasm.c (initializer_constant_valid_p_1): Likewise.
6539 * tree.def: New tree code POINTER_DIFF_EXPR.
6541 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6543 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6544 integer suffix to movbe mnemonic.
6545 (*bswaphi2_movbe): Ditto.
6546 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6548 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6551 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6552 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6553 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6555 2017-11-21 Martin Liska <mliska@suse.cz>
6557 * tree-inline.c (expand_call_inline): Remove not needed
6560 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6561 Jakub Jelinek <jakub@redhat.com>
6564 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6565 Remove static keyword from f variable.
6567 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6569 PR tree-optimization/83086
6570 * gimple-ssa-store-merging.c
6571 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6572 rather than n.base_addr.
6574 2017-11-21 Martin Liska <mliska@suse.cz>
6576 PR rtl-optimization/82044
6577 PR tree-optimization/82042
6578 * dse.c (check_mem_read_rtx): Check for overflow.
6580 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6582 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6585 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6587 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6588 and its corresponding call arg location note.
6590 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6591 Andrew Burgess <andrew.burgess@embecosm.com>
6593 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6595 (arc_return_slot_offset): Likewise.
6596 (arc_eh_return_address_location): New declaration.
6597 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6598 (MUST_SAVE_REGISTER): Add exception handler case.
6599 (MUST_SAVE_RETURN_ADDR): Likewise.
6600 (arc_frame_pointer_required): Likewise.
6601 (arc_frame_pointer_needed): New function.
6602 (arc_compute_frame_size): Changed.
6603 (arc_expand_prologue): Likewise.
6604 (arc_expand_epilogue): Likewise.
6605 (arc_initial_elimination_offset): Likewise.
6606 (arc_return_slot_offset): Delete.
6607 (arc_eh_return_address_location): New function.
6608 (arc_builtin_setjmp_frame_value): Likewise.
6609 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6610 (EH_RETURN_STACKADJ_RTX): Define.
6611 (EH_RETURN_HANDLER_RTX): Likewise.
6612 * config/arc/arc.md (eh_return): Delete.
6614 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6616 * print-tree.h (debug_vec_tree): Remove prototype.
6617 * gdbinit.in (pvt): Remove macro.
6619 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6621 PR tree-optimization/83047
6622 * gimple-ssa-store-merging.c
6623 (imm_store_chain_info::output_merged_store): If the loads with the
6624 same vuse are in different basic blocks, for load_gsi pick a load
6625 location that is dominated by the other loads.
6628 * config/i386/i386.c (ix86_memmodel_check): Start
6629 -Winvalid-memory-model diagnostics with lowercase letter.
6632 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6633 set in_cold_section_p to the partition containing loc_list->first.
6634 When seeing loc_list->last_before_switch node, update secname and
6635 perform range_across_switch second partition handling only after that.
6638 * run-rtl-passes.c: Include debug.h.
6639 (run_rtl_passes): Call debug_hooks->assembly_start.
6640 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6644 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6645 OPTAB_DIRECT in calls to expand_simple_binop.
6647 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6650 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6652 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6655 * spellcheck.h (best_match::blithely_get_best_candidate): New
6658 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6660 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6661 with lower case letter.
6663 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6665 * config/i386/i386.md (bswaphi2): New expander.
6666 (*bswaphi2_movbe): New insn pattern.
6667 (bswaphi -> rorhi pepehole2): New peephole pattern.
6669 2017-11-20 Jeff Law <law@redhat.com>
6671 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6672 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6673 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6674 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6675 gimple-ssa-evrp-analyze.h.
6677 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6678 set BB_VISITED here.
6679 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6681 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6683 * tree-predcom.c: Add general comment on Store-Store chains.
6684 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6686 (get_chain_last_ref_at): Rename into...
6687 (get_chain_last_write_at): ...this.
6688 (get_chain_last_write_before_load): New function.
6689 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6690 CT_STORE_STORE when write reference is added.
6691 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6692 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6693 (initialize_root_vars_store_elim_1): Ditto.
6694 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6695 definition is created.
6696 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6697 chain by replacing it with dominant stored value.
6699 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6701 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6703 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6705 * vr-values.c (extract_range_from_binary_expr): Use a full range
6708 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6710 * config/arm/arm.md (R4_REGNUM): Define constant.
6711 (nonsecure_call_internal): Remove r4 clobber.
6712 (nonsecure_call_value_internal): Likewise.
6713 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6714 clobber and resequence match_operands.
6715 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6716 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6717 (nonsecure_call_value_reg_thumb2): Likewise.
6719 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6721 PR tree-optimization/78821
6722 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6723 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6724 to the address of the base rather than the base itself.
6725 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6726 (find_bswap_or_nop_finalize): New function.
6727 (find_bswap_or_nop): Use it.
6728 (bswap_replace): Return a tree rather than bool, change first
6729 argument from gimple * to gimple_stmt_iterator, allow inserting
6730 into an empty sequence, allow ins_stmt to be NULL - then emit
6731 all stmts into gsi. Fix up MEM_REF address gimplification.
6732 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6733 (struct store_immediate_info): Add N and INS_STMT non-static
6735 (store_immediate_info::store_immediate_info): Initialize them
6736 from newly added ctor args.
6737 (merged_store_group::apply_stores): Formatting fixes. Sort by
6739 (stmts_may_clobber_ref_p): For stores call also
6740 refs_anti_dependent_p.
6741 (gather_bswap_load_refs): New function.
6742 (imm_store_chain_info::try_coalesce_bswap): New method.
6743 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6744 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6745 (imm_store_chain_info::output_merged_store): Fail if number of
6746 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6747 and NOP_EXPR rhs_code.
6748 (pass_store_merging::process_store): Compute n and ins_stmt, if
6749 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6750 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6752 (pass_store_merging::execute): Calculate dominators.
6754 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6755 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6756 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6757 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6758 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6759 class pass_optimize_bswap, bswap_replace,
6760 pass_optimize_bswap::execute): Moved to ...
6761 * gimple-ssa-store-merging.c: ... this file.
6762 Include optabs-tree.h.
6763 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6764 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6765 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6766 anonymous namespace, remove static keywords.
6767 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6768 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6770 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6773 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6775 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6777 * vec.h (debug_helper): New function.
6778 (DEFINE_DEBUG_VEC): New macro.
6779 * hash-set.h (debug_helper): New function.
6780 (DEFINE_DEBUG_HASH_SET): New macro.
6781 * cfg.c (debug_slim (edge)): New function.
6782 Call DEFINE_DEBUG_VEC for edges.
6783 Call DEFINE_DEBUG_HASH_SET for edges.
6784 * cfghooks.c (debug_slim (basic_block)): New function.
6785 Call DEFINE_DEBUG_VEC for basic blocks.
6786 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6787 * print-tree.c (debug_slim): New function to handle trees.
6788 Call DEFINE_DEBUG_VEC for trees.
6789 Call DEFINE_DEBUG_HASH_SET for trees.
6790 (debug (vec<tree, va_gc>) &): Remove.
6791 (debug (<vec<tree, va_gc>) *): Remove.
6792 * print-rtl.c (debug_slim): New function to handle const_rtx.
6793 Call DEFINE_DEBUG_VEC for rtx_def.
6794 Call DEFINE_DEBUG_VEC for rtx_insn.
6795 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6796 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6797 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6798 (debug (vec<rtx_insn *> *ptr): Remove.
6799 (debug_insn_vector): Remove.
6800 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6802 2017-11-20 Tom de Vries <tom@codesourcery.com>
6804 PR rtl-optimization/82020
6805 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6806 IF_THEN_ELSE condition.
6808 2017-11-19 Jeff Law <law@redhat.com>
6810 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6811 of degenerates resulting from ignoring an edge.
6813 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6816 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6818 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6821 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6822 for uninitialied counts.
6824 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6827 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6829 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6832 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6834 2017-11-19 Tom de Vries <tom@codesourcery.com>
6836 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6839 2017-11-19 Tom de Vries <tom@codesourcery.com>
6841 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6844 2017-11-19 Tom de Vries <tom@codesourcery.com>
6846 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6849 2017-11-19 Tom de Vries <tom@codesourcery.com>
6851 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6853 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6855 2017-11-19 Tom de Vries <tom@codesourcery.com>
6857 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6859 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6860 semicolon after MCORE_EXPORT_NAME call.
6862 2017-11-19 Tom de Vries <tom@codesourcery.com>
6865 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6866 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6868 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6870 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6871 (cgraph_edge::clone): Cleanup updating of profile.
6872 * ipa-cp.c (update_profiling_info): Likewise.
6873 * ipa-inline-transform.c (inline_transform): Likewise.
6874 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6875 * ipa-split.c (execute_split_functions): Do not split when function
6877 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6878 * profile-count.c (profile_count::dump): Add global0.
6879 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6881 (profile_count::to_sreal_scale): Likewise.
6882 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6883 (profile_count::combine_with_ipa_count): New function.
6884 * profile-count.h (profile_guessed_global0adjusted): New.
6885 (profile_count::adjusted_zero): New.
6886 (profile_count::global0adjusted): New.
6887 (profile_count::combine_with_ipa_count): New.
6888 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6889 correct profile of return block of split functions.
6890 (copy_cfg_body): Remove unused profile_count.
6891 (copy_body): Likewise.
6892 (expand_call_inline): Update.
6893 (tree_function_versioning): Update.
6895 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6897 * hash-set.h (hash_set::empty): New.
6898 * tree-ssa-threadbackward.h: Delete.
6899 * tree-ssa-threadbackward.c (class thread_jumps): New.
6900 Move max_threaded_paths into class.
6901 (fsm_find_thread_path): Remove arguments that are now in class.
6902 (profitable_jump_thread_path): Rename to...
6903 (thread_jumps::profitable_jump_thread_path): ...this.
6904 (convert_and_register_jump_thread_path): Rename to...
6905 (thread_jumps::convert_and_register_current_path): ...this.
6906 (check_subpath_and_update_thread_path): Rename to...
6907 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6908 (register_jump_thread_path_if_profitable): Rename to...
6909 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6910 (handle_phi): Rename to...
6911 (thread_jumps::handle_phi): ...this.
6912 (handle_assignment): Rename to...
6913 (thread_jumps::handle_assignment): ...this.
6914 (fsm_find_control_statement_thread_paths): Rename to...
6915 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6916 (find_jump_threads_backwards): Rename to...
6917 (thread_jumps::find_jump_threads_backwards): ...this.
6918 Initialize path local data.
6919 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6920 from within thread_jumps class.
6921 (pass_early_thread_jumps::execute): Same.
6923 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6925 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6927 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6929 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6930 to canonical location.
6932 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6935 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6937 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6939 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6941 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6942 rcrt1.o%s/grcrt1.o%s for -static-pie.
6944 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6946 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6947 ix86_shift_rotate_cost): Break out from ...
6948 (ix86_rtx_costs): ... here.
6949 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6952 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6954 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6955 when entry block was promoted unlikely.
6956 (estimate_bb_frequencies): Increase frequency scale.
6957 * profile-count.h (profile_count): Export precision info.
6959 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6961 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6962 disturbin profile of entry block.
6964 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6966 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6969 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6971 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6972 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6974 2017-11-17 Jeff Law <law@redhat.com>
6976 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6977 from evrp_dom_walker class. Various methods moved into new class.
6978 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6979 (evrp_range_analyzer::enter): New method.
6980 (evrp_range_analyzer::leave): New method.
6981 (evrp_dom_walker): Remove delegators no longer needed by this class.
6982 Replace vr_values data member with evrp_range_analyzer
6984 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6985 method extracted from evrp_dom_walker::before_dom_children.
6986 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6987 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6989 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6990 Add private copy constructor and move assignment operators.
6991 Privatize methods and class data where trivially possible.
6992 (evrp_dom_walker::cleanup): New function, extracted from
6993 execute_early_vrp. Simplify access to class data.
6995 * vr-values.h (get_output_for_vrp): Prototype.
6996 * vr-values.c (get_output_for_vrp): New function extracted from
6997 vrp_visit_assignment_or_call and extract_range_from_stmt.
6998 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
7000 2017-11-17 Luis Machado <luis.machado@linaro.org>
7002 * config/aarch64/aarch64.c
7003 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
7004 (qdf24xx_tunings) <autoprefetcher_model>: Set to
7005 tune_params::AUTOPREFETCHER_WEAK.
7007 2017-11-17 Tamar Christina <tamar.christina@arm.com>
7010 * config/arm/arm.c (arm_valid_target_attribute_rec):
7011 Parse "arch=" and "+<ext>".
7012 (arm_valid_target_attribute_tree): Re-init global options.
7013 (arm_option_override): Make non-static.
7014 (arm_options_perform_arch_sanity_checks): Make errors fatal.
7015 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
7016 (__ARM_FEATURE_CRC32): Support undef.
7017 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
7018 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
7020 2017-11-17 David Malcolm <dmalcolm@redhat.com>
7022 * gdbinit.in (break-on-diagnostic): New command.
7024 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7026 * config/i386/i386.c (ix86_expand_epilogue): Change simple
7027 return to indirect jump for EH return if control-flow
7028 protection is enabled. Change explicit 'false' argument in
7029 pro_epilogue_adjust_stack with a value of
7031 * config/i386/i386.md (simple_return_indirect_internal):
7032 Remove SImode restriction to support 64-bit.
7034 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
7036 * combine.c (added_notes_insn): New.
7037 (try_combine): Handle added_notes_insn like added_links_insn.
7038 Rewrite return value code.
7039 (distribute_notes): Set added_notes_insn to the earliest insn we added
7042 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
7044 PR rtl-optimization/82621
7045 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
7046 dest of one of those SETs is unused.
7048 2017-11-17 Richard Biener <rguenther@suse.de>
7051 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
7052 * tree-pretty-print.c (dump_generic_node): Handle
7053 annot_expr_parallel_kind.
7054 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
7055 * gimplify.c (gimple_boolify): Likewise.
7057 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7059 * config.gcc (extra_headers): Add cet.h for x86 targets.
7060 * config/i386/cet.h: New file.
7061 * doc/install.texi: Add --enable-cet/--disable-cet.
7063 2017-11-17 Richard Biener <rguenther@suse.de>
7065 PR tree-optimization/83017
7066 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
7067 (gen_parallel_loop): Properly count iterations.
7068 (parallelize_loops): Handle loop->can_be_parallel independent
7069 of flag_loop_parallelize_all. Make static profitability test match
7071 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
7072 * invoke.texi (parloops-min-per-thread): Document.
7074 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
7076 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
7077 upstreaming review comments.
7079 2017-11-17 Tamar Christina <tamar.christina@arm.com>
7081 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
7082 with fast unaligned access.
7083 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
7085 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7087 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
7088 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
7090 2017-11-17 Richard Biener <rguenther@suse.de>
7092 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
7093 folding of references.
7095 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
7098 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
7099 of replacing call to strncmp with corresponding call to strcmp when
7102 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
7104 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
7105 option prefer-avx256 for skylake-avx512 configuration.
7106 * config/i386/i386.c (ix86_option_override_internal): Ditto.
7107 (get_builtin_code_for_version): Ditto.
7109 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7110 Monk Chiang <sh.chiang04@gmail.com>
7112 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
7113 (FIXED_REGISTERS): Reserve more register numbers.
7114 (CALL_USED_REGISTERS): Likewise.
7115 (REG_ALLOC_ORDER): Likewise.
7116 (REG_CLASS_CONTENTS): Likewise.
7117 (REGISTER_NAMES): Likewise.
7119 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7120 Kito Cheng <kito.cheng@gmail.com>
7122 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
7124 * config/nds32/iterators.md: Add vector mode iterators and attributes.
7126 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
7128 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
7129 parameter list for vec_splats.
7131 2017-11-16 Joseph Myers <joseph@codesourcery.com>
7133 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
7135 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
7137 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
7140 * doc/install.texi: Document incompatibility of Perl >=5.6.26
7141 with the required version of automake 1.11.6.
7143 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
7145 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
7147 (DU_C2_3_power9): Correct reservation combinations.
7148 (FP_DIV_power9, VEC_DIV_power9): New.
7149 (power9-alu): Split out rotate/shift...
7150 (power9-rot): ...to here, correct dispatch resource.
7151 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
7153 (power9-fp): Correct latency.
7154 (power9-sdiv): Add div/sqrt resource.
7155 (power9-ddiv): Correct latency, add div/sqrt resource.
7156 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
7157 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
7159 (power9-qpdiv, power9-qpmul): Adjust resource usage.
7161 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
7163 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
7164 switch statement mapping KF built-ins to TF built-ins if we don't
7165 have the proper ISA 3.0 assembler support.
7167 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7169 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
7170 (lower_emutls_function_body): Do not compute it.
7172 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7174 * ipa-split.c (split_bb_info): Turn time to sreal.
7175 (split_point): Likewise.
7176 (dump_split_point): Likewise.
7177 (fine_split_points): Likewise.
7178 (execute_split_functions): Only zero split_bbs; turn time to sreals.
7180 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7182 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
7185 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7187 * predict.c (combine_predictions_for_bb): Preserve zero predicted
7189 (expensive_function_p): Remove useless assert.
7190 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
7192 2017-11-16 Martin Sebor <msebor@redhat.com>
7194 PR tree-optimization/82588
7195 PR tree-optimization/82583
7196 * tree-vrp.c (check_array_ref): Handle flexible array members,
7197 string literals, and inner indices.
7198 (search_for_addr_array): Add detail to diagnostics.
7200 2017-11-16 Nathan Sidwell <nathan@acm.org>
7204 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
7205 (SET_DECL_ASSEMBLER_NAME): Forward to
7206 overwrite_decl_assembler_name.
7207 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
7208 (overwrite_decl_assembler_name): Declare.
7209 * tree.c (overwrite_decl_assembler_name): New.
7210 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
7211 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
7212 (LANG_HOOKS_INITIALIZER): Add it.
7213 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
7214 * langhooks.c (lhd_set_decl_assembler_name): Use
7215 SET_DECL_ASSEMBLER_NAME.
7216 (lhd_overwrite_decl_assembler_name): Default implementation.
7218 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7219 Jackson Woodruff <jackson.woodruff@arm.com>
7221 PR tree-optimization/71026
7222 * match.pd: Canonicalize constant multiplies in division.
7224 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7226 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7227 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7228 Remove OPT_fomit_frame_pointer entry.
7229 * common/config/arc/arc-common.c: Likewise.
7230 * common/config/arm/arm-common.c: Likewise.
7231 * common/config/avr/avr-common.c: Likewise.
7232 * common/config/c6x/c6x-common.c: Likewise.
7233 * common/config/cr16/cr16-common.c: Likewise.
7234 * common/config/cris/cris-common.c: Likewise.
7235 * common/config/epiphany/epiphany-common.c: Likewise.
7236 * common/config/fr30/fr30-common.c: Likewise.
7237 * common/config/frv/frv-common.c: Likewise.
7238 * common/config/ia64/ia64-common.c: Likewise.
7239 * common/config/iq2000/iq2000-common.c: Likewise.
7240 * common/config/lm32/lm32-common.c: Likewise.
7241 * common/config/m32r/m32r-common.c: Likewise.
7242 * common/config/mcore/mcore-common.c: Likewise.
7243 * common/config/microblaze/microblaze-common.c: Likewise.
7244 * common/config/mips/mips-common.c: Likewise.
7245 * common/config/mmix/mmix-common.c: Likewise.
7246 * common/config/mn10300/mn10300-common.c: Likewise.
7247 * common/config/nios2/nios2-common.c: Likewise.
7248 * common/config/pa/pa-common.c: Likewise.
7249 * common/config/pdp11/pdp11-common.c: Likewise.
7250 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7251 * common/config/riscv/riscv-common.c: Likewise.
7252 * common/config/rs6000/rs6000-common.c: Likewise.
7253 * common/config/rx/rx-common.c: Likewise.
7254 * common/config/s390/s390-common.c: Likewise.
7255 * common/config/sh/sh-common.c: Likewise.
7256 * common/config/sparc/sparc-common.c: Likewise.
7257 * common/config/tilegx/tilegx-common.c: Likewise.
7258 * common/config/tilepro/tilepro-common.c: Likewise.
7259 * common/config/v850/v850-common.c: Likewise.
7260 * common/config/visium/visium-common.c: Likewise.
7261 * common/config/xstormy16/xstormy16-common.c: Likewise.
7262 * common/config/xtensa/xtensa-common.c: Likewise.
7263 * invoke.texi (-fomit-frame-pointer): Update documentation.
7265 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7267 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7269 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7271 * cfg.c (scale_bbs_frequencies_int,
7272 cale_bbs_frequencies_gcov_type): Remove.
7273 * cfg.h (scale_bbs_frequencies_int,
7274 cale_bbs_frequencies_gcov_type): Remove.
7276 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7278 * tree-ssa-loop-manip.c
7279 (scale_dominated_blocks_in_loop): Update to profile counts.
7280 (tree_transform_and_unroll_loop): Likewise.
7282 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7284 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7285 scale_bbs_frequencies_int.
7287 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7289 * final.c (compute_alignments): Use counts rather than frequencies.
7291 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7293 * cfgloopanal.c: Include sreal.h
7294 (average_num_loop_insns): Use counts and sreal for accounting.
7296 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7298 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7301 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7302 Kito Cheng <kito.cheng@gmail.com>
7304 * config/nds32/constraints.md: Provide more constraints.
7305 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7306 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7307 support constraints usage.
7309 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7311 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7313 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7314 Kito Cheng <kito.cheng@gmail.com>
7316 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7317 * config/nds32/nds32.opt: Refine the layout.
7318 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7319 TARGET_EXT_STRING): Support new options.
7320 * config/nds32/nds32.h: Likewise.
7321 * config/nds32/nds32.md: Likewise.
7322 * config/nds32/nds32-predicates.c: Likewise.
7323 * config/nds32/constraints.md: Likewise.
7324 * common/config/nds32/nds32-common.c: Likewise.
7326 2017-11-16 Julia Koval <julia.koval@intel.com>
7329 * config/i386/gfniintrin.h: Add sse check.
7330 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7332 2017-11-16 Julia Koval <julia.koval@intel.com>
7334 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7335 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7336 (ix86_handle_option): Handle -mavx512vbmi2.
7337 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7338 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7339 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7340 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7341 (ix86_valid_target_attribute_inner_p): Ditto.
7342 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7343 * config/i386/i386.opt (mavx512vbmi2): New option.
7344 * doc/invoke.texi: Add new option.
7346 2017-11-16 Julia Koval <julia.koval@intel.com>
7348 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7349 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7350 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7351 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7352 _mm512_gf2p8mul_epi8): New intrinsics.
7353 * config/i386/i386-builtin-types.def
7354 (V64QI_FTYPE_V64QI_V64QI): New type.
7355 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7356 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7357 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7358 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7359 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7360 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7362 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7364 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7365 explicitly as a stream of bytes.
7367 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7369 * config/rs6000/altivec.h (vec_xst_be): New #define.
7370 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7371 and externalize from *altivec_vperm_<mode>_internal.
7372 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7374 (XL_BE_V8HI): Likewise.
7375 (XL_BE_V4SI): Likewise.
7376 (XL_BE_V4SI): Likewise.
7377 (XL_BE_V2DI): Likewise.
7378 (XL_BE_V4SF): Likewise.
7379 (XL_BE_V2DF): Likewise.
7380 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7381 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7382 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7383 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7384 VSX_BUILTIN_VEC_XST_BE.
7385 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7386 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7388 (altivec_init_builtins): Replace conditional calls to def_builtin
7389 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7390 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7391 with unconditional calls. Remove calls to def_builtin for
7392 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7393 __builtin_vec_xst_be.
7394 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7395 to define_expand, and add alternate RTL generation for P8.
7396 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7397 vsx_ld_elemrev_v8hi.
7398 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7399 add alternate RTL generation for P8.
7400 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7401 vsx_ld_elemrev_v16qi.
7402 (vsx_st_elemrev_v8hi): Convert define_insn
7403 to define_expand, and add alternate RTL generation for P8.
7404 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7405 vsx_st_elemrev_v8hi.
7406 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7407 add alternate RTL generation for P8.
7408 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7409 vsx_st_elemrev_v16qi.
7411 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7414 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7415 TARGET_AVX512ER check.
7416 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7417 neither -mzeroupper nor -mno-zeroupper is used and
7418 TARGET_EMIT_VZEROUPPER is set.
7419 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7420 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7422 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7424 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7425 folding of vector compares.
7426 (fold_build_vec_cmp): New helper function.
7427 (fold_compare_helper): New helper function.
7428 (builtin_function_type): Add compare builtins to the list of functions
7429 having unsigned arguments. Cosmetic updates to comment indentation.
7430 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7431 the not+eq combination.
7433 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7435 PR tree-optimization/82726
7436 PR tree-optimization/70754
7437 * tree-predcom.c (order_drefs_by_pos): New function.
7438 (combine_chains): Move code setting has_max_use_after to...
7439 (try_combine_chains): ...here. New parameter. Sort combined chains
7440 according to position information.
7441 (tree_predictive_commoning_loop): Update call to above function.
7442 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7444 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7446 PR tree-optimization/82726
7448 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7450 PR tree-optimization/70754
7451 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7452 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7453 combined stmt before it if not NULL.
7454 (combine_chains): Process refs reversely and compute dominance point
7458 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7460 PR tree-optimization/79663
7461 * tree-predcom.c (combine_chains): Process refs in reverse order
7462 only for ZERO length chains, and add explaining comment.
7464 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7466 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7467 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7468 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7470 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7472 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7474 2017-11-15 Martin Liska <mliska@suse.cz>
7476 * tree-cfg.c (pass_warn_function_return::execute):
7477 Compare warn_return_type for greater than zero.
7479 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7483 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7484 to return true on Xeon and not on Xeon Phi.
7485 (ix86_check_avx256_register): Changed to ...
7486 (ix86_check_avx_upper_register): ... this. Add extra check for
7487 VALID_AVX512F_REG_OR_XI_MODE.
7488 (ix86_avx_u128_mode_needed): Changed
7489 ix86_check_avx256_register to ix86_check_avx_upper_register.
7490 (ix86_check_avx256_stores): Changed to ...
7491 (ix86_check_avx_upper_stores): ... this. Changed
7492 ix86_check_avx256_register to ix86_check_avx_upper_register.
7493 (ix86_avx_u128_mode_after): Changed
7494 avx_reg256_found to avx_upper_reg_found. Changed
7495 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7496 (ix86_avx_u128_mode_entry): Changed
7497 ix86_check_avx256_register to ix86_check_avx_upper_register.
7498 (ix86_avx_u128_mode_exit): Ditto.
7499 * config/i386/i386.h: (host_detect_local_cpu): New define.
7501 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7503 * config/arm/xgene1.md (xgene1): Split into automatons
7504 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7505 (xgene1_f_load): Adjust reservations and/or types.
7506 (xgene1_f_store): Likewise.
7507 (xgene1_load_pair): Likewise.
7508 (xgene1_store_pair): Likewise.
7509 (xgene1_fp_load1): Likewise.
7510 (xgene1_load1): Likewise.
7511 (xgene1_store1): Likewise.
7512 (xgene1_move): Likewise.
7513 (xgene1_alu): Likewise.
7514 (xgene1_simd): Likewise.
7515 (xgene1_bfm): Likewise.
7516 (xgene1_neon_load1): Likewise.
7517 (xgene1_neon_store1): Likewise.
7518 (xgene1_neon_logic): Likewise.
7519 (xgene1_neon_st1): Likewise.
7520 (xgene1_neon_ld1r): Likewise.
7521 (xgene1_alu_cond): Added.
7522 (xgene1_shift_reg): Likwise.
7523 (xgene1_bfx): Likewise.
7524 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7526 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7529 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7531 (can_widen_mult_without_libcall): New function.
7532 (expand_mul_overflow): If only checking unsigned mul overflow,
7533 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7534 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7535 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7536 (expand_DIVMOD): Formatting fix.
7537 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7538 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7539 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7541 PR tree-optimization/82977
7542 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7543 constructed temporary to strlen_to_stridx.put.
7545 2017-11-15 Martin Liska <mliska@suse.cz>
7547 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7548 * configure: Regenerate.
7550 2017-11-15 Martin Liska <mliska@suse.cz>
7553 * config/sh/sh-mem.cc: Use proper probability for
7556 2017-11-14 Jeff Law <law@redhat.com>
7558 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7559 the red zone for stack_clash_protection_final_dynamic_probe targets
7560 when the total dynamic stack size is zero bytes.
7562 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7563 blocks is post order.
7565 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7567 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7568 * final.c (rest_of_clean_state): Set it for the
7569 -fcompare-debug dump.
7570 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7571 class when TDF_COMPARE_DEBUG is set.
7573 * dwarf2out.c (gen_producer_string): Discard
7576 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7579 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7580 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7581 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7582 (__TGMATH_REAL_2_3): Remove macros.
7583 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7584 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7586 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7588 (remquo): Define using __TGMATH_REAL_3.
7590 2017-11-14 Jeff Law <law@redhat.com>
7592 * vr-values.c: New file with contents extracted from tree-vrp.c.
7593 * Makefile.in (OBJS): Add vr-values.o
7594 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7595 (set_value_range, set_and_canonicalize_value_range): Likewise.
7596 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7597 (value_range_constant_singleton, symbolic_range_p): Likewise.
7598 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7599 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7600 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7601 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7602 (find_case_label_range, find_case_label_index): Likewise.
7603 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7604 (range_int_cst_singleton_p, value_inside_range): Likewise.
7605 (get_single_symbol): Likewise.
7606 (switch_update): Move structure definition here.
7607 (to_remove_edges, to_update_switch_stmts): Provide externs.
7608 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7609 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7610 (vrp_val_is_min, set_value_range): Likewise.
7611 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7612 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7613 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7614 (range_is_nonnull, range_int_cst_p): Likewwise.
7615 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7616 (get_single_symbol, operand_less_p): Likewise
7617 (compare_values_warnv, compare_values): Likewise.
7618 (value_inside_range, value_range_constant_singleton): Likewise.
7619 (zero_nonzero_bitgs_from_vr): Likewise.
7620 (extract_range_from_binary_expr_1): Likewise.
7621 (overflow_comparison_p): Likewise.
7622 (to_remove_edges, to_update_switch_stmts): Likewise.
7623 (find_case_label-index, find_case_label_range): Likewise.
7624 (switch_update, set_value_range_to_nonnegative): Remove.
7625 (set_value_range_to_truthvalue): Likewise.
7626 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7627 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7628 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7629 (find_case_label_ranges, test_for_singularity): Likewise.
7630 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7631 (x_vr_values): Move to its remaining use site.
7633 2017-11-10 Jeff Law <law@redhat.com>
7635 * vr-values.h (VR_INITIALIZER): Move #define here.
7636 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7637 * Makefile.in (OBJS): Add tree-evrp.o
7638 * tree-vrp.h (assert_info): Move structure definition here.
7639 (set_value_range_to_varying): Prototype.
7640 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7641 (infer_value_range, register_edge_assert_for): Likewise.
7642 (stmt_interesting_for_vrp): Likewise.
7643 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7644 (set_value_range_to_varying): No longer static.
7645 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7646 (infer_value_range, register_edge_assert_for): Likewise.
7648 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7650 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7651 generate the XXBRD instruction.
7653 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7654 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7655 as being compatible if -mabi=ieeelongdouble.
7656 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7657 to setup float128 built-ins with hardware support.
7658 (BU_FLOAT128_HW_2): Likewise.
7659 (BU_FLOAT128_HW_3): Likewise.
7660 (BU_FLOAT128_HW_VSX_1): Likewise.
7661 (BU_FLOAT128_HW_VSX_2): Likewise.
7662 (scalar_extract_expq): Change float128 built-in functions to
7663 accommodate having both KFmode and TFmode functions. Use the
7664 KFmode variant as the default.
7665 (scalar_extract_sigq): Likewise.
7666 (scalar_test_neg_qp): Likewise.
7667 (scalar_insert_exp_q): Likewise.
7668 (scalar_insert_exp_qp): Likewise.
7669 (scalar_test_data_class_qp): Likewise.
7670 (sqrtf128_round_to_odd): Delete processing the round to odd
7671 built-in functions as special built-in functions, and define them
7672 as float128 built-ins. Use the KFmode variant as the default.
7673 (truncf128_round_to_odd): Likewise.
7674 (addf128_round_to_odd): Likewise.
7675 (subf128_round_to_odd): Likewise.
7676 (mulf128_round_to_odd): Likewise.
7677 (divf128_round_to_odd): Likewise.
7678 (fmaf128_round_to_odd): Likewise.
7679 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7680 support for KFmode and TFmode xststdcqp calls.
7681 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7682 point, switch the built-in handlers for the get/set float128
7683 exponent, get float128 mantissa, float128 test built-ins, and the
7684 float128 round to odd built-in functions. Eliminate creating the
7685 float128 round to odd built-in functions as special built-ins.
7686 (rs6000_init_builtins): Eliminate special creation of the float128
7687 round to odd built-in functions.
7688 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7689 function insns to support both TFmode and KFmode variants.
7690 (xsxsigqp_<mode>): Likewise.
7691 (xsiexpqpf_<mode>): Likewise.
7692 (xsiexpqp_<mode>): Likewise.
7693 (xststdcqp_<mode>): Likewise.
7694 (xststdcnegqp_<mode>): Likewise.
7695 (xststdcqp_<mode>): Likewise.
7697 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7699 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7700 unused path_in_freq_ptr parameter.
7701 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7703 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7705 * ipa-inline.c (edge_badness): Dump sreal frequency.
7706 (compute_inlined_call_time): Match natural implementaiton ...
7707 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7708 forgotten division by CGRAPH_FREQ_BASE.
7710 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7712 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7713 Solaris 11. Update comment.
7714 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7716 * config/sol2.h (STARTFILE_SPEC): Likewise.
7717 * configure: Regenerate.
7719 2017-11-14 Carl Love <cel@us.ibm.com>
7721 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7722 le_ and be_ prefixes to swap* variables. Remove
7723 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7726 2017-11-14 Jason Merrill <jason@redhat.com>
7728 Support GTY((cache)) on hash_map.
7729 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7730 (ggc_cache_remove): Override it instead of ggc_mx.
7731 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7732 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7733 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7734 (simple_cache_map_traits): Override maybe_mx.
7735 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7736 (hash_map): Friend gt_cleare_cache.
7737 (gt_cleare_cache): New.
7738 * tree.h (tree_cache_traits): New hash_map traits class.
7739 (tree_cache_map): New typedef.
7741 2017-11-14 Richard Biener <rguenther@suse.de>
7743 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7744 paramter and handling.
7745 (cleanup_control_flow_bb): Likewise.
7746 (cleanup_control_flow_pre): New helper performing a DFS walk
7747 to call cleanup_control_flow_bb in PRE order.
7748 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7749 via cleanup_control_flow_pre.
7751 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7753 * config/aarch64/aarch64-simd.md
7754 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7755 (*aarch64_simd_bsl<mode>_alt): Likewise.
7756 (aarch64_simd_bsldi_internal): New.
7757 (aarch64_simd_bsldi_alt): Likewise.
7759 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7761 * tracer.c (better_p): Do not compare frequencies.
7762 * reg-stack.c (better_edge): Likewise.
7763 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7766 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7768 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7769 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7770 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7771 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7772 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7773 * predict.c (maybe_hot_frequency_p): Inline to ...
7774 (maybe_hot_count_p): ... here; rewrite to counts.
7775 (counts_to_freqs): Rename to ...
7776 (update_max_bb_count): ... this one.
7777 (expensive_function_p): Use counts.
7778 (estimate_bb_frequencies): Update.
7779 (rebuild_frequencies): Update.
7780 * predict.h (counts_to_freqs): Rename to ...
7781 (update_max_bb_count): ... this one.
7782 * profile.c (compute_branch_probabilities): Add debug info
7783 * tree-inline.c (expand_call_inline): Update debug info.
7784 (optimize_inline_calls): Use update_max_bb_count..
7785 (tree_function_versioning): Use update_max_bb_count..
7786 * value-prof.c (gimple_value_profile_transformations):
7787 Do not use update_max_bb_count.
7789 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7791 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7792 always use frequencies.
7794 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7796 * bb-reorder.c: Remove frequencies from comments.
7797 (better_edge_p): Use profile counts.
7798 (find_traces): Dump profile counts.
7799 (rotate_loop): Use profile counts.
7800 (find_traces_1_round): Likewise.
7801 (connect_better_edge_p): Use counts instead of probabilities for
7803 (copy_bb_p): Drop early check for non-0 frequency.
7804 (sanitize_hot_paths): Update comments.
7806 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7808 * ipa-split.c (struct split_point): Add count.
7809 (consider_split): Do not compute incoming frequency; compute incoming
7810 count and store it to split_point.
7811 (split_function): Set count of the call to split part correctly.
7813 2017-11-13 Carl Love <cel@us.ibm.com>
7815 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7817 2017-11-13 Tom Tromey <tom@tromey.com>
7819 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7821 2017-11-13 Carl Love <cel@us.ibm.com>
7823 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7824 Add support for builtins:
7825 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7826 vector {un,}signed {char,int,short},
7827 vector {un,}signed {char,int,short}) arguments.
7828 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7829 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7830 Add BU_P9V_AV_2 expansions for the builtins.
7831 * config/rs6000/altivec.h (vec_first_match_index,
7832 vec_first_mismatch_index, vec_first_match_or_eos_index,
7833 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7834 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7835 new extern declaration.
7836 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7837 * config/rs6000/vsx.md (first_match_index_<mode>,
7838 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7839 first_mismatch_or_eos_index_<mode>): Add define expand.
7840 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7841 * doc/extend.texi: Update the built-in documenation file for the new
7844 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7846 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7847 into the min/max operations for _Float<N> and _Float<N>X types.
7849 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7852 * dwarf2out.c (do_eh_frame): New static variable.
7853 (dwarf2out_begin_prologue): Set it.
7854 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7856 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7858 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7860 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7862 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7865 * debug.h (dwarf2out_do_eh_frame): Declare.
7866 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7867 (dwarf2out_do_frame): Use it.
7868 (dwarf2out_do_cfi_asm): Likewise.
7869 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7870 (dwarf2out_assembly_start): Likewise.
7871 (dwarf2out_begin_prologue): Fix comment.
7872 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7873 if the target needs either debug or unwind DWARF2 info.
7874 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7875 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7876 -ftrapping-math, -ftrapv and -fwrapv.
7878 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7880 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7881 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7882 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7883 (estimate_edge_size_and_time): Likewise.
7884 (ipa_merge_fn_summary_after_inlining): Likewise.
7885 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7886 (compute_uninlined_call_time): Use sreal_frequency.
7887 (compute_inlined_call_time): Likewise.
7888 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7889 * profile-count.c: Include sreal.h.
7890 (profile_count::to_sreal_scale): New.
7891 * profile-count.h: Forward declare sreal.
7892 (profile_count::to_sreal_scale): Declare.
7894 2017-11-13 Nathan Sidwell <nathan@acm.org>
7896 * diagnostic.c (maybe_line_and_column): New.
7897 (diagnostic_get_location_text): Use it.
7898 (diagnostic_report_current_module): Likewise.
7899 (test_diagnostic_get_location_text): Add tests.
7901 2017-11-13 Luis Machado <luis.machado@linaro.org>
7903 * doc/md.texi (Specifying processor pipeline description): Fix
7904 incorrect latency for the div instruction example.
7906 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7908 PR tree-optimization/78821
7909 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7910 that bit_not_p is the same.
7911 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7912 (split_group): Count precisely bit_not_p bits in each statement.
7913 (invert_op): New function.
7914 (imm_store_chain_info::output_merged_store): Use invert_op to
7915 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7916 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7918 2017-11-13 Martin Liska <mliska@suse.cz>
7920 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7921 (struct source_info): Likewise.
7922 (add_branch_counts): Likewise.
7923 (add_line_counts): Likewise.
7924 (function_summary): Likewise.
7925 (output_intermediate_line): Likewise.
7926 (generate_results): Likewise.
7928 2017-11-13 Martin Liska <mliska@suse.cz>
7930 * gcov.c (struct block_info): Remove typedef for block_t.
7931 (struct line_info): Likewise.
7932 (line_info::has_block): Likewise.
7933 (EXIT_BLOCK): Likewise.
7934 (unblock): Likewise.
7935 (circuit): Likewise.
7936 (get_cycles_count): Likewise.
7937 (process_file): Likewise.
7938 (read_graph_file): Likewise.
7939 (solve_flow_graph): Likewise.
7940 (find_exception_blocks): Likewise.
7941 (add_line_counts): Likewise.
7942 (accumulate_line_info): Likewise.
7943 (output_line_details): Likewise.
7945 2017-11-13 Martin Liska <mliska@suse.cz>
7947 * gcov.c (struct arc_info): Remove typedef for arc_t.
7948 (struct line_info): Likewise.
7949 (add_branch_counts): Likewise.
7950 (output_branch_count): Likewise.
7951 (function_info::~function_info): Likewise.
7952 (circuit): Likewise.
7953 (output_intermediate_line): Likewise.
7954 (read_graph_file): Likewise.
7955 (solve_flow_graph): Likewise.
7956 (find_exception_blocks): Likewise.
7957 (add_line_counts): Likewise.
7958 (accumulate_line_info): Likewise.
7959 (output_line_details): Likewise.
7960 (output_function_details): Likewise.
7962 2017-11-13 Martin Liska <mliska@suse.cz>
7964 * gcov.c (struct function_info): Remove typedef for function_t.
7965 (struct source_info): Likewise.
7966 (source_info::get_functions_at_location): Likewise.
7967 (solve_flow_graph): Likewise.
7968 (find_exception_blocks): Likewise.
7969 (add_line_counts): Likewise.
7970 (output_intermediate_file): Likewise.
7971 (process_file): Likewise.
7972 (generate_results): Likewise.
7973 (release_structures): Likewise.
7974 (read_graph_file): Likewise.
7975 (read_count_file): Likewise.
7976 (accumulate_line_counts): Likewise.
7977 (output_lines): Likewise.
7979 2017-11-13 Martin Liska <mliska@suse.cz>
7981 * gcov.c (function_info::function_info): Remove num_counts
7982 and add vector<gcov_type>.
7983 (function_info::~function_info): Use the vector.
7984 (process_file): Likewise.
7985 (read_graph_file): Likewise.
7986 (read_count_file): Likewise.
7987 (solve_flow_graph): Likewise.
7989 2017-11-13 Martin Liska <mliska@suse.cz>
7991 * gcov.c (function_info::is_artificial): New function.
7992 (process_file): Erase all artificial early.
7993 (generate_results): Skip as all artificial are already
7996 2017-11-13 Martin Liska <mliska@suse.cz>
7998 * gcov.c (read_graph_file): Store to global vector of functions.
7999 (read_count_file): Iterate the vector.
8000 (process_file): Likewise.
8001 (generate_results): Likewise.
8002 (release_structures): Likewise.
8004 2017-11-13 Jakub Jelinek <jakub@redhat.com>
8006 PR tree-optimization/82954
8007 * gimple-ssa-store-merging.c
8008 (imm_store_chain_info::coalesce_immediate_stores): If
8009 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
8011 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
8013 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
8014 Upddate call to ENDIAN_LANE_N.
8015 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
8016 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
8017 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
8018 and use aarch64_endian_lane_rtx.
8019 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
8021 2017-11-12 Tom de Vries <tom@codesourcery.com>
8023 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
8025 2017-11-12 Tom de Vries <tom@codesourcery.com>
8027 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
8029 2017-11-12 Tom de Vries <tom@codesourcery.com>
8031 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
8033 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
8034 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
8035 * defaults.h (ASM_OUTPUT_LABELREF): Same.
8037 2017-11-11 Martin Sebor <msebor@redhat.com>
8040 * doc/extend.texi (attribute nonstring): Remove spurious argument.
8043 * prefic.c (translate_name): Replace strncpy with memcpy to
8044 avoid -Wstringop-truncation.
8046 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8048 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
8050 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8052 * predict.c (maybe_hot_frequency_p): Do not use cfun.
8054 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8056 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
8059 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
8061 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
8062 enable generating XXBRH if the value is in a vector register.
8063 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
8064 value is in a vector register.
8065 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
8066 register to register bswap64's instead of doing the GPR sequence
8067 used on previous machines.
8068 (bswapdi2_xxbrd): New insn.
8069 (bswapdi2_reg): Disallow on ISA 3.0.
8070 (register to register bswap64 splitter): Do not split the insn on
8071 ISA 3.0 systems that use XXBRD.
8073 2017-11-10 Martin Sebor <msebor@redhat.com>
8076 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
8078 (find_subframework_file): Same.
8080 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8082 * auto-profile.c (afdo_indirect_call): Drop frequency.
8083 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
8084 (cgraph_node::create_edge): Drop frequency argument.
8085 (cgraph_node::create_indirect_edge): Drop frequency argument.
8086 (cgraph_edge::make_speculative): Drop frequency arguments.
8087 (cgraph_edge::resolve_speculation): Do not update frequencies
8088 (cgraph_edge::dump_edge_flags): Do not dump frequency.
8089 (cgraph_node::dump): Check consistency in IPA mode.
8090 (cgraph_edge::maybe_hot_p): Use IPA counter.
8091 (cgraph_edge::verify_count_and_frequency): Rename to ...
8092 (cgraph_edge::verify_count): ... this one; drop frequency checking.
8093 (cgraph_node::verify_node): Update.
8094 * cgraph.h (struct cgraph_edge): Drop frequency.
8095 (cgraph_edge::frequency): New function.
8096 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
8098 (cgraph_edge::rebuild_edges): Likewise.
8099 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
8100 (duplicate_thunk_for_node): Do not pass frequency.
8101 (cgraph_node::create_clone): Scale only counts.
8102 (cgraph_node::create_virtual_clone): Do not pass frequency.
8103 (cgraph_node::create_edge_including_clones): Do not pass frequency.
8104 (cgraph_node::create_version_clone): Do not pass frequency.
8105 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
8106 (cgraph_node::expand_thunk): Do not pass frequency.
8107 (cgraph_node::create_wrapper): Do not pass frequency.
8108 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
8110 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
8111 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
8112 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
8113 (gather_caller_stats): Use frequency function.
8114 (ipcp_cloning_candidate_p): Use frequency function.
8115 (ipcp_propagate_stage): Use frequency function.
8116 (get_info_about_necessary_edges): Use frequency function.
8117 (update_profiling_info): Update only IPA profile.
8118 (update_specialized_profile): Use frequency functoin.
8119 (perhaps_add_new_callers): Update only IPA profile.
8120 * ipa-devirt.c (ipa_devirt): Use IPA profile.
8121 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
8122 (dump_ipa_call_summary): Use frequency function.
8123 (estimate_edge_size_and_time): Use frequency function.
8124 (ipa_merge_fn_summary_after_inlining): Use frequency function.
8125 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
8126 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
8127 (update_noncloned_counts): ... ths one; scale counts only.
8128 (clone_inlined_nodes): Do not scale frequency.
8129 (inline_call): Do not pass frequency.
8130 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
8131 (compute_inlined_call_time): Use IPA profile.
8132 (want_inline_small_function_p): Use IPA profile.
8133 (want_inline_self_recursive_call_p): Use IPA profile.
8134 (edge_badness): Use IPA profile.
8135 (lookup_recursive_calls): Use IPA profile.
8136 (recursive_inlining): Do not pass frequency.
8137 (resolve_noninline_speculation): Do not update frequency.
8138 (inline_small_functions): Collect max of IPA profile.
8139 (dump_overall_stats): Dump IPA porfile.
8140 (dump_inline_stats): Dump IPA porfile.
8141 (ipa_inline): Collect IPA stats.
8142 * ipa-inline.h (clone_inlined_nodes): Update prototype.
8143 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
8144 (ipa_propagate_frequency): Use frequency function.
8145 (ipa_profile): Cleanup.
8146 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
8147 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
8148 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
8149 (input_node): Do not stream frequency.
8150 (input_edge): Do not stream frequency.
8151 (merge_profile_summaries): Scale only IPA profiles.
8152 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
8153 * predict.c (drop_profile): Do not recompute frequency.
8154 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
8155 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
8156 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
8157 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
8158 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
8159 * tree-inline.c (copy_bb): Do not scale frequency.
8160 (expand_call_inline): Do not scale frequency.
8161 (tree_function_versioning): Do not scale frequency.
8162 * ubsan.c (ubsan_create_edge): Do not pass frequency.
8164 2017-11-10 Julia Koval <julia.koval@intel.com>
8166 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
8167 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
8168 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
8169 (_mm256_maskz_gf2p8affine_epi64_epi8)
8170 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
8171 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
8172 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
8173 (__builtin_ia32_vgf2p8affineqb_v32qi)
8174 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
8175 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
8177 2017-11-10 Tamar Christina <tamar.christina@arm.com>
8181 (arm_option_override): Refactor.
8182 (arm_option_reconfigure_globals): New.
8183 (arm_options_perform_arch_sanity_checks): New.
8184 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
8186 (arm_options_perform_arch_sanity_checks): Likewise
8188 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
8190 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
8191 (power9-qpmul): New.
8192 * rs6000/rs6000.md ("type" attr): Add qmul.
8193 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
8194 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
8195 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
8197 2017-11-10 Martin Sebor <msebor@redhat.com>
8200 * builtins.c (compute_objsize): Handle arrays that
8201 compute_builtin_object_size likes to fail for. Make extern.
8202 * builtins.h (compute_objsize): Declare.
8203 (check_strncpy_sizes): New function.
8204 (expand_builtin_strncpy): Call check_strncpy_sizes.
8205 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
8206 -Wstringop-truncation.
8207 (gimple_fold_builtin_strncat): Same.
8208 * gimple.c (gimple_build_call_from_tree): Set call location.
8209 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
8210 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
8211 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
8212 (handle_builtin_strlen): Use strlen_to_stridx.
8213 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
8215 Use strlen_to_stridx.
8216 (pass_strlen::execute): Release strlen_to_stridx.
8217 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
8218 (-Wstringop-truncation): Document new option.
8220 2017-11-10 Martin Liska <mliska@suse.cz>
8222 PR gcov-profile/82702
8223 * gcov.c (main): Handle intermediate files in a different
8225 (get_gcov_intermediate_filename): New function.
8226 (output_gcov_file): Remove support of intermediate files.
8227 (generate_results): Allocate intermediate file.
8228 (release_structures): Clean-up properly fn_end.
8229 (output_intermediate_file): Start iterating with line 1.
8231 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8233 PR tree-optimization/82929
8234 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8235 ops_swapped_p non-static data member.
8236 (store_immediate_info::store_immediate_info): Clear it.
8237 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8238 ops set ops_swapped_p.
8239 (count_multiple_uses): Handle ops_swapped_p.
8241 2017-11-10 Martin Liska <mliska@suse.cz>
8243 * coverage.c (coverage_init): Stream information about
8244 support of has_unexecuted_blocks.
8245 * doc/gcov.texi: Document that.
8246 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8247 * gcov.c (read_graph_file): Likewise.
8248 (output_line_beginning): Fix a small issue with
8251 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8253 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8254 reference of trivial component.
8256 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8259 * gimple-ssa-store-merging.c
8260 (pass_store_merging::terminate_all_aliasing_chains): For
8261 gimple_store_p stmts also call refs_output_dependent_p.
8263 PR rtl-optimization/82913
8264 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8267 2017-11-09 Jeff Law <law@redhat.com>
8269 * vr-values.h: New file with vr_values class.
8270 * tree-vrp.c: Include vr-values.h
8271 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8272 data objects into the vr_values class.
8273 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8274 (get_value_range): Make it a member function within vr_values class.
8275 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8276 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8277 (op_with_constant_singleton_value_range): Likewise.
8278 (extract_range_for_var_from_comparison_expr): Likewise.
8279 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8280 (extract_range_from_binary_expr): Likewise.
8281 (extract_range_from_unary_expr): Likewise.
8282 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8283 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8284 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8285 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8286 (compare_name_with_value, compare_names): Likewise.
8287 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8288 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8289 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8290 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8291 (extract_range_from_phi_node): Likewise.
8292 (simplify_truth_ops_using_ranges): Likewise.
8293 (simplify_div_or_mod_using_ranges): Likewise.
8294 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8295 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8296 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8297 (simplify_float_conversion_using_ranges): Likewise.
8298 (simplify_internal_call_using_ranges): Likewise.
8299 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8300 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8301 poitner via x_vr_values for calls into gimple folder.
8302 (vrp_initialize_lattice): Make this the vr_values ctor.
8303 (vrp_free_lattice): Make this the vr_values dtor.
8304 (set_vr_value): New function.
8305 (class vrp_prop): Add vr_values data member. Add various member
8306 functions as well as member functions that delegate to vr_values.
8307 (check_array_ref): Make a member function within vrp_prop class.
8308 (search_for_addr_array, vrp_initialize): Likewise.
8309 (vrp_finalize): Likewise. Revamp to avoid direct access to
8310 vr_value, values_propagated, etc.
8311 (check_array_bounds): Extract vrp_prop class instance pointer from
8312 walk info structure. Use it to call member functions.
8313 (check_all_array_refs): Make a member function within vrp_prop class.
8314 Smuggle class instance pointer via walk info structure.
8315 (x_vr_values): New local static.
8316 (vrp_valueize): Use x_vr_values to get class instance.
8317 (vr_valueize_1): Likewise.
8318 (class vrp_folder): Add vr_values data member. Add various member
8319 functions as well as member functions that delegate to vr_values.
8320 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8321 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8322 class instance from vr_values. Use it to call member functions.
8323 (vrp_dom_walker): Add vr_values data member.
8324 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8325 instance via x_vr_values.
8326 (identify_jump_threads): Accept vr_values as argument. Store
8327 it into the walker structure.
8328 (evrp_dom_walker): Add vr_values class data member. Add various
8330 (evrp_dom_walker::try_find_new_range): Use vr_values data
8331 member to access the memory allocator.
8332 (evrp_dom_walker::before_dom_children): Store vr_values class
8333 instance into the vrp_folder class.
8334 (evrp_dom_walker::push_value_range): Rework to avoid direct
8335 access to num_vr_values and vr_value.
8336 (evrp_dom_walker::pop_value_range): Likewise.
8337 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8338 Use vr_values to get to dump_all_value_ranges member function.
8339 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8340 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8341 Pass vr_values class instance down to identify_jump_threads.
8342 Remove call to vrp_free_lattice.
8343 (debug_all_value_ranges): Remove.
8345 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8346 (vrp_folder): Likewise.
8348 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8349 Get it from the existing bitmap instead.
8350 (vrp_intersect_ranges_1): Likewise.
8352 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8354 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8356 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8357 set bit_not_p to it.
8358 (imm_store_chain_info::coalesce_immediate_stores): Break group
8359 if bit_not_p is different.
8360 (count_multiple_uses, split_group,
8361 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8362 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8363 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8364 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8365 to store_immediate_info ctor.
8367 2017-11-09 Jim Wilson <jimw@sifive.com>
8369 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8370 (scan_prog_file): Likewise.
8372 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8374 * bb-reorder.c (max_entry_frequency): Remove.
8375 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8376 connect_traces, push_to_next_round_p): Remove prototypes.
8377 (find_traces_1_round): Use counts only.
8378 (push_to_next_round_p): Likewise.
8379 (find_traces): Likewise.
8380 (rotate_loop): Likewise.
8381 (find_traces_1_round): Likewise.
8382 (connect_traces): Likewise.
8383 (edge_order): Likewise.
8385 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8387 * config/arm/arm.c (output_return_instruction): Add comments to
8388 indicate requirement for cmse_nonsecure_entry return to account
8389 for the size of clearing instruction output here.
8390 (thumb_exit): Likewise.
8391 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8392 return in hardfloat mode.
8394 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8396 * config/rs6000/rs6000.c (machine_function): Add a bool,
8397 "toc_is_wrapped_separately".
8398 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8399 if it wasn't explicitly set or unset, we are optimizing for speed, and
8400 doing separate shrink-wrapping.
8401 (rs6000_get_separate_components): Enable the TOC component if
8402 saving the TOC register in the prologue.
8403 (rs6000_components_for_bb): Handle the TOC component.
8404 (rs6000_emit_prologue_components): Store the TOC register where needed.
8405 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8406 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8408 2017-11-09 Martin Jambor <mjambor@suse.cz>
8410 * ipa-param-manipulation.c: New file.
8411 * ipa-param-manipulation.h: Likewise.
8412 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8413 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8414 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8415 (ipa_parm_adjustment): Likewise.
8416 (ipa_parm_adjustment_vec): Likewise.
8417 (ipa_get_vector_of_formal_parms): Moved declaration to
8418 ipa-param-manipulation.h.
8419 (ipa_get_vector_of_formal_parm_types): Likewise.
8420 (ipa_modify_formal_parameters): Likewise.
8421 (ipa_modify_call_arguments): Likewise.
8422 (ipa_combine_adjustments): Likewise.
8423 (ipa_dump_param_adjustments): Likewise.
8424 (ipa_modify_expr): Likewise.
8425 (ipa_get_adjustment_candidate): Likewise.
8426 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8427 ipa-param-manipulation.c.
8428 (ipa_get_vector_of_formal_parm_types): Likewise.
8429 (ipa_modify_formal_parameters): Likewise.
8430 (ipa_modify_call_arguments): Likewise.
8431 (ipa_modify_expr): Likewise.
8432 (get_ssa_base_param): Likewise.
8433 (ipa_get_adjustment_candidate): Likewise.
8434 (index_in_adjustments_multiple_times_p): Likewise.
8435 (ipa_combine_adjustments): Likewise.
8436 (ipa_dump_param_adjustments): Likewise.
8437 * tree-sra.c: Also include ipa-param-manipulation.h
8438 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8441 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8442 Alan Hayward <alan.hayward@arm.com>
8443 David Sherwood <david.sherwood@arm.com>
8445 * doc/sourcebuild.texi (vect_masked_store): Document.
8447 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8448 Alan Hayward <alan.hayward@arm.com>
8449 David Sherwood <david.sherwood@arm.com>
8451 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8453 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8454 Alan Hayward <alan.hayward@arm.com>
8455 David Sherwood <david.sherwood@arm.com>
8457 * doc/sourcebuild.texi (vect_variable_length): Document.
8459 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8460 Alan Hayward <alan.hayward@arm.com>
8461 David Sherwood <david.sherwood@arm.com>
8463 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8465 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8466 Alan Hayward <alan.hayward@arm.com>
8467 David Sherwood <david.sherwood@arm.com>
8469 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8471 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8472 Alan Hayward <alan.hayward@arm.com>
8473 David Sherwood <david.sherwood@arm.com>
8475 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8476 previously undocumented selectors.
8477 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8479 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8481 * doc/rtl.texi (const_vector): Say that elements can be
8482 const_wide_ints too.
8483 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8484 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8485 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8486 * optabs.c (expand_vector_broadcast): Likewise.
8488 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8489 Alan Hayward <alan.hayward@arm.com>
8490 David Sherwood <david.sherwood@arm.com>
8492 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8493 scaled index even if the unscaled address was invalid.
8494 Don't increase the complexity of using a scale in that case.
8496 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8497 Alan Hayward <alan.hayward@arm.com>
8498 David Sherwood <david.sherwood@arm.com>
8500 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8501 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8502 UNITS_PER_WORD bytes.
8503 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8504 based on REGMODE_NATURAL_SIZE of the inner mode.
8505 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8506 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8507 * expmed.c (lowpart_bit_field_p): Divide the value up into
8508 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8509 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8510 whether something is likely to occupy more than one register.
8512 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8515 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8516 profile_count::adjust_for_ipa_scaling.
8517 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8518 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8520 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8522 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8524 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8525 (split_group): Add total_orig and total_new arguments, estimate the
8526 number of statements related to the store group without store merging
8527 and with store merging.
8528 (imm_store_chain_info::output_merged_store): Adjust split_group
8529 callers, punt if estimated number of statements with store merging
8530 is not smaller than estimated number of statements without it.
8532 (handled_load): Remove has_single_use checks.
8533 (pass_store_merging::process_store): Likewise.
8535 2017-11-09 Richard Biener <rguenther@suse.de>
8537 PR tree-optimization/82902
8538 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8540 2017-11-09 Martin Liska <mliska@suse.cz>
8543 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8546 2017-11-09 Martin Liska <mliska@suse.cz>
8548 PR tree-optimization/82669
8549 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8551 2017-11-09 Martin Liska <mliska@suse.cz>
8553 PR gcov-profile/48463
8554 * coverage.c (coverage_begin_function): Output also end locus
8555 of a function and information whether the function is
8557 * gcov-dump.c (tag_function): Parse and print the information.
8558 * gcov.c (INCLUDE_MAP): Add include.
8559 (INCLUDE_SET): Likewise.
8560 (struct line_info): Move earlier in the source file because
8561 of vector<line_info> in function_info structure.
8562 (line_info::line_info): Likewise.
8563 (line_info::has_block): Likewise.
8564 (struct source_info): Add new member index.
8565 (source_info::get_functions_at_location): New function.
8566 (function_info::group_line_p): New function.
8567 (output_intermediate_line): New function.
8568 (output_intermediate_file): Use the mentioned function.
8569 (struct function_start): New.
8570 (struct function_start_pair_hash): Likewise.
8571 (process_file): Add code that identifies group functions.
8572 Assign lines either to global or function scope.
8573 (generate_results): Skip artificial functions.
8574 (find_source): Assign index for each source file.
8575 (read_graph_file): Read new flag artificial and end_line.
8576 (add_line_counts): Assign it either to global of function scope.
8577 (accumulate_line_counts): Isolate core of the function to
8578 accumulate_line_info and call it for both function and global
8580 (accumulate_line_info): New function.
8581 (output_line_beginning): Fix GNU coding style.
8582 (print_source_line): New function.
8583 (output_line_details): Likewise.
8584 (output_function_details): Likewise.
8585 (output_lines): Iterate both source (global) scope and function
8587 (struct function_line_start_cmp): New class.
8588 * doc/gcov.texi: Reflect changes in documentation.
8590 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8593 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8594 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8595 and similarly for not instead of neg.
8597 2017-11-08 Andi Kleen <ak@linux.intel.com>
8599 * config/i386/i386.opt: Add -mforce-indirect-call.
8600 * config/i386/predicates.md: Check for flag_force_indirect_call.
8601 * doc/invoke.texi: Document -mforce-indirect-call
8603 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8605 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8607 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8608 config/riscv/riscv.c (predict.h): New include.
8609 (riscv_slow_unaligned_access_p): No longer static.
8610 (riscv_block_move_straight): Add require.
8611 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8613 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8616 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8617 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8618 nonimmediate_operand predicate for operand 1 instead of
8621 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8623 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8625 * config/aarch64/constraints.md (Uml): New constraint.
8626 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8629 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8631 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8632 of two vec_duplicates into a vec_concat.
8634 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8636 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8637 Simplify vec_merge of vec_duplicate and vec_concat.
8638 * config/aarch64/constraints.md (Utq): New constraint.
8639 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8642 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8644 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8645 Simplify vec_merge of vec_duplicate and const_vector.
8646 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8648 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8649 mode iterator. Update predicate on operand 1 to
8650 handle non-const_vec constants. Delete constraints.
8651 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8653 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8655 PR tree-optimization/78821
8656 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8658 (store_operand_info::store_operand_info): Initialize it to false.
8659 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8660 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8661 store in the group, and if chain_info is non-NULL, to ignore altogether
8663 (compatible_load_p): Fail if bit_not_p does not match.
8664 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8665 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8666 (pass_store_merging::process_store): Adjust
8667 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8668 call terminate_all_aliasing_chains newly when adding a store into
8669 a chain with non-NULL chain_info.
8671 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8673 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8675 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8677 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8679 (aarch64_layout_frame): Initialise emit_frame_chain.
8680 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8681 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8683 2017-11-08 Martin Liska <mliska@suse.cz>
8685 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8686 of gimple_call_internal_p.
8688 2017-11-07 Tom de Vries <tom@codesourcery.com>
8690 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8692 2017-11-07 Tom de Vries <tom@codesourcery.com>
8694 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8697 2017-11-08 Martin Liska <mliska@suse.cz>
8700 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8702 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8704 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8706 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8709 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8710 (*cmp<mode>_ccz_1): New insn with $k alternative.
8713 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8714 RTX_COMM_COMPARE as commutative as well.
8715 (ix86_binary_operator_ok): Formatting fix.
8716 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8717 *<code><mode>3<mask_name><round_saeonly_name>,
8718 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8719 *<s>mul<mode>3_highpart<mask_name>,
8720 *vec_widen_umult_even_v16si<mask_name>,
8721 *vec_widen_umult_even_v8si<mask_name>,
8722 *vec_widen_umult_even_v4si<mask_name>,
8723 *vec_widen_smult_even_v16si<mask_name>,
8724 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8725 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8726 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8727 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8728 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8729 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8730 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8731 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8732 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8733 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8734 ix86_binary_operator_ok. Formatting fixes.
8735 (*<plusminus_insn><mode>3<mask_name><round_name>,
8736 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8739 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8741 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8742 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8743 eq and ne if TARGET_ISEL.
8744 (cmp): New code_iterator.
8745 (UNS, UNSU_, UNSIK): New code_attrs.
8746 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8747 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8749 ("eq<mode>3"): ... this.
8750 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8752 ("ne<mode>3"): ... this.
8754 2017-11-07 Julia Koval <julia.koval@intel.com>
8757 * common/config/i386/i386-common.c
8758 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8759 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8760 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8761 * config/i386/i386.opt: Ditto.
8762 * config/i386/i386.c (ix86_target_string): Ditto.
8763 (ix86_option_override_internal): Ditto.
8764 (ix86_init_mpx_builtins): Move MPX to args2.
8765 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8766 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8767 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8768 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8769 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8770 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8771 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8773 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8776 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8777 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8778 (zero-extendsidi peephole2): Remove peephole.
8780 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8783 * stor-layout.c: Include attribs.h.
8784 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8785 explicit lookup of "aligned" attribute.
8787 2017-11-07 Andrew Waterman <andrew@sifive.com>
8789 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8790 (riscv_expand_block_move): Likewise.
8791 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8793 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8794 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8795 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8796 (riscv_adjust_block_mem): Likewise.
8797 (riscv_block_move_loop): Likewise.
8798 (riscv_expand_block_move): Likewise.
8799 * config/riscv/riscv.md (movmemsi): New pattern.
8801 2017-11-07 Michael Clark <michaeljclark@mac.com>
8803 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8804 (MUSL_DYNAMIC_LINKER): Likewise.
8806 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8808 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8809 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8812 2017-11-07 Richard Biener <rguenther@suse.de>
8814 * match.pd: Fix build.
8816 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8817 Jackson Woodruff <jackson.woodruff@arm.com>
8819 PR tree-optimization/71026
8820 * match.pd: Canonicalize negate in division.
8822 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8825 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8827 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8829 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8830 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8832 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8834 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8835 non-scalar integral types.
8836 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8837 (-(A-B), -(~A)): New transformations.
8839 2017-11-07 Tom de Vries <tom@codesourcery.com>
8841 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8842 semicolon after "do {} while (0)".
8843 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8844 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8845 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8846 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8847 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8849 2017-11-07 Tom de Vries <tom@codesourcery.com>
8851 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8852 after "do {} while (0)".
8853 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8854 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8855 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8856 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8857 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8859 2017-11-07 Tom de Vries <tom@codesourcery.com>
8862 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8864 (arm_rtx_costs_internal): Add missing semicolon after
8865 HANDLE_NARROW_SHIFT_ARITH call.
8867 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8869 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8870 disable isel if it was not set explicitly.
8872 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8874 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8875 (add_type_attribute) likewise.
8877 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8879 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8880 of struct ix86_frame.
8881 (ix86_initial_elimination_offset): Likewise.
8882 (ix86_expand_split_stack_prologue): Likewise.
8884 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8886 * tree-vrp.h (enum value_range_type): Update stale comment.
8888 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8889 Alan Hayward <alan.hayward@arm.com>
8890 David Sherwood <david.sherwood@arm.com>
8892 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8893 (aarch64_expand_vec_perm_const): Take the number of units too.
8894 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8895 (aarch64_expand_vec_perm_const): Likewise.
8896 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8897 (vec_perm<mode>): Update accordingly.
8899 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8900 Alan Hayward <alan.hayward@arm.com>
8901 David Sherwood <david.sherwood@arm.com>
8903 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8904 Take the number of units too.
8905 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8906 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8907 but check for a vector mode before rather than after the call.
8908 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8909 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8910 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8911 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8912 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8913 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8914 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8915 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8916 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8917 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8918 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8919 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8920 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8921 (aarch64_sqdmull2_n<mode>): Update accordingly.
8923 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8924 Alan Hayward <alan.hayward@arm.com>
8925 David Sherwood <david.sherwood@arm.com>
8927 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8928 the number of units too.
8929 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8930 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8931 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8932 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8933 (vec_store_lanesxi<mode>): Update accordingly.
8935 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8936 Alan Hayward <alan.hayward@arm.com>
8937 David Sherwood <david.sherwood@arm.com>
8939 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8940 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8941 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8942 of units rather than the mode.
8943 * config/aarch64/iterators.md (nunits): New mode attribute.
8944 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8945 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8946 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8947 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8948 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8949 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8950 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8951 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8952 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8953 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8954 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8955 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8956 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8957 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8958 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8959 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8960 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8961 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8962 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8963 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8964 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8965 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8966 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8967 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8968 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8969 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8970 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8971 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8972 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8973 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8974 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8975 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8976 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8977 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8978 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8979 (aarch64_simd_vec_setv2di): Likewise.
8981 2017-11-06 Carl Love <cel@us.ibm.com>
8983 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8985 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8986 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8988 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8990 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8991 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8992 Add power 8 macro expansions.
8993 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8994 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8995 power 8 instructions. (VSX_XXBR): Add iterator.
8997 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8999 * config/arm/arm.md (predicable_short_it): Change default to "no",
9000 improve documentation, remove uses that are identical to the default.
9001 (enabled_for_depr_it): Rename to enabled_for_short_it.
9002 * gcc/config/arm/arm-fixed.md (predicable_short_it):
9003 Remove default uses.
9004 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
9005 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
9006 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
9007 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
9009 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
9012 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
9013 float128 helper macros, which are no longer used after deleting
9014 the old 'q' built-in functions, and moving the round to odd
9015 built-in functions to being special built-in functions.
9016 (BU_FLOAT128_2): Likewise.
9017 (BU_FLOAT128_1_HW): Likewise.
9018 (BU_FLOAT128_2_HW): Likewise.
9019 (BU_FLOAT128_3_HW): Likewise.
9020 (FABSQ): Delete old 'q' built-in functions.
9021 (COPYSIGNQ): Likewise.
9022 (SQRTF128_ODD): Move round to odd built-in functions to be
9023 special built-in functions, so that we can handle
9024 -mabi=ieeelongdouble.
9025 (TRUNCF128_ODD): Likewise.
9026 (ADDF128_ODD): Likewise.
9027 (SUBF128_ODD): Likewise.
9028 (MULF128_ODD): Likewise.
9029 (DIVF128_ODD): Likewise.
9030 (FMAF128_ODD): Likewise.
9031 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
9032 built-in names to 'f128'.
9033 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
9034 old 'q' built-in functions, as the machine independent code for
9035 'f128' built-in functions handles this.
9036 (rs6000_expand_builtin): Add expansion for float128 round to odd
9037 functions, keying off on -mabi=ieeelongdouble of whether to use
9038 the KFmode or TFmode variant.
9039 (rs6000_init_builtins): Initialize the _Float128 round to odd
9041 * doc/extend.texi (PowerPC Built-in Functions): Document the old
9042 _Float128 'q' built-in functions are now mapped into the new
9043 'f128' built-in functions.
9045 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
9047 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
9048 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
9050 2017-11-06 David Malcolm <dmalcolm@redhat.com>
9053 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
9054 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
9055 * toplev.c: Include "ipa-fnsummary.h".
9056 (toplev::finalize): Call ipa_fnsummary_c_finalize.
9058 2017-11-06 Jakub Jelinek <jakub@redhat.com>
9060 PR tree-optimization/82838
9061 * gimple-ssa-store-merging.c
9062 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
9063 on a separate gimple_seq which is then appended to seq.
9065 2017-11-06 Jeff Law <law@redhat.com>
9068 * config/i386/i386.c (PROBE_INTERVAL): Remove.
9069 (get_probe_interval): New functions.
9070 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
9071 (ix86_adjust_stack_and_probe): Likewise.
9072 (output_adjust_stack_and_probe): Likewise.
9073 (ix86_emit_probe_stack_range): Likewise.
9074 (ix86_expand_prologue): Likewise.
9076 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9078 PR tree-optimization/82816
9079 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
9080 if the modes of the two types are the same.
9081 (convert_plusminus_to_widen): Likewise.
9083 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9085 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
9087 (usadv16qi): New define_expand.
9088 (usadv8hi): New define_expand.
9090 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
9093 * ipa-inline-transform.c (update_noncloned_frequencies): Always
9095 (inline_transform): Likewise.
9096 * predict.c (counts_to_freqs): Remove useless conditional.
9097 * profile-count.h (profile_count::apply_scale): Move sanity check.
9098 * tree-inline.c (copy_bb): Always scale.
9099 (copy_cfg_body): Likewise.
9101 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
9104 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
9106 (*cmp_ite0): Add enabled_for_depr_it attribute.
9107 (*cmp_ite1): Likewise.
9109 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
9111 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
9114 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9116 * tree-vrp.c (vrp_int_const_binop): Return true on success and
9117 return the value by pointer.
9118 (extract_range_from_multiplicative_op_1): Update accordingly.
9119 Return as soon as an operation fails.
9121 2017-11-05 Tom de Vries <tom@codesourcery.com>
9124 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
9125 (DEF_SANITIZER_BUILTIN): ... here.
9126 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
9127 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
9129 2017-11-05 Tom de Vries <tom@codesourcery.com>
9132 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
9134 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
9135 ASM_OUTPUT_BEFORE_CASE_LABEL call.
9136 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
9138 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9139 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9140 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9142 2017-11-05 Tom de Vries <tom@codesourcery.com>
9145 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
9148 2017-11-04 Michael Clark <michaeljclark@mac.com>
9150 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
9151 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
9153 (*addsi3_extended): Likewise.
9154 (*addsi3_extended2): Likewise.
9155 (<optab>si3): Likewise.
9156 (<optab>di3): Likewise.
9157 (<optab><mode>3): Likewise.
9158 (<*optabe>si3_internal): Likewise.
9159 (zero_extendqi<SUPERQI:mode>2): Likewise.
9160 (*add<mode>hi3): Likewise.
9161 (*xor<mode>hi3): Likewise.
9162 (<optab>di3): Likewise.
9163 (*<optab>si3_extend): Likewise.
9164 (*sge<u>_<X:mode><GPR:mode>): Likewise.
9165 (*slt<u>_<X:mode><GPR:mode>): Likewise.
9166 (*sle<u>_<X:mode><GPR:mode>): Likewise.
9168 2017-11-04 Andrew Waterman <andrew@sifive.com>
9170 * config/riscv/riscv.c (riscv_option_override): Conditionally set
9171 TARGET_STRICT_ALIGN based upon -mtune argument.
9173 2017-11-04 Andrew Waterman <andrew@sifive.com>
9175 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
9177 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
9179 * config/i386/i386.c (choose_basereg): Use optional scratch
9180 register and add assertion.
9181 (x86_emit_outlined_ms2sysv_save): Use scratch register when
9182 needed, and don't allocate stack.
9183 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
9184 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
9185 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
9187 2017-11-03 Jeff Law <law@redhat.com>
9189 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
9190 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
9191 to probe at the start of a noreturn function.
9193 2017-11-03 Jakub Jelinek <jakub@redhat.com>
9195 PR tree-optimization/78821
9196 * gimple-ssa-store-merging.c: Update the file comment.
9197 (MAX_STORE_ALIAS_CHECKS): Define.
9198 (struct store_operand_info): New type.
9199 (store_operand_info::store_operand_info): New constructor.
9200 (struct store_immediate_info): Add rhs_code and ops data members.
9201 (store_immediate_info::store_immediate_info): Add rhscode, op0r
9202 and op1r arguments to the ctor, initialize corresponding data members.
9203 (struct merged_store_group): Add load_align_base and load_align
9205 (merged_store_group::merged_store_group): Initialize them.
9206 (merged_store_group::do_merge): Update them.
9207 (merged_store_group::apply_stores): Pick the constant for
9208 encode_tree_to_bitpos from one of the two operands, or skip
9209 encode_tree_to_bitpos if neither operand is a constant.
9210 (class pass_store_merging): Add process_store method decl. Remove
9211 bool argument from terminate_all_aliasing_chains method decl.
9212 (pass_store_merging::terminate_all_aliasing_chains): Remove
9213 var_offset_p argument and corresponding handling.
9214 (stmts_may_clobber_ref_p): New function.
9215 (compatible_load_p): New function.
9216 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
9217 if there is overlap and rhs_code is not INTEGER_CST. For
9218 non-overlapping stores terminate group if rhs is not mergeable.
9219 (get_alias_type_for_stmts): Change first argument from
9220 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
9221 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
9222 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9224 (get_location_for_stmts): Change first argument from
9225 auto_vec<gimple *> & to vec<gimple *> &.
9226 (struct split_store): Remove orig_stmts data member, add orig_stores.
9227 (split_store::split_store): Create orig_stores rather than orig_stmts.
9228 (find_constituent_stmts): Renamed to ...
9229 (find_constituent_stores): ... this. Change second argument from
9230 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9231 to info structures rather than the statements.
9232 (split_group): Rename ALLOW_UNALIGNED argument to
9233 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9234 it. Adjust find_constituent_stores caller.
9235 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9236 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9237 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9238 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9239 (mem_valid_for_store_merging): New function.
9240 (handled_load): New function.
9241 (pass_store_merging::process_store): New method.
9242 (pass_store_merging::execute): Use process_store method. Adjust
9243 terminate_all_aliasing_chains caller.
9245 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9247 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9248 Return true for more constants, symbols and label references.
9249 (aarch64_valid_floating_const): Remove unused function.
9251 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9254 * config/aarch64/aarch64.c (aarch64_layout_frame):
9255 Undo forcing of LR at bottom of frame.
9257 2017-11-03 Jeff Law <law@redhat.com>
9260 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9261 for int_registers_saved.
9263 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9264 extracted from tree-ssa-dom.c.
9265 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9266 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9267 (record_equivalences_from_incoming_edge): Add additional argument
9268 to single_pred_edge_ignoring_loop_edges call.
9269 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9270 (uncprop_dom_walker::before_dom_children): Add additional argument
9271 to single_pred_edge_ignoring_loop_edges call.
9272 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9273 single_pred_edge_ignoring_loop_edges rather than open coding.
9274 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9276 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9278 * match.pd (-(-A)): Rewrite.
9280 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9282 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9283 (rs6000_emit_int_cmove): New declaration.
9284 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9285 (rs6000_emit_sISEL): Delete.
9286 (rs6000_emit_int_cmove): Make non-static.
9287 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9288 instead of rs6000_emit_sISEL.
9290 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9292 * asan.c (create_cond_insert_point): Maintain profile.
9293 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9295 * basic-block.h (struct basic_block_def): Remove frequency.
9296 (EDGE_FREQUENCY): Use to_frequency
9297 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9299 (find_traces): Update to use to_frequency.
9300 (find_traces_1_round): Likewise; use only IPA counts.
9301 (bb_to_key): Likewise.
9302 (connect_traces): Use IPA counts only.
9303 (copy_bb_p): Update to use to_frequency.
9304 (fix_up_crossing_landing_pad): Likewise.
9305 (sanitize_hot_paths): Likewise.
9306 * bt-load.c (basic_block_freq): Likewise.
9307 * cfg.c (init_flow): Set count_max to uninitialized.
9308 (check_bb_profile): Remove frequencies; check counts.
9309 (dump_bb_info): Do not dump frequencies.
9310 (update_bb_profile_for_threading): Update counts only.
9311 (scale_bbs_frequencies_int): Likewise.
9312 (MAX_SAFE_MULTIPLIER): Remove.
9313 (scale_bbs_frequencies_gcov_type): Update counts only.
9314 (scale_bbs_frequencies_profile_count): Update counts only.
9315 (scale_bbs_frequencies): Update counts only.
9316 * cfg.h (struct control_flow_graph): Add count-max.
9317 (update_bb_profile_for_threading): Update prototype.
9318 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9319 (find_many_sub_basic_blocks): Likewise.
9320 * cfgcleanup.c (try_forward_edges): Likewise.
9321 (try_crossjump_to_edge): Likewise.
9322 * cfgexpand.c (expand_gimple_cond): Likewise.
9323 (expand_gimple_tailcall): Likewise.
9324 (construct_init_block): Likewise.
9325 (construct_exit_block): Likewise.
9326 * cfghooks.c (verify_flow_info): Check consistency of counts.
9327 (dump_bb_for_graph): Do not dump frequencies.
9328 (split_block_1): Do not update frequencies.
9329 (split_edge): Do not update frequencies.
9330 (make_forwarder_block): Do not update frequencies.
9331 (duplicate_block): Do not update frequencies.
9332 (account_profile_record): Do not update frequencies.
9333 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9334 for global heuristics.
9335 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9336 (expected_loop_iterations_unbounded): Use counts only.
9337 * cfgloopmanip.c (scale_loop_profile): Simplify.
9338 (create_empty_loop_on_edge): Simplify
9340 (duplicate_loop_to_header_edge): Simplify
9341 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9342 (update_br_prob_note): Take care of removing note when profile
9344 (relink_block_chain): Do not dump frequency.
9345 (rtl_account_profile_record): Use to_frequency.
9346 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9347 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9348 (cgraph_update_edges_for_call_stmt_node): Likewise.
9349 (cgraph_edge::verify_count_and_frequency): Update.
9350 (cgraph_node::verify_node): Temporarily disable frequency verification.
9351 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9352 to_cgraph_frequency.
9353 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9354 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9356 (cgraph_node::expand_thunk): Update profile.
9357 * except.c (dw2_build_landing_pads): Do not update frequency.
9358 * final.c (compute_alignments): Use to_frequency.
9359 (dump_basic_block_info): Do not dump frequency.
9360 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9361 (dump_gimple_bb_header): Do not dump frequency.
9362 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9364 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9365 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9366 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9367 (init_before_recovery): Do not update frequency.
9368 (sched_create_recovery_edges): Do not update frequency.
9369 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9370 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9371 (ipa_cp_c_finalize): Set max_count to uninitialized.
9372 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9373 (param_change_prob): Use counts.
9374 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9376 * ipa-split.c (consider_split): Use to_frequency.
9377 (split_function): Use to_frequency.
9378 * ira-build.c (loop_compare_func): Likewise.
9379 (mark_loops_for_removal): Likewise.
9380 (mark_all_loops_for_removal): Likewise.
9381 * loop-doloop.c (doloop_modify): Do not update frequency.
9382 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9384 * lto-streamer-in.c (input_function): Update count_max.
9385 * omp-expand.c (expand_omp_taskreg): Update count_max.
9386 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9387 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9388 (maybe_hot_count_p): Use ipa counts only.
9389 (maybe_hot_bb_p): Simplify.
9390 (maybe_hot_edge_p): Simplify.
9391 (probably_never_executed): Do not take frequency argument.
9392 (probably_never_executed_bb_p): Do not pass frequency.
9393 (probably_never_executed_edge_p): Likewise.
9394 (combine_predictions_for_bb): Check that profile is nonzero.
9395 (propagate_freq): Do not set frequency.
9396 (drop_profile): Simplify.
9397 (counts_to_freqs): Simplify.
9398 (expensive_function_p): Use to_frequency.
9399 (propagate_unlikely_bbs_forward): Simplify.
9400 (determine_unlikely_bbs): Simplify.
9401 (estimate_bb_frequencies): Add hack to silence graphite issues.
9402 (compute_function_frequency): Use ipa counts.
9403 (pass_profile::execute): Update.
9404 (rebuild_frequencies): Use counts only.
9405 (force_edge_cold): Use counts only.
9406 * profile-count.c (profile_count::dump): Dump new count types.
9407 (profile_count::differs_from_p): Check compatiblity.
9408 (profile_count::to_frequency): New function.
9409 (profile_count::to_cgraph_frequency): New function.
9410 * profile-count.h (struct function): Declare.
9411 (enum profile_quality): Add profile_guessed_local and
9412 profile_guessed_global0.
9413 (class profile_proability): Decrease number of bits to 29;
9414 update from_reg_br_prob_note and to_reg_br_prob_note.
9415 (class profile_count: Update comment; decrease number of bits
9416 to 61. Check compatibility.
9417 (profile_count::compatible_p): New private member function.
9418 (profile_count::ipa_p): New member function.
9419 (profile_count::operator<): Handle global zero correctly.
9420 (profile_count::operator>): Handle global zero correctly.
9421 (profile_count::operator<=): Handle global zero correctly.
9422 (profile_count::operator>=): Handle global zero correctly.
9423 (profile_count::nonzero_p): New member function.
9424 (profile_count::force_nonzero): New member function.
9425 (profile_count::max): New member function.
9426 (profile_count::apply_scale): Handle IPA scalling.
9427 (profile_count::guessed_local): New member function.
9428 (profile_count::global0): New member function.
9429 (profile_count::ipa): New member function.
9430 (profile_count::to_frequency): Declare.
9431 (profile_count::to_cgraph_frequency): Declare.
9432 * profile.c (OVERLAP_BASE): Delete.
9433 (compute_frequency_overlap): Delete.
9434 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9435 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9436 * sched-ebb.c (rank): Use counts only.
9437 * shrink-wrap.c (handle_simple_exit): Use counts only.
9438 (try_shrink_wrapping): Use counts only.
9439 (place_prologue_for_one_component): Use counts only.
9440 * tracer.c (find_best_predecessor): Use to_frequency.
9441 (find_trace): Use to_frequency.
9442 (tail_duplicate): Use to_frequency.
9443 * trans-mem.c (expand_transaction): Do not update frequency.
9444 * tree-call-cdce.c: Do not update frequency.
9445 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9446 (gimple_merge_blocks): Likewise.
9447 (gimple_split_edge): Likewise.
9448 (gimple_duplicate_sese_region): Likewise.
9449 (gimple_duplicate_sese_tail): Likewise.
9450 (move_sese_region_to_fn): Likewise.
9451 (gimple_account_profile_record): Likewise.
9452 (insert_cond_bb): Likewise.
9453 * tree-complex.c (expand_complex_div_wide): Likewise.
9454 * tree-eh.c (lower_resx): Update profile.
9455 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9457 (initialize_cfun): Do not initialize frequencies
9458 (freqs_to_counts): Delete.
9459 (copy_cfg_body): Ignore count parameter.
9460 (copy_body): Update.
9461 (expand_call_inline): Update count_max.
9462 (optimize_inline_calls): Update count_max.
9463 (tree_function_versioning): Update count_max.
9464 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9465 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9467 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9468 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9469 (try_peel_loop): Likewise.
9470 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9472 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9473 (tree_transform_and_unroll_loop): Do not use frequencies
9474 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9475 Use reliable prediction only.
9476 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9477 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9478 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9479 probability scaling.
9480 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9481 not update frequency
9482 (any_remaining_duplicated_blocks): Likewise.
9483 (update_profile): Likewise.
9484 (estimated_freqs_path): Delete.
9485 (freqs_to_counts_path): Delete.
9486 (clear_counts_path): Delete.
9487 (ssa_fix_duplicate_block_edges): Likewise.
9488 (duplicate_thread_path): Likewise.
9489 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9490 * tree-tailcall.c (decrease_profile): Do not update frequency.
9491 (eliminate_tail_call): Likewise.
9492 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9493 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9494 (optimize_mask_stores): Likewise.
9495 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9496 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9497 (ubsan_expand_ptr_ifn): Update profile.
9498 * value-prof.c (gimple_ic): Simplify.
9499 * value-prof.h (gimple_ic): Update prototype.
9500 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9501 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9503 (want_inline_self_recursive_call_p): Likewise.
9504 (resolve_noninline_speculation): Only cummulate defined counts.
9505 (inline_small_functions): Use nonzero_p.
9506 (ipa_inline): Do not access freed node.
9508 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9510 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9511 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9513 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9515 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9516 non-legitimate address.
9518 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9520 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9521 (*lt0_<mode>di, *lt0_<mode>si): New.
9523 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9525 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9526 TARGET_PAIRED_FLOAT.
9528 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9529 Jim Wilson <jim.wilson@linaro.org>
9531 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9532 * config/aarch64/aarch64-tune.md: Regenerated.
9533 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9534 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9536 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9538 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9539 (arc_expand_prologue): Restore blink for millicode.
9540 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9542 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9545 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9546 gen_vec_duplicate after forcing the scalar into a register.
9548 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9550 * combine (try_combine): Print the insns input to try_combine to the
9553 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9556 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9557 Remove second argument from aarch64_process_target_attr call.
9558 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9560 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9562 (aarch64_handle_attr_arch): Remove second argument.
9563 (aarch64_handle_attr_cpu): Ditto.
9564 (aarch64_handle_attr_tune): Ditto.
9565 (aarch64_handle_attr_isa_flags): Ditto.
9566 (aarch64_process_one_target_attr): Ditto.
9567 (aarch64_process_target_attr): Ditto.
9568 (aarch64_option_valid_attribute_p): Remove second argument.
9569 on aarch64_process_target_attr call.
9571 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9573 * diagnostic.c: Include "selftest-diagnostic.h".
9574 (selftest::assert_location_text): New function.
9575 (selftest::test_diagnostic_get_location_text): New function.
9576 (selftest::diagnostic_c_tests): Call it.
9578 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9580 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9581 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9582 (class selftest::test_diagnostic_context): Move to...
9583 * selftest-diagnostic.c: New file.
9584 * selftest-diagnostic.h: New file.
9586 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9588 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9589 offset range for FT32B.
9590 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9591 * config/ft32/ft32.md: Add TARGET_NOPM.
9592 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9593 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9595 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9597 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9599 2017-11-02 Jeff Law <law@redhat.com>
9601 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9602 virtual keyword on FINAL OVERRIDE members.
9604 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9605 virtuals together. Add virtual destructor.
9606 (substitute_and_fold_engine): Similarly.
9608 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9610 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9612 2017-11-02 Richard Biener <rguenther@suse.de>
9614 PR tree-optimization/82795
9615 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9617 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9619 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9621 (gcc_GAS_CHECK_FEATURE): Remove.
9622 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9623 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9624 * configure: Regenerate.
9626 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9628 * config/arc/arc.c (hwloop_optimize): Account for empty
9631 2017-11-02 Richard Biener <rguenther@suse.de>
9634 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9635 Truncate ARRAY_REF index and element size.
9637 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9639 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9641 2017-11-01 Jeff Law <law@redhat.com>
9643 * tree-ssa-ccp.c (ccp_folder): New class derived from
9644 substitute_and_fold_engine.
9645 (ccp_folder::get_value): New member function.
9646 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9647 (ccp_fold_stmt): Remove prototype.
9648 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9649 * tree-ssa-copy.c (copy_folder): New class derived from
9650 substitute_and_fold_engine.
9651 (copy_folder::get_value): Renamed from get_value.
9652 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9653 * tree-vrp.c (vrp_folder): New class derived from
9654 substitute_and_fold_engine.
9655 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9656 (vrp_folder::get_value): New member function.
9657 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9658 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9659 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9660 provide a class interface to folder/substitute routines.
9661 (ssa_prop_fold_stmt_fn): Remove typedef.
9662 (ssa_prop_get_value_fn): Likewise.
9663 (subsitute_and_fold): Remove prototype.
9664 (replace_uses_in): Likewise.
9665 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9666 Renamed from replace_uses_in. Call the virtual member function
9667 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9668 (substitute_and_fold_dom_walker): Remove initialization of
9669 data member entries for calbacks. Add substitute_and_fold_engine
9670 member and initialize it.
9671 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9672 member functions for get_value, replace_phi_args_in c
9673 replace_uses_in, and fold_stmt calls.
9674 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9675 substitute_and_fold. Remove assert. Update ctor call.
9677 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9678 (ssa_prop_visit_phi_fn): Likewise.
9679 (class ssa_propagation_engine): New class to provide an interface
9681 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9683 (ssa_prop_visit_phi): Likewise.
9684 (ssa_propagation_engine::simulate_stmt): Moved into class.
9685 Call visit_phi/visit_stmt from the class rather than via
9686 file scoped static variables.
9687 (ssa_propagation_engine::simulate_block): Moved into class.
9688 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9689 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9690 set file scoped statics for the visit_stmt/visit_phi callbacks.
9691 * tree-complex.c (complex_propagate): New class derived from
9692 ssa_propagation_engine.
9693 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9694 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9695 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9697 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9698 ssa_propagation_engine.
9699 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9700 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9701 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9702 * tree-ssa-copy.c (copy_prop): New class derived from
9703 ssa_propagation_engine.
9704 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9705 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9706 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9707 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9708 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9709 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9710 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9712 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9714 PR rtl-optimization/82778
9715 PR rtl-optimization/82597
9716 * compare-elim.c (struct comparison): Add in_a_setter field.
9717 (find_comparison_dom_walker::before_dom_children): Remove killed
9718 bitmap and df_simulate_find_defs call, instead walk the defs.
9719 Compute last_setter and initialize in_a_setter. Merge definitions
9720 with first initialization for a few variables.
9721 (try_validate_parallel): Use insn_invalid_p instead of
9722 recog_memoized. Return insn rather than just the pattern.
9723 (try_merge_compare): Fix up comment. Don't uselessly test if
9724 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9726 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9729 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9730 Alan Hayward <alan.hayward@arm.com>
9731 David Sherwood <david.sherwood@arm.com>
9733 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9734 aarch64_hard_regno_nregs to get the number of registers
9737 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9738 Alan Hayward <alan.hayward@arm.com>
9739 David Sherwood <david.sherwood@arm.com>
9741 * config/aarch64/constraints.md (Upl): Rename to...
9743 * config/aarch64/aarch64.md
9744 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9747 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9748 Alan Hayward <alan.hayward@arm.com>
9749 David Sherwood <david.sherwood@arm.com>
9751 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9752 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9755 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9756 Alan Hayward <alan.hayward@arm.com>
9757 David Sherwood <david.sherwood@arm.com>
9759 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9760 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9761 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9763 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9764 of a single element.
9765 * config/aarch64/iterators.md: Add a comment above the permute
9766 unspecs to say that they are generated directly by
9767 aarch64_expand_vec_perm_const.
9768 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9770 2017-11-01 Nathan Sidwell <nathan@acm.org>
9772 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9774 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9776 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9777 and medany code models, and describe what they do.
9779 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9781 Revert accidental duplicate:
9783 * combine.c (can_change_dest_mode): Reject changes in
9784 REGMODE_NATURAL_SIZE.
9786 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9788 PR rtl-optimization/64682
9789 PR rtl-optimization/69567
9790 PR rtl-optimization/69737
9791 PR rtl-optimization/82683
9792 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9793 register mentioned in the note, drop the note, unless it came from I3,
9794 in which case it should go to I3 again.
9796 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9798 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9799 and return false if not.
9800 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9802 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9804 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9805 range is known to be empty.
9807 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9808 Alan Hayward <alan.hayward@arm.com>
9809 David Sherwood <david.sherwood@arm.com>
9811 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9812 and CONST_VECTOR_NUNITS instead of computing the number of units from
9813 the byte sizes of the vector and element.
9814 (simplify_binary_operation_1): Likewise.
9815 (simplify_const_binary_operation): Likewise.
9816 (simplify_ternary_operation): Likewise.
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 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9823 (int_mem_offset): ...this new function.
9824 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9825 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9826 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9829 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9830 Alan Hayward <alan.hayward@arm.com>
9831 David Sherwood <david.sherwood@arm.com>
9833 * lower-subreg.c (interesting_mode_p): New function.
9834 (compute_costs, find_decomposable_subregs, decompose_register)
9835 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9836 (resolve_clobber, dump_choices): Use it.
9838 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9839 Alan Hayward <alan.hayward@arm.com>
9840 David Sherwood <david.sherwood@arm.com>
9842 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9844 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9845 Alan Hayward <alan.hayward@arm.com>
9846 David Sherwood <david.sherwood@arm.com>
9848 * alias.c (find_base_value, find_base_term): Only process integer
9849 truncations. Check the precision rather than the size.
9851 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9852 Alan Hayward <alan.hayward@arm.com>
9853 David Sherwood <david.sherwood@arm.com>
9855 * machmode.h (is_narrower_int_mode): New function
9856 * optabs.c (expand_float, expand_fix): Use it.
9857 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9859 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9860 Alan Hayward <alan.hayward@arm.com>
9861 David Sherwood <david.sherwood@arm.com>
9863 * rtl.h (narrower_subreg_mode): New function.
9864 * ira-color.c (update_costs_from_allocno): Use it.
9866 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9867 Alan Hayward <alan.hayward@arm.com>
9868 David Sherwood <david.sherwood@arm.com>
9870 * optabs-query.h (convert_optab_p): New function, split out from...
9871 (convert_optab_handler): ...here.
9872 (widening_optab_handler): Delete.
9873 (find_widening_optab_handler): Remove permit_non_widening parameter.
9874 (find_widening_optab_handler_and_mode): Likewise. Provide an
9875 override that operates on mode class wrappers.
9876 * optabs-query.c (widening_optab_handler): Delete.
9877 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9878 parameter. Assert that the two modes are the same class and that
9879 the "from" mode is narrower than the "to" mode. Use
9880 convert_optab_handler instead of widening_optab_handler.
9881 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9882 instead of widening_optab_handler.
9883 * expr.c (expand_expr_real_2): Update calls to
9884 find_widening_optab_handler.
9885 * optabs.c (expand_widen_pattern_expr): Likewise.
9886 (expand_binop_directly): Take the insn_code as a parameter.
9887 (expand_binop): Only call find_widening_optab_handler for
9888 conversion optabs; use optab_handler otherwise. Update calls
9889 to find_widening_optab_handler and expand_binop_directly.
9890 Use convert_optab_handler instead of widening_optab_handler.
9891 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9892 find_widening_optab_handler and use scalar_mode rather than
9894 (convert_plusminus_to_widen): Likewise.
9896 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9897 Alan Hayward <alan.hayward@arm.com>
9898 David Sherwood <david.sherwood@arm.com>
9900 * machmode.h (fixed_size_mode): New class.
9901 * rtl.h (get_pool_mode): Return fixed_size_mode.
9902 * gengtype.c (main): Add fixed_size_mode.
9903 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9904 (get_raw_arg_mode): Likewise.
9905 * doc/tm.texi: Regenerate.
9906 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9907 * targhooks.c (default_get_reg_raw_mode): Likewise.
9908 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9909 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9910 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9911 (msp430_get_raw_result_mode): Likewise.
9912 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9913 * dbxout.c (dbxout_parms): Require fixed-size modes.
9914 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9915 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9916 * omp-low.c (lower_oacc_reductions): Likewise.
9917 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9918 (simplify_subreg): Update accordingly.
9919 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9920 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9921 that aren't fixed-size.
9922 (get_pool_mode): Return a fixed_size_mode.
9923 (output_constant_pool_2): Take a fixed_size_mode.
9925 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9926 Alan Hayward <alan.hayward@arm.com>
9927 David Sherwood <david.sherwood@arm.com>
9929 * doc/rtl.texi (vec_series): Document.
9930 (const): Say that the operand can be a vec_series.
9931 * rtl.def (VEC_SERIES): New rtx code.
9932 * rtl.h (const_vec_series_p_1): Declare.
9933 (const_vec_series_p): New function.
9934 * emit-rtl.h (gen_const_vec_series): Declare.
9935 (gen_vec_series): Likewise.
9936 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9937 (gen_vec_series): Likewise.
9938 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9939 * simplify-rtx.c (simplify_unary_operation): Handle negations
9941 (simplify_binary_operation_series): New function.
9942 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9943 (test_vector_ops_series): New function.
9944 (test_vector_ops): Call it.
9945 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9946 gen_const_vec_series.
9947 (altivec_lvsr): Likewise.
9948 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9950 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9951 Alan Hayward <alan.hayward@arm.com>
9952 David Sherwood <david.sherwood@arm.com>
9954 * doc/rtl.texi (const): Update description of address constants.
9955 Say that vector constants are allowed too.
9956 * common.md (E, F): Use CONSTANT_P instead of checking for
9958 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9959 checking for CONST_VECTOR.
9960 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9962 * expr.c (expand_expr_real_2): Check for vector modes instead
9963 of checking for CONST_VECTOR.
9964 * rtl.h (const_vec_p): New function.
9965 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9966 (unwrap_const_vec_duplicate): Handle them here too.
9968 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9969 David Malcolm <dmalcolm@redhat.com>
9970 Alan Hayward <alan.hayward@arm.com>
9971 David Sherwood <david.sherwood@arm.com>
9973 * rtl.h (vec_duplicate_p): New function.
9974 * selftest-rtl.c (assert_rtx_eq_at): New function.
9975 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9976 (assert_rtx_eq_at): Declare.
9977 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9978 * selftest-run-tests.c (selftest::run_tests): Call it.
9979 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9980 (simplify_unary_operation_1): Recursively handle vector duplicates.
9981 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9983 (simplify_subreg): Handle subregs of vector duplicates.
9984 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9985 (selftest::simplify_rtx_c_tests): New functions.
9987 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9988 Alan Hayward <alan.hayward@arm.com>
9989 David Sherwood <david.sherwood@arm.com>
9991 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9992 (gen_vec_duplicate): Likewise.
9993 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9995 (gen_const_vector): ...here.
9996 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9997 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9998 whose elements are all equal.
9999 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
10000 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
10001 (simplify_relational_operation): Likewise.
10002 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
10004 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
10005 (aarch64_expand_vector_init): Likewise.
10006 * config/arm/arm.c (neon_vdup_constant): Likewise.
10007 (neon_expand_vector_init): Likewise.
10008 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
10009 (arm_block_set_unaligned_vect): Likewise.
10010 (arm_block_set_aligned_vect): Likewise.
10011 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
10012 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
10013 (expand_vec_perm_even_odd_pack): Likewise.
10014 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
10015 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
10016 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
10017 gen_const_vec_duplicate.
10018 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
10019 * config/mips/mips.c (mips_gen_const_int_vector): Use
10020 gen_const_vec_duplicate.
10021 (mips_expand_vector_init): Use CONST0_RTX.
10022 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
10023 (define_split): Use gen_const_vec_duplicate.
10024 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
10025 (define_split): Use gen_const_vec_duplicate.
10026 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
10027 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
10028 * config/spu/spu.c (spu_const): Likewise.
10030 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
10031 Alan Hayward <alan.hayward@arm.com>
10032 David Sherwood <david.sherwood@arm.com>
10034 * combine.c (can_change_dest_mode): Reject changes in
10035 REGMODE_NATURAL_SIZE.
10037 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
10039 * configure.ac (--enable-libssp): New.
10040 (gcc_cv_libc_provides_ssp): Check for explicit setting before
10041 trying to determine target-specific default. Adjust indentation.
10042 * configure: Regenerated.
10043 * doc/install.texi (Configuration): Expand --disable-libssp
10046 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
10048 config/i386/i386.c (ix86_expand_epilogue): Correct stack
10051 2017-10-31 Martin Jambor <mjambor@suse.cz>
10054 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
10056 2017-10-31 David Malcolm <dmalcolm@redhat.com>
10058 * auto-profile.c (autofdo_source_profile::read): Use
10059 UNKNOWN_LOCATION rather than 0.
10060 * diagnostic-core.h (warning_at_rich_loc): Rename to...
10061 (warning_at): ...this overload.
10062 (warning_at_rich_loc_n): Rename to...
10063 (warning_n): ...this overload.
10064 (error_at_rich_loc): Rename to...
10065 (error_at): ...this overload.
10066 (pedwarn_at_rich_loc): Rename to...
10067 (pedwarn): ...this overload.
10068 (permerror_at_rich_loc): Rename to...
10069 (permerror): ...this overload.
10070 (inform_at_rich_loc): Rename to...
10071 (inform): ...this overload.
10072 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
10073 (diagnostic_n_impl_richloc): Rename to...
10074 (diagnostic_n_impl): ...this rich_location *-based decl.
10075 (inform_at_rich_loc): Rename to...
10076 (inform): ...this, and add an assertion.
10077 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
10078 (warning_at_rich_loc): Rename to...
10079 (warning_at): ...this, and add an assertion.
10080 (warning_at_rich_loc_n): Rename to...
10081 (warning_n): ...this, and add an assertion.
10082 (warning_n): Update location_t-based implementation for removal of
10083 location_t-based diagnostic_n_impl.
10084 (pedwarn_at_rich_loc): Rename to...
10085 (pedwarn): ...this, and add an assertion.
10086 (permerror_at_rich_loc): Rename to...
10087 (permerror): ...this, and add an assertion.
10088 (error_n): Update for removal of location_t-based diagnostic_n_impl.
10089 (error_at_rich_loc): Rename to...
10090 (error_at): ...this, and add an assertion.
10091 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
10092 (driver::do_spec_on_infiles): Likewise.
10093 * substring-locations.c (format_warning_va): Update for renaming
10094 of inform_at_rich_loc.
10096 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
10098 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
10099 _Float<N>X built-in functions so that the variant without the
10100 "__builtin_" prefix is only enabled for the GNU C and Objective C
10101 languages when they are in non-strict ANSI/ISO mode.
10102 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
10103 * target.def (floatn_builtin_p): Add a target hook to control
10104 whether _Float<N> and _Float<N>X built-in functions without the
10105 "__builtin_" prefix are enabled, and return true for C and
10106 Objective C in the default hook. Include langhooks.h in
10108 * targhooks.h (default_floatn_builtin_p): Likewise.
10109 * targhooks.c (default_floatn_builtin_p): Likewise.
10110 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
10111 floatn_builtin_p target hook.
10112 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
10114 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
10115 Eric Botcazou <ebotcazou@adacore.com>
10117 PR rtl-optimization/81803
10118 * lra-constraints.c (curr_insn_transform): Also reload the whole
10119 register for a strict subreg no wider than a word if this is for
10120 a WORD_REGISTER_OPERATIONS target.
10122 2017-10-31 Jason Merrill <jason@redhat.com>
10124 * gdbinit.in: Skip over inlines from timevar.h.
10126 2017-10-31 Martin Liska <mliska@suse.cz>
10128 * doc/gcov.texi: Document new option.
10129 * gcov.c (print_usage): Likewise print it.
10130 (process_args): Support the argument.
10131 (format_count): New function.
10132 (format_gcov): Use the function.
10134 2017-10-31 Martin Liska <mliska@suse.cz>
10136 * gcov.c (struct name_map): do not use typedef.
10137 Define operator== and operator<.
10138 (name_search): Remove.
10139 (name_sort): Remove.
10140 (main): Do not allocate names.
10141 (process_file): Add vertical space.
10142 (generate_results): Use std::find.
10143 (release_structures): Do not release memory.
10144 (find_source): Use std::find.
10146 2017-10-31 Martin Liska <mliska@suse.cz>
10148 * gcov.c (struct line_info): Remove it's typedef.
10149 (line_info::line_info): Add proper ctor.
10150 (line_info::has_block): Do not use a typedef.
10151 (struct source_info): Do not use typedef.
10152 (circuit): Likewise.
10153 (get_cycles_count): Likewise.
10154 (output_intermediate_file): Iterate via vector iterator.
10155 (add_line_counts): Use std::vector methods.
10156 (accumulate_line_counts): Likewise.
10157 (output_lines): Likewise.
10159 2017-10-31 Martin Liska <mliska@suse.cz>
10161 * gcov.c (struct source_info): Remove typedef.
10162 (source_info::source_info): Add proper ctor.
10163 (accumulate_line_counts): Use struct, not it's typedef.
10164 (output_gcov_file): Likewise.
10165 (output_lines): Likewise.
10166 (main): Do not allocate an array.
10167 (output_intermediate_file): Use size of vector container.
10168 (process_file): Resize the vector.
10169 (generate_results): Do not preallocate, use newly added vector
10171 (release_structures): Do not release sources.
10172 (find_source): Use vector methods.
10173 (add_line_counts): Do not use typedef.
10175 2017-10-31 Martin Liska <mliska@suse.cz>
10177 * doc/gcov.texi: Document that.
10178 * gcov.c (add_line_counts): Mark lines with a non-executed
10180 (output_line_beginning): Handle such lines.
10181 (output_lines): Pass new argument.
10182 (output_intermediate_file): Print it in intermediate format.
10184 2017-10-31 Martin Liska <mliska@suse.cz>
10186 * color-macros.h: New file.
10187 * diagnostic-color.c: Factor out color related to macros to
10189 * doc/gcov.texi: Document -k option.
10190 * gcov.c (INCLUDE_STRING): Include string.h.
10191 (print_usage): Add -k option.
10192 (process_args): Parse it.
10193 (pad_count_string): New function.
10194 (output_line_beginning): Likewise.
10195 (DEFAULT_LINE_START): New macro.
10196 (output_lines): Support color output.
10198 2017-10-31 Martin Liska <mliska@suse.cz>
10200 PR gcov-profile/82633
10201 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
10202 their interaction with GCOV infrastructure.
10203 * configure.ac: Add -fkeep-{inline,static}-functions to
10205 * configure: Regenerate.
10207 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
10210 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
10212 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
10215 * config/rs6000/rs6000.md (allocate_stack): Force update interval
10216 into a register if it does not fit into an immediate offset field.
10218 2017-10-31 Olivier Hainque <hainque@adacore.com>
10220 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
10222 2017-10-31 Julia Koval <julia.koval@intel.com>
10224 * config.gcc: Add gfniintrin.h.
10225 * config/i386/gfniintrin.h: New.
10226 * config/i386/i386-builtin-types.def
10227 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10228 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10229 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10230 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10231 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10232 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10233 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10234 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10235 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10236 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10237 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10238 * config/i386/immintrin.h: Include gfniintrin.h.
10239 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10241 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10243 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10245 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10247 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10248 (arm_ashldi3_1bit): Remove pattern.
10249 (ashrdi3): Remove shift by 1 expansion.
10250 (arm_ashrdi3_1bit): Remove pattern.
10251 (lshrdi3): Remove shift by 1 expansion.
10252 (arm_lshrdi3_1bit): Remove pattern.
10253 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10254 cost of ashldi3 by 1.
10255 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10256 (<shift>di3_neon): Likewise.
10258 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10260 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10261 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10262 and (*aarch64_simd_mov<VQ:mode>).
10263 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10264 pattern alternative.
10265 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10266 attributes to match pattern alternative.
10268 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10270 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10271 * config/rs6000/emmintrin.h: New file.
10272 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10274 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10276 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10277 * (movdi_vfp_cortexa8): Remove pattern.
10279 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10281 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10283 (Specific, alpha*-dec-osf5.1): Remove.
10284 (Specific, mips-sgi-irix5): Remove.
10285 (Specific, mips-sgi-irix6): Remove.
10287 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10289 PR middle-end/22141
10290 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10291 arguments to clear_bit_region_be.
10293 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10295 * gimplify.c: Include memmodel.h.
10297 2017-10-30 Martin Jambor <mjambor@suse.cz>
10299 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10300 condition whether loop should be executed at all.
10302 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10304 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10305 gimple folding of vec_madd() intrinsics.
10306 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10307 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10308 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10310 2017-10-30 Richard Biener <rguenther@suse.de>
10312 PR tree-optimization/82762
10314 2017-10-23 Richard Biener <rguenther@suse.de>
10316 PR tree-optimization/82129
10318 2017-08-01 Richard Biener <rguenther@suse.de>
10320 PR tree-optimization/81181
10321 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10322 (compute_antic): ... end of iteration here.
10324 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10326 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10327 -std=iso9899:2017 and -std=gnu17.
10328 * doc/standards.texi (C Language): Document C17 support.
10329 * doc/cpp.texi (Overview): Mention -std=c17.
10330 (Standard Predefined Macros): Document C11 and C17 values of
10331 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10332 * doc/extend.texi (Inline): Do not list individual options for
10333 standards newer than C99.
10334 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10336 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10339 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10341 * asan.c (asan_finish_file): Align asan globals array by shadow
10344 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10346 PR middle-end/22141
10347 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10348 (struct store_immediate_info): Add bitregion_start and bitregion_end
10350 (store_immediate_info::store_immediate_info): Add brs and bre
10351 arguments and initialize bitregion_{start,end} from those.
10352 (struct merged_store_group): Add bitregion_start, bitregion_end,
10353 align_base and mask fields. Drop unnecessary struct keyword from
10354 struct store_immediate_info. Add do_merge method.
10355 (clear_bit_region_be): Use memset instead of loop storing zeros.
10356 (merged_store_group::do_merge): New method.
10357 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10358 stores as long as the surrounding bitregions have no gaps.
10359 (merged_store_group::merge_overlapping): Use do_merge.
10360 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10361 is byte aligned, rather than requiring that start and width are
10362 byte aligned. Drop unnecessary struct keyword from
10363 struct store_immediate_info. Allocate and populate also mask array.
10364 Make start of the arrays relative to bitregion_start rather than
10365 start and size them according to bitregion_{end,start} difference.
10366 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10367 struct store_immediate_info.
10368 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10369 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10370 struct keyword from struct store_immediate_info.
10371 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10372 between stores as long as the surrounding bitregions have no gaps.
10374 (struct split_store): Add orig non-static data member.
10375 (split_store::split_store): Initialize orig to false.
10376 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10377 if there is exactly a single original stmt. Change stmts argument
10378 to pointer from reference, if NULL, don't push anything to it. Add
10379 first argument, use it to optimize skipping over orig stmts that
10380 are known to be before bitpos already. Simplify.
10381 (split_group): Return unsigned int count how many stores are or
10382 would be needed rather than a bool. Add allow_unaligned argument.
10383 Change split_stores argument from reference to pointer, if NULL,
10384 only do a dry run computing how many stores would be produced.
10385 Rewritten algorithm to use both alignment and misalign if
10386 !allow_unaligned and handle bitfield stores with gaps.
10387 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10388 from bitregion_start instead of start. Compute allow_unaligned
10389 here, if true, do 2 split_group dry runs to compute which one
10390 produces fewer stores and prefer aligned if equal. Punt if
10391 new count is bigger or equal than original before emitting any
10392 statements, rather than during that. Remove no longer needed
10393 new_ssa_names tracking. Replace num_stmts with
10394 split_stores.length (). Use 32-bit stack allocated entries
10395 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10396 if possible. Handle bitfields with gaps.
10397 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10398 Compute bitregion_{start,end} for the stores and construct
10399 store_immediate_info with that. Formatting fixes.
10401 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10404 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10405 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10407 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10409 * gimplify.c: Include tm_p.h.
10411 * common.opt (gcoff): Re-add as ignored option.
10412 (gcoff1, gcoff2, gcoff3): Likewise.
10414 * Makefile.in (OBJS): Delete sdbout.o.
10415 (GTFILES): Delete $(srcdir)/sdbout.c.
10416 * debug.h: Delete sdb_debug_hooks.
10417 * final.c: Delete sdbout.h include.
10418 (final_scan_insn): Delete SDB_DEBUG check.
10419 (rest_of_clean_state): Likewise.
10420 * output.h: Delete sdb_begin_function_line.
10421 * sdbout.c: Delete.
10422 * sdbout.h: Delete.
10423 * toplev.c: Delete sdbout.h include.
10424 (process_options): Delete SDB_DEBUG check.
10425 * tree-core.h (tree_type_common): Delete pointer field of
10427 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10428 TYPE_SYMTAB_POINTER.
10429 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10430 (TYPE_SYMTAB_IS_POINTER): Delete.
10431 (TYPE_SYMTAB_IS_DIE): Renumber.
10432 * xcoffout.c: Refer to former sdbout.c file.
10433 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10435 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10436 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10437 (Debugging Options): Delete -gcoff.
10438 (-gstabs): Delete SDB reference.
10440 (-gcoff@var{level}): Delete.
10441 * doc/passes.texi (Debugging information output): Delete SDB and
10442 sdbout.c references.
10443 * doc/tm.texi: Regenerate.
10444 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10445 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10447 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10448 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10449 (SDB_DEBUGGING_INFO): Delete.
10450 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10451 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10452 * target.def (output_source_filename): Delete COFF reference.
10454 * common.opt (gcoff): Delete.
10455 (gxcoff+): Update Negative chain.
10456 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10458 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10459 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10460 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10461 and SDB references.
10462 (expand_function_start): Change sdb reference to past tense.
10463 (expand_function_end): Change sdb reference to past tense.
10464 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10465 * opts.c (debug_type_names): Delete coff entry.
10466 (common_handle_option): Delete OPT_gcoff case.
10467 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10469 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10470 * config/cris/cris.h: Delete SDB reference in comment.
10471 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10472 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10473 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10474 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10476 (ix86_expand_prologue): Likewise.
10477 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10478 * config/ia64/ia64.h: Likewise.
10479 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10480 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10482 * config/mmix/mmix.h: Likewise.
10483 * config/nds32/nds32.c: Likewise.
10484 * config/stormy/storym16.h: Likewise.
10485 * config/visium/visium.h: Likewise.
10486 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10488 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10490 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10491 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10492 FRAME_POINTER_REGNUM point at high end of local var area.
10494 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10496 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10497 Move comment around. Do not reset best_edge for a copiable
10498 destination if the copy would cause a partition change.
10499 (better_edge_p): Remove redundant check.
10501 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10503 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10505 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10507 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10508 for math functions that have _Float<N> and _Float<N>X variants.
10509 (mathfn_built_in_2): Add support for math functions that have
10510 _Float<N> and _Float<N>X variants.
10511 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10512 (expand_builtin_mathfn_ternary): Add support for fma with
10513 _Float<N> and _Float<N>X variants.
10514 (expand_builtin): Likewise.
10515 (fold_builtin_3): Likewise.
10516 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10517 create math function _Float<N> and _Float<N>X variants as external
10519 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10520 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10521 the __builtin_ prefix and if not strict ansi, without the prefix.
10522 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10523 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10524 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10525 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10526 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10527 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10528 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10529 function signatures for fma _Float<N> and _Float<N>X variants.
10530 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10531 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10532 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10533 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10534 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10535 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10536 * gencfn-macros.c (print_case_cfn): Add support for math functions
10537 that have _Float<N> and _Float<N>X variants.
10538 (print_define_operator_list): Likewise.
10539 (fltfn_suffixes): Likewise.
10541 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10542 for math functions that have _Float<N> and _Float<N>X variants.
10543 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10544 and _Float<N>X variants.
10545 (COPYSIGN): Likewise.
10548 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10549 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10551 (integer_valued_read_call_p): Likewise.
10552 * fold-const-call.c (fold_const_call_ss): Likewise.
10553 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10554 _Float<N> and _Float<N>X variants.
10555 (fold_const_call_ssss): Add support for fma _Float<N> and
10556 _Float<N>X variants.
10557 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10558 support for copysign and fma _Float<N> and _Float<N>X variants.
10559 (backprop::process_builtin_call_use): Likewise.
10560 * tree-call-cdce.c (can_test_argument_range); Add support for
10561 sqrt _Float<N> and _Float<N>X variants.
10562 (edom_only_function): Likewise.
10563 (get_no_error_domain): Likewise.
10564 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10565 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10566 copysign _Float<N> and _Float<N>X variants.
10567 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10568 handled by machine independent code.
10569 (FMAF128): Likewise.
10570 * doc/cpp.texi (Common Predefined Macros): Document defining
10571 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10572 fma _Float<N> and _Float<N>X variants.
10574 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10577 * config/i386/i386-modes.def (CCFPU): Remove definition.
10578 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10579 (ix86_cc_modes_compatible): Ditto.
10580 (ix86_expand_carry_flag_compare): Ditto.
10581 (ix86_expand_int_movcc): Ditto.
10582 (ix86_expand_int_addcc): Ditto.
10583 (ix86_reverse_condition): Ditto.
10584 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10585 Return true/false for unordered/ordered fp comparisons.
10586 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10587 (ix86_prepare_fp_compare_args): Update for rename.
10588 (ix86_expand_fp_compare): Update for rename. Generate unordered
10589 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10590 (ix86_expand_sse_compare_and_jump): Ditto.
10591 * config/i386/predicates.md (fcmov_comparison_operator):
10592 Remove CCFPU mode handling.
10593 (ix86_comparison_operator): Ditto.
10594 (ix86_carry_flag_operator): Ditto.
10595 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10596 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10597 (*cmpu<mode>_cc_i387): Ditto.
10598 (FPCMP): Remove mode iterator.
10599 (unord): Remove mode attribute.
10600 (unord_subst): New define_subst transformation
10601 (unord): New define_subst attribute.
10602 (unordered): Ditto.
10603 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10604 (*cmpi<unord>xf_i387): Ditto.
10605 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10606 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10607 using unord_subst transformation.
10608 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10609 (round_saeonly): Also handle CCFP mode.
10610 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10611 Remove UNSPEC_SAHF unspec handling.
10613 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10615 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10617 2017-10-27 Jeff Law <law@redhat.com>
10619 * gimple-ssa-sprintf.c: Include domwalk.h.
10620 (class sprintf_dom_walker): New class, derived from dom_walker.
10621 (sprintf_dom_walker::before_dom_children): New function.
10622 (struct call_info): Moved into sprintf_dom_walker class
10623 (compute_formath_length, handle_gimple_call): Likewise.
10624 (sprintf_length::execute): Call the dominator walker rather
10625 than walking the statements.
10627 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10628 gimple statement locations.
10629 (check_array_bounds): Corresponding changes. Get the statement's
10630 location directly from wi->stmt.
10632 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10635 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10637 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10639 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10640 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10642 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10645 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10646 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10647 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10648 maybe_get_pool_constant.
10649 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10652 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10654 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10655 2.26 caveat. Update gas and gld versions.
10656 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10659 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10661 * cgraph.h (set_malloc_flag): Declare.
10662 * cgraph.c (set_malloc_flag_1): New function.
10663 (set_malloc_flag): Likewise.
10664 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10665 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10667 (read_ipa_call_summary): Add support for reading is_return_callee.
10668 (write_ipa_call_summary): Stream is_return_callee.
10669 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10670 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10671 ipa-prop.h, ipa-fnsummary.h.
10672 (pure_const_names): Change to static.
10673 (malloc_state_e): Define.
10674 (malloc_state_names): Define.
10675 (funct_state_d): Add field malloc_state.
10676 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10677 (check_retval_uses): New function.
10678 (malloc_candidate_p): Likewise.
10679 (analyze_function): Add support for malloc attribute.
10680 (pure_const_write_summary): Stream malloc_state.
10681 (pure_const_read_summary): Add support for reading malloc_state.
10682 (dump_malloc_lattice): New function.
10683 (propagate_malloc): New function.
10684 (warn_function_malloc): New function.
10685 (ipa_pure_const::execute): Call propagate_malloc and
10686 ipa_free_fn_summary.
10687 (pass_local_pure_const::execute): Add support for malloc attribute.
10688 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10689 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10691 2017-10-27 Martin Liska <mliska@suse.cz>
10693 PR gcov-profile/82457
10694 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10695 for fork-like functions to be properly instrumented.
10697 2017-10-27 Richard Biener <rguenther@suse.de>
10699 PR middle-end/81659
10700 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10701 info when we redirected EH.
10703 2017-10-26 Michael Collison <michael.collison@arm.com>
10705 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10707 (<optab>_trunchf<GPI:mode>2: New pattern.
10708 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10709 * config/aarch64/iterators.md (wv): New mode attribute.
10710 (vf, VF): New mode attributes.
10711 (vgp, VGP): New mode attributes.
10712 (s): Update attribute with SImode and DImode prefixes.
10714 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10716 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10717 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10718 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10719 (nios2_option_overide): Initialize it. Don't allow R0-relative
10720 addressing with PIC.
10721 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10722 (nios2_symbolic_constant_p): Likewise.
10723 (nios2_legitimate_address_p): Likewise.
10724 (nios2_r0rel_section_name_p): New.
10725 (nios2_symbol_ref_in_r0rel_data_p): New.
10726 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10727 (r0rel_constant_p): New.
10728 (nios2_print_operand_address): Handle r0rel_constant_p.
10729 (nios2_cdx_narrow_form_p): Likewise.
10730 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10731 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10732 (Nios II Options): Document -mr0rel-sec.
10734 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10736 * config/nios2/nios2.c: Include xregex.h.
10737 (nios2_gprel_sec_regex): New.
10738 (nios2_option_overide): Initialize it. Don't allow GP-relative
10739 addressing with PIC.
10740 (nios2_small_section_name_p): Check for regex match.
10741 * config/nios2/nios2.opt (mgprel-sec=): New option.
10742 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10743 (Nios II Options): Document -mgprel-sec.
10745 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10747 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10749 2017-10-26 Tom de Vries <tom@codesourcery.com>
10751 PR tree-optimization/82707
10752 * gimple.c (gimple_copy): Fix unsharing of
10753 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10755 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10757 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10758 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10759 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10760 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10761 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10762 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10763 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10764 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10765 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10766 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10767 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10768 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10769 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10770 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10771 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10772 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10773 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10775 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10777 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10779 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10780 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10781 warning to rs6000.c. Remove the Undocumented flag, since it has
10783 (-mabi=ibmlongdouble): Likewise.
10784 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10785 already set, set the default format for long double.
10786 (rs6000_debug_reg_global): Print whether long double is IBM or
10788 (rs6000_option_override_internal): Rework setting long double
10789 format. Only warn if the user is changing the long double default
10790 and they did not use -Wno-psabi.
10791 * doc/invoke.texi (PowerPC options): Update the documentation for
10792 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10794 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10795 Alan Hayward <alan.hayward@arm.com>
10796 David Sherwood <david.sherwood@arm.com>
10798 * rtl.h (wider_subreg_mode): New function.
10799 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10800 rather than an unsigned int *.
10801 * ira-color.c (regno_max_ref_width): Replace with...
10802 (regno_max_ref_mode): ...this new variable.
10803 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10804 Use wider_subreg_mode.
10805 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10806 rather than an unsigned int *.
10807 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10808 (process_alt_operands): Likewise.
10809 (invariant_p): Likewise.
10810 * lra-spills.c (assign_mem_slot): Likewise.
10811 (add_pseudo_to_slot): Likewise.
10812 * lra.c (collect_non_operand_hard_regs): Likewise.
10813 (add_regs_to_insn_regno_info): Likewise.
10814 * reload1.c (regno_max_ref_width): Replace with...
10815 (regno_max_ref_mode): ...this new variable.
10816 (reload): Update accordingly. Update call to
10817 ira_sort_regnos_for_alter_reg.
10818 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10819 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10820 (scan_paradoxical_subregs): Likewise.
10822 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10824 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10825 (aarch64_frame): Add emit_frame_chain boolean.
10826 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10827 Move eh_return case to aarch64_layout_frame.
10828 (aarch64_layout_frame): Initialize emit_frame_chain.
10829 (aarch64_expand_prologue): Use emit_frame_chain.
10831 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10833 * config/aarch64/aarch64.c (aarch64_layout_frame):
10834 Ensure LR is always stored at the bottom of the callee-saves.
10835 Remove rarely used frame layout which saves callee-saves at top of
10836 frame, so the store of LR can be used as a valid probe in all cases.
10838 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10840 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10841 Improve unaligned TImode/TFmode base/offset split.
10843 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10844 Alan Hayward <alan.hayward@arm.com>
10845 David Sherwood <david.sherwood@arm.com>
10847 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10848 * combine.c (find_single_use_1): Likewise.
10849 (expand_field_assignment): Likewise.
10850 (move_deaths): Likewise.
10851 * lra-constraints.c (simplify_operand_subreg): Likewise.
10852 (curr_insn_transform): Likewise.
10853 * lra.c (collect_non_operand_hard_regs): Likewise.
10854 (add_regs_to_insn_regno_info): Likewise.
10855 * rtlanal.c (reg_referenced_p): Likewise.
10856 (covers_regno_no_parallel_p): Likewise.
10858 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10860 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10861 Don't print any bits outside the precision of the value.
10862 * wide-int.cc (test_printing): Add some new tests.
10864 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10866 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10867 supports -xbrace_comment option.
10868 * configure: Regenerate.
10869 * config.in: Regenerate.
10870 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10871 (ASM_CPU_SPEC): Use it.
10873 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10875 * target.def (static_rtx_alignment): New hook.
10876 * targhooks.h (default_static_rtx_alignment): Declare.
10877 * targhooks.c (default_static_rtx_alignment): New function.
10878 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10879 * doc/tm.texi: Regenerate.
10880 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10881 instead of targetm.constant_alignment. Remove call to
10882 set_mem_attributes.
10883 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10884 (cris_preferred_mininum_alignment): New function, split out from...
10885 (cris_constant_alignment): ...here.
10886 (cris_static_rtx_alignment): New function.
10887 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10889 (ix86_constant_alignment): ...here.
10890 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10891 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10892 (mmix_static_rtx_alignment): New function.
10893 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10894 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10896 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10899 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10900 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10902 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10904 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10905 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10906 mark as a sign-extending load.
10907 (local_pic_load_u): Define.
10909 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10911 PR middle-end/82062
10912 * fold-const.c (operand_equal_for_comparison_p): Also return true
10913 if ARG0 is a simple variant of ARG1 with narrower precision.
10914 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10916 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10918 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10920 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10921 scatter_static, scatter_per_elt.
10922 * x86-tune-costs.h: Add new cost entries.
10924 2017-10-25 Richard Biener <rguenther@suse.de>
10926 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10927 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10928 class pass_fre): Move to ...
10929 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10930 class pass_fre): ... here and adjust for statistics.
10932 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10935 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10936 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10937 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10938 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10941 2017-10-25 Richard Biener <rguenther@suse.de>
10943 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10944 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10945 eliminate_push_avail, eliminate_insert): Move inside...
10946 (class eliminate_dom_walker): ... this class in preparation
10948 (fini_eliminate): Remove by merging with ...
10949 (eliminate): ... this function. Adjust for class changes.
10950 (pass_pre::execute): Remove fini_eliminate call.
10951 (pass_fre::execute): Likewise.
10953 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10956 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10957 (VPERMI2, VPERMI2I): New mode iterators.
10958 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10959 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10961 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10962 mode iterator. Remove 3 old define_insn patterns.
10963 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10964 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10965 VPERMI2 mode iterator, remove the other two expanders.
10966 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10967 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10968 instructions, remove the other two patterns.
10969 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10970 mode iterator, remove the other two patterns.
10971 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10972 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10973 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10974 and adjust argument order accordingly.
10975 (ix86_expand_vec_perm): Adjust caller.
10976 (expand_vec_perm_1): Likewise.
10977 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10978 (expand_vec_perm_vpermt2_vpshub2): ... this.
10979 (ix86_expand_vec_perm_const_1): Adjust caller.
10980 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10983 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10984 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10985 (vec_shl_<mode>): Remove unused expander.
10986 (avx512bw_<shift_insn><mode>3): New define_insn.
10987 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10988 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10990 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10993 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10996 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10998 PR rtl-optimization/82396
10999 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
11000 (autopref_multipass_init): Simplify initialization.
11001 (autopref_rank_data): Simplify sort order.
11002 * gcc/sched-int.h (autopref_multipass_data_): Remove
11003 multi_mem_insn_p, min_offset and max_offset.
11005 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
11007 PR middle-end/60580
11008 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
11009 Check special value of flag_omit_frame_pointer.
11010 (aarch64_can_eliminate): Likewise.
11011 (aarch64_override_options_after_change_1): Simplify handling of
11012 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
11014 2017-10-24 Richard Biener <rguenther@suse.de>
11016 PR tree-optimization/82697
11017 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
11018 zero for conditional load and unconditional store.
11020 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
11022 * doc/install.texi: Document bootstrap-cet.
11024 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
11027 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
11028 ENDBR instruction at function entrance if function is only
11031 2017-10-24 Jakub Jelinek <jakub@redhat.com>
11034 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
11035 patterns to better describe from which operation the CF is computed.
11036 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
11037 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
11038 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
11039 is 0, use _0 suffixed expanders instead of emitting a comparison
11042 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
11044 * config/i386/i386.md(*movsf_internal, *movdf_internal):
11045 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
11047 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
11049 PR middle-end/82569
11050 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
11051 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
11052 * loop-iv.c (iv_get_reaching_def): Likewise.
11053 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
11054 variable is promoted and the partition contains undefined values.
11056 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11058 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
11060 (nios2_address_cost): Define.
11061 (nios2_legitimize_address): Recognize (exp + constant) directly.
11062 (TARGET_ADDRESS_COST): Define.
11064 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11066 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
11067 (nios2_symbolic_memory_operand_p): Declare.
11068 (nios2_split_large_constant): Declare.
11069 (nios2_split_symbolic_memory_operand): Declare.
11070 * config/nios2/nios2.c: Adjust includes.
11071 (nios2_symbolic_constant_allowed): New.
11072 (nios2_symbolic_constant_p): New.
11073 (nios2_plus_symbolic_constant_p): New.
11074 (nios2_valid_addr_expr_p): Recognize addresses involving
11075 symbolic constants.
11076 (nios2_legitimate_address_p): Likewise, also LO_SUM.
11077 (nios2_symbolic_memory_operand_p): New.
11078 (nios2_large_constant_p): New.
11079 (nios2_split_large_constant): New.
11080 (nios2_split_plus_large_constant): New.
11081 (nios2_split_symbolic_memory_operand): New.
11082 (nios2_legitimize_address): Code refactoring. Handle addresses
11083 involving symbolic constants.
11084 (nios2_emit_move_sequence): Likewise.
11085 (nios2_print_operand): Improve error output.
11086 (nios2_print_operand_address): Handle LO_SUM.
11087 (nios2_cdx_narrow_form_p): Likewise.
11088 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
11089 operands involving symbolic constants.
11090 (movhi_internal, movsi_internal): Likewise.
11091 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
11092 (extendhisi2, extendqi<mode>2): Likewise.
11094 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11096 * tree-pass.h (PROP_rtl_split_insns): Define.
11097 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
11099 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11101 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
11103 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11106 * target.def (const_not_ok_for_debug_p): Default to
11107 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
11108 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
11109 * targhooks.c (default_const_not_ok_for_debug_p): New function.
11110 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
11111 which targetm.const_not_ok_for_debug_p returned true.
11112 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
11114 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
11116 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
11117 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
11118 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
11120 (ix86_const_not_ok_for_debug_p): New function.
11121 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
11122 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
11124 2017-10-23 David Malcolm <dmalcolm@redhat.com>
11127 * system.h: Conditionally include "unique-ptr.h" if
11128 INCLUDE_UNIQUE_PTR is defined.
11129 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
11130 of defining INCLUDE_UNIQUE_PTR before including "system.h".
11132 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
11134 * config/rl78/rl78.md: New define_expand "subdi3".
11136 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
11139 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
11140 DF_REF_INSN if DF_REF_INSN_INFO is false.
11142 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
11144 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
11145 xmm_move instead of sse_move.
11146 (sse_store_index): New function.
11147 (ix86_register_move_cost): Be more sensible about mismatch stall;
11148 model AVX moves correctly; make difference between sse->integer and
11150 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
11151 moves; make difference between SSE and AVX.
11152 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
11153 and zmm_move. Increase size of sse load and store tables;
11154 add unaligned load and store tables; add ssemmx_to_integer.
11155 * x86-tune-costs.h: Update all entries according to real
11156 move latencies from Agner Fog's manual and chip documentation.
11158 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11161 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
11162 * config/i386/constraints.md (Wf): New constraint.
11163 * config/i386/i386.md (UNSPEC_SBB): New unspec.
11164 (cmp<dwi>_doubleword): Removed.
11165 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
11166 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
11167 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
11168 expand with cmp<dwi>_doubleword. For LTU and GEU use
11169 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
11171 * common.opt (gcolumn-info): Enable by default.
11172 * doc/invoke.texi (gcolumn-info): Document new default.
11174 2017-10-23 Richard Biener <rguenther@suse.de>
11176 PR tree-optimization/82672
11177 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
11178 Fold the stmt if we propagated into it.
11180 2017-10-23 Richard Biener <rguenther@suse.de>
11182 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
11183 (bitmap_remove_expr_from_set): ... this. All callers call this
11184 for non-constant values.
11185 (bitmap_set_subtract): Rename to...
11186 (bitmap_set_subtract_expressions): ... this. Adjust and
11188 (bitmap_set_contains_value): Remove superfluous check.
11189 (bitmap_set_replace_value): Inline into single caller ...
11190 (bitmap_value_replace_in_set): ... here and simplify.
11191 (dependent_clean): Merge into ...
11192 (clean): ... this using an overload. Adjust.
11193 (prune_clobbered_mems): Adjust.
11194 (compute_antic_aux): Likewise.
11195 (compute_partial_antic_aux): Likewise.
11197 2017-10-23 Richard Biener <rguenther@suse.de>
11199 PR tree-optimization/82129
11201 2017-08-01 Richard Biener <rguenther@suse.de>
11203 PR tree-optimization/81181
11204 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11205 (compute_antic): ... end of iteration here.
11207 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11209 * target.def (starting_frame_offset): New hook.
11210 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
11211 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
11212 * doc/tm.texi.in: Regenerate.
11213 * hooks.h (hook_hwi_void_0): Declare.
11214 * hooks.c (hook_hwi_void_0): New function.
11215 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
11216 STARTING_FRAME_OFFSET.
11217 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
11218 * reload1.c (reload): Likewise.
11219 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
11220 instead of STARTING_FRAME_OFFSET.
11221 * function.c (try_fit_stack_local): Likewise.
11222 (assign_stack_local_1): Likewise
11223 (instantiate_virtual_regs): Likewise.
11224 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11225 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11226 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11227 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11228 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11229 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11230 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11231 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11232 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11233 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11234 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11235 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11236 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11237 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11238 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11239 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11240 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11241 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11242 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11243 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11244 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11245 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11246 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11247 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11248 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11249 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11250 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11251 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11252 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11253 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11254 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11255 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11256 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11257 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11258 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11259 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11260 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11261 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11262 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11263 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11264 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11265 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11266 return a HOST_WIDE_INT.
11267 (avr_builtin_setjmp_frame_value): Use it instead of
11268 STARTING_FRAME_OFFSET.
11269 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11270 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11271 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11273 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11274 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11275 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11276 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11277 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11278 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11279 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11280 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11281 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11282 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11283 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11284 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11286 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11287 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11288 * config/mips/mips.c (mips_compute_frame_info): Refer to
11289 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11290 (mips_starting_frame_offset): New function.
11291 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11292 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11293 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11294 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11295 and return a HOST_WIDE_INT.
11296 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11297 (mmix_initial_elimination_offset): Refer to
11298 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11299 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11300 * config/pa/pa.c (pa_starting_frame_offset): New function.
11301 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11302 (pa_expand_prologue): Likewise.
11303 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11304 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11305 !FRAME_GROWS_DOWNWARD handling to...
11306 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11307 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11308 !FRAME_GROWS_DOWNWARD handling to...
11309 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11310 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11311 !FRAME_GROWS_DOWNWARD handling to...
11312 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11313 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11315 (rs6000_starting_frame_offset): New function.
11316 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11317 !FRAME_GROWS_DOWNWARD handling to...
11318 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11319 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11320 !FRAME_GROWS_DOWNWARD handling to...
11321 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11322 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11323 !FRAME_GROWS_DOWNWARD handling to...
11324 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11325 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11326 (rs6000_starting_frame_offset): New function.
11327 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11328 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11329 * config/vax/vax.c (vax_starting_frame_offset): New function.
11330 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11331 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11332 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11333 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11334 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11335 * system.h (STARTING_FRAME_OFFSET): Poison.
11337 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11339 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11340 SCALAR_TYPE_MODE instead of TYPE_MODE.
11342 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11343 Alan Hayward <alan.hayward@arm.com>
11344 David Sherwood <david.sherwood@arm.com>
11346 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11348 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11349 Alan Hayward <alan.hayward@arm.com>
11350 David Sherwood <david.sherwood@arm.com>
11352 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11354 2017-10-23 Richard Biener <rguenther@suse.de>
11356 PR tree-optimization/82129
11357 * tree-ssa-pre.c (bitmap_set_and): Remove.
11358 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11359 canonicalizing expressions in the set to those with lowest
11360 ID rather than taking that from the first edge.
11362 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11364 * combine.c (rtx_equal_for_field_assignment_p): Use
11365 byte_lowpart_offset.
11367 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11368 Alan Hayward <alan.hayward@arm.com>
11369 David Sherwood <david.sherwood@arm.com>
11371 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11372 to a SUBREG_PROMOTED_VAR.
11374 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11375 Alan Hayward <alan.hayward@arm.com>
11376 David Sherwood <david.sherwood@arm.com>
11378 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11379 (expand_debug_source_expr): Likewise.
11380 * combine.c (combine_simplify_rtx): Likewise.
11381 * cse.c (fold_rtx): Likewise.
11382 * optabs.c (expand_float): Likewise.
11383 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11384 (simplify_binary_operation_1): Likewise.
11386 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11387 Alan Hayward <alan.hayward@arm.com>
11388 David Sherwood <david.sherwood@arm.com>
11390 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11391 (record_promoted_value): Likewise.
11392 * expr.c (expand_expr_real_2): Likewise.
11393 * ree.c (update_reg_equal_equiv_notes): Likewise.
11394 (combine_set_extension): Likewise.
11395 * rtlanal.c (low_bitmask_len): Likewise.
11396 * simplify-rtx.c (neg_const_int): Likewise.
11397 (simplify_binary_operation_1): Likewise.
11399 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11400 Alan Hayward <alan.hayward@arm.com>
11401 David Sherwood <david.sherwood@arm.com>
11403 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11404 * regcprop.c (maybe_mode_change): Likewise.
11405 * reload1.c (alter_reg): Likewise.
11407 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11409 * inchash.h (inchash::hash::add_wide_int): New function.
11410 * lto-streamer-out.c (hash_tree): Use it.
11412 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11414 * inchash.h (inchash::hash::add_wide_int): Rename to...
11415 (inchash::hash::add_hwi): ...this.
11416 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11417 (polymorphic_call_target_hasher::hash): Likewise.
11418 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11419 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11420 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11421 * lto-streamer-out.c (hash_tree): Likewise.
11422 * optc-save-gen.awk: Likewise.
11423 * tree.c (add_expr): Likewise.
11425 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11428 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11429 for ordered inequality comparisons even with TARGET_IEEE_FP.
11431 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11434 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11435 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11436 Expand with cmp<dwi>_doubleword.
11438 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11440 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11441 List CET intrinsics.
11442 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11443 specific to -fcf-protection option.
11445 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11447 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11448 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11449 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11450 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11451 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11452 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11453 (extra_objs): Add cet.o for Linux/x86 targets.
11454 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11455 * config/i386/cet.c: New file.
11456 * config/i386/cetintrin.h: Likewise.
11457 * config/i386/t-cet: Likewise.
11458 * config/i386/cpuid.h (bit_SHSTK): New.
11459 (bit_IBT): Likewise.
11460 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11461 pass IBT and SHSTK bits.
11462 * config/i386/i386-builtin-types.def
11463 (VOID_FTYPE_UNSIGNED_PVOID): New.
11464 (VOID_FTYPE_UINT64_PVOID): Likewise.
11465 * config/i386/i386-builtin.def: Add CET intrinsics.
11466 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11467 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11468 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11469 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11471 * config/i386/i386.c (rest_of_insert_endbranch): New.
11472 (pass_data_insert_endbranch): Likewise.
11473 (pass_insert_endbranch): Likewise.
11474 (make_pass_insert_endbranch): Likewise.
11475 (ix86_notrack_prefixed_insn_p): Likewise.
11476 (ix86_target_string): Add -mibt, -mshstk flags.
11477 (ix86_option_override_internal): Add flag_cf_protection
11479 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11480 (ix86_print_operand): Add 'notrack' prefix output.
11481 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11482 (ix86_expand_builtin): Expand CET intrinsics.
11483 (x86_output_mi_thunk): Add 'endbranch' instruction.
11484 * config/i386/i386.h (TARGET_IBT): New.
11485 (TARGET_IBT_P): Likewise.
11486 (TARGET_SHSTK): Likewise.
11487 (TARGET_SHSTK_P): Likewise.
11488 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11489 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11490 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11491 (builtin_setjmp_setup): New pattern.
11492 (builtin_longjmp): Likewise.
11493 (rdssp<mode>): Likewise.
11494 (incssp<mode>): Likewise.
11495 (saveprevssp): Likewise.
11496 (rstorssp): Likewise.
11497 (wrss<mode>): Likewise.
11498 (wruss<mode>): Likewise.
11499 (setssbsy): Likewise.
11500 (clrssbsy): Likewise.
11501 (nop_endbr): Likewise.
11502 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11504 * config/i386/immintrin.h: Include <cetintrin.h>.
11505 * config/i386/linux-common.h
11506 (file_end_indicate_exec_stack_and_cet): New prototype.
11507 (TARGET_ASM_FILE_END): New.
11509 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11511 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11512 latencies instead of having separate table; make difference between
11513 integer and float costs.
11514 * i386.h (processor_costs): Remove scalar_stmt_cost,
11515 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11516 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11518 * x86-tune-costs.h: Remove entries which has been removed in
11519 procesor_costs from all tables; make cond_taken_branch_cost
11520 and cond_not_taken_branch_cost COST_N_INSNS based.
11522 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11524 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11526 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11528 * config/i386/i386.md (isa): Remove fma_avx512f.
11529 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11530 <avx512>_fmadd_<mode>_mask3<round_name>,
11531 <avx512>_fmsub_<mode>_mask<round_name>,
11532 <avx512>_fmsub_<mode>_mask3<round_name>,
11533 <avx512>_fnmadd_<mode>_mask<round_name>,
11534 <avx512>_fnmadd_<mode>_mask3<round_name>,
11535 <avx512>_fnmsub_<mode>_mask<round_name>,
11536 <avx512>_fnmsub_<mode>_mask3<round_name>,
11537 <avx512>_fmaddsub_<mode>_mask<round_name>,
11538 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11539 <avx512>_fmsubadd_<mode>_mask<round_name>,
11540 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11541 (*vec_widen_umult_even_v16si<mask_name>,
11542 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11543 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11545 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11547 * extend.texi: Add 'nocf_check' documentation.
11548 * gimple.texi: Add second parameter to
11549 gimple_build_call_from_tree.
11550 * invoke.texi: Add -fcf-protection documentation.
11551 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11553 2017-10-20 Richard Biener <rguenther@suse.de>
11555 PR tree-optimization/82473
11556 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11557 the largest input type.
11559 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11561 * c-attribs.c (handle_nocf_check_attribute): New function.
11562 (c_common_attribute_table): Add 'nocf_check' handling.
11563 * gimple-parser.c: Add second argument NULL to
11564 gimple_build_call_from_tree.
11565 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11566 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11568 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11569 * common.opt: Add fcf-protection flag.
11570 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11571 * flag-types.h: Add enum cf_protection_level.
11572 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11573 Add 'nocf_check' attribute propagation to gimple call.
11574 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11575 (gimple_build_call_from_tree): Update prototype.
11576 (gimple_call_nocf_check_p): New function.
11577 (gimple_call_set_nocf_check): Likewise.
11578 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11579 * ipa-icf.c: Add nocf_check attribute in statement hash.
11580 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11581 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11582 * toplev.c (process_options): Add flag_cf_protection handling.
11584 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11586 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11588 2017-10-20 Richard Biener <rguenther@suse.de>
11590 PR tree-optimization/82603
11591 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11592 remove false predicated stores.
11594 2017-10-20 Richard Biener <rguenther@suse.de>
11596 * graphite-isl-ast-to-gimple.c
11597 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11598 Remove return value and simplify, dump copied stmt after lhs
11600 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11601 Reduce dump verbosity.
11602 (gsi_insert_earliest): Likewise.
11603 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11604 * graphite.c (print_global_statistics): Adjust dumping.
11605 (print_graphite_scop_statistics): Likewise.
11606 (print_graphite_statistics): Do not dump loops here.
11607 (graphite_transform_loops): But here.
11609 2017-10-20 Nicolas Roche <roche@adacore.com>
11611 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11612 * configure: Regenerate.
11614 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11617 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11618 functions when optimizing replace GIMPLE_RETURN stmts with
11619 calls to __builtin_unreachable ().
11622 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11623 for -fsanitize=thread link of executables.
11624 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11625 link of executables.
11628 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11629 New mode iterators.
11630 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11631 define_insns for logical vector shifts to use VI248_AVX512BW
11632 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11633 condition, useless isa and prefix attributes. Change the first
11634 2 of these define_insns to ...
11635 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11636 define_insn for avx512vl.
11637 (<shift_insn><mode>3): ... and this, new define_insn without
11638 masking for non-avx512vl.
11641 * config/i386/sse.md (*andnot<mode>3,
11642 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11643 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11644 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11645 not applied use empty suffix even for TARGET_AVX512VL.
11646 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11647 is applied, supply evex,evex or evex,evex,evex instead of just
11650 2017-10-20 Julia Koval <julia.koval@intel.com>
11652 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11653 (OPTION_MASK_ISA_GFNI_UNSET): New.
11654 (ix86_handle_option): Handle OPT_mgfni.
11655 * config/i386/cpuid.h (bit_GFNI): New.
11656 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11657 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11658 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11659 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11660 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11661 * config/i386/i386.opt: Add mgfni.
11663 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11665 * config/msp430/msp430.c (msp430_option_override): Disable
11666 -fdelete-null-pointer-checks.
11667 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11669 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11671 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11672 of x87 and SSE instructions.
11674 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11676 * asan.c (create_cond_insert_point): Do not update edge count.
11677 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11678 (afdo_propagate_circuit): Likewise.
11679 (afdo_calculate_branch_prob): Likewise.
11680 (afdo_annotate_cfg): Likewise.
11681 * basic-block.h (struct edge_def): Remove count.
11682 (edge_def::count): New accessor.
11683 * bb-reorder.c (rotate_loop): Update.
11684 (find_traces_1_round): Update.
11685 (connect_traces): Update.
11686 (sanitize_hot_paths): Update.
11687 * cfg.c (unchecked_make_edge): Update.
11688 (make_single_succ_edge): Update.
11689 (check_bb_profile): Update.
11690 (dump_edge_info): Update.
11691 (update_bb_profile_for_threading): Update.
11692 (scale_bbs_frequencies_int): Update.
11693 (scale_bbs_frequencies_gcov_type): Update.
11694 (scale_bbs_frequencies_profile_count): Update.
11695 (scale_bbs_frequencies): Update.
11696 * cfganal.c (connect_infinite_loops_to_exit): Update.
11697 * cfgbuild.c (compute_outgoing_frequencies): Update.
11698 (find_many_sub_basic_blocks): Update.
11699 * cfgcleanup.c (try_forward_edges): Update.
11700 (try_crossjump_to_edge): Update
11701 * cfgexpand.c (expand_gimple_cond): Update
11702 (expand_gimple_tailcall): Update
11703 (construct_exit_block): Update
11704 * cfghooks.c (verify_flow_info): Update
11705 (redirect_edge_succ_nodup): Update
11706 (split_edge): Update
11707 (make_forwarder_block): Update
11708 (duplicate_block): Update
11709 (account_profile_record): Update
11710 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11711 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11712 * cfgloopmanip.c (scale_loop_profile): Update.
11714 (lv_adjust_loop_entry_edge): Update.
11715 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11716 (force_nonfallthru_and_redirect): Update.
11717 (purge_dead_edges): Update.
11718 (rtl_flow_call_edges_add): Update.
11719 * cgraphunit.c (init_lowered_empty_function): Update.
11720 (cgraph_node::expand_thunk): Update.
11721 * gimple-pretty-print.c (dump_probability): Update.
11722 (dump_edge_probability): Update.
11723 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11724 * haifa-sched.c (sched_create_recovery_edges): Update.
11725 * hsa-gen.c (convert_switch_statements): Update.
11726 * ifcvt.c (dead_or_predicable): Update.
11727 * ipa-inline-transform.c (inline_transform): Update.
11728 * ipa-split.c (split_function): Update.
11729 * ipa-utils.c (ipa_merge_profiles): Update.
11730 * loop-doloop.c (add_test): Update.
11731 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11732 * lto-streamer-in.c (input_cfg): Update.
11733 (input_function): Update.
11734 * lto-streamer-out.c (output_cfg): Update.
11735 * modulo-sched.c (sms_schedule): Update.
11736 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11737 * predict.c (maybe_hot_edge_p): Update.
11738 (unlikely_executed_edge_p): Update.
11739 (probably_never_executed_edge_p): Update.
11740 (dump_prediction): Update.
11741 (drop_profile): Update.
11742 (propagate_unlikely_bbs_forward): Update.
11743 (determine_unlikely_bbs): Update.
11744 (force_edge_cold): Update.
11745 * profile.c (compute_branch_probabilities): Update.
11746 * reg-stack.c (better_edge): Update.
11747 * shrink-wrap.c (handle_simple_exit): Update.
11748 * tracer.c (better_p): Update.
11749 * trans-mem.c (expand_transaction): Update.
11750 (split_bb_make_tm_edge): Update.
11751 * tree-call-cdce.c: Update.
11752 * tree-cfg.c (gimple_find_sub_bbs): Update.
11753 (gimple_split_edge): Update.
11754 (gimple_duplicate_sese_region): Update.
11755 (gimple_duplicate_sese_tail): Update.
11756 (gimple_flow_call_edges_add): Update.
11757 (insert_cond_bb): Update.
11758 (execute_fixup_cfg): Update.
11759 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11760 * tree-complex.c (expand_complex_div_wide): Update.
11761 * tree-eh.c (lower_resx): Update.
11762 (unsplit_eh): Update.
11763 (cleanup_empty_eh_move_lp): Update.
11764 * tree-inline.c (copy_edges_for_bb): Update.
11765 (freqs_to_counts): Update.
11766 (copy_cfg_body): Update.
11767 * tree-ssa-dce.c (remove_dead_stmt): Update.
11768 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11769 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11770 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11771 (unloop_loops): Update.
11772 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11773 * tree-ssa-loop-split.c (connect_loops): Update.
11774 (split_loop): Update.
11775 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11776 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11777 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11778 * tree-ssa-reassoc.c (branch_fixup): Update.
11779 * tree-ssa-tail-merge.c (replace_block_by): Update.
11780 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11781 (compute_path_counts): Update.
11782 (update_profile): Update.
11783 (recompute_probabilities): Update.
11784 (update_joiner_offpath_counts): Update.
11785 (estimated_freqs_path): Update.
11786 (freqs_to_counts_path): Update.
11787 (clear_counts_path): Update.
11788 (ssa_fix_duplicate_block_edges): Update.
11789 (duplicate_thread_path): Update.
11790 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11791 (case_bit_test_cmp): Update.
11792 (collect_switch_conv_info): Update.
11793 (gen_inbound_check): Update.
11794 (do_jump_if_equal): Update.
11795 (emit_cmp_and_jump_insns): Update.
11796 * tree-tailcall.c (decrease_profile): Update.
11797 (eliminate_tail_call): Update.
11798 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11799 (vect_do_peeling): Update.
11800 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11801 * ubsan.c (ubsan_expand_null_ifn): Update.
11802 (ubsan_expand_ptr_ifn): Update.
11803 * value-prof.c (gimple_divmod_fixed_value): Update.
11804 (gimple_mod_pow2): Update.
11805 (gimple_mod_subtract): Update.
11806 (gimple_ic): Update.
11807 (gimple_stringop_fixed_value): Update.
11809 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11812 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11814 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11816 PR rtl-optimization/82395
11817 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11818 based on non_spilled_static_chain_regno_p.
11820 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11822 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11823 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11824 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11826 2017-10-19 Martin Sebor <msebor@redhat.com>
11828 PR tree-optimization/82596
11829 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11831 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11833 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11834 (get_mem_refs_of_builtin_call): Likewise.
11835 * builtins.c (expand_builtin_apply): Adjust call to
11836 allocate_dynamic_stack_space.
11837 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11838 the third argument to allocate_dynamic_stack_space, otherwise -1.
11839 (expand_builtin): Deal with all alloca variants.
11840 (is_inexpensive_builtin): Likewise.
11841 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11842 * calls.c (special_function_p): Deal with all alloca variants.
11843 (initialize_argument_information): Adjust call to
11844 allocate_dynamic_stack_space.
11845 (expand_call): Likewise.
11846 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11847 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11848 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11849 use it for the stack usage computation.
11850 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11851 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11852 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11853 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11854 (in_loop_p): Remove first argument and useless check.
11855 (pass_walloca::execute): Remove useless test and adjust call to above.
11856 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11857 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11858 (gimplify_call_expr): Deal with all alloca variants.
11859 * hsa-gen.c (gen_hsa_alloca): Likewise.
11860 (gen_hsa_insns_for_call): Likewise.
11861 * ipa-pure-const.c (special_builtin_state): Likewise.
11862 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11863 * tree-object-size.c (alloc_object_size): Likewise.
11864 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11865 (call_may_clobber_ref_p_1): Likewise.
11866 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11867 (ccp_fold_stmt): Likewise.
11868 (optimize_stack_restore): Likewise.
11869 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11870 (mark_all_reaching_defs_necessary_1): Likewise.
11871 (propagate_necessity): Likewise.
11872 (eliminate_unnecessary_stmts): Likewise.
11873 * tree.c (build_common_builtin_nodes): Build
11874 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11875 (build_alloca_call_expr): New function.
11876 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11877 (CASE_BUILT_IN_ALLOCA): Likewise.
11878 (build_alloca_call_expr): Declare.
11879 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11881 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11884 * dwarf2out.c (new_die_raw): New static inline function.
11885 (new_die): Use it to create the DIE.
11886 (add_AT_external_die_ref): Likewise.
11887 (clone_die): Likewise.
11888 (clone_as_declaration): Likewise.
11889 (dwarf2out_vms_debug_main_pointer): Likewise.
11890 (base_type_die): Likewise. Remove early return for corner cases.
11891 Do not call add_pubtype on the DIE here.
11892 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11893 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11894 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11895 native order exists for base types, attach the DIE manually and call
11896 add_pubtype on it. Do not equate a reverse order DIE to the type.
11898 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11900 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11901 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11903 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11905 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11907 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11908 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11909 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11910 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11911 * opts.c (sanitizer_opts): Add builtin.
11912 * ubsan.c (instrument_builtin): New function.
11913 (pass_ubsan::execute): Call it.
11914 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11915 * doc/invoke.texi: Document -fsanitize=builtin.
11917 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11918 builtins, store max (log2 (align), 0) into uchar field instead of
11919 align into uptr field.
11920 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11921 store uchar 0 field instead of uptr 0 field.
11922 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11923 instead of passing one address of struct with 2 locations pass
11924 two addresses of structs with 1 location each.
11925 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11926 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11927 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11928 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11929 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11930 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11931 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11932 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11934 2017-10-19 Martin Liska <mliska@suse.cz>
11937 * file-find.c (remove_prefix): Remove.
11938 * file-find.h (remove_prefix): Likewise.
11939 * gcc-ar.c: Remove smartness of lookup.
11941 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11943 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11944 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11945 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11947 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11950 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11951 (setcc + and to xor + setcc): New peephole2.
11953 2017-10-19 Tom de Vries <tom@codesourcery.com>
11955 * doc/sourcebuild.texi (Test Directives, Variants of
11956 dg-require-support): Add dg-require-stack-size.
11958 2017-10-19 Martin Liska <mliska@suse.cz>
11961 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11962 that have a large alignment.
11963 (gimplify_target_expr): Likewise.
11965 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11967 PR rtl-optimization/82602
11968 * ira.c (rtx_moveable_p): Return false for volatile asm.
11970 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11973 * config/i386/i386-modes.def (CCGZ): New CC mode.
11974 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11975 * config/i386/predicates.md (ix86_comparison_operator):
11977 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11978 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11979 with double-word subtraction.
11980 (put_condition_code): Handle CCGZmode.
11982 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11984 * wide-int.cc (debug (const wide_int &)): New.
11985 (debug (const wide_int *)): New.
11986 (debug (const widest_int &)): New.
11987 (debug (const widest_int *)): New.
11989 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11991 PR middle-end/82556
11992 * lra-constraints.c (curr_insn_transform): Use non-input operand
11993 instead of output one for matched reload.
11995 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11997 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11998 (tree-ssa-loop-ivopts.h): New header file.
11999 (struct builtin_info): New fields.
12000 (classify_builtin_1): Compute and record base and offset parts for
12001 memset builtin partition by calling strip_offset.
12002 (offset_cmp, fuse_memset_builtins): New functions.
12003 (finalize_partitions): Fuse adjacent memset partitions by calling
12005 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
12006 Expose the interface.
12007 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
12009 2017-10-18 Bin Cheng <bin.cheng@arm.com>
12011 PR tree-optimization/82574
12012 * tree-loop-distribution.c (find_single_drs): New parameter. Check
12013 that data reference must be executed exactly once per iteration
12014 against the outermost loop in nest.
12015 (classify_partition): Update call to above function.
12017 2017-10-18 Richard Biener <rguenther@suse.de>
12019 PR tree-optimization/82591
12020 * graphite.c (graphite_transform_loops): Move code gen message
12022 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12023 Here. Handle scop_to_isl_ast failing.
12024 (scop_to_isl_ast): Limit the number of ISL operations.
12026 2017-10-18 Richard Biener <rguenther@suse.de>
12028 * graphite-isl-ast-to-gimple.c
12029 (translate_isl_ast_to_gimple::set_rename): Simplify.
12030 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
12031 (graphite_copy_stmts_from_block): ... here.
12032 (copy_bb_and_scalar_dependences): Simplify.
12033 (add_parameters_to_ivs_params): Canonicalize.
12034 (generate_entry_out_of_ssa_copies): Simplify.
12035 * graphite-sese-to-poly.c (extract_affine_name): Simplify
12036 by passing in ISL dimension.
12037 (parameter_index_in_region_1): Rename to ...
12038 (parameter_index_in_region): ... this.
12039 (extract_affine): Adjust assert, pass down parameter index.
12040 (add_param_constraints): Use range-info when available.
12041 (build_scop_context): Adjust.
12042 * sese.c (new_sese_info): Adjust.
12043 (free_sese_info): Likewise.
12044 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
12045 Remove unused typedefs.
12046 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
12048 2017-10-18 Martin Liska <mliska@suse.cz>
12050 * combine.c (simplify_compare_const): Add gcc_fallthrough.
12052 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12054 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
12055 (s390_sched_init): Do not reset s390_sched_state if we entered the
12056 current basic block via a fallthru edge and all others are unlikely.
12058 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12060 * config/s390/s390.c (NUM_SIDES): New variable.
12061 (LONGRUNNING_THRESHOLD): New variable.
12062 (LATENCY_FACTOR): New variable.
12063 (s390_sched_score): Decrease score for long-running instructions on
12065 (s390_sched_variable_issue): Perform bookkeeping for long-running
12068 2017-10-18 Richard Biener <rguenther@suse.de>
12070 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
12071 Simplify with removal of the parameter rename map.
12072 (set_rename): Likewise.
12073 (should_copy_to_new_region): Likewise.
12074 (graphite_copy_stmts_from_block): Likewise.
12075 (copy_bb_and_scalar_dependences): Remove initialization of
12076 unused copied_bb_map.
12077 (copy_def): Remove.
12078 (copy_internal_parameters): Likewise.
12079 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
12080 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
12081 Use INTEGRAL_TYPE_P.
12082 (parameter_index_in_region_1): Rename to ...
12083 (assign_parameter_index_in_region): ... this. Assert we have
12084 a parameter we handle.
12085 (scan_tree_for_params): Adjust.
12086 * sese.h (parameter_rename_map_t): Remove.
12087 (struct sese_info_t): Remove unused parameter_rename_map and
12088 copied_bb_map members.
12089 * sese.c (new_sese_info): Adjust.
12090 (free_sese_info): Likewise.
12092 2017-10-18 Martin Liska <mliska@suse.cz>
12095 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
12096 on an abnormal edge.
12098 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
12100 * doc/invoke.texi (ffunction-sections and fdata-sections):
12103 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12105 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
12106 the use statement can throw internally.
12108 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12110 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
12111 any RTX present on the RHS of a SET.
12112 * compare-elim.c (try_eliminate_compare): Restore comment.
12114 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12116 * langhooks.h (struct lang_hooks): Document that tree_size langhook
12117 may be also called on tcc_type nodes.
12118 * langhooks.c (lhd_tree_size): Likewise.
12120 2017-10-17 David Malcolm <dmalcolm@redhat.com>
12122 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
12123 format_warning_at_substring.
12124 (maybe_warn): Convert source_range * param to a location_t. Pass
12125 UNKNOWN_LOCATION rather than NULL to fmtwarn.
12126 (format_directive): Remove code to extract source_ranges and
12127 source_range * in favor of just a location_t.
12128 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
12130 * substring-locations.c (format_warning_va): Convert
12131 source_range * param to a location_t.
12132 (format_warning_at_substring): Likewise.
12133 * substring-locations.h (format_warning_va): Likewise.
12134 (format_warning_at_substring): Likewise.
12136 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
12138 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
12140 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
12141 and scatter/gather ops.
12143 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
12144 vec_gather_load and vec_scatter_store.
12145 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
12146 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
12147 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
12148 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
12149 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
12150 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
12152 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
12154 * reg-stack.c (compare_for_stack_reg): Add bool argument.
12155 Detect FTST instruction and handle its register pops. Only pop
12156 second operand if can_pop_second_op is true.
12157 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
12158 set can_pop_second_op to false in the compare_for_stack_reg call.
12160 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
12161 output_fp_compare for stack register operands.
12162 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
12163 instructions here. Do not emit stack register pops here. Assert
12164 that FCOMPP pops next to top stack register. Rewrite function.
12166 2017-10-17 Nathan Sidwell <nathan@acm.org>
12168 PR middle-end/82577
12169 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
12170 use DECL_ASSEMBLER_NAME_RAW.
12172 PR middle-end/82546
12173 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
12176 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
12177 Wilco Dijkstra <wilco.dijkstra@arm.com>
12179 * builtins.c (expand_builtin_update_setjmp_buf): Add a
12180 converstion to Pmode from the buf_addr.
12182 2017-10-17 Richard Biener <rguenther@suse.de>
12184 * graphite-dependences.c (scop_get_reads_and_writes): Change
12185 output parameters to references.
12187 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
12189 PR 71026/tree-optimization
12190 * fold-const.c (distribute_real_division): Removed.
12191 (fold_binary_loc): Remove calls to distribute_real_divison.
12193 2017-10-17 Richard Biener <rguenther@suse.de>
12195 * graphite-scop-detection.c
12196 (scop_detection::stmt_has_simple_data_refs_p): Always use
12197 the full nest as region.
12198 (try_generate_gimple_bb): Likewise.
12199 * sese.c (scalar_evolution_in_region): Simplify now that
12200 SCEV can handle instantiation in regions.
12201 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
12202 in the non-loop part of a function if requested.
12204 2017-10-17 Richard Biener <rguenther@suse.de>
12206 PR tree-optimization/82563
12207 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
12209 (graphite_regenerate_ast_isl): Call it.
12210 * graphite-scop-detection.c (build_scops): Remove entry edge split.
12212 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12214 PR tree-optimization/82549
12215 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
12216 Formatting fixes. Instead of calling make_bit_field_ref with negative
12219 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
12221 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
12222 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12223 _mm_maskz_reduce_ss): New.
12224 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12225 __builtin_ia32_reducess_mask): Ditto..
12226 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12227 * config/i386/sse.md (reduces<mode>): Renamed to ...
12228 (reduces<mode><mask_scalar_name>): ... this.
12229 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12231 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12232 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12234 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12236 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12237 * selftest-run-tests.c (selftest::run_tests): Call
12238 selftest::unique_ptr_tests_cc_tests.
12239 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12240 * unique-ptr-tests.cc: New file.
12242 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12245 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12247 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12248 (make_hard_regno_born, make_hard_regno_dead): Update
12249 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12251 2017-10-16 Jeff Law <law@redhat.com>
12253 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12255 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12257 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12259 2017-10-16 Olivier Hainque <hainque@adacore.com>
12261 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12262 with_cpu if we were configured for an e500v2 target cpu name.
12264 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12266 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12267 * doc/invoke.texi: Document +nodsp as a valid extension for
12270 2017-10-16 Martin Liska <mliska@suse.cz>
12272 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12273 (test_set_range): Likewise.
12274 (test_range_functions): Rename to ...
12275 (test_bit_in_range): ... this.
12276 (sbitmap_c_tests): Add new test.
12278 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12280 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12282 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12283 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12285 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12287 * config/aarch64/aarch64-builtins.c
12288 (aarch64_types_quadopu_lane_qualifiers): New.
12289 (TYPES_QUADOPU_LANE): New.
12290 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12291 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12292 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12293 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12294 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12295 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12296 (Vdottype, DOTPROD): New.
12297 (sur): Add SDOT and UDOT.
12299 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12301 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12302 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12303 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12304 Add TARGET_DOTPROD.
12305 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12306 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12307 Enable TARGET_DOTPROD.
12308 (cortex-a75.cortex-a55): Likewise.
12309 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12311 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12313 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12314 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12316 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12318 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12319 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12320 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12321 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12322 * config/arm/types.md (neon_dot, neon_dot_q): New.
12323 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12325 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12327 * config/arm/arm.h (TARGET_DOTPROD): New.
12328 * config/arm/arm.c (arm_arch_dotprod): New.
12329 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12330 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12331 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12332 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12333 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12334 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12335 * doc/invoke.texi (armv8.2-a): Document dotprod
12337 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12339 * i386.c (ix86_vec_cost): New function.
12340 (ix86_rtx_costs): Handle vector operations better.
12341 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12342 * x86-tune-costs.h: Add new costs to all tables.
12344 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12346 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12348 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12349 divsd, sqrtss and sqrtsd
12350 * x86-tune-costs.h: Add new entries to all costs.
12351 (znver1_cost): Fix to match real instruction latencies.
12353 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12354 Michael Collison <michael.collison@arm.com>
12356 * compare-elim.c: Include emit-rtl.h.
12357 (can_merge_compare_into_arith): New function.
12358 (try_validate_parallel): Likewise.
12359 (try_merge_compare): Likewise.
12360 (try_eliminate_compare): Call the above when no previous clobber
12362 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12365 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12367 PR middle-end/62263
12368 PR middle-end/82498
12369 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12370 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12372 PR middle-end/62263
12373 PR middle-end/82498
12374 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12375 to be any operand_equal_p operands. For & (B - 1) require
12376 B to be power of 2. Recognize
12377 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12379 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12382 * optabs.c (expand_memory_blockage): Fix call of
12383 targetm.have_memory_blockage.
12385 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12388 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12389 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12390 objects to extra_objs instead of overwriting it.
12392 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12394 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12395 Use any_fp_register_operand as operand[3] predicate. Simplify
12396 equality test for operands[2] and operands[4] memory location.
12397 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12398 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12399 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12400 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12401 any_fp_register_operand as operand[1] predicate. Simplify
12402 equality test for operands[0] and operands[3] memory location.
12403 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12404 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12405 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12407 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12409 * target-insns.def: Add memory_blockage.
12410 * optabs.c (expand_memory_blockage): New function.
12411 (expand_asm_memory_barrier): Rename ...
12412 (expand_asm_memory_blockage): ... to this.
12413 (expand_mem_thread_fence): Call expand_memory_blockage
12414 instead of expand_asm_memory_barrier.
12415 (expand_mem_singnal_fence): Ditto.
12416 (expand_atomic_load): Ditto.
12417 (expand_atomic_store): Ditto.
12418 * doc/md.texi (Standard Pattern Names For Generation):
12419 Document memory_blockage instruction pattern.
12421 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12423 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12424 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12425 * config/rl78/rl78.md: New define_expand "adddi3".
12427 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12429 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12432 2017-10-13 Jeff Law <law@redhat.com>
12434 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12436 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12439 * internal-fn.c (expand_mul_overflow): If both operands have
12440 the same highpart of -1 or 0 and the topmost bit of lowpart
12441 is different, overflow is if res <= 0 rather than res < 0.
12443 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12445 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12446 TARGET_P9_VECTOR code for unaligned_load case.
12448 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12450 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12452 2017-10-13 Nathan Sidwell <nathan@acm.org>
12454 * tree-core.h (tree_contains_struct): Make bool.
12455 * tree.c (tree_contains_struct): Likewise.
12456 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12457 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12458 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12459 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12460 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12462 2017-10-13 Richard Biener <rguenther@suse.de>
12464 * graphite-isl-ast-to-gimple.c
12465 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12466 parameters and dominance check.
12467 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12468 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12469 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12470 Do not update SSA form here or do intermediate IL verification.
12471 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12472 (graphite_initialize): Remove check on the number of loops in
12473 the function and inline into graphite_transform_loops.
12474 (graphite_finalize): Inline into graphite_transform_loops.
12475 (graphite_transform_loops): Perform SSA update and IL verification
12477 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12479 2017-10-13 Richard Biener <rguenther@suse.de>
12481 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12482 graphite_expression_type_precision): Avoid global constructor
12484 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12485 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12486 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12487 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12488 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12489 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12491 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12494 * config/i386/i386.h (ix86_red_zone_size): New.
12495 * config/i386/i386.md (push peephole2s): Replace
12496 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12498 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12499 Alan Hayward <alan.hayward@arm.com>
12500 David Sherwood <david.sherwood@arm.com>
12502 * combine.c (can_change_dest_mode): Reject changes in
12503 REGMODE_NATURAL_SIZE.
12505 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12506 Alan Hayward <alan.hayward@arm.com>
12507 David Sherwood <david.sherwood@arm.com>
12509 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12510 (expand_debug_source_expr): Likewise.
12511 * combine.c (combine_simplify_rtx): Likewise.
12512 * cse.c (fold_rtx): Likewise.
12513 * fwprop.c (canonicalize_address): Likewise.
12514 * targhooks.c (default_shift_truncation_mask): Likewise.
12516 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12517 Alan Hayward <alan.hayward@arm.com>
12518 David Sherwood <david.sherwood@arm.com>
12520 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12521 (widened_mode): Likewise.
12522 (expand_unop): Likewise.
12523 * ree.c (transform_ifelse): Likewise.
12524 (merge_def_and_ext): Likewise.
12525 (combine_reaching_defs): Likewise.
12526 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12528 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12529 Alan Hayward <alan.hayward@arm.com>
12530 David Sherwood <david.sherwood@arm.com>
12532 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12533 * combine.c (gen_lowpart_for_combine): Likewise.
12534 * dwarf2out.c (rtl_for_decl_location): Likewise.
12535 * final.c (alter_subreg): Likewise.
12536 * rtlhooks.c (gen_lowpart_general): Likewise.
12537 (gen_lowpart_if_possible): Likewise.
12539 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12540 Alan Hayward <alan.hayward@arm.com>
12541 David Sherwood <david.sherwood@arm.com>
12543 * calls.c (expand_call): Use subreg_lowpart_offset.
12544 * cse.c (cse_insn): Likewise.
12545 * regcprop.c (copy_value): Likewise.
12546 (copyprop_hardreg_forward_1): Likewise.
12548 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12551 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12552 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12553 =Q constraints to +Q and into insn condition add check
12554 that operands[0] and operands[1] are equal.
12555 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12556 =Q constraints to +Q and into insn condition add check
12557 that operands[0] is equal to either operands[1] or operands[2].
12560 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12561 instead of handling MINUS_EXPR twice (once for each argument),
12562 canonicalize operand order and handle just once, use rtype where
12563 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12566 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12567 any values of __C while still being pattern recognizable as a simple
12568 rotate instruction.
12570 2017-10-13 Richard Biener <rguenther@suse.de>
12572 PR tree-optimization/82451
12574 2017-10-02 Richard Biener <rguenther@suse.de>
12576 PR tree-optimization/82355
12577 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12578 a mapping for the enclosing loop but avoid generating one for
12579 the loop tree root.
12580 (copy_bb_and_scalar_dependences): Remove premature codegen
12581 error on PHIs in blocks duplicated into multiple places.
12582 * graphite-scop-detection.c
12583 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12584 in the region use it as loop and nest to analyze the DR in.
12585 (try_generate_gimple_bb): Likewise.
12586 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12587 (add_loop_constraints): For blocks in a loop not in the region
12588 create a dimension with a single iteration.
12589 * sese.h (gbb_loop_at_index): Remove assert.
12591 * cfgloop.c (loop_preheader_edge): For the loop tree root
12592 return the single successor of the entry block.
12593 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12594 Reset the SCEV hashtable and niters.
12595 * graphite-scop-detection.c
12596 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12597 assert that we only have POLYNOMIAL_CHREC that vary in loops
12598 contained in the region.
12599 (scop_detection::graphite_can_represent_expr): Adjust.
12600 (scop_detection::stmt_has_simple_data_refs_p): For loops
12601 not in the region set loop to NULL. The nest is now the
12602 entry edge to the region.
12603 (try_generate_gimple_bb): Likewise.
12604 * sese.c (scalar_evolution_in_region): Adjust for
12605 instantiate_scev change.
12606 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12607 Make nest parameter the edge into the region.
12608 (create_data_ref): Likewise.
12609 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12610 entry edge into a region and adjust instantiate_scev calls.
12611 (create_data_ref): Likewise.
12612 (graphite_find_data_references_in_stmt): Likewise.
12613 (find_data_references_in_stmt): Pass the loop preheader edge
12614 from the nest argument.
12615 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12616 parameter the edge into the region.
12617 (instantiate_parameters): Use the loop preheader edge as entry.
12618 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12620 (get_instantiated_value_entry): Make instantiate_below parameter
12621 the edge into the region.
12622 (instantiate_scev_name): Likewise. Adjust dominance checks,
12623 when we cannot use loop-based instantiation instantiate by
12624 walking use-def chains.
12625 (instantiate_scev_poly): Adjust.
12626 (instantiate_scev_binary): Likewise.
12627 (instantiate_scev_convert): Likewise.
12628 (instantiate_scev_not): Likewise.
12629 (instantiate_array_ref): Remove.
12630 (instantiate_scev_3): Likewise.
12631 (instantiate_scev_2): Likewise.
12632 (instantiate_scev_1): Likewise.
12633 (instantiate_scev_r): Do not blindly handle N-operand trees.
12634 Do not instantiate array-refs. Handle all constants and invariants.
12635 (instantiate_scev): Make instantiate_below parameter
12636 the edge into the region.
12637 (resolve_mixers): Use the loop preheader edge for the region
12638 parameter to instantiate_scev_r.
12639 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12641 2017-10-13 Richard Biener <rguenther@suse.de>
12643 PR tree-optimization/82525
12644 * graphite-isl-ast-to-gimple.c
12645 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12647 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12648 Fail code generation when we cannot represent the isl integer.
12649 (binary_op_to_tree): Elide modulo operations that are no-ops
12650 in the type we code generate. Remove now superfluous code
12653 2017-10-13 Richard Biener <rguenther@suse.de>
12655 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12656 (scop_detection::harmful_loop_in_region): Remove premature
12657 IV type restriction.
12658 (scop_detection::graphite_can_represent_scev): We can handle
12659 pointer IVs just fine.
12661 2017-10-13 Alan Modra <amodra@gmail.com>
12663 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12664 "Clobbers and Scratch Registers". Add paragraph on
12665 alternative to clobbers for scratch registers and OpenBLAS
12668 2017-10-13 Alan Modra <amodra@gmail.com>
12670 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12671 example of a memory input for a string of known length. Move
12672 commentary out of table. Add a number of new examples
12673 covering array memory inputs.
12675 2017-10-12 Martin Liska <mliska@suse.cz>
12677 PR tree-optimization/82493
12678 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12679 (test_range_functions): New function.
12680 (sbitmap_c_tests): Likewise.
12681 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12682 * selftest.h (sbitmap_c_tests): New function.
12684 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12686 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12688 * config/rs6000/amo.h: Fix spacing issue.
12690 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12693 * config/i386/i386.md (*ashl<mode>3_mask_1,
12694 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12695 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12698 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12700 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12701 (profile_probability): Set max_probability
12702 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12704 * profile-count.c (profile_probability::differs_from_p): Do not
12705 rely on max_probaiblity == 10000
12707 2017-10-12 Jeff Law <law@redhat.com>
12709 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12712 2017-10-12 Martin Sebor <msebor@redhat.com>
12716 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12717 (handle_alias_pairs): Call it.
12718 * common.opt (-Wattribute-alias): New option.
12719 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12720 * doc/invoke.texi (-Wattribute-alias): Document.
12722 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12725 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12727 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12729 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12730 (make_hard_regno_born, make_hard_regno_dead): Update
12731 bb_killed_pseudos and bb_gen_pseudos.
12733 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12735 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12737 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12739 * config/alpha/alpha.c (alpha_split_conditional_move):
12740 Use std::swap instead of manually swapping.
12741 (alpha_stdarg_optimize_hook): Ditto.
12742 (alpha_canonicalize_comparison): Ditto.
12744 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12746 * tree-loop-distribution.c (struct builtin_info): New struct.
12747 (struct partition): Refactor fields into struct builtin_info.
12748 (partition_free): Free struct builtin_info.
12749 (build_size_arg_loc, build_addr_arg_loc): Delete.
12750 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12751 information from struct builtin_info.
12752 (find_single_drs): New function refactored from classify_partition.
12753 Also moved builtin validity checks to this function.
12754 (compute_access_range, alloc_builtin): New functions.
12755 (classify_builtin_st, classify_builtin_ldst): New functions.
12756 (classify_partition): Refactor code into functions find_single_drs,
12757 classify_builtin_st and classify_builtin_ldst.
12758 (distribute_loop): Don't do runtime alias check when distributing
12760 (find_seed_stmts_for_distribution): New function.
12761 (pass_loop_distribution::execute): Refactor code finding seed
12762 stmts into above function. Support distribution for the innermost
12763 two-level loop nest. Adjust dump information.
12765 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12767 * tree-loop-distribution.c: Adjust the general comment.
12768 (NUM_PARTITION_THRESHOLD): New macro.
12769 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12770 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12771 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12772 in call to build_partition_graph.
12773 (finalize_partitions): New parameter. Make loop distribution more
12774 conservative by fusing more partitions.
12775 (distribute_loop): Don't do runtime alias check in case of loop nest
12777 (find_seed_stmts_for_distribution): New function.
12778 (prepare_perfect_loop_nest): New function.
12779 (pass_loop_distribution::execute): Refactor code finding seed stmts
12780 and loop nest into above functions. Support loop nest distribution.
12781 Adjust dump information accordingly.
12783 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12785 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12786 and set PTYPE_SEQUENTIAL for merged partition.
12788 2017-10-12 Richard Biener <rguenther@suse.de>
12790 PR tree-optimization/69728
12792 2017-09-19 Richard Biener <rguenther@suse.de>
12794 PR tree-optimization/69728
12795 * graphite-sese-to-poly.c (schedule_error): New global.
12796 (add_loop_schedule): Handle empty domain by failing the
12798 (build_original_schedule): Handle schedule_error.
12800 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12801 domain by returning an unchanged schedule.
12803 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12805 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12806 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12808 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12810 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12813 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12816 * expr.c (store_field): Don't optimize away bitsize == 0 store
12817 from CALL_EXPR with addressable return type.
12819 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12821 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12822 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12823 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12824 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12825 TARGET_ISEL instead of TARGET_ISEL<sel>.
12827 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12829 * config/rs6000/rs6000.c
12830 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12832 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12834 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12836 (reg_or_cint_operand): Fix comment.
12837 (reg_or_zero_operand): New predicate.
12838 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12839 * config/rs6000/rs6000.c (output_isel): Delete.
12840 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12841 instead of reg_or_cint_operand. Output instruction directly (not via
12843 (isel_unsigned_<mode>): Ditto.
12844 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12845 gpc_reg_operand. Add an instruction alternative for this. Output
12846 instruction directly.
12847 (*isel_reversed_unsigned_<mode>): Ditto.
12849 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12851 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12852 (TARGET_CANONICALIZE_COMPARISON): Define.
12854 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12857 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12858 Check whether the dest is REG before adding REG_EQUIV note.
12860 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12863 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12865 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12866 (make_hard_regno_born, make_hard_regno_dead): Update
12867 bb_killed_pseudos and bb_gen_pseudos.
12869 2017-10-11 Nathan Sidwell <nathan@acm.org>
12871 * incpath.h (enum incpath_kind): Name enum, prefix values.
12872 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12873 * incpath.c (heads, tails): Use INC_MAX.
12874 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12875 (merge_include_chains, split_quote_chain,
12876 register_include_chains): Update incpath_kind names.
12877 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12878 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12880 (add_framework_path, darwin_register_objc_includes): Likewise.
12881 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12883 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12885 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12886 Do not use float_operator operator predicate.
12887 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12888 * config/i386/predicates.md (float_operator): Remove predicate.
12890 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12892 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12893 (*jccxf_i387): Ditto.
12894 (*jcc<mode>_i387): Ditto.
12895 (*jccu<mode>_i387): Ditto.
12896 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12897 (*jcc_*_i387 splitters): Remove.
12898 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12899 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12900 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12903 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12905 * profile-count.h (slow_safe_scale_64bit): New function.
12906 (safe_scale_64bit): New inline.
12907 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12908 * profile-count.c: Include wide-int.h
12909 (slow_safe_scale_64bit): New.
12911 2017-10-11 Nathan Sidwell <nathan@acm.org>
12913 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12914 HAS_DECL_ASSEMBLER_NAME_P.
12915 * gimple-expr.c (gimple_decl_printable_name: Check
12916 HAS_DECL_ASSEMBLER_NAME_P too.
12917 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12918 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12919 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12920 * passes.c (rest_of_decl_compilation): Check
12921 HAS_DECL_ASSEMBLER_NAME_P too.
12922 * recog.c (verify_changes): Likewise.
12923 * tree-pretty-print.c (dump_decl_name): Likewise.
12924 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12926 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12927 (SET_DECL_ASSEMBLER_NAME): Use it.
12928 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12929 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12930 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12932 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12934 * config.gcc (i386, x86_64): Add extra objects.
12935 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12936 (ix86_min_insn_size): Declare.
12937 (ix86_issue_rate): Declare.
12938 (ix86_adjust_cost): Declare.
12939 (ia32_multipass_dfa_lookahead): Declare.
12940 (ix86_macro_fusion_p): Declare.
12941 (ix86_macro_fusion_pair_p): Declare.
12942 (ix86_bd_has_dispatch): Declare.
12943 (ix86_bd_do_dispatch): Declare.
12944 (ix86_core2i7_init_hooks): Declare.
12945 (ix86_atom_sched_reorder): Declare.
12946 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12947 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12948 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12949 (rip_relative_addr_p): Rename to ...
12950 (ix86_rip_relative_addr_p): ... this one; export.
12951 (memory_address_length): Update.
12952 (ix86_issue_rate): Move to x86-tune-sched.c.
12953 (ix86_flags_dependent): Move to x86-tune-sched.c.
12954 (ix86_agi_dependent): Move to x86-tune-sched.c.
12955 (exact_dependency_1): Move to x86-tune-sched.c.
12956 (exact_store_load_dependency): Move to x86-tune-sched.c.
12957 (ix86_adjust_cost): Move to x86-tune-sched.c.
12958 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12959 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12960 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12961 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12962 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12963 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12964 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12965 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12966 (min_insn_size): Rename to ...
12967 (ix86_min_insn_size): ... this one; export.
12968 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12969 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12970 (core2i7_first_cycle_multipass_backtrack): Move to
12971 x86-tune-sched-core.c.
12972 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12973 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12974 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12975 (ix86_avoid_jump_mispredicts): Update.
12976 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12977 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12978 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12979 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12980 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12981 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12982 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12983 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12984 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12985 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12986 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12987 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12988 (BIG): Move to ix86-tune-sched-bd.c.
12989 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12990 (enum insn_path): Move to ix86-tune-sched-bd.c.
12991 (get_mem_group): Move to ix86-tune-sched-bd.c.
12992 (is_cmp): Move to ix86-tune-sched-bd.c.
12993 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12994 (is_branch): Move to ix86-tune-sched-bd.c.
12995 (is_prefetch): Move to ix86-tune-sched-bd.c.
12996 (init_window): Move to ix86-tune-sched-bd.c.
12997 (allocate_window): Move to ix86-tune-sched-bd.c.
12998 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12999 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
13000 (process_end_window): Move to ix86-tune-sched-bd.c.
13001 (allocate_next_window): Move to ix86-tune-sched-bd.c.
13002 (find_constant): Move to ix86-tune-sched-bd.c.
13003 (get_num_immediates): Move to ix86-tune-sched-bd.c.
13004 (has_immediate): Move to ix86-tune-sched-bd.c.
13005 (get_insn_path): Move to ix86-tune-sched-bd.c.
13006 (get_insn_group): Move to ix86-tune-sched-bd.c.
13007 (count_num_restricted): Move to ix86-tune-sched-bd.c.
13008 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
13009 (add_insn_window): Move to ix86-tune-sched-bd.c.
13010 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
13011 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
13012 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
13013 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
13014 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
13015 (do_dispatch): Move to ix86-tune-sched-bd.c.
13016 (has_dispatch): Move to ix86-tune-sched-bd.c.
13017 * config/i386/t-i386: Add new object files.
13018 * config/i386/x86-tune-costs.h: New file.
13019 * config/i386/x86-tune-sched-atom.c: New file.
13020 * config/i386/x86-tune-sched-bd.c: New file.
13021 * config/i386/x86-tune-sched-core.c: New file.
13022 * config/i386/x86-tune-sched.c: New file.
13024 2017-10-11 Liu Hao <lh_mouse@126.com>
13026 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
13027 the generic version below instead.
13028 (should_colorize): Recognize Windows consoles as terminals
13030 * pretty-print.c [__MINGW32__] (write_all): New function.
13031 [__MINGW32__] (find_esc_head): Likewise.
13032 [__MINGW32__] (find_esc_terminator): Likewise.
13033 [__MINGW32__] (eat_esc_sequence): Likewise.
13034 [__MINGW32__] (mingw_ansi_fputs): New function that handles
13036 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
13039 2017-10-11 Richard Biener <rguenther@suse.de>
13041 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
13042 Properly call analyze_scalar_evolution with the loop of the stmt.
13044 2017-10-11 Richard Biener <rguenther@suse.de>
13046 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
13047 * tree-core.h (tree_base): Add chrec_var union member.
13048 * tree.h (CHREC_VAR): Remove.
13049 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
13050 * tree-chrec.h (build_polynomial_chrec): Adjust.
13051 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
13052 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
13054 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
13056 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
13057 * match.pd: ... here.
13058 ((T) X == (T) Y): Relax condition.
13060 2017-10-11 Bin Cheng <bin.cheng@arm.com>
13062 PR tree-optimization/82472
13063 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
13065 (break_alias_scc_partitions): Update postorder number.
13067 2017-10-11 Martin Liska <mliska@suse.cz>
13070 * opts.c (parse_no_sanitize_attribute): Do not use error_value
13072 * opts.h (parse_no_sanitize_attribute): Remove last argument.
13074 2017-10-11 Martin Liska <mliska@suse.cz>
13076 * print-rtl.c (print_insn): Move declaration of idbuf
13077 to same scope as name.
13079 2017-10-11 Martin Liska <mliska@suse.cz>
13084 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13085 volatile arguments.
13087 2017-10-11 Martin Liska <mliska@suse.cz>
13090 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13091 volatile arguments.
13093 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
13095 * config.gcc (default_gnu_indirect_function): Default to yes for
13096 arm*-*-linux* with glibc.
13098 2017-10-11 Richard Biener <rguenther@suse.de>
13100 * tree-scalar-evolution.c (get_scalar_evolution): Handle
13101 default-defs and types we do not want to analyze.
13102 (interpret_loop_phi): Replace unreachable code with an assert.
13103 (compute_scalar_evolution_in_loop): Remove and inline ...
13104 (analyze_scalar_evolution_1): ... here, replacing condition with
13105 what makes the intent clearer. Remove handling of cases
13106 get_scalar_evolution now handles.
13108 2017-10-10 Jim Wilson <wilson@tuliptree.org>
13110 PR rtl-optimization/81434
13111 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
13112 comment for main loop. In sched_group_found if, also add checks for
13113 pass and min_cost_group.
13115 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
13117 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
13118 (rs6000_insn_cost): New function.
13119 * config/rs6000/rs6000.md (cost): New attribute.
13121 2017-10-10 Jakub Jelinek <jakub@redhat.com>
13122 H.J. Lu <hongjiu.lu@intel.com>
13126 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
13127 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
13128 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
13129 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
13130 to OPTION_MASK_ISA_AVX512VL - builtins that have both
13131 OPTION_MASK_ISA_MMX and some other bit set require both
13132 mmx and the ISAs without the mmx bit.
13133 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
13134 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
13135 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
13136 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
13137 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
13138 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
13139 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
13140 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
13141 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
13142 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
13143 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
13144 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
13145 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
13146 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
13147 Add OPTION_MASK_ISA_MMX.
13149 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
13151 * config.gcc (armv7*-*-freebsd*): New target.
13152 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
13154 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
13156 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
13157 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
13160 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
13162 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
13163 a template parameter.
13164 (WIDE_INT_REF_FOR): Update accordingly.
13165 * tree.h (wi::int_traits <const_tree>): Delete.
13166 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
13167 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
13168 (wi::tree_to_wide_ref): New typedef.
13169 (wi::to_wide): New function.
13170 * calls.c (get_size_range): Use wi::to_wide when operating on
13171 trees as wide_ints.
13172 * cgraph.c (cgraph_node::create_thunk): Likewise.
13173 * config/i386/i386.c (ix86_data_alignment): Likewise.
13174 (ix86_local_alignment): Likewise.
13175 * dbxout.c (stabstr_O): Likewise.
13176 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
13177 * expr.c (const_vector_from_tree): Likewise.
13178 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
13179 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
13180 (fold_negate_expr_1, int_const_binop_1, const_binop)
13181 (fold_convert_const_int_from_real, optimize_bit_field_compare)
13182 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
13183 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
13184 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
13185 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
13186 (fold_not_const, round_up_loc): Likewise.
13187 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
13188 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
13189 (alloca_call_type): Likewise.
13190 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
13191 * godump.c (go_output_typedef): Likewise.
13192 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
13193 * internal-fn.c (get_min_precision): Likewise.
13194 * ipa-cp.c (ipcp_store_vr_results): Likewise.
13195 * ipa-polymorphic-call.c
13196 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
13197 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
13198 (ipa_modify_call_arguments): Likewise.
13199 * match.pd: Likewise.
13200 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
13201 * print-tree.c (print_node_brief, print_node): Likewise.
13202 * stmt.c (expand_case): Likewise.
13203 * stor-layout.c (layout_type): Likewise.
13204 * tree-affine.c (tree_to_aff_combination): Likewise.
13205 * tree-cfg.c (group_case_labels_stmt): Likewise.
13206 * tree-data-ref.c (dr_analyze_indices): Likewise.
13207 (prune_runtime_alias_test_list): Likewise.
13208 * tree-dump.c (dequeue_and_dump): Likewise.
13209 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
13210 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
13211 * tree-pretty-print.c (dump_generic_node): Likewise.
13212 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
13213 (simple_iv_with_niters): Likewise.
13214 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
13215 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
13216 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
13217 * tree-ssa-loop-niter.c (split_to_var_and_offset)
13218 (refine_value_range_using_guard, number_of_iterations_ne_max)
13219 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
13220 (get_cst_init_from_scev, record_nonwrapping_iv)
13221 (scev_var_range_cant_overflow): Likewise.
13222 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13223 * tree-ssa-pre.c (compute_avail): Likewise.
13224 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13225 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13226 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13227 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13228 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13229 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13230 (dump_case_nodes, try_switch_expansion): Likewise.
13231 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13232 (vect_do_peeling): Likewise.
13233 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13234 * tree-vect-stmts.c (vectorizable_load): Likewise.
13235 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13236 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13237 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13238 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13239 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13240 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13241 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13242 (evrp_dom_walker::before_dom_children): Likewise.
13243 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13244 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13245 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13246 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13247 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13248 (get_range_pos_neg): Likewise.
13249 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13250 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13251 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13252 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13253 * config/avr/avr.c (avr_fold_builtin): Likewise.
13254 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13255 * config/msp430/msp430.c (msp430_attr): Likewise.
13256 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13257 * config/powerpcspe/powerpcspe-c.c
13258 (altivec_resolve_overloaded_builtin): Likewise.
13259 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13260 (rs6000_expand_ternop_builtin): Likewise.
13261 * config/rs6000/rs6000-c.c
13262 (altivec_resolve_overloaded_builtin): Likewise.
13263 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13264 (rs6000_expand_ternop_builtin): Likewise.
13265 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13267 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13269 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13270 when copying loop nest with only one inner loop.
13272 2017-10-10 Richard Biener <rguenther@suse.de>
13274 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13275 blocks if SCEV is active.
13276 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13278 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13279 (scev_initialize): Assert we are not yet initialized.
13281 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13283 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13284 inner loop's exit stmt by making it always exit the loop, otherwise
13285 we would generate an infinite empty loop.
13287 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13289 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13290 renaming variables in new preheader if it's deleted.
13292 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13294 * tree-loop-distribution.c (struct partition): Remove unused field
13295 loops of the structure.
13296 (partition_alloc, partition_free): Ditto.
13297 (build_rdg_partition_for_vertex): Ditto.
13299 2017-10-09 Jeff Law <law@redhat.com>
13301 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13302 return type to match prototype and documentation.
13304 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13306 * config/rs6000/rs6000.c (processor_costs): Move to ...
13307 * config/rs6000/rs6000.h: ... here.
13308 (rs6000_cost): Declare.
13310 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13312 * except.c (setjmp_fn): New global variable.
13313 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13314 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13315 if DONT_USE_BUILTIN_SETJMP is defined.
13317 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13319 * target.def (insn_cost): New hook.
13320 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13321 * doc/tm.texi: Regenerate.
13322 * rtlanal.c (insn_cost): Use the new hook.
13324 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13326 * combine.c (combine_validate_cost): Compute the new insn_cost,
13327 not just pattern_cost.
13328 (try_combine): Adjust comment.
13330 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13332 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13334 * combine.c (uid_insn_cost): Adjust comment.
13335 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13337 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13338 * dse.c (find_shift_sequence): Ditto.
13339 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13340 (bb_valid_for_noce_process_p): Use pattern_cost.
13341 * rtl.h (insn_rtx_cost): Delete.
13342 (pattern_cost): New prototype.
13343 (insn_cost): New prototype.
13344 * rtlanal.c (insn_rtx_cost): Rename to...
13345 (pattern_cost): ... this.
13348 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13350 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13351 (*jcc<mode>_0_r_i387): Ditto.
13352 (*jccxf_r_i387): Ditto.
13353 (*jcc<mode>_r_i387): Ditto.
13354 (*jccu<mode>_r_i387): Ditto.
13355 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13356 (*jcc): Rename from *jcc_1.
13358 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13360 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13361 deferred rescans after the lvx/stvx recombination pre-pass.
13363 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13365 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13366 memory operation instruction support.
13367 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13368 (rs6000-ibm-aix[789]*): Likewise.
13369 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13370 Document new functions.
13372 2017-10-09 Richard Biener <rguenther@suse.de>
13374 PR tree-optimization/82397
13375 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13376 equality only for semantically equal trees.
13378 2017-10-09 Richard Biener <rguenther@suse.de>
13380 PR tree-optimization/82449
13381 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13382 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13383 allow constant addresses.
13384 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13387 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13389 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13392 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13395 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13398 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13401 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13403 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13406 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13407 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13409 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13411 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13412 (WI_BINARY_PREDICATE_RESULT): Likewise.
13413 (wi::binary_traits::operator_result): New type.
13414 (wi::binary_traits::predicate_result): Likewise.
13415 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13416 (generic_wide_int::operator==, generic_wide_int::operator!=)
13417 (generic_wide_int::operator&, generic_wide_int::and_not)
13418 (generic_wide_int::operator|, generic_wide_int::or_not)
13419 (generic_wide_int::operator^, generic_wide_int::operator+
13420 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13422 (operator~, unary operator-, operator==, operator!=, operator&)
13423 (operator|, operator^, operator+, binary operator-, operator*): New
13425 * expr.c (get_inner_reference): Use wi::bit_and_not.
13426 * fold-const.c (fold_binary_loc): Likewise.
13427 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13428 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13429 (bit_value_binop): Likewise.
13430 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13431 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13432 (extract_range_from_binary_expr_1): Likewise.
13433 (masked_increment): Likewise.
13434 (simplify_bit_ops_using_ranges): Likewise.
13436 2017-10-09 Martin Jambor <mjambor@suse.cz>
13439 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13440 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13441 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13443 (hsa_fixup_mov_insn_type): New function.
13444 (hsa_op_with_type::get_in_type): Use it.
13445 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13446 immediates in an assert.
13447 (hsa_op_with_type::extend_int_to_32bit): New method.
13448 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13449 types. Convert to dest type if necessary.
13450 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13451 (reg_for_gimple_ssa): Pass false as min32int to
13452 hsa_type_for_scalar_tree_type.
13453 (gen_hsa_addr): Fixup type when creating addresable temporary.
13454 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13455 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13456 necessary. Call hsa_fixup_mov_insn_type.
13457 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13458 extend operands and convert to dest type if necessary.
13459 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13460 to dest type if necessary.
13461 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13462 if conversion nt necessary and size matches.
13463 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13464 to dest type if necessary.
13465 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13466 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13468 (gen_hsa_clrsb): Likewise.
13469 (gen_hsa_ffs): Likewise.
13470 (gen_hsa_divmod): Extend operands and convert to dest type if
13472 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13474 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13476 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13477 Remove empty default arguments. Use a brace block as output
13479 (conditional return): Ditto.
13481 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13482 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13484 (group_ending_nop): Ditto.
13485 (doloop_end): Ditto.
13486 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13487 (splitters for those): Ditto.
13489 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13491 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13492 a conditional jump (and the compare for it) so that pc_rtx is the
13494 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13495 for the deleted and renamed ctr<mode>_internal[234] patterns.
13496 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13497 Delete second conditional return pattern.
13498 (ctr<mode>_internal2): Delete this second bdnz pattern.
13499 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13500 (ctr<mode>_internal4): Delete this second bdz pattern.
13502 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13504 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13505 (always_initialized_rtx_for_ssa_name_p): New predicate.
13506 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13507 (finish_out_of_ssa): Free new field of SA.
13508 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13509 * tree-ssa-coalesce.c: Include tree-ssa.h.
13510 (get_parm_default_def_partitions): Remove extern keyword.
13511 (get_undefined_value_partitions): New function.
13512 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13513 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13514 uninitialized bits.
13515 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13517 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13519 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13521 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13523 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13524 for targets that preffer 128bit.
13526 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13528 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13530 2017-10-08 Olivier Hainque <hainque@adacore.com>
13532 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13533 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13535 (thumb_set_return_address): Likewise.
13537 2017-10-08 Olivier Hainque <hainque@adacore.com>
13539 * common/config/arm/arm-common.c (arm_except_unwind_info):
13540 Handle DWARF2_UNWIND_INFO.
13542 2017-10-07 Michael Collison <michael.collison@arm.com>
13544 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13547 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13549 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13550 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13551 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13552 defined, force the creation of a new block for a dispatch label.
13554 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13556 * invoke.texi (Wsuggest-attribute=cold): Document.
13557 * common.opt (Wsuggest-attribute=cold): New
13558 * ipa-pure-const.c (warn_function_cold): New function.
13559 * predict.c (compute_function_frequency): Use it.
13560 * predict.h (warn_function_cold): Declare.
13562 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13564 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13567 2017-10-06 Martin Liska <mliska@suse.cz>
13569 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13570 keyword for member functions.
13571 (struct sanopt_tree_couple): New struct.
13572 (struct sanopt_tree_couple_hash): New function.
13573 (struct sanopt_ctx): Add new hash_map.
13574 (has_dominating_ubsan_ptr_check): New function.
13575 (record_ubsan_ptr_check_stmt): Likewise.
13576 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13577 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13578 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13580 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13583 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13584 aarch64_simd_valid_immediate on CONST_VECTORs.
13585 (aarch64_reg_or_bic_imm): Likewise.
13587 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13589 PR rtl-optimization/82396
13590 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13592 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13594 * graphite-dependences.c (scop_get_reads): Move code to...
13595 (scop_get_must_writes): Move code to...
13596 (scop_get_may_writes): Move code to...
13597 (scop_get_reads_and_writes): ... here.
13598 (scop_get_dependences): Call scop_get_reads_and_writes.
13600 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13602 PR tree-optimization/82434
13603 * fold-const.h (can_native_encode_type_p,
13604 can_native_encode_string_p): Remove.
13605 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13606 don't encode anything, just return what would be otherwise returned.
13607 (native_encode_fixed, native_encode_complex, native_encode_vector):
13609 (native_encode_string): Likewise. Inline by hand
13610 can_native_encode_string_p.
13611 (can_native_encode_type_p): Remove.
13612 (can_native_encode_string_p): Remove.
13613 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13614 STRING_CSTs using can_native_encode_string_p, test all
13615 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13616 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13617 argument from native_encode_expr.
13618 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13619 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13622 2017-10-06 Richard Biener <rguenther@suse.de>
13624 PR tree-optimization/82397
13625 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13626 operand_equal_p but rely on data_ref_compare_tree for detecting
13628 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13629 to match up with dr_group_sort_cmp.
13631 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13634 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13636 * config/s390/s390-builtin-types.def: Regenerate.
13638 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13641 * config/s390/s390-builtin-types.def: Regenerate.
13642 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13643 Change flag from B_VXE to B_VX.
13644 (s390_vec_min_dbl): Remove B_VXE flag.
13646 2017-10-06 Richard Biener <rguenther@suse.de>
13648 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13649 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13650 translate_isl_ast_to_gimple::is_valid_rename,
13651 translate_isl_ast_to_gimple::get_rename,
13652 translate_isl_ast_to_gimple::get_def_bb_for_const,
13653 translate_isl_ast_to_gimple::get_new_name,
13654 translate_isl_ast_to_gimple::collect_all_ssa_names,
13655 translate_isl_ast_to_gimple::copy_loop_phi_args,
13656 translate_isl_ast_to_gimple::collect_all_ssa_names,
13657 translate_isl_ast_to_gimple::copy_loop_phi_args,
13658 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13659 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13660 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13661 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13662 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13663 translate_isl_ast_to_gimple::copy_cond_phi_args,
13664 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13665 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13666 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13667 translate_isl_ast_to_gimple::rename_uses,
13668 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13669 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13670 (set_rename_for_each_def): Likewise.
13671 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13672 here. Handle rewriting SCEV analyzable uses here.
13673 (copy_bb_and_scalar_dependences): Generate code for PHI
13675 (graphite_regenerate_ast_isl): Adjust.
13676 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13677 (add_write, add_read): New functions.
13678 (build_cross_bb_scalars_def): Use it and simplify.
13679 (build_cross_bb_scalars_use): Likewise.
13680 (graphite_find_cross_bb_scalar_vars): Inline into...
13681 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13682 simulating out-of-SSA. Compute liveout and add dependencies.
13683 (build_scops): Force an empty entry block.
13684 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13686 (sese_build_liveouts): Declare.
13687 (sese_trivially_empty_bb_p): Likewise.
13688 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13689 compute liveout and debug_liveout.
13690 (sese_bad_liveouts_use): Remove.
13691 (sese_reset_debug_liveouts_bb): Likewise.
13692 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13693 (sese_build_liveouts): Build liveout and debug_liveout and store
13695 (new_sese_info): Adjust.
13696 (free_sese_info): Likewise.
13697 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13698 do not build liveout here.
13699 (move_sese_in_condition): Adjust region entry.
13700 (scev_analyzable_p): Match up with chrec_apply requirements.
13701 (sese_trivially_empty_bb_p): New.
13702 * tree-into-ssa.c (get_reaching_def): Properly support generating
13703 default-defs for incremental rewrite of anonymous names.
13705 2017-10-06 Richard Biener <rguenther@suse.de>
13707 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13708 precision do not perform modulo reduction.
13710 2017-10-06 Richard Biener <rguenther@suse.de>
13712 PR tree-optimization/82436
13713 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13714 conservatively choose the vectorization factor when checking
13715 whether we can perform the required load permutation.
13716 (vect_transform_slp_perm_load): Assert when we may not fail.
13718 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13720 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13721 message for incompatible -msdata=* and -mcall-* options.
13723 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13725 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13726 rate for post-reload scheduling.
13728 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13730 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13732 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13734 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13735 to improve monte carlo in scimark.
13737 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13739 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13740 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13741 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13742 pentium4_cost, nocona_cost): Set reassociation width to 1.
13743 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13744 width to 2 for fp operations and 1 otherwise.
13745 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13747 (atom_cost): Set reassociation width to 2.
13748 (slm_cost, generic_cost): Set fp reassociation width
13749 to 2 and 1 otherwise.
13750 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13751 (core_cost): Set fp reassociation width to 4 and vector to 2.
13752 (ix86_reassociation_width): Rewrite using cost table; special case
13753 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13754 and TARGET_AVX128_OPTIMAL.
13755 * config/i386/i386.h (processor_costs): Add
13756 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13757 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13758 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13759 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13760 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13761 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13763 2017-10-05 Nathan Sidwell <nathan@acm.org>
13765 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13767 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13769 * config/arm/arm.c (arm_test_fpu_data): New.
13770 (arm_run_selftests): Call arm_test_fpu_data.
13772 2017-10-04 Nathan Sidwell <nathan@acm.org>
13774 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13776 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13778 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13780 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13781 check type for aarch64_simd_valid_immediate.
13782 (aarch64_output_simd_mov_immediate): Update prototype.
13783 (aarch64_simd_valid_immediate): Update prototype.
13784 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13785 support for ORR-immediate.
13786 (and<mode>3): modified pattern to add support for BIC-immediate.
13787 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13788 now checks for valid immediate for BIC and ORR based on new enum
13790 (aarch64_output_simd_mov_immediate): Function now used to output
13791 BIC/ORR imm as well based on new enum argument.
13792 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13794 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13795 (aarch64_reg_or_bic_imm): Likewise.
13797 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13799 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13800 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13802 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13806 PR rtl-optimization/82396
13807 * haifa-sched.c (autopref_multipass_init): Simplify
13809 (autopref_rank_data): Simplify sort order.
13810 * sched-int.h (autopref_multipass_data_): Remove
13811 multi_mem_insn_p, min_offset and max_offset.
13813 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13815 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13817 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13819 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13820 vect_doubleint_cvt.
13822 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13824 * doc/sourcebuild.texi: Document vect_long_mult.
13826 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13828 PR tree-optimization/82413
13829 * fold-const.c (build_range_check): Use widest_int when comparing
13830 the maximum ETYPE value with HIGH.
13832 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13834 PR rtl-optimization/82396
13835 * haifa-sched.c (autopref_multipass_init): Simplify
13837 (autopref_rank_data): Simplify sort order.
13838 * sched-int.h (autopref_multipass_data_): Remove
13839 multi_mem_insn_p, min_offset and max_offset.
13841 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13843 PR tree-optimization/82381
13844 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13845 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13848 PR tree-optimization/82374
13849 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13850 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13851 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13852 current_function_decl to the new decl.
13854 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13856 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13857 helper macro for IEEE float128 hardware built-in functions.
13858 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13860 (TRUNCF128_ODD): Likewise.
13861 (ADDF128_ODD): Likewise.
13862 (SUBF128_ODD): Likewise.
13863 (MULF128_ODD): Likewise.
13864 (DIVF128_ODD): Likewise.
13865 (FMAF128_ODD): Likewise.
13866 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13867 UNSPEC_TRUNC_ROUND_TO_ODD.
13868 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13869 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13870 floating point round to odd instructions.
13871 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13872 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13873 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13874 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13875 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13876 (trunc<mode>sf2_hw): Change the truncate with round to odd
13877 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13878 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13879 to odd hardware instructions.
13880 (sub<mode>3_odd): Likewise.
13881 (mul<mode>3_odd): Likewise.
13882 (div<mode>3_odd): Likewise.
13883 (sqrt<mode>2_odd): Likewise.
13884 (fma<mode>4_odd): Likewise.
13885 (fms<mode>4_odd): Likewise.
13886 (nfma<mode>4_odd): Likewise.
13887 (nfms<mode>4_odd): Likewise.
13888 (trunc<mode>df2_odd): Change the truncate with round to odd
13889 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13891 * doc/extend.texi (PowerPC built-in functions): Update documentation
13892 for existing IEEE float128-bit built-in functions. Add built-in
13893 functions that generate the IEEE 128-bit floating point round to
13896 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13898 PR rtl-optimization/77729
13899 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13900 to (X&(C1&~C2))|C2 transformations.
13902 2017-10-03 Martin Jambor <mjambor@suse.cz>
13904 PR tree-optimization/82363
13905 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13906 mismatch, mark lacc written regardless of racc.
13908 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13910 PR tree-optimization/82381
13911 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13912 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13913 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13914 fallthrough into reassoc_stmt_dominates_stmt_p.
13917 * combine.c (combine_instructions): Don't combine in unreachable
13920 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13923 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13924 function to not use the have_cpu variable. Do not set cpu_index,
13925 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13926 or the default cpu.
13927 (rs6000_valid_attribute_p): Remove duplicate initializations of
13928 old_optimize and func_optimize.
13929 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13930 (rs6000_activate_target_options): Make global.
13931 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13934 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13936 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13937 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13938 Set *poffset to 0 and *psize and *pmax_size to -1 if
13939 *poffset + *psize overflows in HOST_WIDE_INT.
13941 PR tree-optimization/82387
13942 PR tree-optimization/82388
13943 PR tree-optimization/82389
13944 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13945 instead of live_bytes non-NULL.
13947 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13950 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13953 2017-10-02 Richard Biener <rguenther@suse.de>
13955 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13956 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13957 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13959 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13961 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13962 requested precision matches the type's.
13963 * calls.c (alloc_max_size): Calculate the new candidate size as
13964 a widest_int and use wi::to_widest when comparing it with the
13965 current candidate size.
13966 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13967 zero rather than integer_zero_node.
13968 * match.pd: Check for a no-op conversion before using wi::add
13969 rather than after. Use tree_to_uhwi when summing small shift
13970 counts into an unsigned int.
13972 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13973 Alan Hayward <alan.hayward@arm.com>
13974 David Sherwood <david.sherwood@arm.com>
13977 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13978 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13979 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13980 POINTER_AND_FP_REGS.
13982 2017-10-02 Richard Biener <rguenther@suse.de>
13984 PR tree-optimization/82355
13985 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13986 a mapping for the enclosing loop but avoid generating one for
13987 the loop tree root.
13988 (copy_bb_and_scalar_dependences): Remove premature codegen
13989 error on PHIs in blocks duplicated into multiple places.
13990 * graphite-scop-detection.c
13991 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13992 in the region use it as loop and nest to analyze the DR in.
13993 (try_generate_gimple_bb): Likewise.
13994 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13995 (add_loop_constraints): For blocks in a loop not in the region
13996 create a dimension with a single iteration.
13997 * sese.h (gbb_loop_at_index): Remove assert.
13999 2017-10-01 Kevin Buettner <kevinb@redhat.com>
14001 * omp-expand.c (adjust_context_scope): New function.
14002 (expand_parallel_call): Call adjust_context_scope.
14004 2017-10-01 Jeff Law <law@redhat.com>
14006 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
14007 dom_opt_dom_walker class with direct access to private members.
14008 Add comments. Call test_for_singularity.
14009 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
14010 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
14011 m_dummy_cond anymore.
14012 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
14014 (pass_dominator:execute): Build the dummy_cond here and pass it
14015 to the dom_opt_dom_walker ctor.
14016 (test_for_singularity): New function.
14018 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
14019 Maya Rashish <coypu@sdf.org>
14021 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
14022 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
14023 (alpha*-*-netbsd*) Use nbsd_tm_file.
14024 (arm*-*-netbsdelf*) Likewise.
14025 (i[34567]86-*-netbsdelf*) Likewise.
14026 (x86_64-*-netbsd*) Likewise.
14027 (mips*-*-netbsd*) Likewise.
14028 (powerpc-*-netbsd*) Likewise.
14029 (sh*-*-netbsd*) Likewise.
14030 (sparc-*-netbsdelf*) Likewise.
14031 (sparc64-*-netbsd*) Likewise.
14032 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
14034 (vax-*-netbsdelf*) Likewise.
14035 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
14036 (UINT_FAST8_TYPE) Likewise.
14037 (INT_FAST16_TYPE) Check CHAR_FAST16.
14038 (UINT_FAST16_TYPE) Likewise.
14040 2017-09-30 Jakub Jelinek <jakub@redhat.com>
14043 * config/i386/i386.md
14044 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
14045 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
14046 *divmodsi4_zext_2): New define_insn_and_split.
14047 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
14048 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
14049 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
14050 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
14051 New define_insn_and_split.
14052 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
14053 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
14054 operands[1] having DImode when mode is SImode.
14056 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
14057 always SImode for DIV and MOD in REG_EQUAL notes.
14059 2017-09-29 Yury Gribov <tetra2005@gmail.com>
14061 PR middle-end/82319
14062 * match.pd: Fix handling of NaNs in pattern.
14064 2017-09-29 Jeff Law <law@redhat.com>
14066 * sbitmap.c (bitmap_bit_in_range_p): New function.
14067 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
14068 * tree-ssa-dse.c (live_bytes_read): New function.
14069 (dse_classify_store): Ignore reads of dead bytes.
14071 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
14072 typos and whitespace errors.
14073 * config/i386/predicates.md (address_no_seg_operand): Likewise.
14074 * config/s390/s390.c (s390_emit_prologue): Likewise.
14076 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14079 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
14080 with a symbol for LRA.
14082 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14084 PR rtl-optimization/82338
14085 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
14087 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
14089 * genmodes.c (calc_wider_mode): Suppress qsort macro.
14090 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
14091 (qsort_chk): Declare.
14092 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
14093 (qsort_chk): New function.
14095 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14097 PR tree-optimization/82337
14098 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
14099 phi definition if the PHI result appears in an abnormal PHI.
14100 (find_basis_for_base_expr): Don't record a basis if the LHS of the
14101 basis appears in an abnormal PHI.
14103 2017-09-29 Richard Biener <rguenther@suse.de>
14105 * graphite-isl-ast-to-gimple.c
14106 (translate_isl_ast_to_gimple::set_codegen_error): New function.
14107 (binary_op_to_tree): Use it.
14108 (get_rename_from_scev): Likewise.
14109 (copy_loop_phi_nodes): Likewise.
14110 (copy_bb_and_scalar_dependences): Likewise.
14111 (translate_pending_phi_nodes): Likewise.
14113 2017-09-29 Jakub Jelinek <jakub@redhat.com>
14116 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
14117 for movabsq $(i32 << shift), r64.
14119 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
14121 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
14122 index when encoding %esp as %rsp to avoid 0x67 prefix.
14124 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
14126 * config/i386/i386.md (*movsf_internal, *movdf_internal):
14127 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
14129 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
14131 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
14132 Extensions with more than 16 double VFP registers.
14133 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
14134 to_clear_mask and all code related to it. Replace the remaining
14135 entry by a sbitmap and adapt code accordingly.
14137 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
14139 * brig-builtins.def: Change pure attributes to const.
14141 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14143 * config.gcc (default_gnu_indirect_function): Default to yes for
14144 sparc*-*-linux* with glibc.
14146 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14148 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
14149 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
14150 when creating .init_array and .fini_array sections with priority
14153 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
14156 * config/aarch64/aarch64.c
14157 (aarch64_builtin_support_vector_misalignment): Always return false
14158 when misalignment is unknown.
14160 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
14162 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
14163 this function to return false if the definition used by the swap
14164 instruction is artificial, or if the memory address from which the
14165 constant value is loaded is not represented by a base address held
14166 in a register or if the base address register is a frame or stack
14167 pointer. Additionally, return false if the base address of the
14168 loaded constant is a SYMBOL_REF but is not considered to be a
14170 (replace_swapped_load_constant): New function.
14171 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
14172 loaded constant vector with a load of a swapped constant vector.
14174 2017-09-27 Carl Love <cel@us.ibm.com>
14176 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14177 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14178 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14181 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
14183 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
14184 first, always call autopref_rank_data otherwise.
14186 2017-09-27 Richard Biener <rguenther@suse.de>
14188 * graphite-scop-detection.c (find_scop_parameters): Move
14189 loop bound handling ...
14190 (gather_bbs::before_dom_children): ... here, avoiding the need
14191 to build scop_info->loop_nest.
14192 (record_loop_in_sese): Remove.
14193 * sese.h (sese_info_t::loop_nest): Remove.
14194 * sese.c (new_sese_info): Do not allocate loop_nest.
14195 (free_sese_info): Do not free loop_nest.
14197 2017-09-27 Jakub Jelinek <jakub@redhat.com>
14200 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
14201 lhs from calls if the lhs has addressable type.
14203 2017-09-27 Richard Biener <rguenther@suse.de>
14205 * graphite.h (scop::max_alias_set): New member.
14206 * graphite-scop-detection.c: Remove references to non-existing
14207 --param in comments.
14208 (build_alias_sets): Record the maximum alias set used for drs.
14209 (build_scops): Support zero as unlimited for
14210 --param graphite-max-arrays-per-scop.
14211 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
14212 and inline into ...
14213 (build_poly_sr_1): ... here. Compute alias set based on the
14214 maximum alias set used for drs rather than
14215 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
14217 2017-09-27 Richard Biener <rguenther@suse.de>
14219 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
14220 --param loop-block-tile-size=0 to disable tiling.
14222 2017-09-27 Richard Biener <rguenther@suse.de>
14224 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14225 (graphite-max-nb-scop-params): Document special value zero.
14226 * domwalk.h (dom_walker::STOP): New symbolical constant.
14227 (dom_walker::dom_walker): Add optional parameter for bb to
14229 (dom_walker::~dom_walker): Declare.
14230 (dom_walker::before_dom_children): Document STOP return value.
14231 (dom_walker::m_user_bb_to_rpo): New member.
14232 (dom_walker::m_bb_to_rpo): Likewise.
14233 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14234 mapping here if not provided by the user.
14235 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14236 provided by the user.
14237 (dom_walker::STOP): Define.
14238 (dom_walker::walk): Do not compute bb to RPO mapping here.
14239 Support STOP return value from before_dom_children to stop
14241 * graphite-optimize-isl.c (optimize_isl): If the schedule
14242 is the same still generate code if -fgraphite-identity
14243 or -floop-parallelize-all are given.
14244 * graphite-scop-detection.c: Include cfganal.h.
14245 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14247 (gather_bbs::before_dom_children): Return STOP for BBs
14249 (build_scops): Compute bb to RPO mapping and pass it to
14250 the domwalk. Treat --param graphite-max-nb-scop-params=0
14251 as not limiting the number of params.
14252 * graphite.c (graphite_initialize): Remove limit on the
14253 number of basic-blocks in a function.
14254 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14255 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14256 default value of 10.
14258 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14260 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14261 Adjust code to eliminate needing to do the shift right 32-bits
14262 operation after XSCVDPSPN.
14264 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14266 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14267 ((X / Y) != 0 -> X >= Y): Likewise.
14269 2017-09-26 Carl Love <cel@us.ibm.com>
14271 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14272 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14273 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14274 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14275 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14276 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14277 definitions and overloading.
14278 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14279 statement for P9V_BUILTIN_XST_LEN_R.
14280 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14281 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14282 define_expand and define_insn for the instructions and builtins.
14283 * doc/extend.texi: Update the built-in documentation file for the new
14284 built-in functions.
14285 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14286 define_insn for the instructions
14288 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14291 * gcc/config/netbsd-protos.h: New file.
14292 * gcc/config/netbsd.c: New file.
14293 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14294 * gcc/config/t-netbsd: New file.
14295 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14296 (tmake_file) Add t-netbsd.
14297 (extra_objs) Add netbsd.o.
14299 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14303 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14305 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14307 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14308 sign extension from a vector register to a GPR by doing a 32-bit
14309 direct move and then an EXTSW.
14310 (extendsi<mode>2 splitter): Likewise.
14311 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14312 right or vector extract after doing XSCVDPSPN. Use
14313 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14315 (movdi_from_sf_zero_ext): Likewise.
14316 (reload_gpr_from_vsxsf): Likewise.
14317 (p8_mfvsrd_4_disf): Delete, no longer used.
14318 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14319 then needing to move the SFmode to a GPR to use the XSCVDPSP
14320 instruction instead of FRSP and XSCVDPSPN.
14321 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14322 it is adjacent to the other XSCVSPDP insns.
14323 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14324 SFmode to be in traditional Altivec registers.
14325 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14326 (vsx_xscvspdpn): Likewise.
14327 (vsx_xscvspdpn_scalar): Likewise.
14329 2017-09-26 Martin Jambor <mjambor@suse.cz>
14331 * tree-sra.c (compare_access_positions): Put integral types first,
14332 stabilize sorting of integral types, remove conditions putting
14333 non-full-precision integers last.
14334 (sort_and_splice_var_accesses): Disable scalarization if a
14335 non-integert would be represented by a non-full-precision integer.
14337 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14339 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14340 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14341 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14342 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14343 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14344 conditionals inside the function instead of around it. Call
14345 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14346 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14348 2017-09-26 Richard Biener <rguenther@suse.de>
14350 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14352 (scop_detection::build_scop_breadth): ... this. Removed.
14353 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14354 (scop_detection::harmful_stmt_in_bb): Likewise.
14355 (scop_detection::graphite_can_represent_stmt): Likewise.
14356 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14357 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14358 (scop_detection::can_represent_loop_1): ... this. Removed.
14359 (scop_detection::harmful_loop_in_region): Simplify after inlining
14360 the above and remove more quadraticness.
14361 (build_scops): Adjust.
14362 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14365 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14368 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14369 REGNO (base) == SP_REG if base is a REG.
14371 PR middle-end/35691
14372 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14373 if it is different SSA_NAME.
14374 (optimize_range_tests_cmp_bitwise): New function.
14375 (optimize_range_tests): Call it.
14377 2017-09-26 Richard Biener <rguenther@suse.de>
14379 PR tree-optimization/82321
14380 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14381 for the def being inside the loop.
14383 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14385 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14387 * config/s390/s390-builtins.def: Fix constraint on op4.
14389 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14391 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14392 independent expanders.
14393 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14394 ("vec_ordered", "vec_unordered"): New expanders.
14396 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14398 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14401 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14403 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14404 vec_unpacks_lo_v16qi.
14405 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14407 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14409 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14410 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14411 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14413 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14415 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14417 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14419 ("*vec_slb<mode>"): New insn pattern.
14420 ("vec_shr_<mode>"): New expander.
14421 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14422 and force the shift count operand to V16QImode.
14423 ("vec_srb<mode>"): Set shift count mode to V16QI.
14425 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14427 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14428 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14429 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14431 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14434 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14436 2017-09-26 Richard Biener <rguenther@suse.de>
14438 PR tree-optimization/82320
14439 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14442 2017-09-25 Jeff Law <law@redhat.com>
14444 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14445 prototype for new argument.
14446 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14447 mostly extracted from rs6000_emit_allocate_stack.
14448 (rs6000_emit_probe_stack_range_stack_clash): New function.
14449 (rs6000_emit_allocate_stack): Call
14450 rs6000_emit_probe_stack_range_stack_clash as needed.
14451 (rs6000_emit_probe_stack_range): Add additional argument
14452 to call to gen_probe_stack_range{si,di}.
14453 (output_probe_stack_range): New.
14454 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14455 (output_probe_stack_range_stack_clash): New.
14456 (rs6000_emit_prologue): Emit notes into dump file as requested.
14457 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14458 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14459 Add additional operand and pass it to output_probe_stack_range.
14461 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14463 PR tree-optimization/82163
14464 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14465 (checking_verify_loop_closed_ssa): New parameter.
14466 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14467 (check_loop_closed_ssa_stmt): Delete.
14468 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14469 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14470 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14473 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14475 * brig-builtins.def: Treat HSAIL barrier builtins as
14476 setjmp/longjump style functions.
14478 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14480 * target.def (constant_alignment): New hook.
14481 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14482 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14483 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14484 * doc/tm.texi: Regenerate.
14485 * targhooks.h (default_constant_alignment): Declare.
14486 (constant_alignment_word_strings): Likewise.
14487 * targhooks.c (default_constant_alignment): New function.
14488 (constant_alignment_word_strings): Likewise.
14489 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14490 instead of CONSTANT_ALIGNMENT.
14491 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14492 (force_const_mem): Likewise.
14493 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14494 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14495 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14496 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14497 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14499 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14500 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14501 constant_alignment_word_strings.
14502 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14503 (CONSTANT_ALIGNMENT): Likewise.
14504 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14505 (arm_constant_alignment): New function.
14506 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14507 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14508 constant_alignment_word_strings.
14509 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14510 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14511 constant_alignment_word_strings.
14512 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14513 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14514 (cris_constant_alignment): New function.
14515 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14516 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14517 (epiphany_constant_alignment): New function.
14518 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14519 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14520 constant_alignment_word_strings.
14521 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14522 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14523 constant_alignment_word_strings.
14524 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14525 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14526 constant_alignment_word_strings.
14527 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14528 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14529 * config/i386/i386.c (ix86_constant_alignment): Make static.
14530 Use the same interface as the target hook.
14531 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14532 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14533 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14534 constant_alignment_word_strings.
14535 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14536 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14537 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14538 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14539 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14540 constant_alignment_word_strings.
14541 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14542 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14543 constant_alignment_word_strings.
14544 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14545 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14546 constant_alignment_word_strings.
14547 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14548 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14550 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14551 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14552 * config/mips/mips.c (mips_constant_alignment): New function.
14553 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14554 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14555 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14556 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14557 (mmix_constant_alignment): Make static. Use the same interface
14558 as the target hook.
14559 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14560 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14561 constant_alignment_word_strings.
14562 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14563 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14564 constant_alignment_word_strings.
14565 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14566 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14567 constant_alignment_word_strings.
14568 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14569 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14570 (rs6000_constant_alignment): New function.
14571 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14572 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14573 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14574 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14575 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14576 (rs6000_constant_alignment): New function.
14577 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14578 * config/s390/s390.c (s390_constant_alignment): New function.
14579 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14580 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14581 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14582 constant_alignment_word_strings.
14583 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14584 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14585 (sparc_constant_alignment): New function.
14586 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14587 * config/spu/spu.c (spu_constant_alignment): New function.
14588 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14589 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14590 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14591 constant_alignment_word_strings.
14592 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14593 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14594 constant_alignment_word_strings.
14595 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14596 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14597 constant_alignment_word_strings.
14598 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14599 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14600 (visium_constant_alignment): New function.
14601 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14602 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14603 (xtensa_constant_alignment): New function.
14604 * system.h (CONSTANT_ALIGNMENT): Poison.
14606 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14608 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14609 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14610 (rs6000_builtin_valid_without_lhs): New helper function.
14611 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14612 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14614 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14616 * target.h (vec_perm_indices): Use unsigned short rather than
14618 (auto_vec_perm_indices): Likewise.
14619 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14620 Use unsigned int rather than unsigned char.
14621 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14623 2017-09-25 Richard Biener <rguenther@suse.de>
14625 * cfgloop.h (sort_sibling_loops): Declare.
14626 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14627 (sort_sibling_loops): New function sorting the sibling loop list
14629 * graphite.c (graphite_transform_loops): Sort sibling loops.
14631 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14633 * target.def (vec_perm_const_ok): Change sel parameter to
14635 * optabs-query.c (can_vec_perm_p): Update accordingly.
14636 * doc/tm.texi: Regenerate.
14637 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14638 auto_vec_perm_indices and remove separate nelt field.
14639 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14640 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14641 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14642 (aarch64_expand_vec_perm_const): Update accordingly.
14643 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14644 to vec_perm_indices.
14645 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14646 auto_vec_perm_indices and remove separate nelt field.
14647 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14648 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14649 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14651 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14652 to vec_perm_indices.
14653 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14654 sel to vec_perm_indices.
14655 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14656 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14657 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14659 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14662 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14665 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14666 on the FUNCTION_DECL function context if it has a DIE that is a
14669 2017-09-25 Richard Biener <rguenther@suse.de>
14671 PR tree-optimization/82285
14672 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14675 2017-09-25 Tom de Vries <tom@codesourcery.com>
14679 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14682 2017-09-25 Richard Biener <rguenther@suse.de>
14684 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14685 ISL errors other than isl_error_quota happen. Dump if the
14686 schedule is the same.
14687 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14688 errors instead of aborting inside ISL.
14690 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14693 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14694 of libgcc_eh for m64.
14695 * config/i386/darwin64.h: Likewise.
14697 2017-09-25 Richard Biener <rguenther@suse.de>
14699 PR middle-end/82144
14700 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14701 attribute for incomplete types nor twice for complete ones.
14703 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14706 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14707 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14710 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14713 * config/i386/i386.opt (mprefer-avx256): Use
14714 ix86_target_flags variable.
14715 * config/i386/i386.c (ix86_target_string): Move
14716 -mprefer-avx256 to flag2_opts.
14718 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14720 PR middle-end/35691
14721 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14722 and x != -1 | y != -1 into (x & y) != -1.
14724 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14726 * config.gcc: Add new case statement to set
14727 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14728 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14729 s390x-*-linux* case statements. Added aarch64 to the list of
14730 supported architectures.
14732 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14734 PR tree-optimization/82289
14735 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14736 STMT_VINFO_RELEVANT_P.
14738 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14739 Alan Hayward <alan.hayward@arm.com>
14740 David Sherwood <david.sherwood@arm.com>
14742 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14743 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14744 (extract_range_from_binary_expr_1): Don't call
14745 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14747 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14748 Alan Hayward <alan.hayward@arm.com>
14749 David Sherwood <david.sherwood@arm.com>
14751 * target.def (preferred_vector_alignment): New hook.
14752 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14754 * doc/tm.texi: Regenerate.
14755 * targhooks.h (default_preferred_vector_alignment): Declare.
14756 * targhooks.c (default_preferred_vector_alignment): New function.
14757 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14759 (DR_TARGET_ALIGNMENT): New macro.
14760 (aligned_access_p): Update commentary.
14761 (vect_known_alignment_in_bytes): New function.
14762 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14764 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14765 Calculate the misalignment based on the target alignment rather than
14767 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14768 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14769 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14770 the target alignment, rather than masking the element misalignment
14771 with the number of elements in a vector. Also use the target
14772 alignment when calculating the maximum number of peels.
14773 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14774 instead of TYPE_ALIGN_UNIT.
14775 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14776 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14777 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14778 (vect_create_data_ref_ptr): Likewise.
14779 (vect_setup_realignment): Realign by ANDing with
14780 -DR_TARGET_MISALIGNMENT.
14781 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14782 the number of peels based on DR_TARGET_ALIGNMENT.
14783 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14784 with the guaranteed alignment boundary when deciding whether
14786 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14787 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14788 (ensure_base_align): Remove stmt_info parameter. Get the
14789 target base alignment from DR_TARGET_ALIGNMENT.
14790 (vectorizable_store): Update call accordingly. Interpret
14791 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14793 (vectorizable_load): Likewise.
14795 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14796 Alan Hayward <alan.hayward@arm.com>
14797 David Sherwood <david.sherwood@arm.com>
14799 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14800 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14801 (vect_enhance_data_refs_alignment): Likewise.
14803 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14805 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14806 error. Only quit immediately if parsing is complete.
14807 (BEGIN): Initialize fatal_err and parse_done.
14808 (begin fpu, end fpu): Check number of arguments.
14809 (begin arch, end arch): Likewise.
14810 (begin cpu, end cpu): Likewise.
14811 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14812 (optalias): Likewise.
14814 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14816 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14817 * config/arm/arm-isa.h: Delete. Move definitions to ...
14818 * arm-cpus.in: ... here. Use new feature and fgroup values.
14819 * config/arm/arm.c (arm_option_override): Use lower case for feature
14821 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14822 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14823 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14825 (print_isa_bits_for): New function.
14826 (gen_isa): New function.
14827 (gen_comm_data): Use print_isa_bits_for.
14828 (define feature): New keyword.
14829 (define fgroup): New keyword.
14830 * config/arm/t-arm (TM_H): Remove.
14831 (GTM_H): Add arm-isa.h.
14832 (arm-isa.h): Add rule to generate file.
14833 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14834 case for feature bit names.
14836 2017-09-22 Richard Biener <rguenther@suse.de>
14838 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14840 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14841 print of no dependency loops ...
14842 * graphite.c (graphite_transform_loops): ... here.
14843 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14845 (same_close_phi_node, remove_duplicate_close_phi,
14846 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14847 (canonicalize_loop_closed_ssa): ... here and simplify.
14848 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14849 (optimize_isl): Use dump_printf_loc to tell when we stopped
14850 optimizing because of an ISL timeout.
14852 2017-09-22 Richard Biener <rguenther@suse.de>
14854 PR tree-optimization/82291
14855 * tree-if-conv.c (predicate_mem_writes): Make sure to
14856 remove writes in blocks predicated with false.
14858 2017-09-22 Richard Biener <rguenther@suse.de>
14860 * sese.c: Include cfganal.h.
14861 (if_region_set_false_region): Remove.
14862 (create_if_region_on_edge): Likewise.
14863 (move_sese_in_condition): Re-implement without destroying
14866 2017-09-22 Richard Biener <rguenther@suse.de>
14868 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14869 Verify both BBs contain loop PHI nodes before dispatching to
14870 copy_loop_phi_args.
14871 (graphite_regenerate_ast_isl): Do not recompute dominators,
14872 do not verify three times. Restructure for clarity.
14873 * graphite-scop-detection.c (same_close_phi_node,
14874 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14875 defined_in_loop_p, canonicalize_loop_closed_ssa,
14876 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14877 checking and SSA rewrite, move to ...
14878 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14879 (graphite_initialize): Do not pass in ctx, do not reset the
14880 SCEV cache, compute only dominators.
14881 (graphite_transform_loops): Allocate ISL ctx after
14882 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14883 Maintain post-dominators only around build_scops.
14884 * sese.c (if_region_set_false_region): Make static. Free
14885 and recompute dominators.
14886 (move_sese_in_condition): Assert we don't get called with
14887 post-dominators computed.
14888 * sese.h (if_region_set_false_region): Remove.
14890 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14892 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14893 mode attribute for TARGET_AVX512VL.
14895 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14897 * config/i386/i386.opt (mprefer-avx256): New option.
14898 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14900 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14901 for TARGET_PREFER_AVX256.
14902 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14904 2017-09-21 Jeff Law <law@redhat.com>
14906 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14907 Fix dump output if the only stack space is for pushed registers.
14909 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14911 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14914 2017-09-21 Martin Sebor <msebor@redhat.com>
14917 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14918 code (in C++) or code that triggers warnings.
14920 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14922 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14924 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14926 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14927 * sched-rgn.c: Ditto.
14928 * sel-sched-ir.c: Ditto.
14930 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14932 * toplev.h (set_random_seed): Adjust return type.
14933 * toplev.c (init_local_tick): Move eager initialization of random_seed
14934 to get_random_seed. Adjust comment.
14935 (init_random_seed): Inline to get_random_seed, delete.
14936 (get_random_seed): Initialize random_seed lazily.
14937 (set_random_seed): Do not return previous value.
14938 (print_switch_value): Do not call get_random_seed.
14940 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14942 * cgraph.c (delete_function_version): New, broken out from...
14943 (cgraph_node::delete_function_version): ...here. Rename to
14944 cgraph_node::delete_function_version_by_decl. Update all uses.
14945 (cgraph_node::remove): Call delete_function_version.
14947 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14950 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14951 VAR_DECLs to which addressable non-volatile parameters are mapped
14952 and for id->retvar after the return value assignment. Clear
14953 id->retval and id->retbnd after inlining.
14955 2017-09-21 Richard Biener <rguenther@suse.de>
14957 PR tree-optimization/82276
14958 PR tree-optimization/82244
14959 * tree-vrp.c (build_assert_expr_for): Set
14960 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14962 (remove_range_assertions): Revert earlier change.
14964 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14967 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14969 2017-09-21 Richard Biener <rguenther@suse.de>
14971 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14972 Restore valid IL after code generation errors.
14973 * graphite.c (graphite_transform_loops): Diagnose code
14974 generation issues as MSG_MISSED_OPTIMIZATION and continue
14975 with processing SCOPs.
14977 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14978 Alan Hayward <alan.hayward@arm.com>
14979 David Sherwood <david.sherwood@arm.com>
14981 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14982 rather than choosing between plus_constant and gen_rtx_<CODE>.
14983 * expr.c (emit_push_insn): Likewise.
14984 (expand_expr_real_2): Likewise.
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 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14992 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14993 Alan Hayward <alan.hayward@arm.com>
14994 David Sherwood <david.sherwood@arm.com>
14996 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14997 calling tree_to_uhwi.
14999 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
15000 Alan Hayward <alan.hayward@arm.com>
15001 David Sherwood <david.sherwood@arm.com>
15003 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
15004 INTEGER_CST rather than a negative test for ADDR_EXPR.
15006 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
15007 Alan Hayward <alan.hayward@arm.com>
15008 David Sherwood <david.sherwood@arm.com>
15010 * tree-vrp.c (extract_range_from_binary_expr_1): Check
15011 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
15013 2017-09-21 Richard Biener <rguenther@suse.de>
15015 PR tree-optimization/71351
15016 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
15017 graphite_create_new_loop_guard): Remove, fold remaining parts
15019 (translate_isl_ast_node_for): ... here and simplify.
15021 2017-09-21 Jakub Jelinek <jakub@redhat.com>
15024 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
15025 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
15026 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
15027 alternative always use QI mode, for -Os imov (=R,R) alternative
15028 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
15031 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15032 Jeff Law <law@redhat.com>
15034 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
15035 (allocate_stack_space): New function, partially extracted from
15036 s390_emit_prologue.
15037 (s390_emit_prologue): Track offset to most recent stack probe.
15038 Code to allocate space moved into allocate_stack_space.
15039 Dump actions when no stack is allocated.
15040 (s390_prologue_plus_offset): New function.
15041 (s390_emit_stack_probe): Likewise.
15043 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
15045 * common.opt (Wa, Wl, Wp, g, gz=): Add
15047 (gno-column-info): Remove.
15048 (gcolumn-info): Drop RejectNegative.
15049 (gno-): New prefix.
15050 (gno-record-gcc-switches): Remove.
15051 (grecord-gcc-switches): Drop RejectNegative.
15052 (gno-split-dwarf): Remove.
15053 (gsplit-dwarf): Drop RejectNegative.
15054 (gno-strict-dwarf): Remove.
15055 (gstrict-dwarf): Drop RejectNegative.
15056 * config/darwin.opt (gfull, gused): Add RejectNegative.
15057 * dwarf2out.c (gen_producer_string): Drop
15058 gno-record-gcc-switches handler.
15059 * optc-gen.awk: Add g to prefixes with negative forms.
15060 * opts-common.c (remapping_prefix_p): New.
15061 (find_opt): Check it.
15062 (generate_canonical_option): Test g prefix.
15063 (option_map): Add -gno- mapping.
15064 (add_misspelling_candidates): Check remapping_prefix_p.
15066 2017-09-20 Jeff Law <law@redhat.com>
15068 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
15069 thinko in stack clash protection support.
15071 * explow.c (compute_stack_clash_protection_loop_data): Use
15072 CONST_INT_P instead of explicit test. Verify object is a
15073 CONST_INT_P before looking at INTVAL.
15074 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
15075 instead of explicit test.
15077 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
15080 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
15081 address instead of to r1 and r11.
15083 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
15085 * config.gcc: Support "knm".
15086 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
15087 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15089 * config/i386/i386.c (m_KNM): Define.
15090 (processor_target_table): Add "knm".
15092 (ix86_option_override_internal): Add "knm".
15093 (ix86_issue_rate): Add PROCESSOR_KNM.
15094 (ix86_adjust_cost): Ditto.
15095 (ia32_multipass_dfa_lookahead): Ditto.
15096 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
15097 (fold_builtin_cpu): Add M_INTEL_KNM.
15098 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
15099 (processor_type): Add PROCESSOR_KNM.
15100 * config/i386/x86-tune.def: Add m_KNM.
15101 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
15103 2017-09-20 Richard Biener <rguenther@suse.de>
15105 PR tree-optimization/80213
15106 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
15107 are allowed in empty BBs as well.
15108 (canonicalize_loop_closed_ssa): Also look for other complex
15110 (scop_detection::get_sese): Include the loop-closed PHI block
15112 (scop_detection::merge_sese): Remove code adding extra blocks.
15113 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
15114 (build_scops): Assert the final returned scop is invalid.
15116 2017-09-20 Richard Biener <rguenther@suse.de>
15118 PR tree-optimization/82264
15119 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
15121 (vn_phi_lookup): Likewise.
15122 (vn_phi_insert): Likewise.
15124 2017-09-20 Jakub Jelinek <jakub@redhat.com>
15126 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
15127 that fits into uhwi or shwi, add DW_AT_const_value regardless
15128 of early_dwarf without going through RTL, using add_AT_unsigned
15131 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
15132 (DEBUG_LTO_ABBREV_SECTION): Likewise.
15133 (DEBUG_LTO_MACINFO_SECTION): Likewise.
15134 (DEBUG_MACRO_SECTION): Likewise.
15135 (DEBUG_LTO_MACRO_SECTION): Likewise.
15136 (DEBUG_STR_DWO_SECTION): Likewise.
15137 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
15138 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
15139 (DEBUG_LTO_DWO_LINE_SECTION): Define.
15140 (DEBUG_LTO_LINE_STR_SECTION): Define.
15141 (init_sections_and_labels): Initialize debug_line_str_section
15142 variable. Initialize debug_loc_section for -gdwarf-5 to
15143 DEBUG_LOCLISTS_SECTION. Formatting fixes.
15145 2017-09-20 Richard Biener <rguenther@suse.de>
15147 * graphite-sese-to-poly.c (extract_affine): Properly handle
15148 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
15150 2017-09-20 Richard Biener <rguenther@suse.de>
15152 PR tree-optimization/81373
15153 * graphite-scop-detection.c (build_cross_bb_scalars_def):
15154 Force SESE live-out defs to be handled even if they are
15157 2017-09-19 Jeff Law <law@redhat.com>
15159 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
15160 nothing for stack adjustments with REG_STACK_CHECK.
15161 * sched-deps.c (parse_add_or_inc): Reject insns with
15162 REG_STACK_CHECK from dependency breaking.
15163 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
15164 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
15165 * reg-notes.def (STACK_CHECK): New note.
15167 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
15168 (ix86_expand_prologue): Dump stack clash info as needed.
15169 Call ix86_adjust_stack_and_probe_stack_clash as needed.
15171 * function.c (dump_stack_clash_frame_info): New function.
15172 * function.h (dump_stack_clash_frame_info): Prototype.
15173 (enum stack_clash_probes): New enum.
15175 * config/alpha/alpha.c (alpha_expand_prologue): Also check
15176 flag_stack_clash_protection.
15177 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
15178 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
15179 (arm_frame_pointer_required): Likewise.
15180 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
15181 (ia64_expand_prologue): Likewise.
15182 * config/mips/mips.c (mips_expand_prologue): Likewise.
15183 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
15184 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15185 (sparc_flat_expand_prologue): Likewise.
15186 * config/spu/spu.c (spu_expand_prologue): Likewise.
15188 * explow.c: Include "params.h".
15189 (anti_adjust_stack_and_probe_stack_clash): New function.
15190 (get_stack_check_protect): Likewise.
15191 (compute_stack_clash_protection_loop_data): Likewise.
15192 (emit_stack_clash_protection_loop_start): Likewise.
15193 (emit_stack_clash_protection_loop_end): Likewise.
15194 (allocate_dynamic_stack_space): Use get_stack_check_protect.
15195 Use anti_adjust_stack_and_probe_stack_clash.
15196 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
15197 (emit_stack_clash_protection_loop_start): Likewise.
15198 (emit_stack_clash_protection_loop_end): Likewise.
15199 * rtl.h (get_stack_check_protect): Prototype.
15200 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
15201 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
15202 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
15204 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
15206 * doc/tm.texi: Rebuilt.
15207 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
15208 get_stack_check_protect.
15209 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
15210 * config/arm/arm.c (arm_expand_prologue): Likewise.
15211 (arm_frame_pointer_required): Likewise.
15212 * config/i386/i386.c (ix86_expand_prologue): Likewise.
15213 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
15214 * config/mips/mips.c (mips_expand_prologue): Likewise.
15215 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
15216 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
15217 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15218 (sparc_flat_expand_prologue): Likewise.
15220 * common.opt (-fstack-clash-protection): New option.
15221 * flag-types.h (enum stack_check_type): Note difference between
15222 -fstack-check= and -fstack-clash-protection.
15223 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15224 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15225 * toplev.c (process_options): Issue warnings/errors for cases
15226 not handled with -fstack-clash-protection.
15227 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15228 (-fstack-check): Note additional problem with -fstack-check=generic.
15229 Note that -fstack-check is primarily for Ada and refer users
15230 to -fstack-clash-protection for stack-clash-protection.
15231 Document new params for stack clash protection.
15233 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15235 * config/i386/i386.c (ix86_split_long_move): Do not handle
15236 address used for LEA in a special way.
15238 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15240 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15242 2017-09-19 Martin Sebor <msebor@redhat.com>
15245 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15246 of incompatible types.
15248 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15250 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15251 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15252 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15253 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15255 2017-09-19 Richard Biener <rguenther@suse.de>
15257 PR tree-optimization/82244
15258 * tree-vrp.c (remove_range_assertions): Do not propagate
15259 a constant to abnormals but replace the assert with a copy.
15261 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15263 PR rtl-optimization/57878
15264 PR rtl-optimization/68988
15265 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15266 avoidance test involving non_reload_pseudos. Move frequency test
15267 below the general fragmentation avoidance test.
15269 2017-09-19 Richard Biener <rguenther@suse.de>
15271 PR tree-optimization/69728
15272 * graphite-sese-to-poly.c (schedule_error): New global.
15273 (add_loop_schedule): Handle empty domain by failing the
15275 (build_original_schedule): Handle schedule_error.
15277 2017-09-19 Richard Biener <rguenther@suse.de>
15279 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15280 Do not iterate to sibling loops but only to siblings of inner
15283 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15286 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15287 that the registers are different.
15289 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15291 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15292 to processor_model and "amdfam17h" to arch_names_table.
15293 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15295 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15298 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15300 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15301 Alan Hayward <alan.hayward@arm.com>
15302 David Sherwood <david.sherwood@arm.com>
15304 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15306 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15308 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15309 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15310 vect_schedule_slp_instance.
15311 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15312 Update call to vect_slp_analyze_node_operations. Simplify return
15314 (vect_slp_analyze_bb_1): Update call accordingly.
15315 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15316 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15317 (vect_schedule_slp): Update call accordingly.
15319 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15320 Alan Hayward <alan.hayward@arm.com>
15321 David Sherwood <david.sherwood@arm.com>
15323 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15324 with types that aren't in fact scalar.
15326 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15328 * tree-vect-slp.c (vect_record_max_nunits): New function,
15330 (vect_build_slp_tree_1): ...here.
15331 (vect_build_slp_tree_2): Call it for phis too.
15333 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15335 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15336 to vect_get_vec_def_for_operand when getting the mask operand.
15338 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15339 Alan Hayward <alan.hayward@arm.com>
15340 David Sherwood <david.sherwood@arm.com>
15342 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15345 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15346 Alan Hayward <alan.hayward@arm.com>
15347 David Sherwood <david.sherwood@arm.com>
15349 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15350 calculation for vector booleans.
15352 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15353 Alan Hayward <alan.hayward@arm.com>
15354 David Sherwood <david.sherwood@arm.com>
15356 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15358 (vect_transform_stmt): ...here.
15359 (vect_analyze_stmt): Use it instead of calling
15360 vectorizable_live_operation directly.
15362 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15364 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15365 non-SIMT targets in acc vector loops.
15367 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15369 * configure.ac: Add arc and check if assembler supports gdwarf2.
15370 * configure: Regenerate.
15372 2017-09-18 Richard Biener <rguenther@suse.de>
15374 PR tree-optimization/82220
15375 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15376 epilogue niters from the min_profitable_iters compute.
15378 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15381 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15382 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15383 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15385 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15388 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15389 switching to a new text section.
15391 2017-09-18 Richard Biener <rguenther@suse.de>
15393 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15395 (build_alias_set): Reject aliases with no access function.
15397 2017-09-18 Richard Biener <rguenther@suse.de>
15399 PR tree-optimization/79622
15400 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15402 (build_cross_bb_scalars_use): Likewise.
15404 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15406 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15408 2017-09-18 Alan Modra <amodra@gmail.com>
15411 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15412 stack_pointer_rtx for count 0. Update comments. Break up
15413 large rtl expression.
15415 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15417 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15418 Increase to 20 bytes.
15419 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15420 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15421 or avx version of the stub.
15423 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15426 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15427 compute the minimum stack alignment. Also update preferred stack
15428 boundary for leaf functions.
15430 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15432 PR tree-optimization/82228
15433 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15436 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15438 * common/config/nds32/nds32-common.c
15439 (nds32_option_optimization_table): Refine formatting.
15440 (nds32_option_optimization_table): Use -fsched-pressure and
15441 -fomit-frame-pointer for specific optimization level.
15443 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15445 * config/nds32/nds32.c: Refine formatting and comments.
15446 * config/nds32/nds32.h: Likewise.
15447 * config/nds32/nds32.md: Likewise.
15448 * config/nds32/nds32-cost.c: Likewise.
15449 * config/nds32/nds32-isr.c: Likewise.
15450 * config/nds32/nds32-md-auxiliary.c: Likewise.
15451 * config/nds32/nds32-multiple.md: Likewise.
15452 * config/nds32/nds32-predicates.c: Likewise.
15454 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15455 Jakub Jelinek <jakub@redhat.com>
15457 Add support for -std=c++2a.
15458 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15460 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15462 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15465 * doc/extend.texi (Common Function Attributes): Add
15466 references to ARM, AArch64, and S/390 specific attributes.
15467 (Function Specific Option Pragmas): Add AArch64 and S/390
15468 to list of back ends that support the target pragma.
15470 2017-09-15 Nathan Sidwell <nathan@acm.org>
15472 * doc/standards.texi: Fix C++17 description. Update URLs for
15475 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15477 * common.opt (Wcast-align=strict): New warning option.
15478 * doc/invoke.texi: Document -Wcast-align=strict.
15480 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15482 * cgraph.h (cgraph_thunk_info): Add comments.
15483 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15484 assert for VIRTUAL_* arguments stricter.
15486 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15488 PR tree-optimization/71026
15489 * match.pd: Move RDIV patterns from fold-const.c
15490 * fold-const.c (distribute_real_division): Removed.
15491 (fold_binary_loc): Remove calls to distribute_real_divison.
15493 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15495 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15496 c++1z and gnu++1z as deprecated. Change other references to
15497 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15498 Change -Wc++1z-compat to -Wc++17-compat.
15499 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15500 * dwarf2out.c (highest_c_language): Handle C++17.
15501 (gen_compile_unit_die): Likewise.
15503 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15505 PR rtl-optimization/82192
15506 * combine.c (make_extraction): Don't look through non-paradoxical
15507 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15510 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15511 Alan Hayward <alan.hayward@arm.com>
15512 David Sherwood <david.sherwood@arm.com>
15514 * target.def (function_arg_offset): New hook.
15515 * targhooks.h (default_function_arg_offset): Declare.
15516 * targhooks.c (default_function_arg_offset): New function.
15517 * function.c (locate_and_pad_parm): Use
15518 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15519 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15520 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15521 * doc/tm.texi: Regenerate.
15522 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15523 * config/spu/spu.c (spu_function_arg_offset): New function.
15524 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15525 * system.h (FUNCTION_ARG_OFFSET): Poison.
15527 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15528 Alan Hayard <alan.hayward@arm.com>
15529 David Sherwood <david.sherwood@arm.com>
15531 * target.def (truly_noop_truncation): New hook.
15532 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15533 than TRULY_NOOP_TRUNCATION.
15534 * hooks.h (hook_bool_uint_uint_true): Declare.
15535 * hooks.c (hook_bool_uint_uint_true): New function.
15536 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15537 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15538 * doc/tm.texi: Regenerate.
15539 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15540 rather than TRULY_NOOP_TRUNCATION in comments.
15541 (simplify_comparison): Likewise.
15542 (record_truncated_value): Likewise.
15543 * expmed.c (extract_bit_field_1): Likewise.
15544 (extract_split_bit_field): Likewise.
15545 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15546 instead of TRULY_NOOP_TRUNCATION.
15547 * function.c (assign_parm_setup_block): Likewise.
15548 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15549 * rtlhooks.c: Include target.h.
15550 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15551 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15552 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15553 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15554 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15555 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15556 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15557 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15558 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15559 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15560 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15561 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15562 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15563 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15564 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15565 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15566 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15567 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15568 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15569 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15570 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15571 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15572 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15573 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15574 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15575 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15576 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15577 rather than TRULY_NOOP_TRUNCATION in comments.
15578 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15579 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15580 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15581 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15582 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15583 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15584 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15585 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15586 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15587 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15588 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15589 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15590 rather than TRULY_NOOP_TRUNCATION in comments.
15591 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15592 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15593 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15594 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15595 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15596 TRULY_NOOP_TRUNCATION condition.
15597 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15598 (TRULY_NOOP_TRUNCATION): Delete.
15599 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15600 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15601 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15602 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15603 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15604 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15605 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15606 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15607 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15608 rather than TRULY_NOOP_TRUNCATION in comments.
15609 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15610 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15611 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15612 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15613 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15614 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15616 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15619 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15620 (*cmp_ior): Likewise.
15621 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15622 (*ior_scc_scc_cmp): Likewise.
15623 (*and_scc_scc): Likewise.
15624 (*and_scc_scc_cmp): Likewise.
15626 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15627 Alan Hayard <alan.hayward@arm.com>
15628 David Sherwood <david.sherwood@arm.com>
15630 * target.def (can_change_mode_class): New hook.
15631 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15632 (hard_regno_nregs): Likewise.
15633 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15634 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15635 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15636 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15637 (LOAD_EXTEND_OP): Update accordingly.
15638 * doc/tm.texi: Regenerate.
15639 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15640 CANNOT_CHANGE_MODE_CLASS.
15641 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15642 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15643 * combine.c (simplify_set): Update accordingly.
15644 * emit-rtl.c (validate_subreg): Likewise.
15645 * recog.c (general_operand): Likewise.
15646 * regcprop.c (mode_change_ok): Likewise.
15647 * reload1.c (choose_reload_regs): Likewise.
15648 (inherit_piecemeal_p): Likewise.
15649 * rtlanal.c (simplify_subreg_regno): Likewise.
15650 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15651 instead of CANNOT_CHANGE_MODE_CLASS.
15652 (reload_cse_simplify_operands): Likewise.
15653 * reload.c (push_reload): Use targetm.can_change_mode_class
15654 instead of CANNOT_CHANGE_MODE_CLASS.
15655 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15656 REG_CANNOT_CHANGE_MODE_P.
15657 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15658 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15659 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15660 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15661 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15662 (arm_can_change_mode_class): New function.
15663 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15664 than CANNOT_CHANGE_MODE_CLASS in comments.
15665 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15666 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15667 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15668 (ix86_can_change_mode_class): ...this new function, inverting the
15669 sense of the return value.
15670 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15671 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15672 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15673 (ia64_can_change_mode_class): New function.
15674 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15675 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15676 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15677 (m32c_can_change_mode_class): ...this new function, inverting the
15678 sense of the return value.
15679 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15680 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15681 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15682 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15683 (mips_can_change_mode_class): ...this new function, inverting the
15684 sense of the return value.
15685 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15686 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15687 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15688 (msp430_can_change_mode_class): New function.
15689 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15690 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15691 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15692 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15693 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15694 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15695 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15696 (pa_cannot_change_mode_class): Replace with...
15697 (pa_can_change_mode_class): ...this new function, inverting the
15698 sense of the return value.
15699 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15700 than CANNOT_CHANGE_MODE_CLASS in comments.
15701 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15702 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15703 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15704 (pdp11_cannot_change_mode_class): Replace with...
15705 (pdp11_can_change_mode_class): ...this new function, inverting the
15706 sense of the return value.
15707 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15708 * config/powerpcspe/powerpcspe-protos.h
15709 (rs6000_cannot_change_mode_class_ptr): Delete.
15710 * config/powerpcspe/powerpcspe.c
15711 (rs6000_cannot_change_mode_class_ptr): Delete.
15712 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15713 (rs6000_option_override_internal): Assign to
15714 targetm.can_change_mode_class instead of
15715 rs6000_cannot_change_mode_class_ptr.
15716 (rs6000_cannot_change_mode_class): Replace with...
15717 (rs6000_can_change_mode_class): ...this new function, inverting the
15718 sense of the return value.
15719 (rs6000_debug_cannot_change_mode_class): Replace with...
15720 (rs6000_debug_can_change_mode_class): ...this new function.
15721 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15722 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15723 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15724 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15725 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15727 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15728 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15729 (rs6000_option_override_internal): Assign to
15730 targetm.can_change_mode_class instead of
15731 rs6000_cannot_change_mode_class_ptr.
15732 (rs6000_cannot_change_mode_class): Replace with...
15733 (rs6000_can_change_mode_class): ...this new function, inverting the
15734 sense of the return value.
15735 (rs6000_debug_cannot_change_mode_class): Replace with...
15736 (rs6000_debug_can_change_mode_class): ...this new function.
15737 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15738 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15739 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15740 (s390_can_change_mode_class): ...this new function, inverting the
15741 sense of the return value.
15742 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15743 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15744 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15745 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15746 (sh_cannot_change_mode_class): Replace with...
15747 (sh_can_change_mode_class): ...this new function, inverting the
15748 sense of the return value.
15749 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15750 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15751 (sparc_can_change_mode_class): New function.
15752 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15753 * config/spu/spu.c (spu_can_change_mode_class): New function.
15754 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15755 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15756 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15757 (visium_can_change_mode_class): New function.
15758 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15760 2017-09-15 Richard Biener <rguenther@suse.de>
15762 PR tree-optimization/82217
15763 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15764 but not undefined case.
15766 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15769 * postreload.c (reload_cse_simplify_operands): Skip
15770 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15772 2017-09-15 Richard Biener <rguenther@suse.de>
15774 PR tree-optimization/68823
15775 * graphite-scop-detection.c (build_alias_set): If we have a
15776 possible dependence check whether we can handle them by just
15777 looking at the DRs DR_ACCESS_FNs.
15778 (build_scops): If build_alias_set fails, fail the SCOP.
15780 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15782 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15783 to support float128 built-in functions that require the ISA 3.0
15785 (BU_FLOAT128_3_HW): Likewise.
15786 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15787 built-in functions.
15788 (FMAF128): Likewise.
15790 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15791 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15792 floating point instructions.
15793 (rs6000_invalid_builtin): Likewise.
15794 (rs6000_builtin_mask_names): Likewise.
15795 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15796 (RS6000_BTM_FLOAT128_HW): Likewise.
15797 (RS6000_BTM_COMMON): Likewise.
15798 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15800 * doc/extend.texi (RS/6000 built-in functions): Document the
15801 IEEE 128-bit floating point square root and fused multiply-add
15802 built-in functions.
15804 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15806 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15807 reg (r2) isn't in the set of registers defined in the prologue.
15809 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15810 Alan Hayward <alan.hayward@arm.com>
15811 David Sherwood <david.sherwood@arm.com>
15813 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15814 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15815 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15816 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15817 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15819 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15820 max_vectorization_factor.
15821 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15823 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15824 Alan Hayward <alan.hayward@arm.com>
15825 David Sherwood <david.sherwood@arm.com>
15827 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15828 (vect_worthwhile_without_simd_p): Declare.
15829 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15830 (vectorizable_reduction): Use it.
15831 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15832 (vectorizable_operation): Likewise.
15834 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15835 Alan Hayward <alan.hayward@arm.com>
15836 David Sherwood <david.sherwood@arm.com>
15838 * tree-vectorizer.h (vect_get_num_copies): New function.
15839 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15840 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15841 (vectorizable_induction): Likewise.
15842 (vectorizable_live_operation): Likewise.
15843 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15844 (vectorizable_bswap): Likewise.
15845 (vectorizable_call): Likewise.
15846 (vectorizable_conversion): Likewise.
15847 (vectorizable_assignment): Likewise.
15848 (vectorizable_shift): Likewise.
15849 (vectorizable_operation): Likewise.
15850 (vectorizable_store): Likewise.
15851 (vectorizable_load): Likewise.
15852 (vectorizable_condition): Likewise.
15853 (vectorizable_comparison): Likewise.
15854 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15856 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15857 Alan Hayward <alan.hayward@arm.com>
15858 David Sherwood <david.sherwood@arm.com>
15860 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15861 of vect_init_vector.
15863 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15864 Alan Hayward <alan.hayward@arm.com>
15865 David Sherwood <david.sherwood@arm.com>
15867 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15868 an inline wrapper that provides a location.
15869 (gimple_build_vector): Likewise.
15870 * gimple-fold.c (gimple_build_vector_from_val): New function.
15871 (gimple_build_vector): Likewise.
15872 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15873 functions to build the initial value. Always return a gimple value.
15874 (get_initial_defs_for_reduction): Likewise. Only compute
15876 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15877 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15878 (vectorizable_induction): Use gimple_build_vector rather than
15881 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15882 Alan Hayward <alan.hayward@arm.com>
15883 David Sherwood <david.sherwood@arm.com>
15885 * target.h (vec_perm_indices): New typedef.
15886 (auto_vec_perm_indices): Likewise.
15887 * optabs-query.h: Include target.h
15888 (can_vec_perm_p): Take a vec_perm_indices *.
15889 * optabs-query.c (can_vec_perm_p): Likewise.
15890 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15891 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15892 * tree-vect-generic.c (lower_vec_perm): Likewise.
15893 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15894 (vect_grouped_load_supported): Likewise.
15895 (vect_shift_permute_load_chain): Likewise.
15896 (vect_permute_store_chain): Use auto_vec_perm_indices.
15897 (vect_permute_load_chain): Likewise.
15898 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15899 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15900 Update uses of can_vec_perm_p.
15901 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15902 mode with a number of elements. Take a vec_perm_indices *.
15903 (vect_create_epilog_for_reduction): Update accordingly.
15904 Use auto_vec_perm_indices.
15905 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15906 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15907 (vect_transform_slp_perm_load): Likewise.
15908 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15909 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15910 (vect_gen_perm_mask_checked): Likewise.
15911 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15912 (vect_gen_perm_mask_checked): Likewise.
15913 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15914 (vectorizable_store): Likewise.
15915 (vectorizable_load): Likewise.
15916 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15917 (vectorizable_bswap): Likewise.
15919 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15920 Alan Hayward <alan.hayward@arm.com>
15921 David Sherwood <david.sherwood@arm.com>
15923 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15924 * tree.c (build_vector): Likewise.
15925 (build_vector_from_ctor): Update accordingly.
15926 (build_vector_from_val): Likewise.
15927 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15928 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15929 * tree-vect-generic.c (add_rshift): Likewise.
15930 (expand_vector_divmod): Likewise.
15931 (optimize_vector_constructor): Likewise.
15932 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15933 (vect_transform_slp_perm_load): Likewise.
15934 (vect_schedule_slp_instance): Likewise.
15935 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15936 (vectorizable_call): Likewise.
15937 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15938 * expmed.c (make_tree): Likewise.
15939 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15940 a vector passed to build_vector.
15941 (fold_convert_const): Likewise.
15942 (exact_inverse): Likewise.
15943 (fold_ternary_loc): Likewise.
15944 (fold_relational_const): Likewise.
15945 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15946 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15947 (const_unop): Likewise. Store the reduction accumulator in a
15948 variable rather than an array.
15949 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15950 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15951 the new vector, rather than constructing it after the input arrays.
15952 (native_interpret_vector): Use auto_vec<tree> when building
15953 a vector passed to build_vector. Add elements in order.
15954 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15955 auto_vec<tree> when building a vector passed to build_vector.
15956 (vect_create_epilog_for_reduction): Likewise.
15957 (vectorizable_induction): Likewise.
15958 (get_initial_def_for_reduction): Likewise. Fix indentation of
15960 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15962 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15963 passed to build_vector.
15965 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15966 Alan Hayward <alan.hayward@arm.com>
15967 David Sherwood <david.sherwood@arm.com>
15969 * tree-core.h (tree_base::u): Add an "nelts" field.
15970 (tree_vector): Use VECTOR_CST_NELTS as the length.
15971 * tree.c (tree_size): Likewise.
15972 (make_vector): Initialize VECTOR_CST_NELTS.
15973 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15974 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15975 TYPE_VECTOR_SUBPARTS.
15976 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15977 as the number of units, setting it from VECTOR_CST_NELTS.
15978 (const_vector_from_tree): Likewise.
15979 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15980 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15981 (fold_negate_expr_1): Likewise.
15982 (fold_convert_const): Likewise.
15983 (const_binop): Likewise. Differentiate the number of output and
15985 (const_unop): Likewise.
15986 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15987 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15988 in cases that did the opposite.
15990 2017-09-14 Richard Biener <rguenther@suse.de>
15992 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15995 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15997 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15999 2017-09-14 Jakub Jelinek <jakub@redhat.com>
16002 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
16003 if and where to split a bb, except for splitting before debug insn
16004 sequences followed by non-label real insn. Delete debug insns
16005 in between basic blocks.
16007 * combine.c (make_compound_operation_int): Formatting fixes.
16009 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
16010 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16011 * config/netbsd.h (LINK_EH_SPEC): Likewise.
16012 * config/sol2.h (LINK_EH_SPEC): Likewise.
16013 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16014 * config/s390/linux.h (LINK_SPEC): Likewise.
16015 * config/freebsd.h (LINK_EH_SPEC): Likewise.
16016 * config/openbsd.h (LINK_EH_SPEC): Likewise.
16017 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16018 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
16019 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
16020 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16021 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
16022 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
16024 2017-09-13 Jakub Jelinek <jakub@redhat.com>
16026 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
16028 (ENDFILE_LINUX_SPEC): Likewise.
16029 (LINK_EH_SPEC): Likewise.
16030 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
16031 (LINK_OS_LINUX_SPEC32): Likewise.
16032 (LINK_OS_LINUX_SPEC64): Likewise.
16033 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
16034 (LINK_OS_LINUX_SPEC): Likewise.
16036 2017-09-13 Martin Liska <mliska@suse.cz>
16038 PR middle-end/82154
16039 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
16040 CASE_HIGH is NULL_TREE.
16042 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16043 Alan Hayward <alan.hayward@arm.com>
16044 David Sherwood <david.sherwood@arm.com>
16046 * target.def (secondary_memory_needed): New hook.
16047 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
16048 instead of SECONDARY_MEMORY_NEEDED.
16049 (secondary_memory_needed_mode): Likewise.
16050 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
16051 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
16052 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
16053 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
16054 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
16055 * doc/tm.texi: Regenerate.
16056 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
16057 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
16058 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16059 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
16060 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
16061 * config/i386/i386.c (inline_secondary_memory_needed): Put the
16062 mode argument first and change the reg_class arguments to reg_class_t.
16063 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
16064 Make static. Update the call to inline_secondary_memory_needed.
16065 (ix86_register_move_cost): Update the call to
16066 inline_secondary_memory_needed.
16067 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16068 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
16070 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
16071 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16073 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
16074 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
16075 * config/mips/mips.c (mips_secondary_memory_needed): Make static
16076 and match hook interface. Add comment from mips.h.
16077 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16078 * config/mmix/mmix.md (truncdfsf2): Refer to
16079 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16081 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
16082 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
16083 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16084 (pa_secondary_memory_needed): New function.
16085 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
16086 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
16087 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16088 (pdp11_secondary_memory_needed): Make static and match hook interface.
16089 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
16090 * config/powerpcspe/powerpcspe-protos.h
16091 (rs6000_secondary_memory_needed_ptr): Delete.
16092 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
16094 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16095 (rs6000_option_override_internal): Assign to
16096 targetm.secondary_memory_needed rather than
16097 rs6000_secondary_memory_needed_ptr.
16098 (rs6000_secondary_memory_needed): Match hook interface.
16099 (rs6000_debug_secondary_memory_needed): Likewise.
16100 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
16101 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
16102 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
16103 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16104 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
16105 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
16107 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
16108 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16109 (rs6000_option_override_internal): Assign to
16110 targetm.secondary_memory_needed rather than
16111 rs6000_secondary_memory_needed_ptr.
16112 (rs6000_secondary_memory_needed): Match hook interface.
16113 (rs6000_debug_secondary_memory_needed): Likewise.
16114 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
16115 * config/s390/s390.c (s390_secondary_memory_needed): New function.
16116 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16117 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
16118 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16119 (sparc_secondary_memory_needed): New function.
16120 * lra-constraints.c (check_and_process_move): Refer to
16121 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16123 (curr_insn_transform): Likewise.
16124 (process_alt_operands): Use targetm.secondary_memory_needed
16125 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16126 (check_secondary_memory_needed_p): Likewise.
16127 (choose_split_class): Likewise.
16128 * reload.c: Unconditionally include code that was previously
16129 conditional on SECONDARY_MEMORY_NEEDED.
16130 (push_secondary_reload): Use targetm.secondary_memory_needed
16131 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16132 (push_reload): Likewise.
16133 * reload1.c: Unconditionally include code that was previously
16134 conditional on SECONDARY_MEMORY_NEEDED.
16135 (choose_reload_regs): Use targetm.secondary_memory_needed
16136 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16137 (gen_reload): Likewise.
16138 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
16140 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16141 Alan Hayward <alan.hayward@arm.com>
16142 David Sherwood <david.sherwood@arm.com>
16144 * target.def (secondary_memory_needed_mode): New hook:
16145 * targhooks.c (default_secondary_memory_needed_mode): Declare.
16146 * targhooks.h (default_secondary_memory_needed_mode): New function.
16147 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
16148 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
16149 * doc/tm.texi: Regenerate.
16150 * lra-constraints.c (check_and_process_move): Use
16151 targetm.secondary_memory_needed_mode instead of
16152 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
16153 (curr_insn_transform): Likewise.
16154 * reload.c (get_secondary_mem): Likewise.
16155 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16156 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
16158 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16159 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16160 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
16161 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16162 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
16164 * config/powerpcspe/powerpcspe-protos.h
16165 (rs6000_secondary_memory_needed_mode): Delete.
16166 * config/powerpcspe/powerpcspe.c
16167 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16168 (rs6000_secondary_memory_needed_mode): Make static.
16169 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16170 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
16172 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16174 (rs6000_secondary_memory_needed_mode): Make static.
16175 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16176 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
16177 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16178 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16179 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16181 (sparc_secondary_memory_needed_mode): New function.
16182 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
16184 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
16186 * config/aarch64/constraints.md (Umq): New constraint.
16187 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
16189 (mov<mode>): Update condition.
16191 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16193 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
16194 when bitposition is the same.
16196 2017-09-13 Richard Biener <rguenther@suse.de>
16198 * dwarf2out.c (output_die_symbol): Remove.
16199 (output_die): Do not output a DIEs symbol.
16201 2017-09-13 Richard Biener <rguenther@suse.de>
16203 PR middle-end/82128
16204 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
16205 default-def to avoid breaking iterator update with the weird
16206 interaction with cgraph_update_edges_for_call_stmt_node.
16208 2017-09-13 Richard Biener <rguenther@suse.de>
16210 * tree-cfg.c (verify_gimple_assign_binary): Add verification
16211 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
16212 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
16213 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
16215 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
16217 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
16218 Disable pc relative literal load irrespective of
16219 TARGET_FIX_ERR_A53_84341 for default.
16221 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16223 * config/sparc/sparc.c (output_return): Output the source location of
16224 the insn in the delay slot, if any.
16225 (output_sibcall): Likewise.
16227 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16230 * common.opt (-static-pie): New alias.
16231 (shared): Negate static-pie.
16232 (-no-pie): Update help text.
16234 (static-pie): New option.
16235 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16236 -static-pie support.
16237 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16238 (LINK_EH_SPEC): Likewise.
16239 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16240 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16241 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16242 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16243 (init_gcc_specs): Likewise.
16244 (init_spec): Likewise.
16245 (display_help): Update help message for -pie.
16246 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16249 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16251 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16252 (movdi_aarch64): Likewise.
16253 (movti_aarch64): Likewise.
16255 2017-09-12 Simon Wright <simon@pushface.org>
16258 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16259 calculation of the minor version, always output as 0.
16261 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16264 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16265 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16266 on it early, rather than manual conversion late. For
16267 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16268 instead of performing manual conversion.
16270 2017-09-12 Carl Love <cel@us.ibm.com>
16272 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16273 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16275 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16276 VMULOSW): Add definitions.
16277 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16278 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16279 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16280 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16281 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16283 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16285 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16287 (movti_aarch64): Likewise.
16288 (movdf_aarch64): Likewise.
16289 (movtf_aarch64): Likewise.
16290 (load_pairdi): Likewise.
16291 (store_pairdi): Likewise.
16292 (load_pairdf): Likewise.
16293 (store_pairdf): Likewise.
16294 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16295 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16296 (ldr_got_small_<mode>): Likewise.
16297 (ldr_got_small_28k_<mode>): Likewise.
16298 (ldr_got_tiny): Likewise.
16299 * config/aarch64/iterators.md (ldst_sz): New.
16300 (ldpstp_sz): Likewise.
16301 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16303 (thunderx_load): Split load_8 to load_16.
16304 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16306 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16307 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16308 (xgene1_store_pair): Split store_8 to store_16.
16309 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16310 (falkor_st_0_st_sd): Split store_8 to store_16.
16312 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16314 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16315 and store1/2/3/4 to store_4/8/12/16.
16316 * config/aarch64/aarch64.md: Update for rename.
16317 * config/arm/arm.md: Likewise.: Likewise.
16318 * config/arm/arm.c: Likewise.
16319 * config/arm/thumb1.md: Likewise.
16320 * config/arm/thumb2.md: Likewise.
16321 * config/arm/vfp.md: Likewise.
16322 * config/arm/arm-generic.md: Likewise.
16323 * config/arm/arm1020e.md: Likewise.
16324 * config/arm/arm1026ejs.md: Likewise.
16325 * config/arm/arm1136jfs.md: Likewise.
16326 * config/arm/arm926ejs.md: Likewise.
16327 * config/arm/cortex-a15.md: Likewise.
16328 * config/arm/cortex-a17.md: Likewise.
16329 * config/arm/cortex-a5.md: Likewise.
16330 * config/arm/cortex-a53.md: Likewise.
16331 * config/arm/cortex-a57.md: Likewise.
16332 * config/arm/cortex-a7.md: Likewise.
16333 * config/arm/cortex-a8.md: Likewise.
16334 * config/arm/cortex-a9.md: Likewise.
16335 * config/arm/cortex-m4.md: Likewise.
16336 * config/arm/cortex-m7.md: Likewise.
16337 * config/arm/cortex-r4.md: Likewise.
16338 * config/arm/exynos-m1.md: Likewise.
16339 * config/arm/fa526.md: Likewise.
16340 * config/arm/fa606te.md: Likewise.
16341 * config/arm/fa626te.md: Likewise.
16342 * config/arm/fa726te.md: Likewise.
16343 * config/arm/fmp626.md: Likewise.
16344 * config/arm/iwmmxt.md: Likewise.
16345 * config/arm/ldmstm.md: Likewise.
16346 * config/arm/marvell-pj4.md: Likewise.
16347 * config/arm/xgene1.md: Likewise.
16348 * config/aarch64/thunderx.md: Likewise.
16349 * config/aarch64/thunderx2t99.md: Likewise.
16350 * config/aarch64/falkor.md: Likewise.
16352 2017-09-12 Martin Liska <mliska@suse.cz>
16354 * attribs.c (private_lookup_attribute): New function.
16355 * attribs.h (private_lookup_attribute): Declared here.
16356 (lookup_attribute): Called from this place.
16358 2017-09-12 Richard Biener <rguenther@suse.de>
16360 PR tree-optimization/82157
16361 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16362 stmts with side-effects.
16364 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16365 Alan Hayward <alan.hayward@arm.com>
16366 David Sherwood <david.sherwood@arm.com>
16368 * target.def (hard_regno_nregs): New hook.
16369 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16370 * targhooks.h (default_hard_regno_nregs): Declare.
16371 * targhooks.c (default_hard_regno_nregs): New function.
16372 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16373 (TARGET_HARD_REGNO_NREGS): ...this hook.
16374 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16375 (CLASS_MAX_NREGS): Likewise.
16376 * doc/tm.texi: Regenerate.
16377 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16378 instead of HARD_REGNO_NREGS.
16379 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16380 HARD_REGNO_NREGS in the comment.
16381 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16382 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16383 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16384 Return an unsigned int.
16385 (TARGET_HARD_REGNO_NREGS): Redefine.
16386 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16387 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16388 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16389 (arc_hard_regno_nregs): New function.
16390 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16391 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16392 (arm_hard_regno_nregs): New function.
16393 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16394 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16395 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16396 (TARGET_HARD_REGNO_NREGS): Redefine.
16397 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16398 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16399 (HARD_REGNO_NREGS): Delete.
16400 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16401 (cr16_hard_regno_nregs): New function.
16402 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16403 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16404 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16405 (cris_hard_regno_nregs): New function.
16406 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16407 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16408 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16409 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16410 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16411 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16412 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16413 (frv_hard_regno_nregs): Make static. Take and return an
16415 (frv_class_max_nregs): Remove outdated copy of documentation.
16416 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16417 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16418 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16419 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16420 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16421 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16422 (TARGET_HARD_REGNO_NREGS): Redefine.
16423 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16424 (CLASS_MAX_NREGS): Update comment.
16425 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16426 (ia64_hard_regno_nregs): New function.
16427 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16428 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16429 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16430 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16431 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16433 (m32c_hard_regno_nregs): Likewise. Make static.
16434 (TARGET_HARD_REGNO_NREGS): Redefine.
16435 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16436 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16437 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16438 (m68k_hard_regno_nregs): New function.
16439 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16440 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16441 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16442 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16443 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16444 Take and return an unsigned int.
16445 (TARGET_HARD_REGNO_NREGS): Redefine.
16446 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16447 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16448 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16449 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16450 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16451 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16452 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16453 (msp430_hard_regno_nregs): Make static. Take and return an
16455 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16456 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16457 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16458 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16459 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16460 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16461 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16462 (TARGET_HARD_REGNO_NREGS): Redefine.
16463 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16464 (PA_HARD_REGNO_NREGS): ...this.
16465 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16466 (PA_HARD_REGNO_NREGS): ...this.
16467 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16468 (pa_hard_regno_nregs): New function.
16469 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16470 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16471 (pdp11_hard_regno_nregs): New function.
16472 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16473 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16474 (rs6000_hard_regno_nregs_hook): New function.
16475 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16476 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16477 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16478 Take and return an unsigned int. Move earlier in file.
16479 (TARGET_HARD_REGNO_NREGS): Redefine.
16480 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16481 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16482 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16483 (rl78_hard_regno_nregs): Make static. Take and return an
16485 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16486 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16487 (rs6000_hard_regno_nregs_hook): New function.
16488 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16489 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16490 (TARGET_HARD_REGNO_NREGS): Redefine.
16491 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16492 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16493 instead of HARD_REGNO_NREGS.
16494 (s390_hard_regno_nregs): New function.
16495 (s390_hard_regno_mode_ok): Add comment from s390.h.
16496 (TARGET_HARD_REGNO_NREGS): Redefine.
16497 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16498 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16499 (sh_hard_regno_nregs): New function.
16500 (sh_pass_in_reg_p): Use it.
16501 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16502 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16503 (sparc_hard_regno_nregs): New function.
16504 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16505 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16506 (spu_function_arg_advance): Use it, supplying a valid register number.
16507 (TARGET_HARD_REGNO_NREGS): Redefine.
16508 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16509 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16510 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16511 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16512 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16513 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16514 (CLASS_MAX_NREGS): Remove copy of old documentation.
16515 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16516 (visium_hard_regno_nregs): New function.
16517 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16518 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16519 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16520 xtensa_hard_regno_nregs): New function.
16521 * system.h (HARD_REGNO_NREGS): Poison.
16523 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16525 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16526 hard_regno_nregs instead of HARD_REGNO_NREGS.
16527 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16528 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16529 (c6x_expand_epilogue): Likewise.
16530 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16531 (frv_read_iacc_argument): Likewise.
16532 * config/sh/sh.c: Include regs.h.
16533 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16534 (regs_used): Likewise.
16535 (output_stack_adjust): Likewise.
16536 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16537 * expmed.c: Include regs.h.
16538 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16539 * ree.c: Include regs.h.
16540 (combine_reaching_defs): Use hard_regno_nregs instead of
16542 (add_removable_extension): Likewise.
16544 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16546 * regs.h (hard_regno_nregs): Turn into a function.
16547 (end_hard_regno): Update accordingly.
16548 * caller-save.c (setup_save_areas): Likewise.
16549 (save_call_clobbered_regs): Likewise.
16550 (replace_reg_with_saved_mem): Likewise.
16551 (insert_restore): Likewise.
16552 (insert_save): Likewise.
16553 * combine.c (can_change_dest_mode): Likewise.
16554 (move_deaths): Likewise.
16555 (distribute_notes): Likewise.
16556 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16557 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16558 (rs6000_split_multireg_move): Likewise.
16559 (rs6000_register_move_cost): Likewise.
16560 (rs6000_memory_move_cost): Likewise.
16561 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16562 (rs6000_split_multireg_move): Likewise.
16563 (rs6000_register_move_cost): Likewise.
16564 (rs6000_memory_move_cost): Likewise.
16565 * cselib.c (cselib_reset_table): Likewise.
16566 (cselib_lookup_1): Likewise.
16567 * emit-rtl.c (set_mode_and_regno): Likewise.
16568 * function.c (aggregate_value_p): Likewise.
16569 * ira-color.c (setup_profitable_hard_regs): Likewise.
16570 (check_hard_reg_p): Likewise.
16571 (calculate_saved_nregs): Likewise.
16572 (assign_hard_reg): Likewise.
16573 (improve_allocation): Likewise.
16574 (calculate_spill_cost): Likewise.
16575 * ira-emit.c (modify_move_list): Likewise.
16576 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16577 (ira_hard_reg_in_set_p): Likewise.
16578 * ira.c (setup_reg_mode_hard_regset): Likewise.
16579 (clarify_prohibited_class_mode_regs): Likewise.
16580 (check_allocation): Likewise.
16581 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16582 (lra_setup_reg_renumber): Likewise.
16583 (setup_try_hard_regno_pseudos): Likewise.
16584 (spill_for): Likewise.
16585 (assign_hard_regno): Likewise.
16586 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16587 * lra-constraints.c (in_class_p): Likewise.
16588 (lra_constraint_offset): Likewise.
16589 (simplify_operand_subreg): Likewise.
16590 (lra_constraints): Likewise.
16591 (split_reg): Likewise.
16592 (split_if_necessary): Likewise.
16593 (invariant_p): Likewise.
16594 (inherit_in_ebb): Likewise.
16595 * lra-lives.c (process_bb_lives): Likewise.
16596 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16597 (get_hard_regs): Likewise.
16598 (do_remat): Likewise.
16599 * lra-spills.c (assign_spill_hard_regs): Likewise.
16600 * mode-switching.c (create_pre_exit): Likewise.
16601 * postreload.c (reload_combine_recognize_pattern): Likewise.
16602 * recog.c (peep2_find_free_register): Likewise.
16603 * regcprop.c (kill_value_regno): Likewise.
16604 (set_value_regno): Likewise.
16605 (copy_value): Likewise.
16606 (maybe_mode_change): Likewise.
16607 (find_oldest_value_reg): Likewise.
16608 (copyprop_hardreg_forward_1): Likewise.
16609 * regrename.c (check_new_reg_p): Likewise.
16610 (regrename_do_replace): Likewise.
16611 * reload.c (push_reload): Likewise.
16612 (combine_reloads): Likewise.
16613 (find_dummy_reload): Likewise.
16614 (operands_match_p): Likewise.
16615 (find_reloads): Likewise.
16616 (find_equiv_reg): Likewise.
16617 (reload_adjust_reg_for_mode): Likewise.
16618 * reload1.c (count_pseudo): Likewise.
16619 (count_spilled_pseudo): Likewise.
16620 (find_reg): Likewise.
16621 (clear_reload_reg_in_use): Likewise.
16622 (free_for_value_p): Likewise.
16623 (allocate_reload_reg): Likewise.
16624 (choose_reload_regs): Likewise.
16625 (reload_adjust_reg_for_temp): Likewise.
16626 (emit_reload_insns): Likewise.
16627 (delete_output_reload): Likewise.
16628 * rtlanal.c (subreg_get_info): Likewise.
16629 * sched-deps.c (sched_analyze_reg): Likewise.
16630 * sel-sched.c (init_regs_for_mode): Likewise.
16631 (mark_unavailable_hard_regs): Likewise.
16632 (choose_best_reg_1): Likewise.
16633 (verify_target_availability): Likewise.
16634 * valtrack.c (dead_debug_insert_temp): Likewise.
16635 * var-tracking.c (track_loc_p): Likewise.
16636 (emit_note_insn_var_location): Likewise.
16637 * varasm.c (make_decl_rtl): Likewise.
16638 * reginfo.c (choose_hard_reg_mode): Likewise.
16639 (init_reg_modes_target): Refer directly to
16640 this_target_regs->x_hard_regno_nregs.
16642 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16644 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16645 instead of hard_regno_nregs.
16647 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16649 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16650 end_hard_regno instead of hard_regno_nregs.
16651 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16652 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16653 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16654 * ira-color.c (improve_allocation): Likewise.
16655 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16656 * lra-lives.c (mark_regno_live): Likewise.
16657 (mark_regno_dead): Likewise.
16658 * lra-remat.c (operand_to_remat): Likewise.
16659 * lra.c (collect_non_operand_hard_regs): Likewise.
16660 * postreload.c (reload_combine_note_store): Likewise.
16661 (move2add_valid_value_p): Likewise.
16662 * reload.c (regno_clobbered_p): Likewise.
16664 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16666 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16668 * config/v850/v850.c (v850_reorg): Likewise.
16669 * reload.c (refers_to_regno_for_reload_p): Likewise.
16670 (find_equiv_reg): Likewise.
16671 * reload1.c (reload_reg_reaches_end_p): Likewise.
16673 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16675 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16677 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16678 * config/arm/arm.c (output_move_neon): Likewise.
16679 (arm_attr_length_move_neon): Likewise.
16680 (neon_split_vcombine): Likewise.
16681 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16682 (c6x_mark_reg_written): Likewise.
16683 (c6x_dwarf_register_span): Likewise.
16684 * config/i386/i386.c (ix86_save_reg): Likewise.
16685 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16686 (rws_access_reg): Likewise.
16687 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16688 * mode-switching.c (create_pre_exit): Likewise.
16689 * ree.c (combine_reaching_defs): Likewise.
16690 (add_removable_extension): Likewise.
16691 * regcprop.c (find_oldest_value_reg): Likewise.
16692 (copyprop_hardreg_forward_1): Likewise.
16693 * reload.c (reload_inner_reg_of_subreg): Likewise.
16694 (push_reload): Likewise.
16695 (combine_reloads): Likewise.
16696 (find_dummy_reload): Likewise.
16697 (reload_adjust_reg_for_mode): Likewise.
16698 * reload1.c (find_reload_regs): Likewise.
16699 (forget_old_reloads_1): Likewise.
16700 (reload_reg_free_for_value_p): Likewise.
16701 (reload_adjust_reg_for_temp): Likewise.
16702 (emit_reload_insns): Likewise.
16703 (delete_output_reload): Likewise.
16704 * sel-sched.c (choose_best_reg_1): Likewise.
16705 (choose_best_pseudo_reg): Likewise.
16707 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16708 Alan Hayward <alan.hayward@arm.com>
16709 David Sherwood <david.sherwood@arm.com>
16711 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16712 * target.def (slow_unaligned_access): New hook.
16713 * targhooks.h (default_slow_unaligned_access): Declare.
16714 * targhooks.c (default_slow_unaligned_access): New function.
16715 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16716 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16717 * doc/tm.texi: Regenerate.
16718 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16719 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16720 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16722 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16723 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16725 (rs6000_slow_unaligned_access): New function.
16726 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16727 (expand_block_compare): Likewise.
16728 (expand_strn_compare): Likewise.
16729 (rs6000_rtx_costs): Likewise.
16730 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16731 (riscv_slow_unaligned_access): Likewise.
16732 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16733 (riscv_slow_unaligned_access_p): ...this and make static.
16734 (riscv_option_override): Update accordingly.
16735 (riscv_slow_unaligned_access): New function.
16736 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16737 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16738 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16739 (rs6000_slow_unaligned_access): New function.
16740 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16741 (rs6000_rtx_costs): Likewise.
16742 * config/rs6000/rs6000-string.c (expand_block_compare)
16743 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16744 of SLOW_UNALIGNED_ACCESS.
16745 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16746 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16747 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16748 of SLOW_UNALIGNED_ACCESS.
16749 * expmed.c (simple_mem_bitfield_p): Likewise.
16750 * expr.c (alignment_for_piecewise_move): Likewise.
16751 (emit_group_load_1): Likewise.
16752 (emit_group_store): Likewise.
16753 (copy_blkmode_from_reg): Likewise.
16754 (emit_push_insn): Likewise.
16755 (expand_assignment): Likewise.
16756 (store_field): Likewise.
16757 (expand_expr_real_1): Likewise.
16758 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16759 * lra-constraints.c (simplify_operand_subreg): Likewise.
16760 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16761 * gimple-ssa-store-merging.c: Likewise in block comment at start
16763 * tree-ssa-strlen.c: Include target.h.
16764 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16765 of SLOW_UNALIGNED_ACCESS.
16766 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16768 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16770 PR rtl-optimization/82185
16771 * expmed.c (emit_store_flag_int): Only test tem if it has been
16774 2017-09-12 Richard Biener <rguenther@suse.de>
16776 PR middle-end/82149
16777 * match.pd ((FTYPE) N CMP CST): Fix typo.
16779 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16781 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16783 (mips_near_type_p): Add 'short_call' attribute as a synonym
16785 * doc/extend.texi (short_call): Document new function attribute.
16787 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16790 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16791 assertion check that in the condition.
16792 (get_atomic_generic_size): Likewise. Before testing if parameter
16793 has pointer type, if it has array type, call for C++
16794 default_conversion to perform array-to-pointer conversion.
16796 2017-09-12 Richard Biener <rguenther@suse.de>
16798 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16799 for operations we cannot scalarize.
16801 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16803 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16804 vectors heap vectors. Clean up comments.
16805 Make visited_bbs a reference.
16806 (profitable_jump_thread_path): Make GC
16807 vectors heap vectors. Clean up comments.
16809 (convert_and_register_jump_thread_path): Make GC vectors heap
16811 (check_subpath_and_update_thread_path): Same. Clean up comments.
16812 Make visited_bbs a reference.
16813 (handle_phi): Abstract common code to to
16814 register_jump_thread_path_if_profitable.
16815 Rename VAR_BB to DEF_BB.
16817 Make GC vectors heap vectors.
16818 Make visited_bbs a reference.
16819 (handle_assignment): Same.
16820 (register_jump_thread_path_if_profitable): New.
16821 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16823 Make GC vectors heap vectors. Clean up comments.
16824 Make visited_bbs a reference.
16825 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16826 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16829 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16832 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16833 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16835 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16837 Revert r251800 and r251799.
16839 2017-09-11 Martin Jambor <mjambor@suse.cz>
16842 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16843 arguments in advance.
16844 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16845 use it to find predecessor edges.
16846 (naive_outof_ssa): Collect vector of predecessors.
16848 2017-09-08 Jason Merrill <jason@redhat.com>
16850 PR c++/70029 - ICE with ref-qualifier and -flto
16851 * langhooks.h (struct lang_hooks_for_types): Add
16852 copy_lang_qualifiers.
16853 * attribs.c (build_type_attribute_qual_variant): Use it.
16854 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16856 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16857 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16859 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16862 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16863 (*mulsi3_sp64): New instruction.
16864 (mulsi3): New expander.
16866 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16868 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16870 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16872 * sancov.c: Include memmodel.h.
16874 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16877 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16880 2017-09-07 Carl Love <cel@us.ibm.com>
16882 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16883 the sldi instruction.
16885 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16887 * sancov.c: Include tm_p.h.
16889 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16892 * output.h (switch_to_other_text_partition): New declaration.
16893 * varasm.c (switch_to_other_text_partition): New function.
16894 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16895 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16896 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16897 to the other text partition before emitting LCL label and switch back
16898 after emitting the word after it.
16900 2017-09-07 Richard Biener <rguenther@suse.de>
16902 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16903 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16904 critical edges here, after loop init.
16905 (pass_data_pre): Remove PROP_no_crit_edges flags.
16906 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16907 for valueization of call args to avoid leaking VN_TOP.
16908 (visit_use): Assert we do not visit default defs.
16909 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16910 Use error_mark_node to more easily detect leaking VN_TOP.
16911 All default-defs are varying, not VN_TOP. Mark them visited.
16912 (run_scc_vn): Make code match comment.
16914 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16916 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16917 OPTION_MASK_FLOAT128_KEYWORD.
16918 (POWERPC_MASKS): Likewise.
16919 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16920 support for the -mfloat128-type option, and make -mfloat128
16921 default on PowerPC Linux systems. Define or undefine
16922 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16923 Define __float128 to be __ieee128 if IEEE 128-bit support is
16924 enabled, or undefine it.
16925 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16926 Delete defining __FLOAT128_TYPE__.
16927 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16928 -mfloat128-type option and make -mfloat128 default on PowerPC
16930 (TARGET_FLOAT128_TYPE): Likewise.
16931 (-mfloat128-type): Likewise.
16932 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16933 Delete the -mfloat128-type option and make -mfloat128 default on
16934 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16935 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16936 128-bit floating point is enabled. Change tests from using
16937 -mfloat128-type to -mfloat128.
16938 (rs6000_mangle_type): Use the correct mangling for the __float128
16939 type even if normal long double is restricted to 64-bits.
16940 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16942 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16943 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16944 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16946 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16947 documentation for -mfloat128.
16949 2017-09-06 Olivier Hainque <hainque@adacore.com>
16951 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16953 2017-09-06 Wish Wu <wishwu007@gmail.com>
16954 Jakub Jelinek <jakub@redhat.com>
16956 * asan.c (initialize_sanitizer_builtins): Add
16957 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16958 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16959 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16960 BT_FN_VOID_UINT64_PTR variables.
16961 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16962 (BT_FN_VOID_UINT16_UINT16): Likewise.
16963 (BT_FN_VOID_UINT32_UINT32): Likewise.
16964 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16965 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16966 (BT_FN_VOID_UINT64_PTR): Likewise.
16967 * common.opt (flag_sanitize_coverage): New variable.
16968 (fsanitize-coverage=trace-pc): Remove.
16969 (fsanitize-coverage=): Add.
16970 * flag-types.h (enum sanitize_coverage_code): New enum.
16971 * fold-const.c (fold_range_test): Disable non-short-circuit
16972 optimization if flag_sanitize_coverage.
16973 (fold_truth_andor): Likewise.
16974 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16975 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16976 (coverage_sanitizer_opts): New array.
16977 (get_closest_sanitizer_option): Add OPTS argument, handle also
16978 OPT_fsanitize_coverage_.
16979 (parse_sanitizer_options): Adjusted to also handle
16980 OPT_fsanitize_coverage_.
16981 (common_handle_option): Add OPT_fsanitize_coverage_.
16982 * sancov.c (instrument_comparison, instrument_switch): New function.
16983 (sancov_pass): Add trace-cmp support.
16984 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16985 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16986 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16987 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16988 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16989 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16990 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16991 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16992 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16993 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16995 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16997 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16998 error. Only quit immediately if parsing is complete.
16999 (BEGIN): Initialize fatal_err and parse_done.
17000 (begin fpu, end fpu): Check number of arguments.
17001 (begin arch, end arch): Likewise.
17002 (begin cpu, end cpu): Likewise.
17003 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
17004 (optalias): Likewise.
17006 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
17008 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
17009 * config/arm/arm-isa.h: Delete. Move definitions to ...
17010 * arm-cpus.in: ... here. Use new feature and fgroup values.
17011 * config/arm/arm.c (arm_option_override): Use lower case for feature
17013 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
17014 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
17015 * config/arm/parsecpu.awk (END): Add new command 'isa'.
17017 (print_isa_bits_for): New function.
17018 (gen_isa): New function.
17019 (gen_comm_data): Use print_isa_bits_for.
17020 (define feature): New keyword.
17021 (define fgroup): New keyword.
17022 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
17023 (arm-isa.h): Add rule to generate file.
17024 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
17025 case for feature bit names.
17027 2017-09-06 Richard Biener <rguenther@suse.de>
17029 * tree-ssa-pre.c (NECESSARY): Remove.
17030 (create_expression_by_pieces): Do not touch pass-local flags.
17031 (insert_into_preds_of_block): Likewise.
17032 (do_pre_regular_insertion): Likewise.
17033 (eliminate_insert): Likewise.
17034 (eliminate_dom_walker::before_dom_children): Likewise.
17035 (fini_eliminate): Do not look at inserted_exprs.
17036 (mark_operand_necessary): Remove.
17037 (remove_dead_inserted_code): Replace with simple work-list
17038 algorithm based on inserted_exprs and SSA uses.
17039 (pass_pre::execute): Re-order fini_eliminate and
17040 remove_dead_inserted_code.
17042 2017-09-06 Olivier Hainque <hainque@adacore.com>
17044 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
17045 for VxWorks 7. Adjust surrounding comments.
17047 2017-09-06 Richard Biener <rguenther@suse.de>
17049 * gimple-ssa-strength-reduction.c
17050 (find_candidates_dom_walker::before_dom_children): Also allow
17053 2017-09-06 Richard Biener <rguenther@suse.de>
17055 PR tree-optimization/82108
17056 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
17057 for gap in the non-permutation SLP case.
17059 2017-09-06 Martin Jambor <mjambor@suse.cz>
17061 PR tree-optimization/82078
17062 * tree-sra.c (sort_and_splice_var_accesses): Move call to
17063 add_access_to_work_queue...
17064 (build_accesses_from_assign): ...here.
17065 (propagate_all_subaccesses): Make sure racc is the group
17066 representative, if there is one.
17068 2017-09-06 Jakub Jelinek <jakub@redhat.com>
17070 PR middle-end/82095
17071 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
17074 2017-09-06 Richard Biener <rguenther@suse.de>
17076 * gimple-ssa-strength-reduction.c
17077 (find_candidates_dom_walker::before_doom_children): Use a
17078 type and not a mode check.
17080 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
17083 * config/arm/predicates.md (arm_general_adddi_operand): Create new
17085 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
17087 2017-09-05 Jeff Law <law@redhat.com>
17089 PR tree-optimization/64910
17090 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
17091 cases where we have 3 or more operands.
17093 2017-09-05 Jakub Jelinek <jakub@redhat.com>
17095 PR middle-end/81768
17096 * omp-low.c (lower_omp_for): Recompute tree invariant if
17097 gimple_omp_for_initial/final is ADDR_EXPR.
17099 PR middle-end/81768
17100 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
17101 into gimple val before gimplification fo the COND_EXPR.
17103 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
17105 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
17106 REGION_COPY argument.
17107 (thread_through_all_blocks): Remove unused argument to
17108 duplicate_thread_path.
17110 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17111 Alan Hayward <alan.hayward@arm.com>
17112 David Sherwood <david.sherwood@arm.com>
17114 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
17115 Take a scalar_mode rather than a machine_mode.
17116 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17117 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
17118 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17119 (aarch64_gen_adjusted_ldpstp): Likewise.
17120 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
17122 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17123 Alan Hayward <alan.hayward@arm.com>
17124 David Sherwood <david.sherwood@arm.com>
17126 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
17127 Take a scalar_int_mode instead of a machine_mode.
17128 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17129 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17130 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17131 (aarch64_simd_attr_length_rglist): Delete.
17132 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
17133 a scalar_int_mode instead of a machine_mode.
17134 (aarch64_add_offset): Likewise.
17135 (aarch64_internal_mov_immediate): Likewise
17136 (aarch64_add_constant_internal): Likewise.
17137 (aarch64_add_constant): Likewise.
17138 (aarch64_movw_imm): Likewise.
17139 (aarch64_rtx_arith_op_extract_p): Likewise.
17140 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17141 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17142 Remove assert that the mode isn't a vector.
17143 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17144 (aarch64_expand_mov_immediate): Update calls after above changes.
17145 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
17146 (aarch64_and_bitmask_imm): Check for scalar integer modes.
17147 (aarch64_move_imm): Likewise.
17148 (aarch64_can_const_movi_rtx_p): Likewise.
17149 (aarch64_strip_extend): Likewise.
17150 (aarch64_extr_rtx_p): Likewise.
17151 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
17152 a CONST_INT when the mode parameter is VOIDmode.
17153 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
17155 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17157 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
17158 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17159 (bitwise_type_for_mode): Update accordingly.
17161 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17163 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
17164 * stor-layout.c (mode_for_size_tree): Likewise.
17165 (mode_for_array): Update accordingly.
17166 (layout_decl): Likewise.
17167 (compute_record_mode): Likewise. Only set the mode once.
17169 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17171 * target.def (get_mask_mode): Change return type to opt_mode.
17173 * doc/tm.texi: Regenerate.
17174 * targhooks.h (default_get_mask_mode): Return an opt_mode.
17175 * targhooks.c (default_get_mask_mode): Likewise.
17176 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17177 * optabs-query.c (can_vec_mask_load_store_p): Update use of
17178 targetm.get_mask_mode.
17179 * tree.c (build_truth_vector_type): Likewise.
17181 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17183 * machmode.h (mode_for_vector): Return an opt_mode.
17184 * stor-layout.c (mode_for_vector): Likewise.
17185 (mode_for_int_vector): Update accordingly.
17186 (layout_type): Likewise.
17187 * config/i386/i386.c (emit_memmov): Likewise.
17188 (ix86_expand_set_or_movmem): Likewise.
17189 (ix86_expand_vector_init): Likewise.
17190 (ix86_get_mask_mode): Likewise.
17191 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
17193 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
17194 * expmed.c (extract_bit_field_1): Likewise.
17195 * expr.c (expand_expr_real_2): Likewise.
17196 * optabs-query.c (can_vec_perm_p): Likewise.
17197 (can_vec_mask_load_store_p): Likewise.
17198 * optabs.c (expand_vec_perm): Likewise.
17199 * targhooks.c (default_get_mask_mode): Likewise.
17200 * tree-vect-stmts.c (vectorizable_store): Likewise.
17201 (vectorizable_load): Likewise.
17202 (get_vectype_for_scalar_type_and_size): Likewise.
17204 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17206 * machmode.h (mode_for_int_vector): New function.
17207 * stor-layout.c (mode_for_int_vector): Likewise.
17208 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
17209 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
17210 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17211 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17212 (s390_expand_vcond): Likewise.
17214 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17216 * machmode.h (opt_machine_mode): New type.
17217 (opt_mode<T>): Allow construction from anything that can be
17219 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17220 (mode_for_size): Return an opt_machine_mode.
17221 * stor-layout.c (mode_for_size): Likewise.
17222 (mode_for_size_tree): Update call accordingly.
17223 (bitwise_mode_for_mode): Likewise.
17224 (make_fract_type): Likewise.
17225 (make_accum_type): Likewise.
17226 * caller-save.c (replace_reg_with_saved_mem): Update call
17228 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17229 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17230 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17231 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17232 * expmed.c (extract_bit_field_1): Likewise.
17233 * reload.c (get_secondary_mem): Likewise.
17234 * varasm.c (assemble_integer): Likewise.
17235 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17238 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17240 * machmode.h (decimal_float_mode_for_size): New function.
17241 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17242 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17243 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17244 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17245 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17246 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17248 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17250 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17251 (get_builtin_sync_mode): Likewise.
17252 (expand_ifn_atomic_compare_exchange): Likewise.
17253 (expand_builtin_atomic_clear): Likewise.
17254 (expand_builtin_atomic_test_and_set): Likewise.
17255 (fold_builtin_atomic_always_lock_free): Likewise.
17256 * calls.c (compute_argument_addresses): Likewise.
17257 (emit_library_call_value_1): Likewise.
17258 (store_one_arg): Likewise.
17259 * combine.c (combine_instructions): Likewise.
17260 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17261 * config/arm/arm.c (arm_function_value): Likewise.
17262 (aapcs_allocate_return_reg): Likewise.
17263 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17264 * config/i386/i386.c (construct_container): Likewise.
17265 (ix86_gimplify_va_arg): Likewise.
17266 (ix86_expand_sse_cmp): Likewise.
17267 (emit_memmov): Likewise.
17268 (emit_memset): Likewise.
17269 (expand_small_movmem_or_setmem): Likewise.
17270 (ix86_expand_pextr): Likewise.
17271 (ix86_expand_pinsr): Likewise.
17272 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17273 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17275 * config/mips/mips.c (mips_function_value_1) Likewise.
17276 (mips_block_move_straight): Likewise.
17277 (mips_expand_ins_as_unaligned_store): Likewise.
17278 * config/powerpcspe/powerpcspe.c
17279 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17280 (rs6000_darwin64_record_arg_flush): Likewise.
17281 * config/rs6000/rs6000.c
17282 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17283 (rs6000_darwin64_record_arg_flush): Likewise.
17284 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17285 (sparc_function_value_1): Likewise.
17286 * config/spu/spu.c (adjust_operand): Likewise.
17287 (spu_emit_branch_or_set): Likewise.
17288 (arith_immediate_p): Likewise.
17289 * emit-rtl.c (gen_lowpart_common): Likewise.
17290 * expr.c (expand_expr_real_1): Likewise.
17291 * function.c (assign_parm_setup_block): Likewise.
17292 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17293 * reload1.c (alter_reg): Likewise.
17294 * stor-layout.c (mode_for_vector): Likewise.
17295 (layout_type): Likewise.
17297 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17299 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17300 (spu_convert_move): Likewise.
17301 * lower-subreg.c (resolve_simple_move): Likewise.
17303 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17306 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17307 define_insn to a define_expand.
17308 (altivec_vsum2sws_direct): New define_insn.
17309 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17311 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17313 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17316 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17321 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17322 realign stack if stack alignment needed is less than incoming
17325 2017-09-05 Marek Polacek <polacek@redhat.com>
17328 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17331 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17333 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17335 2017-09-05 Richard Biener <rguenther@suse.de>
17337 PR tree-optimization/82084
17338 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17340 2017-09-05 Richard Biener <rguenther@suse.de>
17342 PR tree-optimization/82102
17343 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17345 2017-09-05 Martin Liska <mliska@suse.cz>
17347 PR tree-optimization/82032
17348 * tree-cfg.c (generate_range_test): New function.
17349 * tree-cfg.h (generate_range_test): Declared here.
17350 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17351 (cleanup_control_expr_graph): Use it.
17352 * tree-switch-conversion.c (try_switch_expansion): Remove
17354 (emit_case_nodes): Use generate_range_test.
17356 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17359 * config/i386/i386.md (*<btsc><mode>_mask): Add
17360 TARGET_USE_BT to insn constraint.
17361 (*btr<mode>_mask): Ditto.
17363 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17365 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17366 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17368 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17371 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17372 TARGET_NEON and TARGET_IWMMXT.
17373 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17374 TARGET_NEON and TARGET_IWMMXT.
17375 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17377 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17379 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17380 (ix86_rewrite_tls_address): Ditto.
17381 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17382 (ix86_rewrite_tls_address_1): Ditto.
17383 (ix86_rewrite_tls_address): Ditto.
17384 * config/i386/predicates.md (tls_address_pattern): New predicate.
17385 * config/i386/i386.md (TLS address splitter): New splitter.
17387 2017-09-04 Richard Biener <rguenther@suse.de>
17389 PR tree-optimization/82084
17390 * fold-const.h (can_native_encode_string_p): Declare.
17391 * fold-const.c (can_native_encode_string_p): Factor out from ...
17392 (native_encode_string): ... here.
17393 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17394 vectorizing stores from constants we later cannot handle.
17396 2017-09-04 Marek Polacek <polacek@redhat.com>
17399 * doc/invoke.texi: Update -Wtautological-compare documentation.
17401 2017-09-04 Jeff Law <law@redhat.com>
17403 PR tree-optimization/64910
17404 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17405 swap the first and last operand if the last is a constant.
17407 2017-09-04 Marek Polacek <polacek@redhat.com>
17410 * convert.c (do_narrow): When sanitizing signed integer overflows,
17411 bail out for signed types.
17412 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17414 2017-09-04 Richard Biener <rguenther@suse.de>
17416 PR tree-optimization/82060
17417 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17418 Move devirtualization after stmt folding and before EH/AB/noreturn
17419 cleanup to get the stmt refs canonicalized. Use a bool instead
17420 of gimple_modified_p since that doesn't work for NOPs. Schedule
17421 NOPs generated by folding for removal.
17423 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17424 Alan Hayward <alan.hayward@arm.com>
17425 David Sherwood <david.sherwood@arm.com>
17427 * coretypes.h (pad_direction): New enum.
17428 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17429 (FUNCTION_ARG_PADDING): Likewise.
17430 * target.def (function_arg_padding): New hook.
17431 * targhooks.h (default_function_arg_padding): Declare.
17432 * targhooks.c (default_function_arg_padding): New function.
17433 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17434 (TARGET_FUNCTION_ARG_PADDING): ...this.
17435 * doc/tm.texi: Regenerate.
17436 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17437 instead of direction.
17438 (compute_argument_addresses): Likewise.
17439 (load_register_parameters): Likewise.
17440 (emit_library_call_value_1): Likewise.
17441 (store_one_arg): Use targetm.calls.function_arg_padding instead
17442 of FUNCTION_ARG_PADDING.
17443 (must_pass_in_stack_var_size_or_pad): Likewise.
17444 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17445 (emit_group_store): Likewise.
17446 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17447 instead of FUNCTION_ARG_PADDING.
17448 (emit_push_insn): Likewise, and propagate enum change throughout
17450 * function.h (direction): Delete.
17451 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17453 * function.c (assign_parm_find_stack_rtl): Likewise.
17454 (assign_parm_setup_block_p): Likewise.
17455 (assign_parm_setup_block): Likewise.
17456 (gimplify_parameters): Likewise.
17457 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17458 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17460 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17461 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17462 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17463 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17464 (aarch64_function_arg_padding): ...this new function.
17465 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17466 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17467 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17468 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17469 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17470 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17471 (arm_pad_arg_upward): Replace with...
17472 (arm_function_arg_padding): ...this new function.
17473 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17475 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17476 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17477 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17478 (ia64_hpux_function_arg_padding): Replace with...
17479 (ia64_function_arg_padding): ...this new function. Use pad_direction
17480 instead of direction. Check for TARGET_HPUX.
17481 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17482 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17483 (iq2000_function_arg_padding): New function.
17484 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17485 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17486 (mips_function_arg_padding): ...this new function.
17487 (mips_pad_reg_upward): Update accordingly.
17488 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17489 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17490 targetm.calls.function_arg_padding.
17491 (FUNCTION_ARG_PADDING): Delete.
17492 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17493 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17494 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17495 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17496 (nios2_block_reg_padding): Return pad_direction instead of direction.
17497 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17498 instead of direction.
17499 (nios2_function_arg_padding): Likewise. Make static.
17500 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17501 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17502 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17503 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17504 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17505 (pa_function_arg_padding): Make static. Return pad_direction instead
17507 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17508 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17509 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17510 instead of direction. Use targetm.calls.function_arg_padding.
17511 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17512 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17513 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17514 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17515 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17517 (function_arg_padding): Rename to...
17518 (rs6000_function_arg_padding): ...this. Make static. Return
17519 pad_direction instead of direction.
17520 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17521 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17522 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17523 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17524 instead of direction. Use targetm.calls.function_arg_padding.
17525 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17526 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17527 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17528 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17529 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17530 (function_arg_padding): Rename to...
17531 (rs6000_function_arg_padding): ...this. Make static. Return
17532 pad_direction instead of direction.
17533 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17534 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17535 * config/s390/s390.c (s390_function_arg_padding): New function.
17536 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17537 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17538 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17539 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17540 (function_arg_padding): Rename to...
17541 (sparc_function_arg_padding): ...this. Make static. Return
17542 pad_direction instead of direction.
17543 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17544 * config/spu/spu.c (spu_function_arg_padding): New function.
17545 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17546 * system.h (FUNCTION_ARG_PADDING): Poison.
17548 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17549 Alan Hayward <alan.hayward@arm.com>
17550 David Sherwood <david.sherwood@arm.com>
17552 * target.def (modes_tieable_p): New hook.
17553 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17554 (TARGET_MODES_TIEABLE_P): ...this.
17555 * doc/tm.texi.in: Regenerate.
17556 * hooks.h (hook_bool_mode_mode_true): Declare.
17557 * hooks.c (hook_bool_mode_mode_true): New function.
17558 * combine.c (subst): Use targetm.modes_tieable_p instead of
17560 * dse.c (find_shift_sequence): Likewise.
17561 * expmed.c (extract_low_bits): Likewise.
17562 * lower-subreg.c: Include target.h.
17563 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17565 * rtlanal.c (rtx_cost): Likewise.
17566 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17567 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17568 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17569 (TARGET_MODES_TIEABLE_P): Redefine.
17570 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17571 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17572 (TARGET_MODES_TIEABLE_P): Redefine.
17573 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17574 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17575 (arc_modes_tieable_p): New function.
17576 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17577 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17578 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17579 (arm_modes_tieable_p): Make static.
17580 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17581 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17582 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17583 (TARGET_MODES_TIEABLE_P): Redefine.
17584 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17585 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17586 (TARGET_MODES_TIEABLE_P): Redefine.
17587 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17588 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17589 (cr16_modes_tieable_p): New function.
17590 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17591 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17592 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17593 (TRULY_NOOP_TRUNCATION): Update comment.
17594 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17595 (TRULY_NOOP_TRUNCATION): Update comment.
17596 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17597 (frv_modes_tieable_p): New function.
17598 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17599 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17600 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17601 (TARGET_MODES_TIEABLE_P): Redefine.
17602 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17603 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17604 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17605 (TARGET_MODES_TIEABLE_P): Redefine.
17606 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17607 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17608 (ia64_modes_tieable_p): New function.
17609 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17610 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17611 (iq2000_modes_tieable_p): New function.
17612 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17613 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17614 (lm32_modes_tieable_p): New function.
17615 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17616 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17617 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17618 (TARGET_MODES_TIEABLE_P): Redefine.
17619 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17620 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17621 (m32r_modes_tieable_p): New function.
17622 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17623 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17624 (m68k_modes_tieable_p): New function.
17625 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17626 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17627 (mcore_modes_tieable_p): New function.
17628 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17629 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17631 (TARGET_MODES_TIEABLE_P): Redefine.
17632 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17633 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17634 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17635 (TARGET_MODES_TIEABLE_P): Redefine.
17636 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17637 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17638 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17639 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17640 (mn10300_modes_tieable_p): ...this and make static.
17641 (TARGET_MODES_TIEABLE_P): Redefine.
17642 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17643 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17644 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17645 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17646 (msp430_modes_tieable_p): Make static.
17647 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17648 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17649 (TARGET_MODES_TIEABLE_P): Redefine.
17650 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17651 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17652 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17653 (TARGET_MODES_TIEABLE_P): Redefine.
17654 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17655 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17656 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17657 (TARGET_MODES_TIEABLE_P): Redefine.
17658 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17659 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17660 (pdp11_modes_tieable_p): New function.
17661 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17662 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17663 (rs6000_modes_tieable_p): New function.
17664 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17665 * config/powerpcspe/powerpcspe.md: Update comment.
17666 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17667 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17668 (TARGET_MODES_TIEABLE_P): Redefine.
17669 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17670 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17671 (rl78_modes_tieable_p): New function.
17672 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17673 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17674 (rs6000_modes_tieable_p): New function.
17675 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17676 * config/rs6000/rs6000.md: Update comment.
17677 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17678 * config/rx/rx.c (rx_modes_tieable_p): New function.
17679 (TARGET_MODES_TIEABLE_P): Redefine.
17680 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17681 * config/s390/s390.c (s390_modes_tieable_p): New function.
17682 (TARGET_MODES_TIEABLE_P): Redefine.
17683 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17684 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17685 (sh_modes_tieable_p): New function.
17686 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17687 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17688 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17689 (sparc_modes_tieable_p): Make static.
17690 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17691 * config/spu/spu.c (spu_modes_tieable_p): New function.
17692 (TARGET_MODES_TIEABLE_P): Redefine.
17693 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17694 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17695 (TARGET_MODES_TIEABLE_P): Redefine.
17696 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17697 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17698 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17699 * config/v850/v850.c (v850_modes_tieable_p): New function.
17700 (TARGET_MODES_TIEABLE_P): Redefine.
17701 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17702 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17703 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17704 (visium_modes_tieable_p): New function.
17705 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17706 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17707 (xtensa_modes_tieable_p): New function.
17708 * system.h (MODES_TIEABLE_P): Poison.
17710 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17711 Alan Hayward <alan.hayward@arm.com>
17712 David Sherwood <david.sherwood@arm.com>
17714 * target.def (hard_regno_mode_ok): New hook.
17715 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17716 (TARGET_HARD_REGNO_MODE_OK): ...this.
17717 * doc/tm.texi.in: Regenerate.
17718 * hooks.h (hook_bool_uint_mode_true): Declare.
17719 * hooks.c (hook_bool_uint_mode_true): New function.
17720 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17721 HARD_REGNO_MODE_OK.
17722 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17723 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17724 instead of HARD_REGNO_MODE_OK.
17725 * caller-save.c: Include target.h.
17726 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17727 HARD_REGNO_MODE_OK.
17728 * combine.c (can_combine_p): Likewise.
17729 (combinable_i3pat): Likewise.
17730 (can_change_dest_mode): Likewise.
17731 * expr.c (init_expr_target): Likewise.
17732 (convert_move): Likewise.
17733 (convert_modes): Likewise.
17734 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17735 (setup_prohibited_mode_move_regs): Likewise.
17736 * ira.h (target_ira): Likewise.
17737 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17738 * lra-constraints.c (process_alt_operands): Likewise.
17739 (split_reg): Likewise.
17740 * recog.c (peep2_find_free_register): Likewise.
17741 * ree.c (combine_reaching_defs): Likewise.
17742 * regcprop.c (maybe_mode_change): Likewise.
17743 * reginfo.c (init_reg_sets_1): Likewise.
17744 (choose_hard_reg_mode): Likewise.
17745 (simplifiable_subregs): Likewise.
17746 * regrename.c (check_new_reg_p): Likewise.
17747 * reload.c (find_valid_class): Likewise.
17748 (find_valid_class_1): Likewise.
17749 (reload_inner_reg_of_subreg): Likewise.
17750 (push_reload): Likewise.
17751 (combine_reloads): Likewise.
17752 (find_dummy_reload): Likewise.
17753 (find_reloads): Likewise.
17754 * reload1.c (find_reg): Likewise.
17755 (set_reload_reg): Likewise.
17756 (allocate_reload_reg): Likewise.
17757 (choose_reload_regs): Likewise.
17758 (reload_adjust_reg_for_temp): Likewise.
17759 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17760 (simplify_subreg_regno): Likewise.
17761 * sel-sched.c (init_regs_for_mode): Likewise.
17762 * varasm.c (make_decl_rtl): Likewise.
17763 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17764 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17765 HARD_REGNO_MODE_OK.
17766 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17767 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17768 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17769 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17770 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17771 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17772 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17773 (arc_mode_class): Delete.
17774 (HARD_REGNO_MODE_OK): Delete.
17775 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17776 (arc_hard_regno_mode_ok): Rename old array to...
17777 (arc_hard_regno_mode_ok_modes): ...this.
17778 (arc_conditional_register_usage): Update accordingly.
17779 (arc_mode_class): Make static.
17780 (arc_hard_regno_mode_ok): New function.
17781 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17782 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17783 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17784 (arm_hard_regno_mode_ok): Make static.
17785 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17786 HARD_REGNO_MODE_OK.
17787 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17788 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17789 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17791 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17792 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17793 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17794 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17795 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17796 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17797 * config/bfin/predicates.md (valid_reg_operand): Use
17798 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17799 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17800 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17801 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17802 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17803 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17804 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17805 (cr16_hard_regno_mode_ok): Make static and return a bool.
17806 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17807 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17808 (cris_hard_regno_mode_ok): New function.
17809 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17810 (epiphany_mode_class): Delete.
17811 (HARD_REGNO_MODE_OK): Delete.
17812 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17813 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17814 (hard_regno_mode_ok): Rename to...
17815 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17816 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17817 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17818 HARD_REGNO_MODE_OK.
17819 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17820 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17821 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17822 (frv_hard_regno_mode_ok): Make static and return a bool.
17823 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17824 HARD_REGNO_MODE_OK.
17825 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17826 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17827 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17828 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17830 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17831 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17832 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17833 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17835 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17836 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17837 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17838 (ia64_hard_regno_mode_ok): New function.
17839 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17840 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17841 (iq2000_hard_regno_mode_ok): New function.
17842 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17843 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17844 (lm32_hard_regno_mode_ok): New function.
17845 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17846 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17847 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17848 instead of HARD_REGNO_MODE_OK.
17849 (m32c_hard_regno_ok): Rename to...
17850 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17851 (m32c_cannot_change_mode_class): Update accordingly.
17852 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17853 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17854 (m32r_mode_class): Delete.
17855 (HARD_REGNO_MODE_OK): Delete.
17856 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17857 (m32r_hard_regno_mode_ok): Rename to...
17858 (m32r_hard_regno_modes): ...this.
17859 (m32r_mode_class): Make static.
17860 (m32r_hard_regno_mode_ok): New function.
17861 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17862 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17863 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17864 (m68k_hard_regno_mode_ok): Make static.
17865 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17866 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17867 (mcore_hard_regno_mode_ok): New function.
17868 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17869 (HARD_REGNO_MODE_OK): Delete.
17870 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17872 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17873 (microblaze_hard_regno_mode_ok): New function.
17874 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17875 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17876 (mips_hard_regno_mode_ok): Delete.
17877 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17878 (mips_hard_regno_mode_ok_p): ...this and make static.
17879 (mips_hard_regno_mode_ok_p): Rename to...
17880 (mips_hard_regno_mode_ok_uncached): ...this.
17881 (mips_hard_regno_mode_ok): New function.
17882 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17883 of HARD_REGNO_MODE_OK.
17884 (mips_option_override): Update after above name changes.
17885 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17886 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17887 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17888 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17889 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17890 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17891 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17892 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17893 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17894 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17895 (msp430_hard_regno_mode_ok): Make static and return a bool.
17896 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17897 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17898 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17900 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17901 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17902 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17903 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17904 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17905 (PA_HARD_REGNO_MODE_OK): ...this
17906 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17907 (PA_HARD_REGNO_MODE_OK): ...this.
17908 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17909 (pa_hard_regno_mode_ok): New function.
17910 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17911 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17912 (pdp11_hard_regno_mode_ok): New function.
17913 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17914 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17916 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17918 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17919 (rs6000_hard_regno_mode_ok): Rename to...
17920 (rs6000_hard_regno_mode_ok_uncached): ...this.
17921 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17922 (rs6000_hard_regno_mode_ok): New function.
17923 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17924 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17925 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17926 (riscv_hard_regno_mode_ok): ...this and make static.
17927 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17928 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17929 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17930 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17931 (rl78_hard_regno_mode_ok): Make static and return bool.
17932 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17933 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17935 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17936 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17937 (rs6000_hard_regno_mode_ok): Rename to...
17938 (rs6000_hard_regno_mode_ok_uncached): ...this.
17939 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17940 (rs6000_hard_regno_mode_ok): New function.
17941 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17942 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17943 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17944 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17945 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17946 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17947 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17948 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17949 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17950 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17951 (sh_hard_regno_mode_ok): Make static.
17952 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17953 instead of HARD_REGNO_MODE_OK.
17954 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17955 (sparc_mode_class): Delete.
17956 (HARD_REGNO_MODE_OK): Delete.
17957 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17958 (hard_regno_mode_classes): Make static.
17959 (sparc_mode_class): Likewise.
17960 (sparc_hard_regno_mode_ok): New function.
17961 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17962 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17963 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17965 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17966 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17967 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17968 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17969 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17970 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17971 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17972 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17973 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17974 (visium_hard_regno_mode_ok): New function.
17975 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17976 instead of HARD_REGNO_MODE_OK.
17977 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17978 (HARD_REGNO_MODE_OK): Delete.
17979 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17980 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17981 (xtensa_option_override): Update accordingly.
17982 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17983 (xtensa_hard_regno_mode_ok): New function.
17984 * system.h (HARD_REGNO_MODE_OK): Poison.
17986 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17987 Alan Hayward <alan.hayward@arm.com>
17988 David Sherwood <david.sherwood@arm.com>
17990 * target.def (hard_regno_call_part_clobbered): New hook.
17991 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17992 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17993 * doc/tm.texi: Regenerate.
17994 * hooks.h (hook_bool_uint_mode_false): Declare.
17995 * hooks.c (hook_bool_uint_mode_false): New function.
17996 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17997 * cselib.c (cselib_process_insn): Use
17998 targetm.hard_regno_call_part_clobbered instead of
17999 HARD_REGNO_CALL_PART_CLOBBERED.
18000 * ira-conflicts.c (ira_build_conflicts): Likewise.
18001 * ira-costs.c (ira_tune_allocno_costs): Likewise.
18002 * lra-constraints.c (need_for_call_save_p): Likewise.
18003 * lra-lives.c: Include target.h.
18004 (check_pseudos_live_through_calls): Use
18005 targetm.hard_regno_call_part_clobbered instead of
18006 HARD_REGNO_CALL_PART_CLOBBERED.
18007 * regcprop.c: Include target.h.
18008 (copyprop_hardreg_forward_1): Use
18009 targetm.hard_regno_call_part_clobbered instead of
18010 HARD_REGNO_CALL_PART_CLOBBERED.
18011 * reginfo.c (choose_hard_reg_mode): Likewise.
18012 * regrename.c (check_new_reg_p): Likewise.
18013 * reload.c (find_equiv_reg): Likewise.
18014 * reload1.c (emit_reload_insns): Likewise.
18015 * sched-deps.c (deps_analyze_insn): Likewise.
18016 * sel-sched.c (init_regs_for_mode): Likewise.
18017 (mark_unavailable_hard_regs): Likewise.
18018 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
18019 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18020 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
18022 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18023 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18024 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
18026 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
18028 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18029 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18030 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
18032 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18033 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18034 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
18036 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18037 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
18039 * config/powerpcspe/powerpcspe.c
18040 (rs6000_hard_regno_call_part_clobbered): New function.
18041 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18042 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18043 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
18045 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18046 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18047 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
18049 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18050 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18051 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
18053 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18054 Alan Hayward <alan.hayward@arm.com>
18055 David Sherwood <david.sherwood@arm.com>
18057 * rtl.h (subreg_memory_offset): Declare.
18058 * emit-rtl.c (subreg_memory_offset): New function.
18059 * expmed.c (store_bit_field_1): Use it.
18060 * expr.c (undefined_operand_subword_p): Likewise.
18061 * simplify-rtx.c (simplify_subreg): Likewise.
18063 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
18065 PR rtl-optimization/57448
18068 * optabs.c (expand_atomic_load): Place compiler memory barriers if
18069 using atomic_load pattern.
18070 (expand_atomic_store): Likewise.
18072 2017-09-04 Jakub Jelinek <jakub@redhat.com>
18075 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
18076 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
18077 handling. Use replace_call_with_value with NULL instead of
18078 gsi_replace, unlink_stmt_vdef and release_defs.
18080 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
18083 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
18085 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18088 * rtl.h (emit_library_call_value_1): Declare.
18089 (emit_library_call): Replace declaration with a series of overloads.
18090 Remove the parameter count argument.
18091 (emit_library_call_value): Likewise.
18092 * calls.c (emit_library_call_value_1): Make global. Replace varargs
18093 with an "rtx_mode_t *".
18094 (emit_library_call_value): Delete.
18095 (emit_library_call): Likewise.
18096 * asan.c (asan_emit_stack_protection): Update calls accordingly.
18097 (asan_emit_allocas_unpoison): Likewise.
18098 * builtins.c (expand_builtin_powi): Likewise.
18099 (expand_asan_emit_allocas_unpoison): Likewise.
18100 * cfgexpand.c (expand_main_function): Likewise.
18101 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
18102 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
18103 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
18104 * config/arm/arm.c (arm_trampoline_init): Likewise.
18105 (arm_call_tls_get_addr): Likewise.
18106 (arm_expand_divmod_libfunc): Likewise.
18107 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
18108 (smulsi3_highpart): Likewise.
18109 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
18110 (c6x_expand_compare): Likewise.
18111 (c6x_expand_movmem): Likewise.
18112 * config/frv/frv.c (frv_trampoline_init): Likewise.
18113 * config/i386/i386.c (ix86_trampoline_init): Likewise.
18114 (ix86_expand_divmod_libfunc): Likewise.
18115 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
18116 (ia64_expand_compare): Likewise.
18117 (ia64_profile_hook): Likewise.
18118 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
18119 (nonlocal_goto): Likewise.
18120 (restore_stack_nonlocal): Likewise.
18121 * config/m32r/m32r.c (block_move_call): Likewise.
18122 (m32r_trampoline_init): Likewise.
18123 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
18124 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
18125 (m68k_call_m68k_read_tp): Likewise.
18126 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
18127 (microblaze_expand_divide): Likewise.
18128 * config/mips/mips.h (mips_args): Likewise.
18129 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
18130 (MIPS_ICACHE_SYNC): Likewise.
18131 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
18132 (nios2_trampoline_init): Likewise.
18133 * config/pa/pa.c (hppa_tls_call): Likewise.
18134 (pa_trampoline_init): Likewise.
18135 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
18136 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
18137 (expand_strn_compare): Likewise.
18138 (rs6000_generate_compare): Likewise.
18139 (rs6000_expand_float128_convert): Likewise.
18140 (output_profile_hook): Likewise.
18141 (rs6000_trampoline_init): Likewise.
18142 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
18143 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
18144 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
18145 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
18146 (rs6000_generate_compare): Likewise.
18147 (rs6000_expand_float128_convert): Likewise.
18148 (output_profile_hook): Likewise.
18149 (rs6000_trampoline_init): Likewise.
18150 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
18151 * config/sh/sh.c (sh_trampoline_init): Likewise.
18152 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
18153 (sparc_emit_float_lib_cmp): Likewise.
18154 (sparc32_initialize_trampoline): Likewise.
18155 (sparc64_initialize_trampoline): Likewise.
18156 (sparc_profile_hook): Likewise.
18157 * config/spu/spu.c (ea_load_store): Likewise.
18158 * config/spu/spu.md (floatunssidf2): Likewise.
18159 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
18160 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
18161 * config/visium/visium.c (expand_block_move_4): Likewise.
18162 (expand_block_move_2): Likewise.
18163 (expand_block_move_1): Likewise.
18164 (expand_block_set_4): Likewise.
18165 (expand_block_set_2): Likewise.
18166 (expand_block_set_1): Likewise.
18167 (visium_trampoline_init): Likewise.
18168 (visium_profile_hook): Likewise.
18169 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
18170 (xtensa_setup_frame_addresses): Likewise.
18171 (xtensa_trampoline_init): Likewise.
18172 * except.c (sjlj_emit_function_enter): Likewise.
18173 (sjlj_emit_function_exit): Likewise.
18174 * explow.c (allocate_dynamic_stack_space): Likewise.
18175 (probe_stack_range): Likewise.
18176 * expr.c (convert_mode_scalar): Likewise.
18177 * optabs.c (expand_binop): Likewise.
18178 (expand_twoval_binop_libfunc): Likewise.
18179 (expand_unop): Likewise.
18180 (prepare_cmp_insn): Likewise.
18181 (prepare_float_lib_cmp): Likewise.
18182 (expand_float): Likewise.
18183 (expand_fix): Likewise.
18184 (expand_fixed_convert): Likewise.
18185 (maybe_emit_sync_lock_test_and_set): Likewise.
18186 (expand_atomic_compare_and_swap): Likewise.
18187 (expand_mem_thread_fence): Likewise.
18188 (expand_atomic_fetch_op): Likewise.
18190 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
18192 * doc/generic.texi (OpenACC): Adjust URL.
18193 * doc/invoke.texi (C Dialect Options): Ditto.
18195 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
18197 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
18198 predicate for operand 1. Add (m,<S>) constraint.
18199 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
18200 Prevent memory operand 1 with register operand 2.
18202 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
18204 PR rtl-optimization/82024
18205 * combine.c (try_combine): If the combination result is a PARALLEL,
18206 and we only need to retain the SET in there that would be placed
18207 at I2, check that we can place that at I3 instead, before doing so.
18209 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18212 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
18214 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
18215 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
18218 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
18219 Jeff Law <law@redhat.com>
18221 * varasm.c (bss_initializer_p): Do not put constants into .bss
18222 (categorize_decl_for_section): Handle bss_initializer_p returning
18223 false when DECL_INITIAL is NULL.
18225 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18228 * config/s390/s390.c (s390_can_inline_p): New function.
18230 2017-09-01 Jeff Law <law@redhat.com>
18232 PR tree-optimization/82052
18233 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18234 Always initialize the returned slot after a hash table miss
18235 when INSERT is true.
18237 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18239 * config/s390/s390.md (mem_signal_fence): Remove.
18240 * doc/md.texi (mem_signal_fence): Remove.
18241 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18243 * target-insns.def (mem_signal_fence): Remove.
18245 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18248 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18251 * asan.c (create_odr_indicator): Strip name encoding from assembler
18252 name before appending it after __odr_asan_.
18254 2017-09-01 Martin Liska <mliska@suse.cz>
18256 PR tree-optimization/82059
18257 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18258 frequency only when an edge is redirected.
18260 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18262 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18263 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18264 (arc_conditional_register_usage): Remove ARC600 lp_count
18266 (arc_file_start): Emit Tag_ARC_CPU_variation.
18267 (arc_can_use_doloop_p): New conditions to use ZOLs.
18268 (hwloop_fail): New function.
18269 (hwloop_optimize): Likewise.
18270 (hwloop_pattern_reg): Likewise.
18271 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18272 (arc_reorg_loops): New function, calls reorg_loops.
18273 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18274 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18276 (arc_loop_hazard): Remove function, functionality moved into
18278 (arc_hazard): Remove arc_loop_hazard call.
18279 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18280 into hwloop_optimize.
18281 (arc_label_align): Remove ZOL handling.
18282 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18283 * config/arc/arc.md (doloop_begin): Remove pattern.
18284 (doloop_begin_i): Likewise.
18285 (doloop_end_i): Likewise.
18286 (doloop_fallback): Likewise.
18287 (doloop_fallback_m): Likewise.
18288 (doloop_end): Reimplement expand.
18289 (arc_lp): New pattern for LP instruction.
18290 (loop_end): New pattern.
18291 (loop_fail): Likewise.
18292 (decrement_and_branch_until_zero): Likewise.
18293 * config/arc/arc.opt (mlpc-width): New option.
18294 * doc/invoke.texi (mlpc-width): Document option.
18296 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18298 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18299 (arc_ccfsm_advance): Fix checking for delay slots.
18300 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18302 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18304 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18306 (movhi_insn): Update store instruction constraint which are saving
18307 6-bit short immediates.
18308 (movsi_insn): Consider also short scaled load operations.
18309 (zero_extendhisi2_i): Use Usd constraint instead of T.
18310 (extendhisi2_i): Add q constraint.
18311 (arc_clzsi2): Add type and length attributes.
18312 (arc_ctzsi2): Likewise.
18313 * config/arc/constraints.md (Usc): Update constraint, the
18314 assembler can parse two relocations for a single instruction.
18316 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18318 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18319 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18321 2017-08-31 Olivier Hainque <hainque@adacore.com>
18323 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18324 match as powerpc-wrs-vxworks*.
18326 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18328 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18329 register constraint for by-element operand.
18330 (aarch64_mls_elt_merge<mode>): Likewise.
18332 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18334 * config/arc/arc.c (arc_can_follow_jump): Check for short
18337 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18339 * config.gcc: Use g.opt for arc.
18340 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18341 functionality moved to ...
18342 (legitimate_scaled_address_p): New function, ...here.
18343 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18344 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18345 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18347 (arc_override_options): Handle G option.
18348 (arc_output_pic_addr_const): Correct function definition.
18349 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18350 (arc_decl_anon_ns_mem_p): Delete.
18351 (arc_in_small_data_p): Overhaul this function to take into
18352 consideration the value given via G option.
18353 (arc_rewrite_small_data_1): Renamed and corrected old
18354 arc_rewrite_small_data function.
18355 (arc_rewrite_small_data): New function.
18356 (small_data_pattern): Don't use pic_offset_table_rtx.
18357 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18358 * config/arc/simdext.md (movmisalignv2hi): Use
18359 prepare_move_operands function.
18361 (movmisalign*): Likewise.
18362 * doc/invoke.texi (ARC options): Document -G option.
18364 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18366 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18368 * config/arc/arc.c (arc_print_operand): Output scalled address for
18369 sdata whenever is possible.
18370 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18371 load/stores are available.
18372 (compact_sda_memory_operand): Check for the alignment required by
18373 code density instructions.
18374 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18376 * config/arc/constraints.md (Usd): Update constraint.
18377 (Us0): New constraint.
18378 (Usc): Update constraint.
18380 2017-08-31 Richard Biener <rguenther@suse.de>
18382 PR middle-end/82054
18383 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18384 function only once.
18386 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18388 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18389 Resize type_signature.
18391 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18392 Alan Hayward <alan.hayward@arm.com>
18393 David Sherwood <david.sherwood@arm.com>
18395 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18396 subregs whose inner modes can be stored in GPRs.
18397 (aarch64_classify_index): Likewise.
18399 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18400 Alan Hayward <alan.hayward@arm.com>
18401 David Sherwood <david.sherwood@arm.com>
18403 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18404 (V_INT_EQUIV): ...this.
18405 (v_cmp_result): Rename to...
18406 (v_int_equiv): ...this.
18407 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18408 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18409 (copysign<mode>3): Likewise.
18410 (aarch64_simd_bsl<mode>_internal): Likewise.
18411 (aarch64_simd_bsl<mode>): Likewise.
18412 (vec_cmp<mode><mode>): Likewise.
18413 (vcond<mode><mode>): Likewise.
18414 (vcond<v_cmp_mixed><mode>): Likewise.
18415 (vcondu<mode><v_cmp_mixed>): Likewise.
18416 (aarch64_cm<optab><mode>): Likewise.
18417 (aarch64_cmtst<mode>): Likewise.
18418 (aarch64_fac<optab><mode>): Likewise.
18419 (vec_perm_const<mode>): Likewise.
18420 (vcond_mask_<mode><v_cmp_result>): Rename to...
18421 (vcond_mask_<mode><v_int_equiv>): ...this.
18422 (vec_cmp<mode><v_cmp_result>): Rename to...
18423 (vec_cmp<mode><v_int_equiv>): ...this.
18425 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18426 Alan Hayward <alan.hayward@arm.com>
18427 David Sherwood <david.sherwood@arm.com>
18429 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18431 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18432 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18433 (UNSPEC_LD4_DREG): New unspecs.
18434 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18435 (aarch64_ld2<mode>_dreg_be): Replace with...
18436 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18438 (aarch64_ld3<mode>_dreg_le)
18439 (aarch64_ld3<mode>_dreg_be): Replace with...
18440 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18442 (aarch64_ld4<mode>_dreg_le)
18443 (aarch64_ld4<mode>_dreg_be): Replace with...
18444 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18447 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18449 PR tree-optimization/81987
18450 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18451 insert an initializer in a location not dominated by the stride
18454 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18456 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18457 on the entire header of the finally block in the fallthru case.
18459 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18461 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18463 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18465 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18466 rs6000_emit_move_from_cr and call renamed function.
18467 (rs6000_emit_prologue): Call renamed functions.
18468 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18469 movesi_from_cr, remove volatile CRs.
18471 2017-08-30 Jon Beniston <jon@beniston.com>
18472 Richard Biener <rguenther@suse.de>
18474 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18475 of VECTOR_MODE_P check.
18476 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18477 element vector types.
18479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18481 * df.h (df_read_modify_subreg_p): Remove in favor of...
18482 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18483 const_rtx instead of an rtx.
18484 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18485 * df-problems.c (df_word_lr_mark_ref): Likewise.
18486 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18487 (mark_pseudo_reg_dead): Likewise.
18488 (mark_ref_dead): Likewise.
18489 * reginfo.c (init_subregs_of_mode): Likewise.
18490 * sched-deps.c (sched_analyze_1): Likewise.
18491 * df-scan.c (df_def_record_1): Likewise.
18492 (df_uses_record): Likewise.
18493 (df_read_modify_subreg_p): Remove in favor of...
18494 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18495 const_rtx instead of an rtx.
18497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18498 Alan Hayward <alan.hayward@arm.com>
18499 David Sherwood <david.sherwood@arm.com>
18501 * rtl.h (partial_subreg_p): New function.
18502 * caller-save.c (save_call_clobbered_regs): Use it.
18503 * calls.c (expand_call): Likewise.
18504 * combine.c (combinable_i3pat): Likewise.
18505 (simplify_set): Likewise.
18506 (make_extraction): Likewise.
18507 (make_compound_operation_int): Likewise.
18508 (gen_lowpart_or_truncate): Likewise.
18509 (force_to_mode): Likewise.
18510 (make_field_assignment): Likewise.
18511 (reg_truncated_to_mode): Likewise.
18512 (record_truncated_value): Likewise.
18513 (move_deaths): Likewise.
18514 * cse.c (record_jump_cond): Likewise.
18515 (cse_insn): Likewise.
18516 * cselib.c (cselib_lookup_1): Likewise.
18517 * expmed.c (extract_bit_field_using_extv): Likewise.
18518 * function.c (assign_parm_setup_reg): Likewise.
18519 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18520 * ira-build.c (create_insn_allocnos): Likewise.
18521 * lra-coalesce.c (merge_pseudos): Likewise.
18522 * lra-constraints.c (match_reload): Likewise.
18523 (simplify_operand_subreg): Likewise.
18524 (curr_insn_transform): Likewise.
18525 * lra-lives.c (process_bb_lives): Likewise.
18526 * lra.c (new_insn_reg): Likewise.
18527 (lra_substitute_pseudo): Likewise.
18528 * regcprop.c (mode_change_ok): Likewise.
18529 (maybe_mode_change): Likewise.
18530 (copyprop_hardreg_forward_1): Likewise.
18531 * reload.c (push_reload): Likewise.
18532 (find_reloads): Likewise.
18533 (find_reloads_subreg_address): Likewise.
18534 * reload1.c (alter_reg): Likewise.
18535 (eliminate_regs_1): Likewise.
18536 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18538 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18540 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18541 back to if statements, including unpack.
18543 2017-08-30 Martin Liska <mliska@suse.cz>
18545 PR inline-asm/82001
18546 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18548 (func_checker::compare_asm_inputs_outputs): ... this function.
18549 (func_checker::compare_gimple_asm): Use the function to compare
18550 also ASM constrains.
18551 * ipa-icf-gimple.h: Rename the function.
18553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18554 Alan Hayward <alan.hayward@arm.com>
18555 David Sherwood <david.sherwood@arm.com>
18557 * coretypes.h (complex_mode): New type.
18558 * gdbhooks.py (build_pretty_printer): Handle it.
18559 * machmode.h (complex_mode): New class.
18560 (complex_mode::includes_p): New function.
18561 (is_complex_int_mode): Likewise.
18562 (is_complex_float_mode): Likewise.
18563 * genmodes.c (get_mode_class): Handle complex mode classes.
18564 * function.c (expand_function_end): Use is_complex_int_mode.
18566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18567 Alan Hayward <alan.hayward@arm.com>
18568 David Sherwood <david.sherwood@arm.com>
18570 * coretypes.h (scalar_mode_pod): New typedef.
18571 * gdbhooks.py (build_pretty_printer): Handle it.
18572 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18573 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18574 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18575 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18576 as_a <scalar_mode>.
18578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18579 Alan Hayward <alan.hayward@arm.com>
18580 David Sherwood <david.sherwood@arm.com>
18582 * machmode.h (mode_for_vector): Take a scalar_mode instead
18584 * stor-layout.c (mode_for_vector): Likewise.
18585 * explow.c (promote_mode): Use as_a <scalar_mode>.
18586 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18589 Alan Hayward <alan.hayward@arm.com>
18590 David Sherwood <david.sherwood@arm.com>
18592 * target.def (preferred_simd_mode): Take a scalar_mode
18593 instead of a machine_mode.
18594 * targhooks.h (default_preferred_simd_mode): Likewise.
18595 * targhooks.c (default_preferred_simd_mode): Likewise.
18596 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18597 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18598 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18599 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18600 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18601 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18602 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18603 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18605 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18606 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18607 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18608 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18609 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18610 * doc/tm.texi: Regenerate.
18611 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18615 Alan Hayward <alan.hayward@arm.com>
18616 David Sherwood <david.sherwood@arm.com>
18618 * target.def (scalar_mode_supported_p): Take a scalar_mode
18619 instead of a machine_mode.
18620 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18621 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18622 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18623 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18624 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18625 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18626 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18627 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18628 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18629 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18630 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18631 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18632 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18633 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18635 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18636 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18637 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18638 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18639 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18641 * doc/tm.texi: Regenerate.
18643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18644 Alan Hayward <alan.hayward@arm.com>
18645 David Sherwood <david.sherwood@arm.com>
18647 * coretypes.h (opt_scalar_mode): New typedef.
18648 * gdbhooks.py (build_pretty_printers): Handle it.
18649 * machmode.h (mode_iterator::get_2xwider): Add overload for
18651 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18653 * expr.c (convert_mode_scalar): Likewise.
18654 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18655 * optabs.c (expand_float): Likewise.
18656 (expand_fix): Likewise.
18657 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18660 Alan Hayward <alan.hayward@arm.com>
18661 David Sherwood <david.sherwood@arm.com>
18663 * optabs.c (expand_float): Explicitly check for scalars before
18664 using a branching expansion.
18665 (expand_fix): Likewise.
18667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18668 Alan Hayward <alan.hayward@arm.com>
18669 David Sherwood <david.sherwood@arm.com>
18671 * expr.c (convert_mode): Split scalar handling out into...
18672 (convert_mode_scalar): ...this new function. Treat the modes
18675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18676 Alan Hayward <alan.hayward@arm.com>
18677 David Sherwood <david.sherwood@arm.com>
18679 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18681 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18684 Alan Hayward <alan.hayward@arm.com>
18685 David Sherwood <david.sherwood@arm.com>
18687 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18688 rather than a machine_mode.
18689 (fixed_from_string): Likewise.
18690 (fixed_convert): Likewise.
18691 (fixed_convert_from_int): Likewise.
18692 (fixed_convert_from_real): Likewise.
18693 (real_convert_from_fixed): Likewise.
18694 * fixed-value.c (fixed_from_double_int): Likewise.
18695 (fixed_from_string): Likewise.
18696 (fixed_convert): Likewise.
18697 (fixed_convert_from_int): Likewise.
18698 (fixed_convert_from_real): Likewise.
18699 (real_convert_from_fixed): Likewise.
18700 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18703 Alan Hayward <alan.hayward@arm.com>
18704 David Sherwood <david.sherwood@arm.com>
18706 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18707 of separate mode class checks. Do not allow vector modes here.
18708 (immed_wide_int_const): Use as_a <scalar_mode>.
18709 * explow.c (trunc_int_for_mode): Likewise.
18710 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18711 (wi::shwi): Likewise.
18712 (wi::min_value): Likewise.
18713 (wi::max_value): Likewise.
18714 * dwarf2out.c (loc_descriptor): Likewise.
18715 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18716 for CONST_WIDE_INT.
18718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18719 Alan Hayward <alan.hayward@arm.com>
18720 David Sherwood <david.sherwood@arm.com>
18722 * tree.h (SCALAR_TYPE_MODE): New macro.
18723 * expr.c (expand_expr_addr_expr_1): Use it.
18724 (expand_expr_real_2): Likewise.
18725 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18726 (fold_convert_const_fixed_from_int): Likewise.
18727 (fold_convert_const_fixed_from_real): Likewise.
18728 (native_encode_fixed): Likewise
18729 (native_encode_complex): Likewise
18730 (native_encode_vector): Likewise.
18731 (native_interpret_fixed): Likewise.
18732 (native_interpret_real): Likewise.
18733 (native_interpret_complex): Likewise.
18734 (native_interpret_vector): Likewise.
18735 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18736 (simd_clone_adjust_argument_types): Likewise.
18737 (simd_clone_init_simd_arrays): Likewise.
18738 (simd_clone_adjust): Likewise.
18739 * stor-layout.c (layout_type): Likewise.
18740 * tree.c (build_minus_one_cst): Likewise.
18741 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18742 * tree-inline.c (estimate_move_cost): Likewise.
18743 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18744 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18745 (vectorizable_reduction): Likewise.
18746 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18747 (vect_recog_mixed_size_cond_pattern): Likewise.
18748 (check_bool_pattern): Likewise.
18749 (adjust_bool_pattern): Likewise.
18750 (search_type_for_mask_1): Likewise.
18751 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18752 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18753 (vectorizable_load): Likewise.
18754 (vectorizable_store): Likewise.
18755 * ubsan.c (ubsan_encode_value): Likewise.
18756 * varasm.c (output_constant): Likewise.
18758 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18759 Alan Hayward <alan.hayward@arm.com>
18760 David Sherwood <david.sherwood@arm.com>
18762 * coretypes.h (scalar_mode): New class.
18763 * machmode.h (scalar_mode): Likewise.
18764 (scalar_mode::includes_p): New function.
18765 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18766 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18767 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18768 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18769 * expmed.c (store_bit_field_1): Likewise.
18770 (extract_bit_field_1): Likewise.
18771 * expr.c (write_complex_part): Likewise.
18772 (read_complex_part): Likewise.
18773 (emit_move_complex_push): Likewise.
18774 (expand_expr_real_2): Likewise.
18775 * function.c (assign_parm_setup_reg): Likewise.
18776 (assign_parms_unsplit_complex): Likewise.
18777 * optabs.c (expand_binop): Likewise.
18778 * rtlanal.c (subreg_get_info): Likewise.
18779 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18780 * varasm.c (output_constant_pool_2): Likewise.
18782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18783 Alan Hayward <alan.hayward@arm.com>
18784 David Sherwood <david.sherwood@arm.com>
18786 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18788 (expmed_mult_highpart_optab): Likewise.
18789 (expmed_mult_highpart): Likewise.
18791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18792 Alan Hayward <alan.hayward@arm.com>
18793 David Sherwood <david.sherwood@arm.com>
18795 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18796 instead of a machine_mode.
18797 (builtin_memset_read_str): Likewise.
18798 * builtins.c (c_readstr): Likewise.
18799 (builtin_memcpy_read_str): Likewise.
18800 (builtin_strncpy_read_str): Likewise.
18801 (builtin_memset_read_str): Likewise.
18802 (builtin_memset_gen_str): Likewise.
18803 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18804 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18805 instead of a machine_mode.
18806 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18808 (make_extraction): Likewise.
18809 (try_widen_shift_mode): Take and return scalar_int_modes instead
18811 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18812 a scalar_int_mode instead of a machine_mode.
18813 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18814 (avr_addr_space_pointer_mode): Likewise.
18815 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18816 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18817 (msp430_unwind_word_mode): Likewise.
18818 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18819 (spu_addr_space_pointer_mode): Likewise.
18820 (spu_addr_space_address_mode): Likewise.
18821 (spu_libgcc_cmp_return_mode): Likewise.
18822 (spu_libgcc_shift_count_mode): Likewise.
18823 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18824 (rl78_addr_space_pointer_mode): Likewise.
18825 (fl78_unwind_word_mode): Likewise.
18826 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18828 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18829 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18830 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18831 (mips_valid_pointer_mode): Likewise.
18832 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18833 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18834 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18836 (ft32_addr_space_address_mode): Likewise.
18837 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18838 scalar_int_mode instead of a machine_mode.
18839 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18841 (m32c_addr_space_address_mode): Likewise.
18842 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18843 (rs6000_eh_return_filter_mode): Likewise.
18844 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18845 (rs6000_eh_return_filter_mode): Likewise.
18846 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18847 (s390_libgcc_shift_count_mode): Likewise.
18848 (s390_unwind_word_mode): Likewise.
18849 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18851 * target.def (mode_rep_extended): Likewise.
18852 (valid_pointer_mode): Likewise.
18853 (addr_space.valid_pointer_mode): Likewise.
18854 (eh_return_filter_mode): Return a scalar_int_mode rather than
18856 (libgcc_cmp_return_mode): Likewise.
18857 (libgcc_shift_count_mode): Likewise.
18858 (unwind_word_mode): Likewise.
18859 (addr_space.pointer_mode): Likewise.
18860 (addr_space.address_mode): Likewise.
18861 * doc/tm.texi: Regenerate.
18862 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18864 (do_jump): Use scalar_int_mode for local variables.
18865 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18866 rather than a machine_mode.
18867 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18868 (scompare_loc_descriptor_wide): Likewise.
18869 (scompare_loc_descriptor_narrow): Likewise.
18870 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18872 * except.c (sjlj_emit_dispatch_table): Likewise.
18873 (expand_builtin_eh_copy_values): Likewise.
18874 * explow.c (convert_memory_address_addr_space_1): Likewise.
18875 Take a scalar_int_mode rather than a machine_mode.
18876 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18877 than a machine_mode.
18878 (memory_address_addr_space): Use scalar_int_mode for local variables.
18879 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18880 rather than a machine_mode.
18881 * expmed.c (mask_rtx): Likewise.
18882 (init_expmed_one_conv): Likewise.
18883 (expand_mult_highpart_adjust): Likewise.
18884 (extract_high_half): Likewise.
18885 (expmed_mult_highpart_optab): Likewise.
18886 (expmed_mult_highpart): Likewise.
18887 (expand_smod_pow2): Likewise.
18888 (expand_sdiv_pow2): Likewise.
18889 (emit_store_flag_int): Likewise.
18890 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18892 (extract_low_bits): Likewise.
18893 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18895 * expr.c (pieces_addr::adjust): Likewise.
18896 (can_store_by_pieces): Likewise.
18897 (store_by_pieces): Likewise.
18898 (clear_by_pieces_1): Likewise.
18899 (expand_expr_addr_expr_1): Likewise.
18900 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18901 (expand_expr_real_1): Likewise.
18902 (try_casesi): Likewise.
18903 * final.c (shorten_branches): Likewise.
18904 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18905 type of "mode" to machine_mode.
18906 * internal-fn.c (expand_arith_overflow_result_store): Take a
18907 scalar_int_mode rather than a machine_mode.
18908 (expand_mul_overflow): Use scalar_int_mode for local variables.
18909 * loop-doloop.c (doloop_modify): Likewise.
18910 (doloop_optimize): Likewise.
18911 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18912 than a machine_mode.
18913 (expand_doubleword_shift_condmove): Likewise.
18914 (expand_doubleword_shift): Likewise.
18915 (expand_doubleword_clz): Likewise.
18916 (expand_doubleword_popcount): Likewise.
18917 (expand_doubleword_parity): Likewise.
18918 (expand_absneg_bit): Use scalar_int_mode for local variables.
18919 (prepare_float_lib_cmp): Likewise.
18920 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18921 rather than a machine_mode.
18922 (convert_memory_address_addr_space): Likewise.
18923 (get_mode_bounds): Likewise.
18924 (get_address_mode): Return a scalar_int_mode rather than a
18926 * rtlanal.c (get_address_mode): Likewise.
18927 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18928 than a machine_mode.
18929 * targhooks.c (default_mode_rep_extended): Likewise.
18930 (default_valid_pointer_mode): Likewise.
18931 (default_addr_space_valid_pointer_mode): Likewise.
18932 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18933 than a machine_mode.
18934 (default_libgcc_cmp_return_mode): Likewise.
18935 (default_libgcc_shift_count_mode): Likewise.
18936 (default_unwind_word_mode): Likewise.
18937 (default_addr_space_pointer_mode): Likewise.
18938 (default_addr_space_address_mode): Likewise.
18939 * targhooks.h (default_eh_return_filter_mode): Likewise.
18940 (default_libgcc_cmp_return_mode): Likewise.
18941 (default_libgcc_shift_count_mode): Likewise.
18942 (default_unwind_word_mode): Likewise.
18943 (default_addr_space_pointer_mode): Likewise.
18944 (default_addr_space_address_mode): Likewise.
18945 (default_mode_rep_extended): Take a scalar_int_mode rather than
18947 (default_valid_pointer_mode): Likewise.
18948 (default_addr_space_valid_pointer_mode): Likewise.
18949 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18951 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18952 rather than a machine_mode.
18953 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18954 for local variables.
18955 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18956 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18957 than a machine_mode.
18959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18960 Alan Hayward <alan.hayward@arm.com>
18961 David Sherwood <david.sherwood@arm.com>
18963 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18964 the mode argument to scalar_int_mode.
18965 (do_jump_by_parts_zero_rtx): Likewise.
18966 (do_jump_by_parts_equality_rtx): Likewise.
18967 (do_jump_by_parts_greater): Take a mode argument.
18968 (do_jump_by_parts_equality): Likewise.
18969 (do_jump_1): Update calls accordingly.
18971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18972 Alan Hayward <alan.hayward@arm.com>
18973 David Sherwood <david.sherwood@arm.com>
18975 * is-a.h (safe_dyn_cast): New function.
18976 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18977 (jump_table_for_label): Likewise.
18978 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18979 instead of an rtx_insn *.
18980 (shorten_branches): Use dyn_cast instead of LABEL_P and
18981 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18982 rtx_jump_table_data::get_data_mode.
18983 (final_scan_insn): Likewise.
18985 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18986 Alan Hayward <alan.hayward@arm.com>
18987 David Sherwood <david.sherwood@arm.com>
18989 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18990 trying to combine a full-register integer set with a subreg
18993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18994 Alan Hayward <alan.hayward@arm.com>
18995 David Sherwood <david.sherwood@arm.com>
18997 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18998 that is always either address_mode or pointer_mode.
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 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
19005 when the two are known to be equal.
19007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19008 Alan Hayward <alan.hayward@arm.com>
19009 David Sherwood <david.sherwood@arm.com>
19011 * simplify-rtx.c (simplify_const_unary_operation): Use
19012 is_a <scalar_int_mode> instead of checking for a nonzero
19013 precision. Forcibly convert op_mode to a scalar_int_mode
19014 in that case. More clearly differentiate the operand and
19015 result modes and use the former when deciding what the value
19016 of a count-bits operation should be. Use is_int_mode instead
19017 of checking for a MODE_INT. Remove redundant check for whether
19018 this mode has a zero precision.
19020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19021 Alan Hayward <alan.hayward@arm.com>
19022 David Sherwood <david.sherwood@arm.com>
19024 * optabs.c (widen_leading): Change the type of the mode argument
19025 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
19026 (widen_bswap): Likewise.
19027 (expand_parity): Likewise.
19028 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
19029 (expand_ffs): Likewise.
19030 (epand_unop): Check for scalar integer modes before calling the
19033 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19034 Alan Hayward <alan.hayward@arm.com>
19035 David Sherwood <david.sherwood@arm.com>
19037 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
19039 (expand_expr_real_1): Update call accordingly.
19041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19042 Alan Hayward <alan.hayward@arm.com>
19043 David Sherwood <david.sherwood@arm.com>
19045 * expmed.c (store_bit_field_using_insv): Add op0_mode and
19046 value_mode arguments. Use scalar_int_mode internally.
19047 (store_bit_field_1): Rename the new integer mode from imode
19048 to op0_mode and use it instead of GET_MODE (op0). Update calls
19049 to store_split_bit_field, store_bit_field_using_insv and
19050 store_fixed_bit_field.
19051 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
19052 Use scalar_int_mode internally. Use a bit count rather than a mode
19053 when calculating the largest bit size for get_best_mode.
19054 Update calls to store_split_bit_field and store_fixed_bit_field_1.
19055 (store_fixed_bit_field_1): Add mode and value_mode arguments.
19056 Remove assertion that OP0 has a scalar integer mode.
19057 (store_split_bit_field): Add op0_mode and value_mode arguments.
19058 Update calls to extract_fixed_bit_field.
19059 (extract_bit_field_using_extv): Add an op0_mode argument.
19060 Use scalar_int_mode internally.
19061 (extract_bit_field_1): Rename the new integer mode from imode to
19062 op0_mode and use it instead of GET_MODE (op0). Update calls to
19063 extract_split_bit_field, extract_bit_field_using_extv and
19064 extract_fixed_bit_field.
19065 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
19066 to extract_split_bit_field and extract_fixed_bit_field_1.
19067 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
19068 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
19069 on the target mode.
19070 (extract_split_bit_field): Add an op0_mode argument. Update call
19071 to extract_fixed_bit_field.
19073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19074 Alan Hayward <alan.hayward@arm.com>
19075 David Sherwood <david.sherwood@arm.com>
19077 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
19078 * explow.c (hard_function_value): Likewise.
19079 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
19080 convert_to_mode call outside the loop.
19081 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
19082 for the mode iterator. Require the mode specified by max_pieces
19084 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
19086 (copy_blkmode_to_reg): Likewise.
19087 (set_storage_via_setmem): Likewise.
19088 * optabs.c (prepare_cmp_insn): Likewise.
19089 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19090 * stor-layout.c (finish_bitfield_representative): Likewise.
19092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19093 Alan Hayward <alan.hayward@arm.com>
19094 David Sherwood <david.sherwood@arm.com>
19096 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
19097 * expr.c (convert_move): Use them.
19098 (convert_modes): Likewise.
19099 (store_expr_with_bounds): Likewise.
19101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19102 Alan Hayward <alan.hayward@arm.com>
19103 David Sherwood <david.sherwood@arm.com>
19105 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
19106 parameter for the mode of "x". Remove the "known_x", "known_mode"
19107 and "known_ret" arguments. Change the type of the mode argument
19108 to scalar_int_mode.
19109 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
19110 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
19111 (reg_num_sign_bit_copies_for_combine): Likewise.
19112 * rtlanal.c (nonzero_bits1): Likewise.
19113 (num_sign_bit_copies1): Likewise.
19114 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
19115 (reg_num_sign_bit_copies_general): Likewise.
19116 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
19117 (reg_nonzero_bits_general): Likewise.
19119 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19120 Alan Hayward <alan.hayward@arm.com>
19121 David Sherwood <david.sherwood@arm.com>
19123 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
19124 than in subroutines. Return 1 for non-integer modes.
19125 (cached_num_sign_bit_copies): Change the type of the mode parameter
19126 to scalar_int_mode.
19127 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
19128 classes. Handle CONST_INT_P first and then check whether X also
19129 has a scalar integer mode. Check the same thing for inner registers
19130 of a SUBREG and for values that are being extended or truncated.
19132 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19133 Alan Hayward <alan.hayward@arm.com>
19134 David Sherwood <david.sherwood@arm.com>
19136 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
19137 in subroutines. Return the mode mask for non-integer modes.
19138 (cached_nonzero_bits): Change the type of the mode parameter
19139 to scalar_int_mode.
19140 (nonzero_bits1): Likewise. Remove early exit for other mode
19141 classes. Handle CONST_INT_P first and then check whether X
19142 also has a scalar integer mode.
19144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19145 Alan Hayward <alan.hayward@arm.com>
19146 David Sherwood <david.sherwood@arm.com>
19148 * expr.c (widest_int_mode_for_size): Make the comment match the code.
19149 Return a scalar_int_mode and assert that the size is greater than
19151 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
19152 (op_by_pieces_d::op_by_pieces_d): Likewise.
19153 (op_by_pieces_d::run): Likewise.
19154 (can_store_by_pieces): Likewise.
19156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19157 Alan Hayward <alan.hayward@arm.com>
19158 David Sherwood <david.sherwood@arm.com>
19160 * combine.c (extract_left_shift): Add a mode argument and update
19162 (make_compound_operation_int): Change the type of the mode parameter
19163 to scalar_int_mode and update the call to extract_left_shift.
19165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19166 Alan Hayward <alan.hayward@arm.com>
19167 David Sherwood <david.sherwood@arm.com>
19169 * combine.c (simplify_and_const_int): Change the type of the mode
19170 parameter to scalar_int_mode.
19171 (simplify_and_const_int_1): Likewise. Update recursive call.
19173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19174 Alan Hayward <alan.hayward@arm.com>
19175 David Sherwood <david.sherwood@arm.com>
19177 * combine.c (simplify_compare_const): Check that the mode is a
19178 scalar_int_mode (rather than VOIDmode) before testing its
19180 (simplify_comparison): Move COMPARISON_P handling out of the
19181 loop and restrict the latter part of the loop to scalar_int_modes.
19182 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
19183 and when considering SUBREG_REGs. Use is_int_mode instead of
19184 checking GET_MODE_CLASS against MODE_INT.
19186 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19187 Alan Hayward <alan.hayward@arm.com>
19188 David Sherwood <david.sherwood@arm.com>
19190 * combine.c (try_widen_shift_mode): Move check for equal modes to...
19191 (simplify_shift_const_1): ...here. Use scalar_int_mode for
19192 shift_unit_mode and for modes involved in scalar shifts.
19194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19195 Alan Hayward <alan.hayward@arm.com>
19196 David Sherwood <david.sherwood@arm.com>
19198 * combine.c (force_int_to_mode): New function, split out from...
19199 (force_to_mode): ...here. Keep xmode up-to-date and use it
19200 instead of GET_MODE (x).
19202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19203 Alan Hayward <alan.hayward@arm.com>
19204 David Sherwood <david.sherwood@arm.com>
19206 * optabs-query.h (extraction_insn::struct_mode): Change type to
19207 opt_scalar_int_mode and update comment.
19208 (extraction_insn::field_mode): Change type to scalar_int_mode.
19209 (extraction_insn::pos_mode): Likewise.
19210 * combine.c (make_extraction): Update accordingly.
19211 * optabs-query.c (get_traditional_extraction_insn): Likewise.
19212 (get_optab_extraction_insn): Likewise.
19213 * recog.c (simplify_while_replacing): Likewise.
19214 * expmed.c (narrow_bit_field_mem): Change the type of the mode
19215 parameter to opt_scalar_int_mode.
19217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19218 Alan Hayward <alan.hayward@arm.com>
19219 David Sherwood <david.sherwood@arm.com>
19221 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
19222 to a scalar_int_mode instead of a machine_mode.
19223 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19224 (get_best_mode): Return a boolean and use a pointer argument to store
19225 the selected mode. Replace the limit mode parameter with a bit limit.
19226 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19227 for the values returned by bit_field_mode_iterator::next_mode.
19228 (store_bit_field): Update call to get_best_mode.
19229 (store_fixed_bit_field): Likewise.
19230 (extract_fixed_bit_field): Likewise.
19231 * expr.c (optimize_bitfield_assignment_op): Likewise.
19232 * fold-const.c (optimize_bit_field_compare): Likewise.
19233 (fold_truth_andor_1): Likewise.
19234 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19235 Update for new type of m_mode.
19236 (get_best_mode): As above.
19238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19239 Alan Hayward <alan.hayward@arm.com>
19240 David Sherwood <david.sherwood@arm.com>
19242 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19243 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19244 (store_bit_field): Check is_a <scalar_int_mode> before calling
19245 strict_volatile_bitfield_p.
19246 (extract_bit_field): Likewise.
19248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19249 Alan Hayward <alan.hayward@arm.com>
19250 David Sherwood <david.sherwood@arm.com>
19252 * target.def (cstore_mode): Return a scalar_int_mode.
19253 * doc/tm.texi: Regenerate.
19254 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19255 * targhooks.h (default_cstore_mode): Likewise.
19256 * targhooks.c (default_cstore_mode): Likewise, using a forced
19258 * expmed.c (emit_cstore): Expect the target of the cstore to be
19261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19262 Alan Hayward <alan.hayward@arm.com>
19263 David Sherwood <david.sherwood@arm.com>
19265 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19267 (niter_desc): Likewise mode.
19268 (iv_analyze): Add a mode parameter.
19270 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19271 and change its type to scalar_int_mode.
19272 * loop-iv.c: Update commentary at head of file.
19273 (iv_constant): Pass the mode paraeter before the rtx it describes
19274 and change its type to scalar_int_mode. Remove VOIDmode handling.
19275 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19276 (iv_extend): Likewise.
19277 (shorten_into_mode): Likewise.
19278 (iv_add): Use scalar_int_mode.
19279 (iv_mult): Likewise.
19280 (iv_shift): Likewise.
19281 (canonicalize_iv_subregs): Likewise.
19282 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19283 it describes and change its mode to scalar_int_mode. Also change
19284 the type of the returned inner_mode to scalar_int_mode.
19285 (get_biv_step): Likewise, turning outer_mode from a pointer
19286 into a direct parameter. Update call to get_biv_step_1.
19287 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19288 iv_constant and get_biv_step.
19289 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19290 and change its type to scalar_int_mode. Don't initialise iv->mode
19291 to VOIDmode and remove later checks for its still being VOIDmode.
19292 Update calls to iv_analyze_op and iv_analyze_expr. Check
19293 is_a <scalar_int_mode> when changing the mode under consideration.
19294 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19295 Update call to iv_analyze_expr.
19296 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19297 inner register is not also a scalar_int_mode. Update call to
19299 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19300 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19301 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19302 separate mode class checks. Update calls to iv_analyze. Remove
19303 fix-up of VOIDmodes after iv_analyze_biv.
19304 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19305 don't have a scalar_int_mode. Update call to biv_p.
19307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19308 Alan Hayward <alan.hayward@arm.com>
19309 David Sherwood <david.sherwood@arm.com>
19311 * cfgexpand.c (convert_debug_memory_address): Use
19312 as_a <scalar_int_mode>.
19313 * combine.c (expand_compound_operation): Likewise.
19314 (make_extraction): Likewise.
19315 (change_zero_ext): Likewise.
19316 (simplify_comparison): Likewise.
19317 * cse.c (cse_insn): Likewise.
19318 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19319 (mem_loc_descriptor): Likewise.
19320 (loc_descriptor): Likewise.
19321 * expmed.c (init_expmed_one_mode): Likewise.
19322 (synth_mult): Likewise.
19323 (emit_store_flag_1): Likewise.
19324 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19325 of a comparison with size.
19326 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19327 (reduce_to_bit_field_precision): Likewise.
19328 * function.c (expand_function_end): Likewise.
19329 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19330 * loop-doloop.c (doloop_modify): Likewise.
19331 * optabs.c (expand_binop): Likewise.
19332 (expand_unop): Likewise.
19333 (expand_copysign_absneg): Likewise.
19334 (prepare_cmp_insn): Likewise.
19335 (maybe_legitimize_operand): Likewise.
19336 * recog.c (const_scalar_int_operand): Likewise.
19337 * rtlanal.c (get_address_mode): Likewise.
19338 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19339 (simplify_cond_clz_ctz): Likewise.
19340 * tree-nested.c (get_nl_goto_field): Likewise.
19341 * tree.c (build_vector_type_for_mode): Likewise.
19342 * var-tracking.c (use_narrower_mode): Likewise.
19344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19345 Alan Hayward <alan.hayward@arm.com>
19346 David Sherwood <david.sherwood@arm.com>
19348 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19349 * builtins.c (expand_builtin_signbit): Use it.
19350 * cfgexpand.c (expand_debug_expr): Likewise.
19351 * dojump.c (do_jump): Likewise.
19352 (do_compare_and_jump): Likewise.
19353 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19354 * expmed.c (make_tree): Likewise.
19355 * expr.c (expand_expr_real_2): Likewise.
19356 (expand_expr_real_1): Likewise.
19357 (try_casesi): Likewise.
19358 * fold-const-call.c (fold_const_call_ss): Likewise.
19359 * fold-const.c (unextend): Likewise.
19360 (extract_muldiv_1): Likewise.
19361 (fold_single_bit_test): Likewise.
19362 (native_encode_int): Likewise.
19363 (native_encode_string): Likewise.
19364 (native_interpret_int): Likewise.
19365 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19366 * internal-fn.c (expand_addsub_overflow): Likewise.
19367 (expand_neg_overflow): Likewise.
19368 (expand_mul_overflow): Likewise.
19369 (expand_arith_overflow): Likewise.
19370 * match.pd: Likewise.
19371 * stor-layout.c (layout_type): Likewise.
19372 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19373 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19374 * tree-ssanames.c (get_range_info): Likewise.
19375 * tree-switch-conversion.c (array_value_type) Likewise.
19376 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19377 (vect_recog_divmod_pattern): Likewise.
19378 (vect_recog_mixed_size_cond_pattern): Likewise.
19379 * tree-vrp.c (extract_range_basic): Likewise.
19380 (simplify_float_conversion_using_ranges): Likewise.
19381 * tree.c (int_fits_type_p): Likewise.
19382 * ubsan.c (instrument_bool_enum_load): Likewise.
19383 * varasm.c (mergeable_string_section): Likewise.
19384 (narrowing_initializer_constant_valid_p): Likewise.
19385 (output_constant): Likewise.
19387 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19388 Alan Hayward <alan.hayward@arm.com>
19389 David Sherwood <david.sherwood@arm.com>
19391 * machmode.h (NARROWEST_INT_MODE): New macro.
19392 * expr.c (alignment_for_piecewise_move): Use it instead of
19393 GET_CLASS_NARROWEST_MODE (MODE_INT).
19394 (push_block): Likewise.
19395 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19397 * tree-vrp.c (simplify_float_conversion_using_ranges): 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 * postreload.c (move2add_valid_value_p): Change the type of the
19404 mode parameter to scalar_int_mode.
19405 (move2add_use_add2_insn): Add a mode parameter and use it instead
19407 (move2add_use_add3_insn): Likewise.
19408 (reload_cse_move2add): Update accordingly.
19410 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19411 Alan Hayward <alan.hayward@arm.com>
19412 David Sherwood <david.sherwood@arm.com>
19414 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19416 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19417 * optabs.c (expand_unop): Likewise.
19419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19420 Alan Hayward <alan.hayward@arm.com>
19421 David Sherwood <david.sherwood@arm.com>
19423 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19424 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19425 (popcount_loc_descriptor): Likewise.
19426 (bswap_loc_descriptor): Likewise.
19427 (rotate_loc_descriptor): Likewise.
19428 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19429 calling the functions above.
19431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19432 Alan Hayward <alan.hayward@arm.com>
19433 David Sherwood <david.sherwood@arm.com>
19435 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19437 (try_combine): Likewise.
19438 (simplify_if_then_else): Likewise.
19439 * cse.c (cse_insn): Likewise.
19440 * dwarf2out.c (mem_loc_descriptor): Likewise.
19441 * emit-rtl.c (gen_lowpart_common): Likewise.
19442 * simplify-rtx.c (simplify_truncation): Likewise.
19443 (simplify_binary_operation_1): Likewise.
19444 (simplify_const_relational_operation): Likewise.
19445 (simplify_ternary_operation): Likewise.
19446 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19449 Alan Hayward <alan.hayward@arm.com>
19450 David Sherwood <david.sherwood@arm.com>
19452 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19453 * reload.c (push_reload): Likewise.
19454 (find_reloads): Likewise.
19456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19457 Alan Hayward <alan.hayward@arm.com>
19458 David Sherwood <david.sherwood@arm.com>
19460 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19461 (make_compound_operation_int): Likewise.
19462 (change_zero_ext): Likewise.
19463 * expr.c (convert_move): Likewise.
19464 (convert_modes): Likewise.
19465 * fwprop.c (forward_propagate_subreg): Likewise.
19466 * loop-iv.c (get_biv_step_1): Likewise.
19467 * optabs.c (widen_operand): Likewise.
19468 * postreload.c (move2add_valid_value_p): Likewise.
19469 * recog.c (simplify_while_replacing): Likewise.
19470 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19471 (simplify_binary_operation_1): Likewise. Remove redundant
19472 mode equality check.
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 * combine.c (combine_simplify_rtx): Add checks for
19479 is_a <scalar_int_mode>.
19480 (simplify_if_then_else): Likewise.
19481 (make_field_assignment): Likewise.
19482 (simplify_comparison): Likewise.
19483 * ifcvt.c (noce_try_bitop): Likewise.
19484 * loop-invariant.c (canonicalize_address_mult): Likewise.
19485 * simplify-rtx.c (simplify_unary_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 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19492 is_a <scalar_int_mode> instead of != BLKmode.
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 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19499 instead of != VOIDmode.
19500 * combine.c (if_then_else_cond): Likewise.
19501 (change_zero_ext): Likewise.
19502 * dwarf2out.c (mem_loc_descriptor): Likewise.
19503 (loc_descriptor): Likewise.
19504 * rtlanal.c (canonicalize_condition): Likewise.
19505 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19508 Alan Hayward <alan.hayward@arm.com>
19509 David Sherwood <david.sherwood@arm.com>
19511 * simplify-rtx.c (simplify_binary_operation_1): Use
19512 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19515 Alan Hayward <alan.hayward@arm.com>
19516 David Sherwood <david.sherwood@arm.com>
19518 * wide-int.h (int_traits<unsigned char>) New class.
19519 (int_traits<unsigned short>) Likewise.
19520 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19521 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19523 * combine.c (set_nonzero_bits_and_sign_copies): Use
19524 is_a <scalar_int_mode>.
19525 (find_split_point): Likewise.
19526 (combine_simplify_rtx): Likewise.
19527 (simplify_logical): Likewise.
19528 (expand_compound_operation): Likewise.
19529 (expand_field_assignment): Likewise.
19530 (make_compound_operation): Likewise.
19531 (extended_count): Likewise.
19532 (change_zero_ext): Likewise.
19533 (simplify_comparison): Likewise.
19534 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19535 (ucompare_loc_descriptor): Likewise.
19536 (minmax_loc_descriptor): Likewise.
19537 (mem_loc_descriptor): Likewise.
19538 (loc_descriptor): Likewise.
19539 * expmed.c (init_expmed_one_mode): Likewise.
19540 * lra-constraints.c (lra_constraint_offset): Likewise.
19541 * optabs.c (prepare_libcall_arg): Likewise.
19542 * postreload.c (move2add_note_store): Likewise.
19543 * reload.c (operands_match_p): Likewise.
19544 * rtl.h (load_extend_op): Likewise.
19545 * rtlhooks.c (gen_lowpart_general): Likewise.
19546 * simplify-rtx.c (simplify_truncation): Likewise.
19547 (simplify_unary_operation_1): Likewise.
19548 (simplify_binary_operation_1): Likewise.
19549 (simplify_const_binary_operation): Likewise.
19550 (simplify_const_relational_operation): Likewise.
19551 (simplify_subreg): Likewise.
19552 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19553 * var-tracking.c (adjust_mems): Likewise.
19554 (prepare_call_arguments): Likewise.
19556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19557 Alan Hayward <alan.hayward@arm.com>
19558 David Sherwood <david.sherwood@arm.com>
19560 * machmode.h (is_int_mode): New fuction.
19561 * combine.c (find_split_point): Use it.
19562 (combine_simplify_rtx): Likewise.
19563 (simplify_if_then_else): Likewise.
19564 (simplify_set): Likewise.
19565 (simplify_shift_const_1): Likewise.
19566 (simplify_comparison): Likewise.
19567 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19568 * cse.c (notreg_cost): Likewise.
19569 (cse_insn): Likewise.
19570 * cselib.c (cselib_lookup_1): Likewise.
19571 * dojump.c (do_jump_1): Likewise.
19572 (do_compare_rtx_and_jump): Likewise.
19573 * dse.c (get_call_args): Likewise.
19574 * dwarf2out.c (rtl_for_decl_init): Likewise.
19575 (native_encode_initializer): Likewise.
19576 * expmed.c (emit_store_flag_1): Likewise.
19577 (emit_store_flag): Likewise.
19578 * expr.c (convert_modes): Likewise.
19579 (store_field): Likewise.
19580 (expand_expr_real_1): Likewise.
19581 * fold-const.c (fold_read_from_constant_string): Likewise.
19582 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19583 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19584 * optabs.c (expand_binop): Likewise.
19585 (expand_unop): Likewise.
19586 (expand_abs_nojump): Likewise.
19587 (expand_one_cmpl_abs_nojump): Likewise.
19588 * simplify-rtx.c (mode_signbit_p): Likewise.
19589 (val_signbit_p): Likewise.
19590 (val_signbit_known_set_p): Likewise.
19591 (val_signbit_known_clear_p): Likewise.
19592 (simplify_relational_operation_1): Likewise.
19593 * tree.c (vector_type_mode): Likewise.
19595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19596 Alan Hayward <alan.hayward@arm.com>
19597 David Sherwood <david.sherwood@arm.com>
19599 * machmode.h (smallest_mode_for_size): Fix formatting.
19600 (smallest_int_mode_for_size): New function.
19601 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19602 instead of smallest_mode_for_size.
19603 * combine.c (make_extraction): Likewise.
19604 * config/arc/arc.c (arc_expand_movmem): Likewise.
19605 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19606 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19607 * config/s390/s390.c (s390_expand_insv): Likewise.
19608 * config/sparc/sparc.c (assign_int_registers): Likewise.
19609 * config/spu/spu.c (spu_function_value): Likewise.
19610 (spu_function_arg): Likewise.
19611 * coverage.c (get_gcov_type): Likewise.
19612 (get_gcov_unsigned_t): Likewise.
19613 * dse.c (find_shift_sequence): Likewise.
19614 * expmed.c (store_bit_field_1): Likewise.
19615 * expr.c (convert_move): Likewise.
19616 (store_field): Likewise.
19617 * internal-fn.c (expand_arith_overflow): Likewise.
19618 * optabs-query.c (get_best_extraction_insn): Likewise.
19619 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19620 * stor-layout.c (layout_type): Likewise.
19621 (initialize_sizetypes): Likewise.
19622 * targhooks.c (default_get_mask_mode): Likewise.
19623 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19626 Alan Hayward <alan.hayward@arm.com>
19627 David Sherwood <david.sherwood@arm.com>
19629 * machmode.h (opt_mode::else_blk): New function.
19630 (int_mode_for_mode): Declare.
19631 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19632 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19634 * cfgexpand.c (expand_debug_expr): Likewise.
19635 * combine.c (gen_lowpart_or_truncate): Likewise.
19636 (gen_lowpart_for_combine): Likewise.
19637 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19638 * config/avr/avr.c (avr_to_int_mode): Likewise.
19639 (avr_out_plus_1): Likewise.
19640 (avr_out_plus): Likewise.
19641 (avr_out_round): Likewise.
19642 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19643 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19644 (s390_expand_vcond): Likewise.
19645 * config/spu/spu.c (spu_split_immediate): Likewise.
19646 (spu_expand_mov): Likewise.
19647 * dse.c (get_stored_val): Likewise.
19648 * expmed.c (store_bit_field_1): Likewise.
19649 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19651 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19652 (extract_low_bits): Likewise.
19653 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19654 handling rather than repeating the check.
19655 (emit_group_store): Likewise.
19656 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19657 * optabs.c (expand_absneg_bit): Likewise.
19658 (expand_copysign_absneg): Likewise.
19659 (expand_copysign_bit): Likewise.
19660 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19661 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19662 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19663 * var-tracking.c (prepare_call_arguments): Likewise.
19664 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19665 int_mode_for_mode instead of mode_for_size.
19666 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19669 Alan Hayward <alan.hayward@arm.com>
19670 David Sherwood <david.sherwood@arm.com>
19672 * machmode.h (int_mode_for_size): New function.
19673 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19674 instead of mode_for_size.
19675 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19677 * combine.c (expand_field_assignment): Use int_mode_for_size
19678 instead of mode_for_size.
19679 (make_extraction): Likewise.
19680 (simplify_shift_const_1): Likewise.
19681 (simplify_comparison): Likewise.
19682 * dojump.c (do_jump): Likewise.
19683 * dwarf2out.c (mem_loc_descriptor): Likewise.
19684 * emit-rtl.c (init_derived_machine_modes): Likewise.
19685 * expmed.c (flip_storage_order): Likewise.
19686 (convert_extracted_bit_field): Likewise.
19687 * expr.c (copy_blkmode_from_reg): Likewise.
19688 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19689 * internal-fn.c (expand_mul_overflow): Likewise.
19690 * lower-subreg.c (simple_move): Likewise.
19691 * optabs-libfuncs.c (init_optabs): Likewise.
19692 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19693 * tree.c (vector_type_mode): Likewise.
19694 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19695 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19696 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19697 * tree-vect-stmts.c (vectorizable_load): Likewise.
19698 (vectorizable_store): Likewise.
19700 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19701 Alan Hayward <alan.hayward@arm.com>
19702 David Sherwood <david.sherwood@arm.com>
19704 * coretypes.h (pod_mode): New type.
19705 (scalar_int_mode_pod): New typedef.
19706 * machmode.h (pod_mode): New class.
19707 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19708 * genmodes.c (emit_mode_int_n): Update accordingly.
19709 * lower-subreg.h (target_lower_subreg): Change type to
19710 scalar_int_mode_pod.
19711 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19712 scalar_int_mode_pod.
19714 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19715 Alan Hayward <alan.hayward@arm.com>
19716 David Sherwood <david.sherwood@arm.com>
19718 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19719 machine_mode to scalar_int_mode.
19720 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19721 (rs6000_option_override_internal): Remove cast to int.
19722 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19723 machine_mode to scalar_int_mode.
19724 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19725 (rs6000_option_override_internal): Remove cast to int.
19726 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19727 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19729 * config/s390/s390.c (s390_expand_builtin): Likewise.
19730 * coretypes.h (scalar_int_mode): New type.
19731 (opt_scalar_int_mode): New typedef.
19732 * machmode.h (scalar_int_mode): New class.
19733 (scalar_int_mode::includes_p): New function.
19734 (byte_mode): Change type to scalar_int_mode.
19735 (word_mode): Likewise.
19736 (ptr_mode): Likewise.
19737 * emit-rtl.c (byte_mode): Likewise.
19738 (word_mode): Likewise.
19739 (ptr_mode): Likewise.
19740 (init_derived_machine_modes): Update accordingly.
19741 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19742 and MODE_PARTIAL_INT.
19743 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19744 opt_scalar_int_mode.
19746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19747 Alan Hayward <alan.hayward@arm.com>
19748 David Sherwood <david.sherwood@arm.com>
19750 * target.def (libgcc_floating_mode_supported_p): Take a
19752 * doc/tm.texi: Regenerate.
19753 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19755 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19756 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19760 Alan Hayward <alan.hayward@arm.com>
19761 David Sherwood <david.sherwood@arm.com>
19763 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19764 * doc/tm.texi: Regenerate.
19765 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19766 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19767 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19768 * targhooks.h (default_floatn_mode): Likewise.
19769 * targhooks.c (default_floatn_mode): Likewise.
19770 * tree.c (build_common_tree_nodes): Update accordingly.
19772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19773 Alan Hayward <alan.hayward@arm.com>
19774 David Sherwood <david.sherwood@arm.com>
19776 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19777 (mode_iterator::iterate_p): Likewise.
19778 (mode_iterator::get_wider): Likewise.
19779 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19782 Alan Hayward <alan.hayward@arm.com>
19783 David Sherwood <david.sherwood@arm.com>
19785 * coretypes.h (opt_scalar_float_mode): New typedef.
19786 * machmode.h (float_mode_for_size): New function.
19787 * emit-rtl.c (double_mode): Delete.
19788 (init_emit_once): Use float_mode_for_size.
19789 * stor-layout.c (layout_type): Likewise.
19790 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19793 Alan Hayward <alan.hayward@arm.com>
19794 David Sherwood <david.sherwood@arm.com>
19796 * output.h (assemble_real): Take a scalar_float_mode.
19797 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19798 * config/arm/arm.md (consttable_4): Likewise.
19799 (consttable_8): Likewise.
19800 (consttable_16): Likewise.
19801 * config/mips/mips.md (consttable_float): Likewise.
19802 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19803 * varasm.c (assemble_real): Take a scalar_float_mode.
19804 (output_constant_pool_2): Update accordingly.
19805 (output_constant): 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 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19812 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19813 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19814 (native_encode_real): Likewise.
19815 (native_interpret_real): Likewise.
19816 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19817 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19820 Alan Hayward <alan.hayward@arm.com>
19821 David Sherwood <david.sherwood@arm.com>
19823 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19824 <scalar_float_mode>. Simplify.
19825 (gen_extend_conv_libfunc): Likewise.
19827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19828 Alan Hayward <alan.hayward@arm.com>
19829 David Sherwood <david.sherwood@arm.com>
19831 * coretypes.h (scalar_float_mode): New type.
19832 * machmode.h (mode_traits::from_int): Use machine_mode if
19833 USE_ENUM_MODES is defined.
19834 (is_a): New function.
19836 (dyn_cast): Likewise.
19837 (scalar_float_mode): New class.
19838 (scalar_float_mode::includes_p): New function.
19839 (is_float_mode): Likewise.
19840 * gdbhooks.py (MachineModePrinter): New class.
19841 (build_pretty_printer): Use it for scalar_float_mode.
19842 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19843 (format_helper::format_helper): Turn into a template.
19844 * genmodes.c (get_mode_class): New function.
19845 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19846 or machine_mode if none.
19847 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19848 as_a <scalar_float_mode>.
19849 * dwarf2out.c (mem_loc_descriptor): Likewise.
19850 (insert_float): Likewise.
19851 (add_const_value_attribute): Likewise.
19852 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19853 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19854 (expand_unop): Update accordingly.
19855 (expand_abs_nojump): Likewise.
19856 (expand_copysign_absneg): Take a scalar_float_mode.
19857 (expand_copysign_bit): Likewise.
19858 (expand_copysign): Update accordingly.
19860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19861 Alan Hayward <alan.hayward@arm.com>
19862 David Sherwood <david.sherwood@arm.com>
19864 * coretypes.h (opt_mode): New class.
19865 * machmode.h (opt_mode): Likewise.
19866 (opt_mode::else_void): New function.
19867 (opt_mode::require): Likewise.
19868 (opt_mode::exists): Likewise.
19869 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19870 (GET_MODE_2XWIDER_MODE): Likewise.
19871 (mode_iterator::get_wider): Update accordingly.
19872 (mode_iterator::get_2xwider): Likewise.
19873 (mode_iterator::get_known_wider): Likewise, turning into a template.
19874 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19875 forcing a wider mode to exist.
19876 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19877 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19878 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19879 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19880 * lower-subreg.c (init_lower_subreg): Likewise.
19881 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19882 on the final iteration.
19883 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19884 a wider mode exists before asking for a move pattern.
19885 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19886 forcing a wider mode to exist.
19887 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19888 returning false if no such mode exists.
19889 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19890 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19891 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19892 Avoid checking for a MODE_INT if we already know the mode is not a
19894 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19895 forcing a wider mode to exist.
19896 (expmed_mult_highpart_optab): Likewise.
19897 (expmed_mult_highpart): Likewise.
19898 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19900 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19901 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19902 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19903 * internal-fn.c (expand_mul_overflow): Update use of
19904 GET_MODE_2XWIDER_MODE.
19905 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19906 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19907 GET_MODE_WIDER_MODE.
19908 (convert_plusminus_to_widen): Likewise.
19909 * tree-switch-conversion.c (array_value_type): Likewise.
19910 * var-tracking.c (emit_note_insn_var_location): Likewise.
19911 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19912 Return false inside rather than outside the loop if no wider mode
19914 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19915 and GET_MODE_2XWIDER_MODE
19916 (can_compare_p): Use else_void.
19917 * gdbhooks.py (OptMachineModePrinter): New class.
19918 (build_pretty_printer): Use it for opt_mode.
19920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19921 Alan Hayward <alan.hayward@arm.com>
19922 David Sherwood <david.sherwood@arm.com>
19924 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19925 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19928 Alan Hayward <alan.hayward@arm.com>
19929 David Sherwood <david.sherwood@arm.com>
19931 * machmode.h (mode_traits): New structure.
19932 (get_narrowest_mode): New function.
19933 (mode_iterator::start): Likewise.
19934 (mode_iterator::iterate_p): Likewise.
19935 (mode_iterator::get_wider): Likewise.
19936 (mode_iterator::get_known_wider): Likewise.
19937 (mode_iterator::get_2xwider): Likewise.
19938 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19939 (FOR_EACH_MODE): Likewise.
19940 (FOR_EACH_MODE_FROM): Likewise.
19941 (FOR_EACH_MODE_UNTIL): Likewise.
19942 (FOR_EACH_WIDER_MODE): Likewise.
19943 (FOR_EACH_2XWIDER_MODE): Likewise.
19944 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19945 * combine.c (simplify_comparison): Likewise
19946 * config/i386/i386.c (type_natural_mode): Likewise.
19947 * cse.c (cse_insn): Likewise.
19948 * dse.c (find_shift_sequence): Likewise.
19949 * emit-rtl.c (init_derived_machine_modes): Likewise.
19950 (init_emit_once): Likewise.
19951 * explow.c (hard_function_value): Likewise.
19952 * expmed.c (extract_fixed_bit_field_1): Likewise.
19953 (extract_bit_field_1): Likewise.
19954 (expand_divmod): Likewise.
19955 (emit_store_flag_1): Likewise.
19956 * expr.c (init_expr_target): Likewise.
19957 (convert_move): Likewise.
19958 (alignment_for_piecewise_move): Likewise.
19959 (widest_int_mode_for_size): Likewise.
19960 (emit_block_move_via_movmem): Likewise.
19961 (copy_blkmode_to_reg): Likewise.
19962 (set_storage_via_setmem): Likewise.
19963 (compress_float_constant): Likewise.
19964 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19965 * optabs-query.c (get_best_extraction_insn): Likewise.
19966 * optabs.c (expand_binop): Likewise.
19967 (expand_twoval_unop): Likewise.
19968 (expand_twoval_binop): Likewise.
19969 (widen_leading): Likewise.
19970 (widen_bswap): Likewise.
19971 (expand_parity): Likewise.
19972 (expand_unop): Likewise.
19973 (prepare_cmp_insn): Likewise.
19974 (prepare_float_lib_cmp): Likewise.
19975 (expand_float): Likewise.
19976 (expand_fix): Likewise.
19977 (expand_sfix_optab): Likewise.
19978 * postreload.c (move2add_use_add2_insn): Likewise.
19979 * reg-stack.c (reg_to_stack): Likewise.
19980 * reginfo.c (choose_hard_reg_mode): Likewise.
19981 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19982 * stor-layout.c (mode_for_size): Likewise.
19983 (smallest_mode_for_size): Likewise.
19984 (mode_for_vector): Likewise.
19985 (finish_bitfield_representative): Likewise.
19986 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19987 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19988 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19989 * var-tracking.c (prepare_call_arguments): Likewise.
19991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19992 Alan Hayward <alan.hayward@arm.com>
19993 David Sherwood <david.sherwood@arm.com>
19995 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19996 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19997 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19998 * machmode.h (mode_size): Move earlier in file.
19999 (mode_precision): Likewise.
20000 (mode_inner): Likewise.
20001 (mode_nunits): Likewise.
20002 (mode_unit_size): Likewise.
20003 (unit_unit_precision): Likewise.
20004 (mode_wider): Likewise.
20005 (mode_2xwider): Likewise.
20006 (machine_mode): New class.
20007 (mode_to_bytes): New function.
20008 (mode_to_bits): Likewise.
20009 (mode_to_precision): Likewise.
20010 (mode_to_inner): Likewise.
20011 (mode_to_unit_size): Likewise.
20012 (mode_to_unit_precision): Likewise.
20013 (mode_to_nunits): Likewise.
20014 (GET_MODE_SIZE): Use mode_to_bytes.
20015 (GET_MODE_BITSIZE): Use mode_to_bits.
20016 (GET_MODE_PRECISION): Use mode_to_precision.
20017 (GET_MODE_INNER): Use mode_to_inner.
20018 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
20019 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
20020 (GET_MODE_NUNITS): Use mode_to_nunits.
20021 * system.h (ALWAYS_INLINE): New macro.
20022 * config/powerpcspe/powerpcspe-c.c
20023 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
20024 int for arg1_mode and arg2_mode.
20026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20027 Alan Hayward <alan.hayward@arm.com>
20028 David Sherwood <david.sherwood@arm.com>
20030 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20031 Prefix mode names with E_ in case statements.
20032 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20033 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
20034 (aarch64_split_simd_move): Likewise.
20035 (aarch64_gen_storewb_pair): Likewise.
20036 (aarch64_gen_loadwb_pair): Likewise.
20037 (aarch64_gen_store_pair): Likewise.
20038 (aarch64_gen_load_pair): Likewise.
20039 (aarch64_get_condition_code_1): Likewise.
20040 (aarch64_constant_pool_reload_icode): Likewise.
20041 (get_rsqrte_type): Likewise.
20042 (get_rsqrts_type): Likewise.
20043 (get_recpe_type): Likewise.
20044 (get_recps_type): Likewise.
20045 (aarch64_gimplify_va_arg_expr): Likewise.
20046 (aarch64_simd_container_mode): Likewise.
20047 (aarch64_emit_load_exclusive): Likewise.
20048 (aarch64_emit_store_exclusive): Likewise.
20049 (aarch64_expand_compare_and_swap): Likewise.
20050 (aarch64_gen_atomic_cas): Likewise.
20051 (aarch64_emit_bic): Likewise.
20052 (aarch64_emit_atomic_swap): Likewise.
20053 (aarch64_emit_atomic_load_op): Likewise.
20054 (aarch64_evpc_trn): Likewise.
20055 (aarch64_evpc_uzp): Likewise.
20056 (aarch64_evpc_zip): Likewise.
20057 (aarch64_evpc_ext): Likewise.
20058 (aarch64_evpc_rev): Likewise.
20059 (aarch64_evpc_dup): Likewise.
20060 (aarch64_gen_ccmp_first): Likewise.
20061 (aarch64_gen_ccmp_next): Likewise.
20062 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
20063 (alpha_emit_xfloating_libcall): Likewise.
20064 (emit_insxl): Likewise.
20065 (alpha_arg_type): Likewise.
20066 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
20067 (arc_preferred_simd_mode): Likewise.
20068 (arc_secondary_reload): Likewise.
20069 (get_arc_condition_code): Likewise.
20070 (arc_print_operand): Likewise.
20071 (arc_legitimate_constant_p): Likewise.
20072 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20073 * config/arc/arc.md (casesi_load): Likewise.
20074 (casesi_compact_jump): Likewise.
20075 * config/arc/predicates.md (proper_comparison_operator): Likewise.
20076 (cc_use_register): Likewise.
20077 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20078 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20079 (arm_init_iwmmxt_builtins): Likewise.
20080 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
20081 (neon_expand_vector_init): Likewise.
20082 (arm_attr_length_move_neon): Likewise.
20083 (maybe_get_arm_condition_code): Likewise.
20084 (arm_emit_vector_const): Likewise.
20085 (arm_preferred_simd_mode): Likewise.
20086 (arm_output_iwmmxt_tinsr): Likewise.
20087 (thumb1_output_casesi): Likewise.
20088 (thumb2_output_casesi): Likewise.
20089 (arm_emit_load_exclusive): Likewise.
20090 (arm_emit_store_exclusive): Likewise.
20091 (arm_expand_compare_and_swap): Likewise.
20092 (arm_evpc_neon_vuzp): Likewise.
20093 (arm_evpc_neon_vzip): Likewise.
20094 (arm_evpc_neon_vrev): Likewise.
20095 (arm_evpc_neon_vtrn): Likewise.
20096 (arm_evpc_neon_vext): Likewise.
20097 (arm_validize_comparison): Likewise.
20098 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
20099 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
20100 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
20101 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
20102 (c6x_preferred_simd_mode): Likewise.
20103 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
20104 (epiphany_rtx_costs): Likewise.
20105 * config/epiphany/predicates.md (proper_comparison_operator):
20107 * config/frv/frv.c (condexec_memory_operand): Likewise.
20108 (frv_emit_move): Likewise.
20109 (output_move_single): Likewise.
20110 (output_condmove_single): Likewise.
20111 (frv_hard_regno_mode_ok): Likewise.
20112 (frv_matching_accg_mode): Likewise.
20113 * config/h8300/h8300.c (split_adds_subs): Likewise.
20114 (h8300_rtx_costs): Likewise.
20115 (h8300_print_operand): Likewise.
20116 (compute_mov_length): Likewise.
20117 (output_logical_op): Likewise.
20118 (compute_logical_op_length): Likewise.
20119 (compute_logical_op_cc): Likewise.
20120 (h8300_shift_needs_scratch_p): Likewise.
20121 (output_a_shift): Likewise.
20122 (compute_a_shift_length): Likewise.
20123 (compute_a_shift_cc): Likewise.
20124 (expand_a_rotate): Likewise.
20125 (output_a_rotate): Likewise.
20126 * config/i386/i386.c (classify_argument): Likewise.
20127 (function_arg_advance_32): Likewise.
20128 (function_arg_32): Likewise.
20129 (function_arg_64): Likewise.
20130 (function_value_64): Likewise.
20131 (ix86_gimplify_va_arg): Likewise.
20132 (ix86_legitimate_constant_p): Likewise.
20133 (put_condition_code): Likewise.
20134 (split_double_mode): Likewise.
20135 (ix86_avx256_split_vector_move_misalign): Likewise.
20136 (ix86_expand_vector_logical_operator): Likewise.
20137 (ix86_split_idivmod): Likewise.
20138 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
20139 (ix86_build_const_vector): Likewise.
20140 (ix86_build_signbit_mask): Likewise.
20141 (ix86_match_ccmode): Likewise.
20142 (ix86_cc_modes_compatible): Likewise.
20143 (ix86_expand_branch): Likewise.
20144 (ix86_expand_sse_cmp): Likewise.
20145 (ix86_expand_sse_movcc): Likewise.
20146 (ix86_expand_int_sse_cmp): Likewise.
20147 (ix86_expand_vec_perm_vpermi2): Likewise.
20148 (ix86_expand_vec_perm): Likewise.
20149 (ix86_expand_sse_unpack): Likewise.
20150 (ix86_expand_int_addcc): Likewise.
20151 (ix86_split_to_parts): Likewise.
20152 (ix86_vectorize_builtin_gather): Likewise.
20153 (ix86_vectorize_builtin_scatter): Likewise.
20154 (avx_vpermilp_parallel): Likewise.
20155 (inline_memory_move_cost): Likewise.
20156 (ix86_tieable_integer_mode_p): Likewise.
20157 (x86_maybe_negate_const_int): Likewise.
20158 (ix86_expand_vector_init_duplicate): Likewise.
20159 (ix86_expand_vector_init_one_nonzero): Likewise.
20160 (ix86_expand_vector_init_one_var): Likewise.
20161 (ix86_expand_vector_init_concat): Likewise.
20162 (ix86_expand_vector_init_interleave): Likewise.
20163 (ix86_expand_vector_init_general): Likewise.
20164 (ix86_expand_vector_set): Likewise.
20165 (ix86_expand_vector_extract): Likewise.
20166 (emit_reduc_half): Likewise.
20167 (ix86_emit_i387_round): Likewise.
20168 (ix86_mangle_type): Likewise.
20169 (ix86_expand_round_sse4): Likewise.
20170 (expand_vec_perm_blend): Likewise.
20171 (canonicalize_vector_int_perm): Likewise.
20172 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
20173 (expand_vec_perm_1): Likewise.
20174 (expand_vec_perm_interleave3): Likewise.
20175 (expand_vec_perm_even_odd_pack): Likewise.
20176 (expand_vec_perm_even_odd_1): Likewise.
20177 (expand_vec_perm_broadcast_1): Likewise.
20178 (ix86_vectorize_vec_perm_const_ok): Likewise.
20179 (ix86_expand_vecop_qihi): Likewise.
20180 (ix86_expand_mul_widen_hilo): Likewise.
20181 (ix86_expand_sse2_abs): Likewise.
20182 (ix86_expand_pextr): Likewise.
20183 (ix86_expand_pinsr): Likewise.
20184 (ix86_preferred_simd_mode): Likewise.
20185 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
20186 * config/i386/sse.md (*andnot<mode>3): Likewise.
20187 (<mask_codefor><code><mode>3<mask_name>): Likewise.
20188 (*<code><mode>3): Likewise.
20189 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
20190 (ia64_expand_atomic_op): Likewise.
20191 (ia64_arg_type): Likewise.
20192 (ia64_mode_to_int): Likewise.
20193 (ia64_scalar_mode_supported_p): Likewise.
20194 (ia64_vector_mode_supported_p): Likewise.
20195 (expand_vec_perm_broadcast): Likewise.
20196 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
20197 (iq2000_function_arg_advance): Likewise.
20198 (iq2000_function_arg): Likewise.
20199 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
20200 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
20201 (m68k_libcall_value): Likewise.
20202 (m68k_function_value): Likewise.
20203 (sched_attr_op_type): Likewise.
20204 * config/mcore/mcore.c (mcore_output_move): Likewise.
20205 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
20207 (microblaze_function_arg): Likewise.
20208 * config/mips/mips.c (mips16_build_call_stub): Likewise.
20209 (mips_print_operand): Likewise.
20210 (mips_mode_ok_for_mov_fmt_p): Likewise.
20211 (mips_vector_mode_supported_p): Likewise.
20212 (mips_preferred_simd_mode): Likewise.
20213 (mips_expand_vpc_loongson_even_odd): Likewise.
20214 (mips_expand_vec_unpack): Likewise.
20215 (mips_expand_vi_broadcast): Likewise.
20216 (mips_expand_vector_init): Likewise.
20217 (mips_expand_vec_reduc): Likewise.
20218 (mips_expand_msa_cmp): Likewise.
20219 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
20220 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
20221 (cc_flags_for_mode): Likewise.
20222 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20223 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20224 (nds32_output_casesi_pc_relative): Likewise.
20225 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20226 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20227 (nvptx_gen_unpack): Likewise.
20228 (nvptx_gen_pack): Likewise.
20229 (nvptx_gen_shuffle): Likewise.
20230 (nvptx_gen_wcast): Likewise.
20231 (nvptx_preferred_simd_mode): Likewise.
20232 * config/pa/pa.c (pa_secondary_reload): Likewise.
20233 * config/pa/predicates.md (base14_operand): Likewise.
20234 * config/powerpcspe/powerpcspe-c.c
20235 (altivec_resolve_overloaded_builtin): Likewise.
20236 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20238 (rs6000_preferred_simd_mode): Likewise.
20239 (output_vec_const_move): Likewise.
20240 (rs6000_expand_vector_extract): Likewise.
20241 (rs6000_split_vec_extract_var): Likewise.
20242 (reg_offset_addressing_ok_p): Likewise.
20243 (rs6000_legitimate_offset_address_p): Likewise.
20244 (rs6000_legitimize_address): Likewise.
20245 (rs6000_emit_set_const): Likewise.
20246 (rs6000_const_vec): Likewise.
20247 (rs6000_emit_move): Likewise.
20248 (spe_build_register_parallel): Likewise.
20249 (rs6000_darwin64_record_arg_recurse): Likewise.
20250 (swap_selector_for_mode): Likewise.
20251 (spe_init_builtins): Likewise.
20252 (paired_init_builtins): Likewise.
20253 (altivec_init_builtins): Likewise.
20254 (do_load_for_compare): Likewise.
20255 (rs6000_generate_compare): Likewise.
20256 (rs6000_expand_float128_convert): Likewise.
20257 (emit_load_locked): Likewise.
20258 (emit_store_conditional): Likewise.
20259 (rs6000_output_function_epilogue): Likewise.
20260 (rs6000_handle_altivec_attribute): Likewise.
20261 (rs6000_function_value): Likewise.
20262 (emit_fusion_gpr_load): Likewise.
20263 (emit_fusion_p9_load): Likewise.
20264 (emit_fusion_p9_store): Likewise.
20265 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20266 (fusion_gpr_mem_load): Likewise.
20267 (fusion_addis_mem_combo_load): Likewise.
20268 (fusion_addis_mem_combo_store): Likewise.
20269 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20270 (fusion_gpr_mem_load): Likewise.
20271 (fusion_addis_mem_combo_load): Likewise.
20272 (fusion_addis_mem_combo_store): Likewise.
20273 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20275 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20276 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20277 (rs6000_preferred_simd_mode): Likewise.
20278 (output_vec_const_move): Likewise.
20279 (rs6000_expand_vector_extract): Likewise.
20280 (rs6000_split_vec_extract_var): Likewise.
20281 (reg_offset_addressing_ok_p): Likewise.
20282 (rs6000_legitimate_offset_address_p): Likewise.
20283 (rs6000_legitimize_address): Likewise.
20284 (rs6000_emit_set_const): Likewise.
20285 (rs6000_const_vec): Likewise.
20286 (rs6000_emit_move): Likewise.
20287 (rs6000_darwin64_record_arg_recurse): Likewise.
20288 (swap_selector_for_mode): Likewise.
20289 (paired_init_builtins): Likewise.
20290 (altivec_init_builtins): Likewise.
20291 (rs6000_expand_float128_convert): Likewise.
20292 (emit_load_locked): Likewise.
20293 (emit_store_conditional): Likewise.
20294 (rs6000_output_function_epilogue): Likewise.
20295 (rs6000_handle_altivec_attribute): Likewise.
20296 (rs6000_function_value): Likewise.
20297 (emit_fusion_gpr_load): Likewise.
20298 (emit_fusion_p9_load): Likewise.
20299 (emit_fusion_p9_store): Likewise.
20300 * config/rx/rx.c (rx_gen_move_template): Likewise.
20301 (flags_from_mode): Likewise.
20302 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20303 (s390_slb_comparison): Likewise.
20304 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20305 (s390_vector_mode_supported_p): Likewise.
20306 (s390_cc_modes_compatible): Likewise.
20307 (s390_match_ccmode_set): Likewise.
20308 (s390_canonicalize_comparison): Likewise.
20309 (s390_emit_compare_and_swap): Likewise.
20310 (s390_branch_condition_mask): Likewise.
20311 (s390_rtx_costs): Likewise.
20312 (s390_secondary_reload): Likewise.
20313 (__SECONDARY_RELOAD_CASE): Likewise.
20314 (s390_expand_cs): Likewise.
20315 (s390_preferred_simd_mode): Likewise.
20316 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20317 * config/sh/sh.c (sh_print_operand): Likewise.
20318 (dump_table): Likewise.
20319 (sh_secondary_reload): Likewise.
20320 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20321 * config/sh/sh.md (casesi_worker_1): Likewise.
20322 (casesi_worker_2): Likewise.
20323 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20324 (fcc_comparison_operator): Likewise.
20325 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20326 (emit_soft_tfmode_cvt): Likewise.
20327 (sparc_preferred_simd_mode): Likewise.
20328 (output_cbranch): Likewise.
20329 (sparc_print_operand): Likewise.
20330 (sparc_expand_vec_perm_bmask): Likewise.
20331 (vector_init_bshuffle): Likewise.
20332 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20333 (spu_vector_mode_supported_p): Likewise.
20334 (spu_expand_insv): Likewise.
20335 (spu_emit_branch_or_set): Likewise.
20336 (spu_handle_vector_attribute): Likewise.
20337 (spu_builtin_splats): Likewise.
20338 (spu_builtin_extract): Likewise.
20339 (spu_builtin_promote): Likewise.
20340 (spu_expand_sign_extend): Likewise.
20341 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20342 (tilegx_simd_int): Likewise.
20343 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20344 (tilepro_simd_int): Likewise.
20345 * config/v850/v850.c (const_double_split): Likewise.
20346 (v850_print_operand): Likewise.
20347 (ep_memory_offset): Likewise.
20348 * config/vax/vax.c (vax_rtx_costs): Likewise.
20349 (vax_output_int_move): Likewise.
20350 (vax_output_int_add): Likewise.
20351 (vax_output_int_subtract): Likewise.
20352 * config/visium/predicates.md (visium_branch_operator): Likewise.
20353 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20354 (visium_print_operand_address): Likewise.
20355 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20356 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20357 (xtensa_expand_conditional_branch): Likewise.
20358 (xtensa_copy_incoming_a7): Likewise.
20359 (xtensa_output_literal): Likewise.
20360 * dfp.c (decimal_real_maxval): Likewise.
20361 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20364 Alan Hayward <alan.hayward@arm.com>
20365 David Sherwood <david.sherwood@arm.com>
20367 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20368 (mode_nunits_inline): Likewise.
20369 (mode_inner_inline): Likewise.
20370 (mode_unit_size_inline): Likewise.
20371 (mode_unit_precision_inline): Likewise.
20372 (emit_insn_modes_h): Likewise. Also emit a #define of the
20374 (emit_mode_wider): Add an E_ prefix to mode names.
20375 (emit_mode_complex): Likewise.
20376 (emit_mode_inner): Likewise.
20377 (emit_mode_adjustments): Likewise.
20378 (emit_mode_int_n): Likewise.
20379 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20380 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20381 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20382 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20383 (CRC32_BUILTIN, ENTRY): Likewise.
20384 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20385 (aarch64_pop_regs): Likewise.
20386 (aarch64_process_components): Likewise.
20387 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20388 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20389 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20390 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20391 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20392 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20394 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20395 (pcmpistr): Likewise.
20396 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20397 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20398 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20400 * config/rl78/rl78.c (mduc_regs): Likewise.
20401 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20402 (htm_expand_builtin): Likewise.
20403 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20404 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20405 * config/xtensa/xtensa.c (print_operand): Likewise.
20406 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20407 (NUM_MODE_VECTOR_INT): Likewise.
20408 * genoutput.c (null_operand): Likewise.
20409 (output_operand_data): Likewise.
20410 * genrecog.c (print_parameter_value): Likewise.
20411 * lra.c (debug_operand_data): Likewise.
20413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20415 * dwarf2out.c (scompare_loc_descriptor_wide)
20416 (scompare_loc_descriptor_narrow): New functions, split out from...
20417 (scompare_loc_descriptor): ...here.
20418 * expmed.c (emit_store_flag_int): New function, split out from...
20419 (emit_store_flag): ...here.
20421 2017-08-30 Richard Biener <rguenther@suse.de>
20423 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20424 (dwarf2out_early_finish): Move setting of AT_pubnames from
20425 early debug output to early finish.
20427 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20429 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20430 and -mdata-region to the assembler.
20432 2017-08-30 Richard Biener <rguenther@suse.de>
20434 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20436 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20437 (dwarf2out_early_global_decl): Only generate a DIE for the
20438 abstract origin if it doesn't already exist or is a declaration DIE.
20439 (resolve_addr): Do not add the linkage name twice when
20440 generating a stub DIE for the DW_TAG_GNU_call_site target.
20442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20444 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20445 Use machine_mode rather than int for arg1_mode.
20447 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20450 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20451 that the second argument of the built-in functions to unpack
20452 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20453 switch statement instead a lot of if statements.
20454 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20455 Allow 64-bit values to be in Altivec registers as well as
20456 traditional floating point registers.
20457 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20459 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20461 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20462 MAX_REGS_PER_ADDRESS == 1.
20464 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20466 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20467 * config/i386/i386.c (ix86_option_override_internal): Simplify
20468 setting of opts->x_flag_entry.
20470 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20471 Jakub Jelinek <jakub@redhat.com>
20472 Richard Biener <rguenther@suse.de>
20474 PR tree-optimization/81503
20475 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20476 folded constant fits in the target type; reorder tests for clarity.
20478 2017-08-29 Martin Liska <mliska@suse.cz>
20480 * passes.def: Include pass_lower_switch.
20481 * stmt.c (dump_case_nodes): Remove and move to
20482 tree-switch-conversion.
20483 (case_values_threshold): Likewise.
20484 (expand_switch_as_decision_tree_p): Likewise.
20485 (emit_case_decision_tree): Likewise.
20486 (expand_case): Likewise.
20487 (balance_case_nodes): Likewise.
20488 (node_has_low_bound): Likewise.
20489 (node_has_high_bound): Likewise.
20490 (node_is_bounded): Likewise.
20491 (emit_case_nodes): Likewise.
20492 (struct simple_case_node): New struct.
20493 (add_case_node): Remove.
20494 (emit_case_dispatch_table): Use vector instead of case_list.
20495 (reset_out_edges_aux): Remove.
20496 (compute_cases_per_edge): Likewise.
20497 (expand_case): Build list of simple_case_node.
20498 (expand_sjlj_dispatch_table): Use it.
20499 * tree-switch-conversion.c (struct case_node): Moved from
20500 stmt.c and adjusted.
20501 (emit_case_nodes): Likewise.
20502 (node_has_low_bound): Likewise.
20503 (node_has_high_bound): Likewise.
20504 (node_is_bounded): Likewise.
20505 (case_values_threshold): Likewise.
20506 (reset_out_edges_aux): Likewise.
20507 (compute_cases_per_edge): Likewise.
20508 (add_case_node): Likewise.
20509 (dump_case_nodes): Likewise.
20510 (balance_case_nodes): Likewise.
20511 (expand_switch_as_decision_tree_p): Likewise.
20512 (emit_jump): Likewise.
20513 (emit_case_decision_tree): Likewise.
20514 (try_switch_expansion): Likewise.
20515 (do_jump_if_equal): Likewise.
20516 (emit_cmp_and_jump_insns): Likewise.
20517 (fix_phi_operands_for_edge): New function.
20518 (record_phi_operand_mapping): Likewise.
20519 (class pass_lower_switch): New pass.
20520 (pass_lower_switch::execute): New function.
20521 (make_pass_lower_switch): Likewise.
20522 (conditional_probability):
20523 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20524 * tree-pass.h: Add make_pass_lower_switch.
20526 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20529 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20532 2017-08-29 Richard Biener <rguenther@suse.de>
20534 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20535 we do not add a DW_AT_inline attribute twice.
20536 (gen_subprogram_die): Remove code setting DW_AT_inline on
20537 DECL_ABSTRACT_P nodes.
20539 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20541 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20542 calls to internal functions.
20543 (gimplify_modify_expr): Likewise.
20544 * tree-call-cdce.c (use_internal_fn): Likewise.
20545 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20546 (convert_to_divmod): Set the nothrow flag.
20547 * tree-if-conv.c (predicate_mem_writes): Likewise.
20548 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20549 (vectorizable_call): Likewise.
20550 (vectorizable_store): Likewise.
20551 (vectorizable_load): Likewise.
20552 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20553 (vect_recog_mask_conversion_pattern): Likewise.
20555 2017-08-29 Martin Liska <mliska@suse.cz>
20558 * gcc.c (driver_handle_option): Add new argument.
20559 * opts-common.c (handle_option): Pass
20560 target_option_override_hook.
20561 * opts-global.c (lang_handle_option): Add new option.
20562 (set_default_handlers): Add new argument.
20563 (decode_options): Likewise.
20564 * opts.c (target_handle_option): Likewise.
20565 (common_handle_option): Call target_option_override_hook.
20566 * opts.h (struct cl_option_handler_func): Add hook for
20567 target option override.
20568 (struct cl_option_handlers): Likewise.
20569 (set_default_handlers): Add new argument.
20570 (decode_options): Likewise.
20571 (common_handle_option): Likewise.
20572 (target_handle_option): Likewise.
20573 * toplev.c (toplev::main): Pass targetm.target_option.override
20576 2017-08-29 Richard Biener <rguenther@suse.de>
20577 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20579 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20580 life to the active subtree.
20582 2017-08-28 Jeff Law <law@redhat.com>
20584 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20585 derive_equivalences.
20586 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20587 Code moved into....
20588 (edge_info::derive_equivalences): New private member function
20590 * tree-ssa-dom.c (class edge_info): Changed from a struct
20591 to a class. Add ctor/dtor, methods and data members.
20592 (edge_info::edge_info): Renamed from allocate_edge_info.
20593 Initialize additional members.
20594 (edge_info::~edge_info): New.
20595 (free_dom_edge_info): Delete the edge info.
20596 (record_edge_info): Use new class & associated member functions.
20597 Tighten forms for testing for edge equivalences.
20598 (record_temporary_equivalences): Iterate over the simple
20599 equivalences rather than assuming there's only one per edge.
20600 (cprop_into_successor_phis): Iterate over the simple
20601 equivalences rather than assuming there's only one per edge.
20602 (optimize_stmt): Use operand_equal_p rather than pointer
20603 equality for mini-DSE code.
20605 2017-08-28 Nathan Sidwell <nathan@acm.org>
20607 * gcc.c (execute): Fold SIGPIPE handling into switch
20608 statement. Adjust internal error message.
20610 2017-08-28 Richard Biener <rguenther@suse.de>
20613 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20614 Do nothing for removed DIEs.
20616 2017-08-28 Richard Biener <rguenther@suse.de>
20618 PR tree-optimization/81977
20619 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20622 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20625 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20626 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20627 using targetm.gen_mem_thread_fence.
20629 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20632 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20633 predicate to register_operand. Reorder operands.
20634 (*btr<mode>): Ditto.
20635 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20636 (*btr<mode>_mask): Ditto.
20638 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20640 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20641 * config/rs6000/xmmintrin.h: New file.
20642 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20644 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20647 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20648 parameter and_insn and return it.
20649 (recombine_lvx_pattern): Insert a copy to ensure availability of
20650 the base register of the copied masking operation at the point of
20651 the instruction replacement.
20652 (recombine_stvx_pattern): Likewise.
20654 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20656 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20657 undocumented switches.
20658 (-mpower9-dform-vector): Likewise.
20659 (-mpower9-dform): Likewise.
20660 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20661 comments to delete references to -mpower9-dform* switches.
20662 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20663 Delete reference to -mpower9-dform* switches, test for
20664 -mpower9-vector instead.
20665 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20666 (OTHER_P9_VECTOR_MASKS): Likewise.
20667 (POWERPC_MASKS): Likewise.
20668 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20669 tests against -mpower9-dform* to -mpower9-vector. Delete code
20670 that checked for -mpower9-dform* consistancy with other options.
20671 Add test for -mpower9-misc to enable other power9 switches.
20672 (rs6000_init_hard_regno_mode_ok): Likewise.
20673 (rs6000_option_override_internal): Likewise.
20674 (rs6000_emit_prologue): Likewise.
20675 (rs6000_emit_epilogue): Likewise.
20676 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20677 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20678 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20680 (emit_fusion_p9_store): Likewise.
20681 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20682 resetting these macros if the assembler does not support ISA 3.0
20684 (TARGET_P9_DFORM_VECTOR): Likewise.
20685 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20686 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20688 2017-08-25 Alan Modra <amodra@gmail.com>
20690 PR rtl-optimization/81747
20691 * cse.c (cse_extended_basic_block): Don't attempt to record
20692 equivalences for degenerate conditional branches that branch
20693 to their fall-through.
20695 2017-08-24 Martin Sebor <msebor@redhat.com>
20697 PR middle-end/81908
20698 * gimple-fold.c (size_must_be_zero_p): New function.
20699 (gimple_fold_builtin_memory_op): Call it.
20701 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20703 * config/rs6000/mm_malloc.h: New file.
20705 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20707 PR tree-optimization/81913
20708 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20709 analysis when either IVs in condition can wrap.
20711 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20713 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20714 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20716 2017-08-24 Richard Biener <rguenther@suse.de>
20719 * targhooks.c (default_target_can_inline_p): Properly
20720 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20721 is present and always compare.
20722 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20723 infer -mfpmath=sse from TARGET_SSE_P.
20724 (ix86_can_inline_p): Properly use target_option_default_node when
20725 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20727 2017-08-24 Richard Biener <rguenther@suse.de>
20730 * dwarf2out.c (output_die): Handle flag_generate_offload like
20732 (output_comp_unit): Likewise.
20733 (gen_array_type_die): Likewise.
20734 (dwarf2out_early_finish): Likewise.
20735 (note_variable_value_in_expr): Likewise.
20736 (dwarf2out_finish): Likewise. Adjust assert.
20737 * cgraphunit.c (symbol_table::compile): Move setting of
20738 flag_generate_offload earlier ...
20739 (symbol_table::finalize_compilation_unit): ... here, before
20740 early debug finalization.
20742 2017-08-24 Richard Biener <rguenther@suse.de>
20744 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20745 and ipa-fnsummary.h.
20746 (ix86_can_inline_p): When ix86_fpmath flags do not match
20747 check whether the callee uses FP math at all.
20749 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20751 PR middle-end/81931
20752 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20753 instead of TYPE_PRECISION.
20755 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20756 Alan Hayward <alan.hayward@arm.com>
20757 David Sherwood <david.sherwood@arm.com>
20759 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20761 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20763 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20765 (ix86_frame::stack_realign_allocate): New field.
20766 (struct machine_frame_state): Modify comments.
20767 (machine_frame_state::sp_realigned_fp_end): New field.
20768 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20769 layout calculation.
20770 (sp_valid_at): Add assertion to assure no attempt to access invalid
20771 offset of a realigned stack.
20772 (fp_valid_at): Likewise.
20773 (choose_baseaddr): Modify comments.
20774 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20775 ix86_expand_prologue.
20776 (ix86_expand_prologue): Modify stack realignment and allocation.
20777 (ix86_expand_epilogue): Modify comments.
20778 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20779 avx2_runtime, avx512f, and avx512f_runtime.
20781 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20783 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20784 (mstackrealign): Do not init to -1.
20785 * config/i386/i386.c (ix86_option_override_internal):
20786 Check opts_set, not opts when setting default value of
20787 opts->x_ix86_force_align_arg_pointer.
20789 2017-08-23 Richard Biener <rguenther@suse.de>
20791 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20792 lang_hooks.decl_printable_name.
20793 * print-rtl-function.c (print_rtx_function): Likewise.
20794 * tree-pretty-print.c (dump_function_header): Likewise.
20796 2017-08-23 Richard Biener <rguenther@suse.de>
20799 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20800 -g0 at compile-time.
20802 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20804 PR middle-end/19706
20805 * doc/sourcebuild.texi (Other hardware attributes):
20808 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20810 PR middle-end/19706
20811 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20812 Fix single-use check.
20814 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20816 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20818 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20820 * doc/install.texi: Modify to add more details on running selected
20823 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20825 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20826 is combined with -mabi=ms.
20827 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20830 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20832 PR tree-optimization/81488
20833 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20834 and cached_basis fields.
20835 (MAX_SPREAD): New constant.
20836 (alloc_cand_and_find_basis): Initialize new fields.
20837 (clear_visited): New function.
20838 (create_phi_basis_1): Rename from create_phi_basis, set visited
20839 and cached_basis fields.
20840 (create_phi_basis): New wrapper function.
20841 (phi_add_costs_1): Rename from phi_add_costs, add spread
20842 parameter, set visited field, short-circuit when limits reached.
20843 (phi_add_costs): New wrapper function.
20844 (record_phi_increments_1): Rename from record_phi_increments, set
20846 (record_phi_increments): New wrapper function.
20847 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20848 (phi_incr_cost): New wrapper function.
20849 (all_phi_incrs_profitable_1): Rename from
20850 all_phi_incrs_profitable, set visited field.
20851 (all_phi_incrs_profitable): New wrapper function.
20853 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20854 Alan Hayward <alan.hayward@arm.com>
20855 David Sherwood <david.sherwood@arm.com>
20857 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20858 that takes the outer and inner modes.
20859 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20860 comparison as the canonical test for a paradoxical subreg.
20861 * combine.c (simplify_set): Use paradoxical_subreg_p.
20862 (make_extraction): Likewise.
20863 (force_to_mode): Likewise.
20864 (rtx_equal_for_field_assignment_p): Likewise.
20865 (gen_lowpart_for_combine): Likewise.
20866 (simplify_comparison): Likewise.
20867 * cse.c (equiv_constant): Likewise.
20868 * expmed.c (store_bit_field_1): Likewise.
20869 * final.c (alter_subreg): Likewise.
20870 * fwprop.c (propagate_rtx): Likewise.
20871 (forward_propagate_subreg): Likewise.
20872 * ira-conflicts.c (ira_build_conflicts): Likewise.
20873 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20874 * lra-constraints.c (curr_insn_transform): Likewise.
20875 (split_reg): Likewise.
20876 * lra-eliminations.c (move_plus_up): Likewise.
20877 (lra_eliminate_regs_1): Likewise.
20878 * recog.c (general_operand): Likewise.
20879 * ree.c (combine_reaching_defs): Likewise.
20880 * reload.c (push_reload): Likewise.
20881 (find_reloads): Likewise.
20882 * reload1.c (elimination_effects): Likewise.
20883 (compute_reload_subreg_offset): Likewise.
20884 (choose_reload_regs): Likewise.
20885 * rtlanal.c (subreg_lsb_1): Likewise.
20886 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20887 (simplify_subreg): Likewise.
20888 * var-tracking.c (track_loc_p): Likewise.
20889 * emit-rtl.c (byte_lowpart_offset): Likewise.
20890 (paradoxical_subreg_p): Delete out-of-line definition.
20892 2017-08-22 Jeff Law <law@redhat.com>
20894 PR tree-optimization/81741
20895 PR tree-optimization/71947
20896 * tree-ssa-dom.c: Include tree-inline.h.
20897 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20898 equivalences if one is more expensive to compute than the other.
20899 * tree-ssa-scopedtables.h (class const_or_copies): Make
20900 record_const_or_copy_raw method private.
20901 (class avail_exprs_stack): New method simplify_binary_operation.
20902 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20903 avail_exprs_stack::simplify_binary_operation as needed.
20904 (avail_exprs_stack::simplify_binary_operation): New function.
20906 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20908 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20909 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20910 (DOT_SYMBOLS): Likewise.
20911 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20912 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20913 (RS6000_ABI_NAME): Likewise.
20914 (TARGET_CMODEL): Likewise.
20915 (TOC_SECTION_ASM_OP): Likewise.
20916 (SET_CMODEL): New macro.
20917 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20919 2017-08-22 Richard Biener <rguenther@suse.de>
20921 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20922 to allow for free-lang-data replacements similar to verify_type_variant.
20924 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20925 Alan Hayward <alan.hayward@arm.com>
20926 David Sherwood <david.sherwood@arm.com>
20928 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20929 VOIDmode for the LABEL_REF.
20931 2017-08-22 Richard Biener <rguenther@suse.de>
20933 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20935 2017-08-22 Richard Biener <rguenther@suse.de>
20937 * common.opt (feliminate-dwarf2-dups): Ignore.
20938 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20939 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20940 same_die_p_wrap, compute_section_prefix,
20941 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20942 (comdat_symbol_id, comdat_symbol_number): Likewise.
20943 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20945 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20946 (output_die): Mark unreachable path unreachable.
20947 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20948 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20949 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20950 (dwarf2out_early_finish): Likewise.
20952 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20954 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20956 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20959 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20960 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20961 (avr_attribute_table) <io, io_low, address>: Initialize
20962 .decl_required with true.
20964 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20966 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20967 undocumented debugging options.
20968 (-mvsx-scalar-double): Likewise.
20969 (-mallow-df-permute): Likewise.
20970 (-mvectorize-builtins): Likewise.
20971 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20972 (rs6000_builtin_vectorized_function): Likewise.
20973 (rs6000_builtin_md_vectorized_function): Likewise.
20974 (rs6000_opt_vars): Likewise.
20976 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20979 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20980 (*btrq_imm): Rename from *btrq.
20981 (*btcq_imm): Rename from *btcq.
20982 (btsc): New code attribute.
20983 (*<btsc><mode>): New insn pattern.
20984 (*btr<mode>): Ditto.
20985 (*<btsc><mode>_mask): New insn_and_split pattern.
20986 (*btr<mode>_mask): Ditto.
20988 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20989 Alan Hayward <alan.hayward@arm.com>
20990 David Sherwood <david.sherwood@arm.com>
20992 * function.c (pad_below): Simplify padding calculation.
20994 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20995 Alan Hayward <alan.hayward@arm.com>
20996 David Sherwood <david.sherwood@arm.com>
20998 * target.def (function_prologue): Remove frame size argument.
20999 (function_epilogue): Likewise.
21000 * doc/tm.texi: Regenerate.
21001 * final.c (final_start_function): Update call to function_prologue.
21002 (final_end_function): Update call to function_epilogue.
21003 (default_function_pro_epilogue): Remove frame size argument.
21004 * output.h (default_function_pro_epilogue): Likewise.
21005 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
21006 (arm_output_function_prologue): Likewise.
21007 * config/frv/frv.c (frv_function_prologue): Likewise.
21008 (frv_function_epilogue): Likewise.
21009 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
21010 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
21011 (ia64_output_function_epilogue): Likewise.
21012 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
21013 (m32r_output_function_epilogue): Likewise.
21014 * config/microblaze/microblaze.c (microblaze_function_prologue)
21015 (microblaze_function_epilogue): Likewise.
21016 * config/mips/mips.c (mips_output_function_prologue): Likewise.
21017 (mips_output_function_epilogue): Likewise.
21018 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
21019 (mmix_target_asm_function_epilogue): Likewise.
21020 * config/msp430/msp430.c (msp430_start_function): Likewise.
21021 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
21022 (nds32_asm_function_epilogue): Likewise.
21023 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
21024 * config/pa/pa.c (pa_output_function_prologue): Likewise.
21025 (pa_output_function_epilogue): Likewise.
21026 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
21027 (rs6000_output_function_epilogue): Likewise.
21028 * config/rl78/rl78.c (rl78_start_function): Likewise.
21029 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
21030 (rs6000_output_function_epilogue): Likewise.
21031 * config/rx/rx.c (rx_output_function_prologue): Likewise.
21032 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
21033 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
21034 (sparc_asm_function_epilogue): Likewise.
21036 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21038 * tree.h (type_has_mode_precision_p): New function.
21039 * convert.c (convert_to_integer_1): Use it.
21040 * expr.c (expand_expr_real_2): Likewise.
21041 (expand_expr_real_1): Likewise.
21042 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
21043 * match.pd: Likewise.
21044 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
21045 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
21046 * tree-tailcall.c (process_assignment): Likewise.
21047 * tree-vect-loop.c (vectorizable_reduction): Likewise.
21048 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
21049 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
21050 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
21051 (vectorizable_assignment): Likewise.
21052 (vectorizable_shift): Likewise.
21053 (vectorizable_operation): Likewise.
21054 * tree-vrp.c (register_edge_assert_for_2): Likewise.
21056 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
21058 * match.pd: Add pow (C, x) simplification.
21060 2017-08-21 Richard Biener <rguenther@suse.de>
21062 PR tree-optimization/81900
21063 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
21064 for blocks with abnormal predecessors.
21065 (compute_antic): Do not set visited flag prematurely.
21067 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
21070 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
21072 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21074 * stor-layout.h (vector_type_mode): Move to...
21075 * tree.h (vector_type_mode): ...here.
21076 * stor-layout.c (vector_type_mode): Move to...
21077 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
21079 2017-08-21 Richard Biener <rguenther@suse.de>
21081 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
21082 register_external_die hooks.
21083 (debug_false_tree_charstarstar_uhwistar): Declare.
21084 (debug_nothing_tree_charstar_uhwi): Likewise.
21085 * debug.c (do_nothing_debug_hooks): Adjust.
21086 (debug_false_tree_charstarstar_uhwistar): New do nothing.
21087 (debug_nothing_tree_charstar_uhwi): Likewise.
21088 * dbxout.c (dbx_debug_hooks): Adjust.
21089 (xcoff_debug_hooks): Likewise.
21090 * sdbout.c (sdb_debug_hooks): Likewise.
21091 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21092 * dwarf2out.c (macinfo_label_base): New global.
21093 (dwarf2out_register_external_die): New function for the
21094 register_external_die hook.
21095 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
21096 (dwarf2_debug_hooks): Use them.
21097 (dwarf2_lineno_debug_hooks): Adjust.
21098 (struct die_struct): Add with_offset flag.
21099 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
21100 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
21101 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
21102 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
21103 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
21104 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
21105 defining section names for the early LTO debug variants.
21106 (reset_indirect_string): New helper.
21107 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
21108 (print_dw_val): Add support for offsetted symbol references.
21109 (get_ultimate_context): Split out from is_cxx.
21110 (is_cxx): Use get_ultimate_context.
21111 (is_fortran): Add decl overload.
21112 (compute_comp_unit_symbol): Split out worker from
21113 compute_section_prefix.
21114 (compute_section_prefix): Call compute_comp_unit_symbol and
21115 set comdat_type_p here.
21116 (output_die): Skip DIE symbol output for the LTO added one.
21117 Handle DIE symbol references with offset.
21118 (output_comp_unit): Guard section name mangling properly.
21119 For LTO debug sections emit a symbol at the section beginning
21120 which we use to refer to its DIEs.
21121 (add_abstract_origin_attribute): For DIEs registered via
21122 dwarf2out_register_external_die directly refer to the early
21123 DIE rather than indirectly through the shadow one we created.
21124 Remove obsolete call to dwarf2out_abstract_function for
21125 non-function/block origins.
21126 (gen_array_type_die): When generating early LTO debug do
21127 not emit DW_AT_string_length.
21128 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
21129 late when in LTO. As suggested place a gcc_unreachable for
21130 the DECL_ABSTRACT_P case.
21131 (gen_subprogram_die): Avoid another specification DIE
21132 for early built declarations/definitions for the late LTO case.
21133 (gen_variable_die): Add type references for late duplicated VLA dies
21135 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
21136 we have the abstract instance already.
21137 (process_scope_var): Adjust decl DIE contexts in LTO which
21138 first puts them in limbo.
21139 (gen_decl_die): Do not generate type DIEs late apart from
21140 types for VLAs or for decls we do not yet have a DIE. Do not
21141 call dwarf2out_abstract_function late.
21142 (dwarf2out_early_global_decl): Make sure to create DIEs
21143 for abstract instances of a decl first.
21144 (dwarf2out_late_global_decl): Adjust comment.
21145 (output_macinfo_op): With multiple macro sections use
21146 macinfo_label_base to distinguish labels.
21147 (output_macinfo): Likewise. Update macinfo_label_base.
21148 Pass in the line info label.
21149 (note_variable_value_in_expr): When generating LTO resolve
21150 all variable values here by generating DIEs as needed.
21151 (init_sections_and_labels): Add early LTO debug flag parameter
21152 and generate different sections and names if set. Add generation
21153 counter for the labels so we can have multiple of them.
21154 (reset_dies): Helper to allow DIEs to be output multiple times.
21155 (dwarf2out_finish): When outputting DIEs to the fat part of an
21156 LTO object first reset DIEs.
21157 (dwarf2out_early_finish): Output early DIEs when generating LTO.
21158 (modified_type_die): Check for decl_ultimate_origin being self
21160 (gen_type_die_with_usage): Likewise.
21161 (gen_typedef_die): Allow decl_ultimate_origin being self.
21162 (set_decl_abstract_flags): Remove.
21163 (set_block_abstract_flags): Likewise.
21164 (dwarf2out_abstract_function): Treat the early generated DIEs
21165 as the abstract copy and only add DW_AT_inline and
21166 DW_AT_artificial here and call set_decl_origin_self.
21167 If the DIE has an abstract origin don't do anything.
21168 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
21169 if we have none yet (Go fails to build one, PR78628).
21170 (variably_modified_type_p): Prevent endless recursion for Ada
21171 cyclic pointer types.
21172 * lto-streamer-in.c: Include debug.h.
21173 (dref_queue): New global.
21174 (lto_read_tree_1): Stream in DIE references.
21175 (lto_input_tree): Register DIE references.
21176 (input_function): Stream DECL_DEBUG_ARGS.
21177 * lto-streamer-out.c: Include debug.h.
21178 (lto_write_tree_1): Output DIE references.
21179 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
21180 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
21181 (output_function): Stream DECL_DEBUG_ARGS.
21182 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
21183 Stream DECL_ABSTRACT_ORIGIN.
21184 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
21185 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
21186 DECL_CONTEXT for file-scope decls.
21187 * lto-streamer.h (struct dref_entry): Declare.
21188 (dref_queue): Likewise.
21189 * cfgexpand.c (pass_expand::execute): Do not call the
21190 outlining_inline_function hook here.
21191 * lto-wrapper.c (debug_obj): New global.
21192 (tool_cleanup): Unlink it if required.
21193 (debug_objcopy): New function.
21194 (run_gcc): Handle early debug sections in the IL files by
21195 extracting them to separate files, partially linkin them and
21196 feeding the result back as result to the linker.
21197 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
21198 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
21199 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
21200 sections into a separate segment.
21201 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
21203 (darwin_asm_dwarf_section): Likewise.
21204 (darwin_asm_output_dwarf_offset): Likewise.
21205 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
21207 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21208 Alan Hayward <alan.hayward@arm.com>
21209 David Sherwood <david.sherwood@arm.com>
21211 * read-md.h (md_reader::record_potential_iterator_use): Replace
21212 pointer argument with an rtx and an index.
21213 * read-rtl.c (iterator_group::apply_iterator): Likewise.
21214 (apply_mode_iterator): Likewise.
21215 (apply_code_iterator): Likewise.
21216 (apply_int_iterator): Likewise.
21217 (apply_subst_iterator): Likewise.
21218 (record_iterator_use): Likewise.
21219 (record_attribute_use): Likewise.
21220 (md_reader::record_potential_iterator_use): Likewise. Update calls
21221 to record_iterator_use and apply_iterator.
21222 (iterator_use): Replace ptr with x and index.
21223 (attribute_use): Likewise.
21224 (apply_attribute_uses): Update calls to apply_iterator.
21225 (apply_iterators): Likewise. Update initialization of iterator_use.
21226 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21227 and record_potential_iterator_use.
21228 (rtx_reader::read_rtx_operand): Likewise.
21230 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21231 Alan Hayward <alan.hayward@arm.com>
21232 David Sherwood <david.sherwood@arm.com>
21234 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21237 2017-08-21 Richard Biener <rguenther@suse.de>
21239 PR middle-end/81884
21240 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21241 at struct end conservatively when comparing common bases.
21243 2017-08-21 Richard Biener <rguenther@suse.de>
21245 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21246 (mem_ref_in_stmt): Remove.
21247 (determine_max_movement): Use ref index to get at the reference.
21248 (invariantness_dom_walker::before_dom_children): Deal with
21249 lim data already initialized.
21250 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21252 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21254 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21255 (TARGET_ISA_ROUND): Ditto.
21256 (TARGET_ROUND): Ditto.
21257 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21258 * config/i386/i386.md: Ditto.
21259 * config/i386/sse.md: Ditto.
21260 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21261 with OPTION_MASK_ISA_SSE4_1.
21263 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21266 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21267 __builtin_ia32_lzcnt_u16.
21269 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21272 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21273 (rs6000_set_current_function): Rewrite function to use it.
21275 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21278 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21279 and TYPE_WARN_IF_NOT_ALIGN.
21280 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21281 (handle_warn_if_not_align): New.
21282 (place_union_field): Call handle_warn_if_not_align.
21283 (place_field): Call handle_warn_if_not_align.
21284 Copy TYPE_WARN_IF_NOT_ALIGN.
21285 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21286 (layout_type): Likewise.
21287 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21289 (tree_decl_common): Add warn_if_not_align.
21290 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21291 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21292 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21293 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21294 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21295 * doc/extend.texi: Document warn_if_not_aligned attribute.
21296 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21298 2017-08-17 Martin Liska <mliska@suse.cz>
21301 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21302 (get_data_dependence): Use it as pointer type.
21303 (distribute_loop): Likewise.
21305 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21307 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21308 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21309 (p8_vmrgow_<mode>_direct): New define_insn.
21310 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21311 handle endianness for vmrgew and vmrgow permute patterns.
21313 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21315 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21316 * config/rs6000/rs6000-cpus.def: Remove comment.
21317 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21318 (POWERPC_MASKS): Likewise.
21319 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21320 use of TARGET_VSX_TIMODE.
21321 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21322 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21323 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21324 (rs6000_option_override_internal): Remove dead code.
21325 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21326 (rs6000_legitimize_reload_address): Likewise.
21327 (rs6000_legitimate_address_p): Likewise.
21328 (rs6000_opt_masks): Delete "vsx-timode".
21329 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21330 from function comment.
21331 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21332 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21333 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21335 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21336 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21337 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21338 (VSX_TI): Likewise.
21340 (define_peephole2): Likewise.
21342 2017-08-17 Martin Sebor <msebor@redhat.com>
21345 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21346 past the end of an array.
21347 (test_pp_format): Add test cases.
21349 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21351 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21352 missing ECF_NOTHROW flags.
21354 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21357 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21358 operands residing in integer registers.
21359 (*vsx_le_perm_load_<mode>): Likewise.
21360 (*vsx_le_perm_store_<mode>): Likewise.
21361 (define_peephole2): Add peepholes to optimize the above.
21363 2017-08-17 Marek Polacek <polacek@redhat.com>
21365 PR middle-end/81814
21366 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21367 to mimic what shorten_compare did. Change the return type to bool.
21368 (fold_cond_expr_with_comparison): Update call to
21369 operand_equal_for_comparison_p.
21370 (fold_ternary_loc): Likewise.
21372 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21374 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21376 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21378 2017-08-17 Richard Biener <rguenther@suse.de>
21380 * tree-ssa-structalias.c (solve_graph): When propagating
21381 to successors update the graphs succ edges and avoid duplicate work.
21383 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21386 * config/i386/i386.c (ix86_option_override_internal): Save target
21387 specific options after ix86_stack_protector_guard_reg was changed.
21389 2017-08-17 Richard Biener <rguenther@suse.de>
21391 PR tree-optimization/81827
21392 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21393 (new_var_info): Initialize it conservatively.
21394 (get_call_vi): Mark register vars.
21395 (new_scalar_tmp_constraint_exp): Likewise.
21396 (handle_rhs_call): Likewise.
21397 (handle_const_call): Likewise.
21398 (create_function_info_for): Likewise.
21399 (solve_constraints): Sort varinfos to separate register from
21400 non-register vars to pack points-to solution bitmaps during
21403 2017-08-17 Marek Polacek <polacek@redhat.com>
21405 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21407 2017-08-17 Richard Biener <rguenther@suse.de>
21409 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21410 to true when overflow is undefined and we saturated the result.
21412 2017-08-17 Alan Modra <amodra@gmail.com>
21415 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21416 Don't use store multiple if only one reg needs saving.
21417 (interesting_frame_related_regno): New function.
21418 (rs6000_frame_related): Don't emit frame info for regs that
21420 (rs6000_emit_epilogue): Likewise.
21422 2017-08-16 Nathan Sidwell <nathan@acm.org>
21424 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21425 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21426 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21427 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21428 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21429 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21430 (verify_type): Adjust for TYPE_BINFO move.
21431 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21432 process TYPE_BINFO directly.
21433 (hash_tree): Likewise.
21434 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21436 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21439 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21441 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21443 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21446 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21447 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21448 Add ix86_binary_operator_ok to insn constraint.
21449 (*iordi_1_bts): Ditto.
21450 (*xordi_1_btc): Ditto.
21451 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21452 Update corresponding peephole2 pattern.
21456 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21458 PR tree-optimization/81832
21459 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21460 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21462 2017-08-16 Marek Polacek <polacek@redhat.com>
21465 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21466 regarding size_zero_node.
21468 2017-08-16 Martin Liska <mliska@suse.cz>
21471 * config.gcc: Respect previously set extra_objs in case
21474 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21476 PR tree-optimization/81835
21477 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21478 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21479 not depend on the phi.
21481 2017-08-16 Alan Modra <amodra@gmail.com>
21483 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21486 2017-08-16 Alan Modra <amodra@gmail.com>
21488 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21489 (save_reg_p): ..into this. Update all callers.
21490 (first_reg_to_save): Simplify.
21492 2017-08-16 Alan Modra <amodra@gmail.com>
21494 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21497 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21501 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21502 constant count if that count is less than 32.
21504 2017-08-15 Nathan Sidwell <nathan@acm.org>
21506 * gcc.c (execute): Emit friendlier message if inferior is killed
21507 by an external cause.
21509 2017-08-15 Richard Biener <rguenther@suse.de>
21511 PR tree-optimization/81790
21512 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21513 CONSTRUCTORs from simplifying and VN.
21515 2017-08-14 Martin Sebor <msebor@redhat.com>
21517 * builtin-attrs.def: Add comments.
21519 2017-08-14 Martin Sebor <msebor@redhat.com>
21522 * doc/extend.texi (attribute nonstring): Document new attribute.
21524 2017-08-14 Martin Sebor <msebor@redhat.com>
21527 * tree-diagnostic.c (default_tree_printer): Handle %G.
21528 * gimple-pretty-print.h (percent_G_format): Declare new function.
21529 * gimple-pretty-print.c (percent_G_format): Define.
21530 * tree-pretty-print.c (percent_K_format): Add argument.
21532 2017-08-14 Martin Sebor <msebor@redhat.com>
21534 PR translation/79998
21535 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21536 Remove a stray space.
21538 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21541 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21542 (*iordi_1_bts): Ditto.
21543 (*xordi_1_btc): Ditto.
21545 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21548 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21549 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21551 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21552 quoted strings, and make more translator-friendly.
21553 (darwin_rs6000_override_options): Likewise.
21554 (rs6000_option_override_internal): Likewise.
21555 (rs6000_return_in_memory): Fix overlong line.
21556 (init_cmulative_args): Use quoted strings, and make more
21557 translator-friendly.
21558 (rs6000_pass_by_reference): Fix overlong line.
21559 (def_builtin): Use quoted strings.
21560 (altivec_expand_predicate_builtin): Use quoted strings, and make
21561 more translator-friendly.
21562 (htm_expand_builtin): Use quoted strings.
21563 (cpu_expand_builtin): Use quoted strings, and make more
21564 translator-friendly.
21565 (altivec_expand_builtin): Likewise.
21566 (paired_expand_predicate_builtin): Likewise.
21567 (rs6000_invalid_builtin): Likewise.
21568 (builtin_function_type): Use quoted strings.
21569 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21570 more translator-friendly.
21571 (rs6000_trampoline_init): Likewise.
21572 (rs6000_handle_altivec_attribute): Likewise.
21573 (rs6000_inner_target_options): Use quoted strings.
21574 (rs6000_disable_incompatible_switches): Likewise.
21575 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21576 strings, and make more translator-friendly.
21577 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21579 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21581 PR tree-optimization/81799
21582 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21583 cond_expr to simple gimple operand.
21585 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21587 PR middle-end/46932
21588 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21590 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21594 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21595 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21596 TARGET_GASISR_PROLOGUES.
21597 * config/avr/avr.c (avr_option_override): Same.
21598 (avr_pass_pre_proep::execute): Same.
21600 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21603 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21604 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21606 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21608 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21609 define_insn_and_split. Split to a memory load from 0 in
21610 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21611 using PTR mode iterator.
21612 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21613 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21614 (*add_tp_<mode>): Redefine as define_insn_and_split.
21615 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21616 address space. Merge with *add_tp_x32 using PTR mode iterator.
21617 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21618 Split to an add with a memory load from 0 in
21619 DEFAULT_TLS_SEG_REG address space.
21621 2017-08-12 Andrew Pinski <apinski@cavium.com>
21623 * config/aarch64/aarch64-option-extensions.def (rdma):
21624 Fix feature string to what Linux prints out in /proc/cpuinfo.
21626 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21629 * dwarf2out.c (modified_type_die): For C typedef types that have
21630 an ultimate origin, process the ultimate origin instead of the
21632 (gen_typedef_die): Assert that input DECLs have no ultimate
21634 (gen_type_die_with_usage): For typedef variants that have an
21635 ultimate origin, just call gen_decl_die on the original DECL.
21636 (process_scope_var): Avoid creating DIEs for local typedefs and
21637 concrete static variables.
21639 2017-08-12 Alan Modra <amodra@gmail.com>
21643 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21644 match gnu-user.h startfile.
21645 (ENDFILE_LINUX_SPEC): Similarly.
21647 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21650 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21652 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21654 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21655 * config/aarch64/aarch64.md (mov<mode>): Change.
21656 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21657 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21658 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21660 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21662 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21663 for storage order barriers.
21665 2017-08-11 Martin Liska <mliska@suse.cz>
21668 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21669 variables of void type.
21671 2017-08-11 Martin Liska <mliska@suse.cz>
21673 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21674 TARGET_SUPPORTS_ALIASES.
21675 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21676 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21677 (optimize_weakref): Likewise.
21678 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21679 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21680 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21682 2017-08-11 Martin Liska <mliska@suse.cz>
21685 * config/i386/i386.c (make_resolver_func): Do complete
21686 refactoring of the function.
21688 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21691 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21692 * config/i386/i386.c (ix86_stack_protect_guard): Use
21693 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21694 * doc/invoke.texi (x86 Options): Document
21695 -mstack-protector-guard-symbol= option.
21697 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21699 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21700 * config/i386/i386.c (ix86_split_stack_guard): New function.
21701 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21702 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21703 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21704 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21705 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21706 (split_stack_space_check): Call ix86_split_stack_guard.
21708 2017-08-10 Martin Sebor <msebor@redhat.com>
21710 * print-tree.c (print_node): Print location using the established
21712 Replace spaces with colons.
21713 (debug_raw, debug): Ditto.
21715 2017-08-10 Martin Sebor <msebor@redhat.com>
21718 * pretty-print.c (pp_format): Correct the handling of %s precision.
21720 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21723 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21725 (ix86_finalize_stack_frame_flags): This. Also clear
21726 frame_pointer_needed if -fno-omit-frame-pointer is used without
21728 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21729 with ix86_finalize_stack_frame_flags.
21730 (ix86_expand_epilogue): Likewise.
21731 (ix86_expand_split_stack_prologue): Likewise.
21732 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21734 2017-08-10 Martin Liska <mliska@suse.cz>
21737 * c-attribs.c (handle_target_attribute):
21738 Report warning for an empty string argument of target attribute.
21740 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21743 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21745 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21746 or DECL_NONLOCAL labels.
21747 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21748 or DECL_NONLOCAL labels here.
21750 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21752 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21753 to indicate when early gimple folding has been disabled.
21754 (rs6000_gimple_fold_builtin): Add debug content.
21755 (rs6000_invalid_builtin): Fix whitespace.
21756 (rs6000_expand_builtin): Fix whitespace.
21757 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21759 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21762 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21763 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21765 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21767 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21768 (qdf24xx): Likewise.
21769 * config/aarch64/aarch64.md: Include falkor.md.
21770 * config/aarch64/falkor.md: New.
21772 2017-08-09 Marek Polacek <polacek@redhat.com>
21775 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21776 * diagnostic.c (emit_diagnostic): Add a comment.
21777 (emit_diagnostic_valist): New function.
21779 2017-08-09 Marek Polacek <polacek@redhat.com>
21782 * input.c (make_location): New overload.
21783 * input.h (make_location): Declare.
21785 2017-08-08 Alan Modra <amodra@gmail.com>
21786 H.J. Lu <hongjiu.lu@intel.com>
21789 * gcc.c (NO_PIE_SPEC): Delete.
21790 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21792 (LINK_PIE_SPEC): ..to here.
21793 (LINK_COMMAND_SPEC): Support -no-pie.
21794 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21795 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21797 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21798 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21799 (ENDFILE_CRTEND_SPEC): Similarly.
21801 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21804 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21805 (mstack-protector-guard-offset=): Ditto.
21806 * config/i386/i386.c (ix86_option_override): Handle
21807 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21809 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21810 ix86_stack_protect_guard_offset variables.
21811 (TARGET_STACK_PROTECT_GUARD): Always define.
21812 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21813 and -mstack-protector-guard-offset= options.
21815 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21817 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21818 boundary case for the last candidate.
21820 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21822 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21823 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21825 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21827 PR middle-end/19706
21828 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21829 * config/aarch64/aarch64-builtins.c
21830 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21831 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21832 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21834 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21835 Andrew Pinski <pinskia@gmail.com>
21837 PR middle-end/19706
21838 * internal-fn.def (XORSIGN): New.
21839 * optabs.def (xorsign_optab): New.
21840 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21841 (convert_expand_mult_copysign): New.
21842 (pass_optimize_widening_mul::execute): Call
21843 convert_expand_mult_copysign.
21845 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21847 PR tree-optimization/81354
21848 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21849 Insert on edges rather than explicitly creating landing pads.
21850 (analyze_candidates_and_replace): Commit edge inserts.
21852 2017-08-08 Richard Biener <rguenther@suse.de>
21854 PR middle-end/81719
21855 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21856 (expand_simple_operations): Also look through ADDR_EXPRs with
21857 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21859 2017-08-08 Richard Biener <rguenther@suse.de>
21861 PR tree-optimization/81723
21862 * tree-vect-slp.c (struct bst_traits): New hash traits.
21863 (bst_fail): New global.
21864 (vect_build_slp_tree_2): New worker, split out from ...
21865 (vect_build_slp_tree): ... this now wrapping it with using
21866 bst_fail set to cache SLP tree build fails. Properly handle
21868 (vect_analyze_slp_instance): Allocate and free bst_fail.
21870 2017-08-08 Martin Liska <mliska@suse.cz>
21873 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21874 LABEL_DECLs that can be from a different function.
21876 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21878 PR tree-optimization/81744
21879 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21880 loop's number of iterations.
21882 2017-08-08 Martin Liska <mliska@suse.cz>
21884 * asan.c: Include header files.
21885 * attribs.c (build_decl_attribute_variant): New function moved
21887 (build_type_attribute_qual_variant): Likewise.
21888 (cmp_attrib_identifiers): Likewise.
21889 (simple_cst_list_equal): Likewise.
21890 (omp_declare_simd_clauses_equal): Likewise.
21891 (attribute_value_equal): Likewise.
21892 (comp_type_attributes): Likewise.
21893 (build_type_attribute_variant): Likewise.
21894 (lookup_ident_attribute): Likewise.
21895 (remove_attribute): Likewise.
21896 (merge_attributes): Likewise.
21897 (merge_type_attributes): Likewise.
21898 (merge_decl_attributes): Likewise.
21899 (merge_dllimport_decl_attributes): Likewise.
21900 (handle_dll_attribute): Likewise.
21901 (attribute_list_equal): Likewise.
21902 (attribute_list_contained): Likewise.
21903 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21904 (lookup_attribute_by_prefix): Likewise.
21905 * bb-reorder.c: Include header files.
21906 * builtins.c: Likewise.
21907 * calls.c: Likewise.
21908 * cfgexpand.c: Likewise.
21909 * cgraph.c: Likewise.
21910 * cgraphunit.c: Likewise.
21911 * convert.c: Likewise.
21912 * dwarf2out.c: Likewise.
21913 * final.c: Likewise.
21914 * fold-const.c: Likewise.
21915 * function.c: Likewise.
21916 * gimple-expr.c: Likewise.
21917 * gimple-fold.c: Likewise.
21918 * gimple-pretty-print.c: Likewise.
21919 * gimple.c: Likewise.
21920 * gimplify.c: Likewise.
21921 * hsa-common.c: Likewise.
21922 * hsa-gen.c: Likewise.
21923 * internal-fn.c: Likewise.
21924 * ipa-chkp.c: Likewise.
21925 * ipa-cp.c: Likewise.
21926 * ipa-devirt.c: Likewise.
21927 * ipa-fnsummary.c: Likewise.
21928 * ipa-inline.c: Likewise.
21929 * ipa-visibility.c: Likewise.
21931 * lto-cgraph.c: Likewise.
21932 * omp-expand.c: Likewise.
21933 * omp-general.c: Likewise.
21934 * omp-low.c: Likewise.
21935 * omp-offload.c: Likewise.
21936 * omp-simd-clone.c: Likewise.
21937 * opts-global.c: Likewise.
21938 * passes.c: Likewise.
21939 * predict.c: Likewise.
21940 * sancov.c: Likewise.
21941 * sanopt.c: Likewise.
21942 * symtab.c: Likewise.
21943 * toplev.c: Likewise.
21944 * trans-mem.c: Likewise.
21945 * tree-chkp.c: Likewise.
21946 * tree-eh.c: Likewise.
21947 * tree-into-ssa.c: Likewise.
21948 * tree-object-size.c: Likewise.
21949 * tree-parloops.c: Likewise.
21950 * tree-profile.c: Likewise.
21951 * tree-ssa-ccp.c: Likewise.
21952 * tree-ssa-live.c: Likewise.
21953 * tree-ssa-loop.c: Likewise.
21954 * tree-ssa-sccvn.c: Likewise.
21955 * tree-ssa-structalias.c: Likewise.
21956 * tree-ssa.c: Likewise.
21957 * tree-streamer-in.c: Likewise.
21958 * tree-vectorizer.c: Likewise.
21959 * tree-vrp.c: Likewise.
21960 * tsan.c: Likewise.
21961 * ubsan.c: Likewise.
21962 * varasm.c: Likewise.
21963 * varpool.c: Likewise.
21964 * tree.c: Remove functions moved to attribs.[ch].
21965 * tree.h: Likewise.
21966 * config/aarch64/aarch64.c: Add attrs.h header file.
21967 * config/alpha/alpha.c: Likewise.
21968 * config/arc/arc.c: Likewise.
21969 * config/arm/arm.c: Likewise.
21970 * config/avr/avr.c: Likewise.
21971 * config/bfin/bfin.c: Likewise.
21972 * config/c6x/c6x.c: Likewise.
21973 * config/cr16/cr16.c: Likewise.
21974 * config/cris/cris.c: Likewise.
21975 * config/darwin.c: Likewise.
21976 * config/epiphany/epiphany.c: Likewise.
21977 * config/fr30/fr30.c: Likewise.
21978 * config/frv/frv.c: Likewise.
21979 * config/ft32/ft32.c: Likewise.
21980 * config/h8300/h8300.c: Likewise.
21981 * config/i386/winnt.c: Likewise.
21982 * config/ia64/ia64.c: Likewise.
21983 * config/iq2000/iq2000.c: Likewise.
21984 * config/lm32/lm32.c: Likewise.
21985 * config/m32c/m32c.c: Likewise.
21986 * config/m32r/m32r.c: Likewise.
21987 * config/m68k/m68k.c: Likewise.
21988 * config/mcore/mcore.c: Likewise.
21989 * config/microblaze/microblaze.c: Likewise.
21990 * config/mips/mips.c: Likewise.
21991 * config/mmix/mmix.c: Likewise.
21992 * config/mn10300/mn10300.c: Likewise.
21993 * config/moxie/moxie.c: Likewise.
21994 * config/msp430/msp430.c: Likewise.
21995 * config/nds32/nds32-isr.c: Likewise.
21996 * config/nds32/nds32.c: Likewise.
21997 * config/nios2/nios2.c: Likewise.
21998 * config/nvptx/nvptx.c: Likewise.
21999 * config/pa/pa.c: Likewise.
22000 * config/pdp11/pdp11.c: Likewise.
22001 * config/powerpcspe/powerpcspe.c: Likewise.
22002 * config/riscv/riscv.c: Likewise.
22003 * config/rl78/rl78.c: Likewise.
22004 * config/rx/rx.c: Likewise.
22005 * config/s390/s390.c: Likewise.
22006 * config/sh/sh.c: Likewise.
22007 * config/sol2.c: Likewise.
22008 * config/sparc/sparc.c: Likewise.
22009 * config/spu/spu.c: Likewise.
22010 * config/stormy16/stormy16.c: Likewise.
22011 * config/tilegx/tilegx.c: Likewise.
22012 * config/tilepro/tilepro.c: Likewise.
22013 * config/v850/v850.c: Likewise.
22014 * config/vax/vax.c: Likewise.
22015 * config/visium/visium.c: Likewise.
22016 * config/xtensa/xtensa.c: Likewise.
22018 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22021 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
22022 constraints since the -mupper-regs-* switches have been
22024 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
22025 into a vector from a double word element that was extracted from
22026 another vector, and eliminate extra XXPERMDI instructions.
22027 (vsx_concat_<mode>_2): Likewise.
22028 (vsx_concat_<mode>_3): Likewise.
22029 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
22030 concat to allow optimizing inserts from previous extracts.
22032 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
22034 * config/i386/i386.c (ix86_stack_protect_guard): Generate
22035 memory reference to a SSP offset in TLS address space.
22036 (ix86_print_operand) <case '@'>: Remove.
22037 (ix86_print_operand_punct_valid_p): Remove '@' code.
22038 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
22039 UNSPEC_SP_TLS_TEST.
22040 (stack_tls_protect_set_<mode>): Remove.
22041 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
22042 (stack_tls_protect_test_<mode>): Remove.
22043 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
22045 2017-08-07 Olivier Hainque <hainque@adacore.com>
22048 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
22050 2017-08-07 Douglas Rupp <rupp@adacore.com>
22052 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
22053 variable was referenced as multidir in command.
22055 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22058 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
22061 2017-08-07 Martin Liska <mliska@suse.cz>
22063 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
22064 * config/rl78/rl78.c: Add include of attribs.h.
22065 * config/sh/sh.c: Likewise.
22066 * config/v850/v850.c: Likewise.
22068 2017-08-07 Tom de Vries <tom@codesourcery.com>
22070 PR middle-end/78266
22071 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
22073 2017-08-07 Martin Liska <mliska@suse.cz>
22075 * config/mips/mips.c: Include attribs.h.
22077 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
22080 * doc/invoke.texi: Document change in behvaior for -Ofast for
22083 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
22085 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
22087 (aarch64_pop_regs): Likewise.
22088 (aarch64_gen_load_pair): Likewise.
22089 (aarch64_save_callee_saves): Likewise.
22090 (aarch64_restore_callee_saves): Likewise.
22092 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22094 * config/i386/i386.c: Revert the last change.
22096 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22099 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
22101 (ix86_finalize_stack_frame_flags): This. Also clear
22102 frame_pointer_needed if -fno-omit-frame-pointer is used without
22104 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
22105 with ix86_finalize_stack_frame_flags.
22106 (ix86_expand_epilogue): Likewise.
22107 (ix86_expand_split_stack_prologue): Likewise.
22109 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22112 * config/i386/i386.c (get_builtin_code_for_version): Set priority
22113 to P_AES for Westmere.
22115 2017-08-07 Jonathan Yong <10walls@gmail.com>
22117 * config/i386/mingw.opt (fset-stack-executable): Removed.
22118 * config/i386/cygming.opt (fset-stack-executable): Moved
22120 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
22122 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
22124 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
22126 2017-08-07 Marek Polacek <polacek@redhat.com>
22128 PR middle-end/81737
22129 * fold-const.c (fold_indirect_ref_1): Check type_domain.
22131 2017-08-07 Martin Liska <mliska@suse.cz>
22133 * attribs.h (canonicalize_attr_name): New function.
22134 (cmp_attribs): Move from c-format.c and adjusted.
22135 (is_attribute_p): Moved from tree.h.
22136 * tree-inline.c: Add new includes.
22137 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
22138 (private_is_attribute_p): Remove.
22139 (private_lookup_attribute): Likewise.
22140 (private_lookup_attribute_by_prefix): Simplify.
22141 (remove_attribute): Use is_attribute_p.
22142 * tree.h: Remove removed declarations.
22144 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22146 PR middle-end/81698
22147 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
22148 instead of computing it in the function. Formatting fix.
22149 (expand_case): Don't rely on default_edge being the first edge,
22150 clear it if removing it, pass default_edge to
22151 emit_case_dispatch_table.
22152 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
22155 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
22157 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
22158 insn in the function, emit NOP after the insn.
22160 2017-08-06 Tom de Vries <tom@codesourcery.com>
22162 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
22165 2017-08-06 Tom de Vries <tom@codesourcery.com>
22167 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
22170 2017-08-04 Yury Gribov <tetra2005@gmail.com>
22172 PR tree-optimization/57371
22173 * match.pd: New pattern.
22175 2017-08-04 Marek Polacek <polacek@redhat.com>
22177 PR middle-end/81695
22178 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
22179 perform the computation in offset_int.
22181 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22183 PR tree-optimization/81136
22184 * tree-vectorizer.h: Include tree-hash-traits.h.
22185 (vec_base_alignments): New typedef.
22186 (vec_info): Add a base_alignments field.
22187 (vect_record_base_alignments): Declare.
22188 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
22190 (DR_IS_CONDITIONAL_IN_STMT): New macro.
22191 (create_data_ref): Add an is_conditional_in_stmt argument.
22192 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
22193 the is_conditional_in_stmt field.
22194 (data_ref_loc): Add an is_conditional_in_stmt field.
22195 (get_references_in_stmt): Set the is_conditional_in_stmt field.
22196 (find_data_references_in_stmt): Update call to create_data_ref.
22197 (graphite_find_data_references_in_stmt): Likewise.
22198 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
22199 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
22200 (vect_record_base_alignment): New function.
22201 (vect_record_base_alignments): Likewise.
22202 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
22203 for nested statements even if we fail to compute a misalignment.
22204 Use pooled base alignments for unconditional references.
22205 (vect_find_same_alignment_drs): Compare base addresses instead
22207 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
22208 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
22210 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22212 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
22213 Add an explicit name for the enum. Use auto_vec for slp_instances
22214 and grouped_stores.
22215 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
22217 (_bb_vec_info): Add a constructor and destructor.
22218 (vinfo_for_stmt): Return NULL for uids of -1 as well.
22219 (destroy_loop_vec_info): Delete.
22220 (vect_destroy_datarefs): Likewise.
22221 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
22222 (vec_info::vec_info): New function.
22223 (vec_info::~vec_info): Likewise.
22224 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22225 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22226 destroy_loop_vec_info.
22227 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22228 (_loop_vec_info::_loop_vec_info): ...this.
22229 (destroy_loop_vec_info): Replace with...
22230 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22231 the stmt_vec_infos. Leave handling of vec_info information to its
22232 destructor. Remove explicit vector releases.
22233 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22234 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22235 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22236 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22237 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22238 (destroy_bb_vec_info): Replace with...
22239 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22240 information to its destructor.
22241 (vect_slp_analyze_bb_1): Use new and delete instead of
22242 new_bb_vec_info and destroy_bb_vec_info.
22243 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22246 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22248 * tree-data-ref.h (subscript): Add access_fn field.
22249 (data_dependence_relation): Add could_be_independent_p.
22250 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22251 (same_access_functions): Move to tree-data-ref.c.
22252 * tree-data-ref.c (ref_contains_union_access_p): New function.
22253 (access_fn_component_p): Likewise.
22254 (access_fn_components_comparable_p): Likewise.
22255 (dr_analyze_indices): Add a reference to access_fn_component_p.
22256 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22258 (constant_access_functions): Likewise.
22259 (add_other_self_distances): Likewise.
22260 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22261 (initialize_data_dependence_relation): Use XCNEW and remove
22262 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22263 of access functions that have the same type. Allow the
22264 subsequence to end with different bases in some circumstances.
22265 Record the chosen access functions in SUB_ACCESS_FN.
22266 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22267 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22268 (subscript_dependence_tester_1): Likewise dra and drb.
22269 (build_classic_dist_vector): Update calls accordingly.
22270 (subscript_dependence_tester): Likewise.
22271 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22272 DDR_COULD_BE_INDEPENDENT_P.
22273 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22274 comp_alias_ddrs instead of may_alias_ddrs.
22275 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22277 (vect_analyze_data_ref_dependence): Use it if
22278 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22279 distance vectors if that fails.
22280 (dependence_distance_ge_vf): New function.
22281 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22282 LOOP_VINFO_MAY_ALIAS_DDRS.
22284 2017-08-04 Richard Biener <rguenther@suse.de>
22286 PR middle-end/81705
22287 * fold-const.c (fold_binary_loc): Properly restrict
22288 minus_var0 && minus_var1 case when associating undefined overflow
22291 2017-08-04 Tom de Vries <tom@codesourcery.com>
22293 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22295 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22297 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22298 Don't start diagnostic messages with a capital letter.
22299 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22301 (rs6000_invalid_builtin): Likewise.
22302 (rs6000_trampoline_init): Likewise.
22304 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22307 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22308 after setting changeable df flags.
22310 2017-08-03 Richard Biener <rguenther@suse.de>
22312 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22313 up if the use is in USE - X.
22315 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22317 * toplev.c (dumpfile.h): New include.
22318 (internal_error_reentered): New static function. Use it...
22319 (internal_error_function): ...here to handle reentered internal_error.
22321 2017-08-03 Richard Biener <rguenther@suse.de>
22323 PR middle-end/81148
22324 * fold-const.c (split_tree): Add minus_var and minus_con
22325 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22326 here but always use minus_*.
22327 (associate_trees): Assert we never associate with MINUS_EXPR
22328 and NULL first operand. Do not recurse for PLUS_EXPR operands
22329 when associating as MINUS_EXPR either.
22330 (fold_binary_loc): Track minus_var and minus_con.
22332 2017-08-03 Tom de Vries <tom@codesourcery.com>
22335 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22336 ACCEL_COMPILER, apply finish_options on
22337 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22339 2017-08-03 Tom de Vries <tom@codesourcery.com>
22342 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22343 function_entry_patch_area_size > 0.
22345 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22348 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22349 instead of 10??LU, perform unit multiplication in wide_int,
22350 don't change alloc_object_size_limit if the limit is larger
22353 PR tree-optimization/81655
22354 PR tree-optimization/81588
22355 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22356 the case when ranges[i].low and high are 1 for unsigned type with
22359 PR middle-end/81052
22360 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22361 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22363 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22365 * tree-vrp.h: Add include guard.
22367 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22370 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22371 (ud2): New insn pattern.
22372 * config/i386/i386.c (ix86_expand_epilogue):
22373 For naked functions, generate ud2 instead of trap insn.
22375 2017-08-02 Marek Polacek <polacek@redhat.com>
22378 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22380 2017-08-02 Tom de Vries <tom@codesourcery.com>
22381 Cesar Philippidis <cesar@codesourcery.com>
22383 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22384 Add missing edge probabilities.
22386 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22388 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22389 Correct endianness.
22391 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22393 PR middle-end/79499
22394 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22395 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22396 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22397 prologue_seq sequences - if any.
22399 2017-08-02 Richard Biener <rguenther@suse.de>
22401 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22402 via vectors if supported, integer extracts via punning if supported
22403 or otherwise vector extracts.
22405 2017-08-02 Richard Biener <rguenther@suse.de>
22407 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22409 (bitmap_insert_into_set): ... this.
22411 2017-08-02 Richard Biener <rguenther@suse.de>
22413 PR tree-optimization/81633
22415 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22417 PR tree-optimization/71752
22418 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22420 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22422 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22423 (machine_function::call_ms2sysv_pad_out): Remove field.
22424 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22425 (ix86_compute_frame_layout): Likewise.
22427 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22430 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22431 attribute with interrupt attribute.
22433 2017-08-01 Andrew Pinski <apinski@cavium.com>
22435 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22436 BIT_INSERT_EXPR's operand 1
22437 to see if the types precision matches.
22439 2017-08-01 Martin Liska <mliska@suse.cz>
22441 PR middle-end/70140
22442 * builtins.c (expand_builtin_memcpy_args): Remove.
22443 (expand_builtin_memcpy): Call newly added function
22444 expand_builtin_memory_copy_args.
22445 (expand_builtin_memcpy_with_bounds): Likewise.
22446 (expand_builtin_mempcpy): Remove last argument.
22447 (expand_builtin_mempcpy_with_bounds): Likewise.
22448 (expand_builtin_memory_copy_args): New function created from
22449 expand_builtin_mempcpy_args with small modifications.
22450 (expand_builtin_mempcpy_args): Remove.
22451 (expand_builtin_stpcpy): Remove unused argument.
22452 (expand_builtin): Likewise.
22453 (expand_builtin_with_bounds): Likewise.
22455 2017-08-01 Martin Liska <mliska@suse.cz>
22458 Make mempcpy more optimal (PR middle-end/70140).
22460 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22463 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22464 __builtin_vec_cmpne verify both arguments are compatible vectors
22465 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22466 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22467 move computation of aligned to after checking the argument types.
22471 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22472 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22475 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22476 Doug Rupp <rupp@adacore.com>
22477 Olivier Hainque <hainque@adacore.com>
22479 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22480 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22482 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22483 for TARGET_OS_CPP_BUILTIN.
22484 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22485 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22487 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22488 passing required abi options to the assembler for EABI configurations.
22489 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22490 of .text.hot and .text.unlikely sections for kernel modules when
22491 using ARM style exceptions.
22492 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22493 options. Add EXTRA_CC1_SPEC.
22494 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22496 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22498 (ARM_TARGET2_DWARF_FORMAT): Define.
22499 * config/arm/t-vxworks: Adjust multilib control to removal of the
22500 Diab command line options.
22502 2017-08-01 Martin Liska <mliska@suse.cz>
22504 PR gcov-profile/81561
22505 * gcov.c (unblock): Make unblocking safe as we need to preserve
22506 index correspondence of blocks and block_lists.
22508 2017-08-01 Richard Biener <rguenther@suse.de>
22510 PR tree-optimization/81181
22511 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22512 (compute_antic): ... end of iteration here.
22514 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22516 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22517 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22518 (ftree-slp-vectorize): Likewise.
22519 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22520 can no longer be set independent of flag_tree_loop_vectorize.
22521 * omp-general.c (emp_max_vf): Likewise.
22522 * opts.c (enable_fdo_optimizations): Remove references to
22523 flag_tree_vectorize, these are now implicit.
22524 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22525 and leave it for the options machinery.
22527 2017-08-01 Martin Liska <mliska@suse.cz>
22529 PR middle-end/70140
22530 * builtins.c (expand_builtin_memcpy_args): Remove.
22531 (expand_builtin_memcpy): Call newly added function
22532 expand_builtin_memory_copy_args.
22533 (expand_builtin_memcpy_with_bounds): Likewise.
22534 (expand_builtin_mempcpy): Remove last argument.
22535 (expand_builtin_mempcpy_with_bounds): Likewise.
22536 (expand_builtin_memory_copy_args): New function created from
22537 expand_builtin_mempcpy_args with small modifications.
22538 (expand_builtin_mempcpy_args): Remove.
22539 (expand_builtin_stpcpy): Remove unused argument.
22540 (expand_builtin): Likewise.
22541 (expand_builtin_with_bounds): Likewise.
22543 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22546 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22547 print "ds:" only for immediates in generic address space.
22549 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22552 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22553 (ix86_function_ok_for_sibcall): Return false for naked functions.
22555 2017-08-01 Richard Biener <rguenther@suse.de>
22557 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22558 (compute_antic): Seed worklist with exit block predecessors.
22559 * cfganal.c (dfs_find_deadend): For a cycle return the source
22560 of the edge closing it.
22562 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22564 * config/aarch64/aarch64.c
22565 (aarch64_can_const_movi_rtx_p): Move 0 check.
22567 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22569 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22570 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22572 * match.pd: Ditto in address comparison pattern.
22574 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22576 PR tree-optimization/81627
22577 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22578 closed ssa form for store-store chain.
22580 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22582 PR tree-optimization/81620
22583 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22584 for store-store chain.
22586 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22588 PR tree-optimization/81588
22589 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22590 ranges[i].in_p, invert comparison code ccode. For >/>=,
22591 swap rhs1 and rhs2 and comparison code unconditionally,
22592 for </<= don't do that. Don't swap rhs1/rhs2 again if
22593 ranges[i].in_p, instead invert comparison code ccode if
22594 opcode or oe->rank is BIT_IOR_EXPR.
22597 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22598 a direct optab to conversion optab.
22599 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22600 with GET_MODE_INNER as last argument instead of optab_handler.
22601 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22602 vector extraction if possible and optab is available.
22603 * expr.c (store_constructor): Use convert_optab_handler instead
22604 of optab_handler. Use vector initialization from smaller
22605 vectors if possible and optab is available.
22606 * tree-vect-stmts.c (vectorizable_load): Likewise.
22607 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22608 now have two modes.
22609 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22610 of vec_init from half-sized vectors with the same element mode.
22611 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22612 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22613 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22614 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22615 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22616 after mode in gen_vec_extract* calls.
22617 (vec_extract<mode>): Renamed to ...
22618 (vec_extract<mode><ssescalarmodelower>): ... this.
22619 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22620 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22621 element mode after mode in gen_vec_init* calls.
22622 (VEC_INIT_HALF_MODE): New mode iterator.
22623 (vec_init<mode>): Renamed to ...
22624 (vec_init<mode><ssescalarmodelower>): ... this.
22625 (vec_init<mode><ssehalfvecmodelower>): New expander.
22626 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22627 (vec_extractv2sfsf): ... this.
22628 (vec_initv2sf): Renamed to ...
22629 (vec_initv2sfsf): ... this.
22630 (vec_extractv2si): Renamed to ...
22631 (vec_extractv2sisi): ... this.
22632 (vec_initv2si): Renamed to ...
22633 (vec_initv2sisi): ... this.
22634 (vec_extractv4hi): Renamed to ...
22635 (vec_extractv4hihi): ... this.
22636 (vec_initv4hi): Renamed to ...
22637 (vec_initv4hihi): ... this.
22638 (vec_extractv8qi): Renamed to ...
22639 (vec_extractv8qiqi): ... this.
22640 (vec_initv8qi): Renamed to ...
22641 (vec_initv8qiqi): ... this.
22642 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22643 (vec_init<mode>): Renamed to ...
22644 (vec_init<mode><VEC_base_l>): ... this.
22645 (vec_extract<mode>): Renamed to ...
22646 (vec_extract<mode><VEC_base_l>): ... this.
22647 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22648 (vec_initv2sfsf): ... this.
22649 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22650 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22651 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22652 element mode after mode in gen_vec_init* calls.
22653 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22654 (vec_init<mode><Vel>): ... this.
22655 (vec_extract<mode>): Renamed to ...
22656 (vec_extract<mode><Vel>): ... this.
22657 * config/aarch64/iterators.md (Vel): New mode attribute.
22658 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22659 Add element mode after mode in gen_vec_extract* calls.
22660 * config/s390/vector.md (non_vec_l): New mode attribute.
22661 (vec_extract<mode>): Renamed to ...
22662 (vec_extract<mode><non_vec_l>): ... this.
22663 (vec_init<mode>): Renamed to ...
22664 (vec_init<mode><non_vec_l>): ... this.
22665 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22666 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22668 * config/arm/iterators.md (V_elem_l): New mode attribute.
22669 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22670 (vec_extract<mode><V_elem_l>): ... this.
22671 (vec_extractv2di): Renamed to ...
22672 (vec_extractv2didi): ... this.
22673 (vec_init<mode>): Renamed to ...
22674 (vec_init<mode><V_elem_l>): ... this.
22675 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22676 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22677 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22678 Add element mode after gen_vec_extract* calls.
22679 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22680 (vec_init<mode><unitmode>): ... this.
22681 (vec_extract<mode>): Renamed to ...
22682 (vec_extract<mode><unitmode>): ... this.
22683 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22684 (vec_init<mode><unitmode>): ... this.
22685 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22686 (vec_initv2sfsf): ... this.
22687 (vec_extractv2sf): Renamed to ...
22688 (vec_extractv2sfsf): ... this.
22689 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22690 Add element mode after gen_vec_extract* calls.
22691 * config/mips/mips.md (unitmode): New mode iterator.
22692 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22693 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22694 * config/spu/spu.md (inner_l): New mode attribute.
22695 (vec_init<mode>): Renamed to ...
22696 (vec_init<mode><inner_l>): ... this.
22697 (vec_extract<mode>): Renamed to ...
22698 (vec_extract<mode><inner_l>): ... this.
22699 * config/sparc/sparc.md (veltmode): New mode iterator.
22700 (vec_init<VMALL:mode>): Renamed to ...
22701 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22702 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22703 (vec_initv2sisi): ... this.
22704 (vec_initv2sf): Renamed to ...
22705 (vec_initv2sfsf): ... this.
22706 (vec_extractv2sf): Renamed to ...
22707 (vec_extractv2sfsf): ... this.
22708 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22709 (vec_init<mode>): Renamed to ...
22710 (vec_init<mode><VEC_base_l>): ... this.
22711 (vec_extract<mode>): Renamed to ...
22712 (vec_extract<mode><VEC_base_l>): ... this.
22713 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22714 (vec_initv2sfsf): ... this.
22715 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22716 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22717 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22718 gen_vec_init* calls.
22720 2017-08-01 Richard Biener <rguenther@suse.de>
22722 PR tree-optimization/81297
22723 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22724 TREE_OVERFLOW from INTEGER_CSTs.
22726 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22728 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22730 2017-07-31 Carl Love <cel@us.ibm.com>
22732 * config/rs6000/rs6000-c: Add support for built-in functions
22733 vector signed char vec_xl_be (signed long long, signed char *);
22734 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22735 vector signed int vec_xl_be (signed long long, signed int *);
22736 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22737 vector signed long long vec_xl_be (signed long long, signed long long *);
22738 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22739 vector signed short vec_xl_be (signed long long, signed short *);
22740 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22741 vector double vec_xl_be (signed long long, double *);
22742 vector float vec_xl_be (signed long long, float *);
22743 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22744 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22745 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22747 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22748 (altivec_expand_builtin): Add switch statement to call
22749 altivec_expand_xl_be for each builtin.
22750 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22751 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22752 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22753 __builtin_vsx_le_be_v16qi.
22754 * doc/extend.texi: Update the built-in documentation file for the
22755 new built-in functions.
22757 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22760 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22762 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22764 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22766 * config.gcc: Add z14.
22767 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22768 CPU model numbers for z13s and z14.
22769 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22771 * config/s390/s390-opts.h (enum processor_type): Rename
22772 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22773 * config/s390/s390.c (processor_table): Add field for CPU name to
22774 be passed to Binutils.
22775 (s390_asm_output_machine_for_arch): Use the new field in
22776 processor_table for Binutils.
22777 (s390_expand_builtin): Replace arch12 with z14.
22778 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22779 (s390_get_sched_attrmask): Likewise.
22780 (s390_get_unit_mask): Likewise.
22781 * config/s390/s390.opt: Add z14 to processor_type enum.
22783 2017-07-31 Martin Jambor <mjambor@suse.cz>
22786 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22787 regardless of optimization level.
22789 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22790 Martin Liska <mliska@suse.cz>
22792 * predict.def: Remove old comment and adjust probability.
22793 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22794 PREDICT statements.
22796 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22799 * config/i386/i386.c (ix86_function_naked): New function.
22800 (ix86_can_use_return_insn_p): Return false for naked functions.
22801 (ix86_expand_prologue): Skip prologue for naked functions.
22802 (ix86_expand_epilogue): Skip epilogue for naked functions
22803 and emit trap instruction.
22804 (ix86_warn_func_return): New function.
22805 (ix86_attribute_table): Add "naked" attribute specification.
22806 (TARGET_WARN_FUNC_RETURN): Define.
22807 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22809 2017-07-31 Martin Liska <mliska@suse.cz>
22811 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22812 (dump_gimple_bb_header): Always dump BB info.
22813 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22815 2017-07-31 Martin Liska <mliska@suse.cz>
22818 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22819 also with current_function_decl non-null equality.
22821 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22824 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22825 change type to the element type, instead add eltype variable and
22826 use it where we are interested in the element type.
22828 PR tree-optimization/81603
22829 * ipa-polymorphic-call.c
22830 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22831 offset arithmetic in offset_int, bail out if the resulting bit offset
22832 doesn't fit into shwi.
22834 2017-07-31 Martin Liska <mliska@suse.cz>
22836 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22837 (gimplify_save_expr): Fix comment.
22839 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22842 * config/i386/i386.c (ix86_function_arg): Update arguments for
22844 (ix86_compute_frame_layout): Set the initial stack offset to
22845 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22846 INCOMING_FRAME_SP_OFFSET.
22847 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22848 stack before exception handler returns.
22849 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22850 the 'ERROR_CODE' for exception handler.
22852 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22854 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22855 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22856 (ASM_OUTPUT_REG_POP): Ditto.
22857 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22858 instead of asm_fprintf to output pure string.
22860 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22862 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22863 to imported_module_or_decl hook.
22864 (debug_nothing_tree_tree_tree_bool): Remove.
22865 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22866 * debug.c (do_nothing_debug_hooks): Use
22867 debug_nothing_tree_tree_tree_bool_bool instead of
22868 debug_nothing_tree_tree_tree_bool.
22869 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22870 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22871 * sdbout.c (sdb_debug_hooks): Likewise.
22872 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22873 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22875 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22876 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22877 attribute, don't add anything.
22879 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22881 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22882 (fold_build2_stat_loc): Likewise.
22883 (fold_build3_stat_loc): Likewise.
22884 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22885 (fold_build1_loc): Remove macro.
22886 (fold_build2_loc): Likewise.
22887 (fold_build3_loc): Likewise.
22889 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22891 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22892 (gimple_build_debug_bind_source_stat): Likewise.
22893 * gimple.h (gimple_build_debug_bind): Remove macro.
22894 (gimple_build_debug_bind_source): Likewise.
22896 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22898 * bitmap.c (bitmap_alloc): Adjust.
22899 (bitmap_gc_alloc): Likewise.
22900 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22902 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22904 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22905 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22906 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22907 (bitmap_gc_alloc_stat): Likewise.
22908 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22910 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22912 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22913 * rtl.h (shallow_copy_rtx): Remove macro.
22915 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22917 * emit-rtl.c (gen_raw_REG): Adjust.
22918 * gengenrtl.c (gendef): Likewise.
22919 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22920 * rtl.h (rtx_alloc): Remove macro.
22922 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22924 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22925 (build_tree_list_stat): Likewise.
22926 * tree.h (build_tree_list): Remove macro.
22927 (build_tree_list_vec): Likewise.
22929 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22931 * tree.c (make_vector_stat): Remove _stat from name.
22932 (build_vector_stat): Likewise.
22933 * tree.h (make_vector_stat): Remove macro.
22934 (build_vector_stat): Likewise.
22936 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22938 * tree.h (build_var_debug_value): Remove prototype.
22940 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22942 * tree.c (tree_cons_stat): Remove _stat from name.
22943 * tree.h (tree_cons): Remove macro.
22945 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22947 * tree.c (build_vl_exp_stat): Remove _stat from name.
22948 * tree.h (build_vl_exp): Remove macro.
22950 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22952 * tree.c (build_decl_stat): Remove _stat from name.
22953 * tree.h (build_decl): Remove macro.
22955 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22957 * gimple.c (gimple_build_with_ops_stat): Adjust.
22958 (gimple_alloc_stat): Remove _stat from name.
22959 * gimple.h (gimple_alloc): Remove macro.
22961 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22963 * tree.c (make_tree_vec_stat): Remove _stat from name.
22964 (grow_tree_vec_stat): Likewise.
22965 * tree.h (make_tree_vec_stat): Adjust prototype.
22966 (grow_tree_vec_stat): Likewise.
22967 (make_tree_vec): Remove macro.
22968 (grow_tree_vec): Likewise.
22970 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22972 * fold-const.c (fold_build1_stat_loc): Adjust.
22973 (fold_build2_stat_loc): Likewise.
22974 (fold_build3_stat_loc): Likewise.
22975 * tree.c (build0_stat): Remove _stat from name.
22976 (build1_stat): Likewise.
22977 (build2_stat): Likewise.
22978 (build3_stat): Likewise.
22979 (build4_stat): Likewise.
22980 (build5_stat): Likewise.
22981 * tree.h (build1_loc): Remove macro, and rename _stat function
22983 (build2_loc): Likewise.
22984 (build3_loc): Likewise.
22985 (build4_loc): Likewise.
22986 (build5_loc): Likewise.
22988 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22990 * tree.c (make_int_cst_stat): Remove _stat from name.
22991 * tree.h (make_int_cst_stat): Adjust prototype.
22992 (make_int_cst): Remove macro.
22994 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22996 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22997 * tree.h (make_tree_binfo_stat): Adjust prototype.
22998 (make_tree_binfo): Remove.
23000 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23002 * tree.c (copy_node_stat): Rename to copy_node.
23003 (build_distinct_type_copy): Adjust.
23004 * tree.h (copy_node_stat): Adjust prototype.
23005 (copy_node): Remove macro.
23007 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23009 * tree.c (make_node_stat): rename to make_node.
23010 (build_tree_list_stat): Adjust.
23011 (build0_stat): Likewise.
23012 (build2_stat): Likewise.
23013 (build3_stat): Likewise.
23014 (build4_stat): Likewise.
23015 (build5_stat): Likewise.
23016 (build_decl_stat): Likewise.
23017 * tree.h (make_node_stat): Adjust prototype.
23018 (make_node): remove macro.
23020 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
23022 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
23023 (PPC_FEATURE2_SCV): Likewise.
23024 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
23026 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23028 * config/aarch64/aarch64.c
23029 (aarch64_internal_mov_immediate): Add new special pattern.
23030 * config/aarch64/aarch64.md (*movdi_aarch64):
23031 Add reg/32bit const mov case.
23033 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23034 Richard Sandiford <richard.sandiford@linaro.org>
23036 * config/aarch64/aarch64.md (mov<mode>): Generalize.
23037 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
23038 Add integer and movi cases.
23039 (movi-split-hf-df-sf split, fp16): New.
23040 (enabled): Added TARGET_FP_F16INST.
23041 * config/aarch64/iterators.md (GPF_HF): New.
23042 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
23044 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23046 * config/aarch64/aarch64.c
23047 (aarch64_simd_container_mode): Add prototype.
23048 (aarch64_expand_mov_immediate): Add HI support.
23049 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
23050 (aarch64_can_const_movi_rtx_p): New.
23051 (aarch64_preferred_reload_class):
23052 Remove restrictions of using FP registers for certain SIMD operations.
23053 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
23054 (aarch64_valid_floating_const): Add integer move validation.
23055 (aarch64_simd_imm_scalar_p): Remove.
23056 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
23057 (aarch64_legitimate_constant_p): Expand list of supported cases.
23058 * config/aarch64/aarch64-protos.h
23059 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
23060 (aarch64_reinterpret_float_as_int): New.
23061 (aarch64_simd_imm_scalar_p): Remove.
23062 * config/aarch64/constraints.md (Uvi): New.
23063 (Dd): Split into Ds and new Dd.
23064 * config/aarch64/aarch64.md (*movsi_aarch64):
23066 (*movdi_aarch64): Add SIMD mov case.
23068 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23070 * tree-predcom.c: (struct chain): Handle store-store chain in which
23071 stores for elimination only store loop invariant values.
23072 (execute_pred_commoning_chain): Ditto.
23073 (prepare_initializers_chain_store_elim): Ditto.
23074 (prepare_finalizers): Ditto.
23075 (is_inv_store_elimination_chain): New function.
23076 (initialize_root_vars_store_elim_1): New function.
23078 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23080 * tree-predcom.c: Revise general description of the pass.
23081 (enum chain_type): New enum type for store elimination.
23082 (struct chain): New field supporting store elimination.
23083 (struct component): Ditto.
23084 (dump_chain): Dump store-stores chain.
23085 (release_chain): Release resources.
23086 (split_data_refs_to_components): Compute and create component
23087 contains only stores for elimination.
23088 (get_chain_last_ref_at): New function.
23089 (make_invariant_chain): Initialization.
23090 (make_rooted_chain): Specify chain type in parameter and record it.
23091 (add_looparound_copies): Skip for store-stores chain.
23092 (determine_roots_comp): Compute type of chain and pass it to
23094 (initialize_root_vars_store_elim_2): New function.
23095 (finalize_eliminated_stores): New function.
23096 (remove_stmt): Handle store for elimination.
23097 (execute_pred_commoning_chain): Execute predictive commoning on
23098 store-store chains.
23099 (determine_unroll_factor): Skip unroll for store-stores chain.
23100 (prepare_initializers_chain_store_elim): New function.
23101 (prepare_initializers_chain): Hanlde store-store chain.
23102 (prepare_finalizers_chain, prepare_finalizers): New function.
23103 (tree_predictive_commoning_loop): Return integer value indicating
23104 if loop is unrolled or lcssa form is corrupted.
23105 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
23107 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23109 * tree-predcom.c (initialize_root): Delete.
23110 (execute_pred_commoning_chain): Initialize root vars and replace
23111 reference of non-combined chain directly, rather than call above
23114 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23116 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
23117 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
23119 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23121 * tree-predcom.c (struct chain): New field init_seq.
23122 (release_chain): Release init_seq.
23123 (prepare_initializers_chain): Record intialization stmts in above
23125 (insert_init_seqs): New function.
23126 (tree_predictive_commoning_loop): Call insert_init_seqs.
23128 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23130 * tree-predcom.c (determine_roots_comp): Skip trivial components.
23132 2017-07-28 Richard Biener <rguenther@suse.de>
23134 * match.pd: Remove superfluous :c.
23135 * genmatch.c (simplify::id): Add member.
23136 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
23138 (current_id): New global.
23139 (dt_node::parent): Move from ...
23140 (dt_operand::parent): ... here. Add for_id member.
23141 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
23142 (decision_tree::find_node): Relax order requirement when
23143 merging DT_TRUE nodes to ones inbetween the current simplify
23144 and the one we try to merge with. Add diagnostic whenever
23145 we need to enforce pattern order by not merging.
23146 (decision_tree::insert): Set current_id.
23147 (decision_tree::print_node): Dump parent node and for_id.
23148 (parser::last_id): Add member.
23149 (parser::push_simplify): Assign unique id.
23150 (parser::parser): Initialize last_id.
23152 2017-07-28 Martin Liska <mliska@suse.cz>
23155 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
23156 gimple_build_debug_bind.
23158 2017-07-28 Richard Biener <rguenther@suse.de>
23160 PR tree-optimization/81502
23161 * match.pd: Add pattern combining BIT_INSERT_EXPR with
23163 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
23165 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
23166 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
23167 for BIT_FIELD_REF args.
23168 * fold-const.c (make_bit_field_ref): Likewise.
23169 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
23171 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23174 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
23175 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
23176 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
23177 Or it into SANITIZER_UNDEFINED.
23178 * ubsan.c: Include gimple-fold.h and varasm.h.
23179 (ubsan_expand_ptr_ifn): New function.
23180 (instrument_pointer_overflow): New function.
23181 (maybe_instrument_pointer_overflow): New function.
23182 (instrument_object_size): Formatting fix.
23183 (pass_ubsan::execute): Call instrument_pointer_overflow
23184 and maybe_instrument_pointer_overflow.
23185 * internal-fn.c (expand_UBSAN_PTR): New function.
23186 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
23187 * sanitizer.def (__ubsan_handle_pointer_overflow,
23188 __ubsan_handle_pointer_overflow_abort): New builtins.
23189 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
23190 * internal-fn.def (UBSAN_PTR): New internal function.
23191 * opts.c (sanitizer_opts): Add pointer-overflow.
23192 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
23193 * fold-const.c (build_range_check): Compute pointer range check in
23194 integral type if pointer arithmetics would be needed. Formatting
23197 2017-07-28 Martin Liska <mliska@suse.cz>
23200 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
23201 parameters that are of a variable-length.
23203 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
23205 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
23207 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
23208 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
23209 (CRT_CALL_STATIC_FUNCTION): Likewise.
23210 (ASM_DEFAULT_SPEC): New define.
23211 (ASM_SPEC32): Likewise.
23212 (ASM_SPEC64): Likewise.
23213 (ASM_SPEC_COMMON): Likewise.
23214 (ASM_SPEC): Likewise.
23215 (INVALID_64BIT): Likewise.
23216 (LINK_OS_DEFAULT_SPEC): Likewise.
23217 (LINK_OS_SPEC32): Likewise.
23218 (LINK_OS_SPEC64): Likewise.
23219 (POWERPC_LINUX): Likewise.
23220 (PTRDIFF_TYPE): Likewise.
23221 (RESTORE_FP_PREFIX): Likewise.
23222 (RESTORE_FP_SUFFIX): Likewise.
23223 (SAVE_FP_PREFIX): Likewise.
23224 (SAVE_FP_SUFFIX): Likewise.
23225 (SIZE_TYPE): Likewise.
23226 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23227 (TARGET_64BIT): Likewise.
23228 (TARGET_64BIT): Likewise.
23229 (TARGET_AIX): Likewise.
23230 (WCHAR_TYPE_SIZE): Likewise.
23231 (WCHAR_TYPE): Undefine.
23232 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23233 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23234 (CPP_OS_RTEMS_SPEC): Delete.
23235 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23236 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23238 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23240 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23242 PR tree-optimization/81578
23243 * tree-parloops.c (build_new_reduction): Bail out if
23244 reduction_code isn't one of the standard OpenMP reductions.
23245 Move the details printing after that decision.
23247 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23249 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23250 related to reload_in_progress.
23251 (splat_input_operand): Likewise.
23252 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23254 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23256 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23257 (TARGET_INSTANTIATE_DECLS): Likewise.
23258 (legitimate_indexed_address_p): Delete reload_in_progress code.
23259 (rs6000_debug_legitimate_address_p): Likewise.
23260 (rs6000_eliminate_indexed_memrefs): Likewise.
23261 (rs6000_emit_le_vsx_store): Likewise.
23262 (rs6000_emit_move_si_sf_subreg): Likewise.
23263 (rs6000_emit_move): Likewise.
23264 (register_to_reg_type): Likewise.
23265 (rs6000_pre_atomic_barrier): Likewise.
23266 (rs6000_machopic_legitimize_pic_address): Likewise.
23267 (rs6000_allocate_stack_temp): Likewise.
23268 (rs6000_address_for_fpconvert): Likewise.
23269 (rs6000_address_for_altivec): Likewise.
23270 (rs6000_secondary_memory_needed_rtx): Delete function.
23271 (rs6000_check_sdmode): Likewise.
23272 (rs6000_alloc_sdmode_stack_slot): Likewise.
23273 (rs6000_instantiate_decls): Likewise.
23274 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23275 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23276 Delete reload_in_progress.
23277 (*vec_reload_and_plus_<mptrsize>): Likewise.
23278 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23279 (vsx_div_v2di): Likewise.
23280 (vsx_udiv_v2di): Likewise.
23282 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23284 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23285 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23286 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23287 (rs6000_debug_reg_global): Delete print of LRA status.
23288 (rs6000_option_override_internal): Delete dead LRA related code.
23289 (rs6000_lra_p): Delete function.
23290 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23292 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23294 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23295 * config/riscv/rtems.h: New file.
23297 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23298 Sudakshina Das <sudi.das@arm.com>
23300 * config/aarch64/aarch64.md
23301 (define_split for and<mode>3nr_compare): Move
23302 non aarch64_logical_operand to a register.
23303 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23304 register immediate operand to a register.
23305 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23307 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23309 PR middle-end/81564
23310 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23312 2017-07-27 Richard Biener <rguenther@suse.de>
23314 PR tree-optimization/81573
23315 PR tree-optimization/81494
23316 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23317 multi defuse cycle case.
23319 2017-07-27 Richard Biener <rguenther@suse.de>
23321 PR tree-optimization/81571
23322 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23325 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23327 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23328 earlier and only if MASK_FPU is set. Adjust formatting.
23330 2017-07-27 Martin Liska <mliska@suse.cz>
23332 * opt-functions.awk: Add validation of value of Init.
23333 * optc-gen.awk: Pass new argument.
23335 2017-07-27 Martin Liska <mliska@suse.cz>
23337 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23338 Fix wrong condition.
23340 2017-07-27 Martin Liska <mliska@suse.cz>
23342 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23343 BBs and edges seen by autoFDO.
23345 2017-07-27 Richard Biener <rguenther@suse.de>
23347 PR tree-optimization/81502
23348 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23349 with incompatible but same sized type.
23350 (execute_update_addresses_taken): Likewise.
23352 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23354 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23355 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23357 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23360 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23361 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23362 Change s_operand to memory_operand.
23364 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23366 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23367 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23368 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23369 Emit instructions rather than returning an expression. Handle TFmode
23370 and KFmode by casting to TImode.
23371 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23372 (rs6000_emit_le_vsx_store): Likewise.
23373 * config/rs6000/vsx.md (VSX_TI): New iterator.
23374 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23375 (*vsx_le_undo_permute_<mode>): Likewise.
23376 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23377 emit the split sequence.
23378 (*vsx_le_perm_store_<mode>): Likewise.
23380 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23382 PR tree-optimization/81555
23383 PR tree-optimization/81556
23384 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23385 if true, force CHANGED for the recursive invocation.
23386 (reassociate_bb): Remember original length of ops array, pass
23387 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23389 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23390 attributes for noipa attribute. For naked attribute use
23391 lookup_attribute first before lookup_attribute_spec.
23392 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23394 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23395 for functions with noipa attribute.
23396 (cgraph_externally_visible_p): Return true for functions with noipa
23398 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23399 for functions with noipa attribute.
23400 * doc/extend.texi: Document noipa function attribute.
23401 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23402 also for functions with noipa attribute.
23403 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23405 2017-07-26 Andrew Pinski <apinski@cavium.com>
23407 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23408 vec_unalign_load_cost and vec_unalign_store_cost.
23410 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23412 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23413 -mvsx-small-integer option.
23414 (ISA_3_0_MASKS_IEEE): Likewise.
23415 (OTHER_VSX_VECTOR_MASKS): Likewise.
23416 (POWERPC_MASKS): Likewise.
23417 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23418 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23419 code, only testing for DImode being allowed in non-VSX floating
23421 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23422 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23424 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23425 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23426 TARGET_P8_VECTOR test.
23427 (rs6000_secondary_reload_simple_move): Likewise.
23428 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23429 since TARGET_P9_VECTOR was already tested.
23430 (rs6000_opt_masks): Remove -mvsx-small-integer.
23431 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23432 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23434 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23435 test for TARGET_VEXTRACTUB was used, and that uses
23437 (p9 extract splitter): Likewise.
23438 (vsx_extract_<mode>_di_p9): Likewise.
23439 (vsx_extract_<mode>_store_p9): Likewise.
23440 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23441 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23442 the elimination of TARGET_VSX_SMALL_INTEGER.
23443 (vsx_extract_<mode>_p8): Likewise.
23444 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23445 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23446 (vsx_set_<mode>_p9): Likewise.
23447 (vsx_set_v4sf_p9): Likewise.
23448 (vsx_set_v4sf_p9_zero): Likewise.
23449 (vsx_insert_extract_v4sf_p9): Likewise.
23450 (vsx_insert_extract_v4sf_p9_2): Likewise.
23451 * config/rs6000/rs6000.md (sign extend splitter): Change
23452 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23453 (floatsi<mode>2_lfiwax_mem): Likewise.
23454 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23455 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23456 since a test for TARGET_P9_VECTOR was used.
23457 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23458 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23459 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23460 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23461 TARGET_P8_VECTOR test.
23462 (fix_trunc<mode>si2_stfiwx): Likewise.
23463 (fix_trunc<mode>si2_internal): Likewise.
23464 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23465 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23467 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23468 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23469 TARGET_P8_VECTOR test.
23470 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23471 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23472 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23474 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23475 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23476 since a test for TARGET_P9_VECTOR was used.
23477 (splitter for loading small constants): Likewise.
23479 2017-07-26 Andrew Pinski <apinski@cavium.com>
23481 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23484 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23487 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23488 (fp_valid_at): Likewise.
23490 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23492 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23493 (qdf24xx_addrcost_table): Likewise.
23494 (cortexa57_tunings): Update to use generic_branch_cost.
23495 (cortexa72_tunings): Likewise.
23496 (cortexa73_tunings): Likewise.
23497 (qdf24xx_tunings): Likewise.
23499 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23501 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23502 (thunderx2t99_branch_cost): Likewise.
23503 (cortexa35_tunings): Update to use generic_branch_cost.
23504 (cortexa53_tunings): Likewise.
23505 (cortexa57_tunings): Likewise.
23506 (cortexa72_tunings): Likewise.
23507 (cortexa73_tunings): Likewise.
23508 (thunderx2t99_tunings): Likewise.
23510 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23512 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23513 (sparc_option_override): Honour MASK_FSMULD.
23514 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23515 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23516 * config/sparc/sparc.opt (mfsmuld): New option.
23517 * doc/invoke.texi (mfsmuld): Document option.
23519 2017-07-26 Marek Polacek <polacek@redhat.com>
23521 PR middle-end/70992
23522 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23524 2017-07-26 Richard Biener <rguenther@suse.de>
23526 * gimple-match-head.c (do_valueize): Return OP if valueize
23528 (get_def): New helper to get at the def stmt of a SSA name
23529 if valueize allows.
23530 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23531 do_valueize to get at the def stmt.
23532 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23534 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23536 PR middle-end/46932
23537 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23539 2017-07-26 Martin Liska <mliska@suse.cz>
23542 * function.c (expand_function_start): Make expansion of
23543 nonlocal_goto_save_area after parm_birth_insn.
23545 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23547 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23548 from all CPU target flags enable members.
23550 2017-07-26 Richard Biener <rguenther@suse.de>
23552 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23553 (decision_tree::gen): Make 'type' const.
23554 (write_predicate): Likewise.
23556 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23558 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23559 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23560 (rs6000_option_override_internal): Likewise.
23561 (rs6000_expand_vector_set): Likewise.
23562 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23563 (TARGET_UPPER_REGS_SF): Likewise.
23564 (TARGET_UPPER_REGS_DI): Likewise.
23565 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23566 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23567 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23568 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23569 (Splitters for DI constants in Altivec registers): Likewise.
23570 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23571 (vsx_set_v4sf_p9): Likewise.
23572 (vsx_set_v4sf_p9_zero): Likewise.
23573 (vsx_insert_extract_v4sf_p9): Likewise.
23574 (vsx_insert_extract_v4sf_p9_2): Likewise.
23576 2017-07-25 Carl Love <cel@us.ibm.com>
23578 * doc/extend.texi: Update the built-in documentation file for the
23579 existing built-in functions
23580 vector signed char vec_cnttz (vector signed char);
23581 vector unsigned char vec_cnttz (vector unsigned char);
23582 vector signed short vec_cnttz (vector signed short);
23583 vector unsigned short vec_cnttz (vector unsigned short);
23584 vector signed int vec_cnttz (vector signed int);
23585 vector unsigned int vec_cnttz (vector unsigned int);
23586 vector signed long long vec_cnttz (vector signed long long);
23587 vector unsigned long long vec_cnttz (vector unsigned long long);
23589 2017-07-25 Andrew Pinski <apinski@cavium.com>
23591 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23592 accesses where the use is for the first operand of a BIT_INSERT.
23594 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23597 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23598 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23600 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23602 * config/i386/gstabs.h: Delete.
23603 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23605 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23607 * config/i386/i386.c (ix86_decompose_address): Do not check for
23608 register RTX when looking at index_reg or base_reg.
23609 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23611 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23613 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23614 to update EH info here.
23616 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23618 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23620 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23622 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23624 2017-07-25 Torsten Duwe <duwe@suse.de>
23626 * common.opt: Introduce -fpatchable-function-entry
23627 command line option, and its variables function_entry_patch_area_size
23628 and function_entry_patch_area_start.
23629 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23630 including a two-value parser.
23631 * target.def (print_patchable_function_entry): New target hook.
23632 * targhooks.h (default_print_patchable_function_entry): New function.
23633 * targhooks.c (default_print_patchable_function_entry): Likewise.
23634 * toplev.c (process_options): Switch off IPA-RA if
23635 patchable function entries are being generated.
23636 * varasm.c (assemble_start_function): Look at the
23637 patchable-function-entry command line switch and current
23638 function attributes and maybe generate NOP instructions by
23639 calling the print_patchable_function_entry hook.
23640 * doc/extend.texi: Document patchable_function_entry attribute.
23641 * doc/invoke.texi: Document -fpatchable_function_entry
23642 command line option.
23643 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23645 * doc/tm.texi: Re-generate.
23647 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23650 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23651 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23653 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23655 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23657 2017-07-25 Richard Biener <rguenther@suse.de>
23659 PR tree-optimization/81455
23660 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23661 not walk in cycles when looking for guards.
23663 2017-07-25 Richard Biener <rguenther@suse.de>
23665 PR tree-optimization/81529
23666 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23667 when optimizing backedge uses.
23669 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23671 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23673 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23674 to dl_section_ref. On AIX, append an expression to subtract
23675 the size of the section length to dl_section_ref.
23677 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23679 * configure.ac: If any of the config.* scripts fail, exit 1.
23680 * configure: Regenerate.
23682 2017-07-25 Richard Biener <rguenther@suse.de>
23684 PR middle-end/81546
23685 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23686 of immediate uses, be more verbose on errors.
23688 2017-07-25 Richard Biener <rguenther@suse.de>
23690 PR tree-optimization/81510
23691 * tree-vect-loop.c (vect_is_simple_reduction): When the
23692 reduction stmt is not inside the loop bail out.
23694 2017-07-25 Richard Biener <rguenther@suse.de>
23696 PR tree-optimization/81303
23697 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23698 profitability check against LOOP_VINFO_NITERSM1.
23700 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23702 * domwalk.c (cmp_bb_postorder): Simplify.
23703 (sort_bbs_postorder): New function. Use it...
23704 (dom_walker::walk): ...here to optimize common cases.
23706 2017-07-25 Martin Liska <mliska@suse.cz>
23709 * ipa-visibility.c (function_and_variable_visibility): Make the
23710 redirection just on target that supports aliasing.
23711 Fix GNU coding style.
23713 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23716 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23718 * config/arm/rtems.h: Likewise.
23719 * config/bfin/rtems.h: Likewise.
23720 * config/i386/rtemself.h: Likewise.
23721 * config/lm32/rtems.h: Likewise.
23722 * config/m32c/rtems.h: Likewise.
23723 * config/m68k/rtemself.h: Likewise.
23724 * config/microblaze/rtems.h: Likewise.
23725 * config/mips/rtems.h: Likewise.
23726 * config/moxie/rtems.h: Likewise.
23727 * config/nios2/rtems.h: Likewise.
23728 * config/powerpcspe/rtems.h: Likewise.
23729 * config/rs6000/rtems.h: Likewise.
23730 * config/rtems.h: Likewise.
23731 * config/sh/rtems.h: Likewise.
23732 * config/sh/rtemself.h: Likewise.
23733 * config/sparc/rtemself.h: Likewise.
23735 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23738 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23739 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23740 * tree-ssa-structalias.c (alias_get_name): Same.
23742 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23745 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23746 instructions if no du chain is found.
23748 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23750 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23752 2017-07-25 Richard Biener <rguenther@suse.de>
23754 PR middle-end/81505
23755 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23758 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23760 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23761 upper-regs options.
23762 (ISA_2_7_MASKS_SERVER): Likewise.
23763 (ISA_3_0_MASKS_IEEE): Likewise.
23764 (OTHER_P8_VECTOR_MASKS): Likewise.
23765 (OTHER_VSX_VECTOR_MASKS): Likewise.
23766 (POWERPC_MASKS): Likewise.
23767 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23768 duplicate list of options.
23769 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23770 explicit -mupper-regs options.
23771 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23772 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23773 alias for -mupper-regs-df.
23774 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23775 (rs6000_init_hard_regno_mode_ok): Likewise.
23776 (rs6000_option_override_internal): Likewise.
23777 (rs6000_opt_masks): Likewise.
23778 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23779 options in terms of whether -mvsx or -mpower8-vector was used.
23780 (TARGET_UPPER_REGS_DI): Likewise.
23781 (TARGET_UPPER_REGS_SF): Likewise.
23782 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23783 -mupper-regs-* options.
23785 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23787 * passes.c (emergency_dump_function): Print some empty lines and a
23788 header before the RTL dump.
23790 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23792 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23794 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23797 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23798 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23800 2017-07-24 Carl Love <cel@us.ibm.com>
23802 * config/rs6000/rs6000-c.c: Add support for built-in functions
23803 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23804 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23805 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23806 vec_extract_fp_from_shortl): Add defines for the two builtins.
23807 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23808 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23810 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23811 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23812 * doc/extend.texi: Update the built-in documentation file for the
23813 new built-in function.
23815 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23818 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23820 * doc/generic.texi: Likewise.
23821 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23822 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23824 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23826 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23827 (aarch64_mls_elt_merge<mode>): Likewise.
23829 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23831 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23832 having __cxa_atexit.
23834 2017-07-23 Michael Collison <michael.collison@arm.com>
23836 * config/arm/arm.c (arm_option_override): Deprecate
23837 use of -mstructure-size-boundary.
23838 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23839 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23841 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23844 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23845 Reduce cost estimate for direct moves.
23847 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23850 * config/i386/i386.c (ix86_option_override_internal): Disable
23851 BMI, BMI2 and TBM instructions for -m16.
23853 2017-07-21 Carl Love <cel@us.ibm.com>
23855 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23856 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23857 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23858 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23859 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23860 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23861 VMULOSW): New enum "unspec" values.
23862 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23863 altivec_vmulosw): New patterns.
23864 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23865 VMULOSW): Add definitions.
23867 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23869 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23870 (qdf24xx): Likewise.
23871 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23872 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23873 (AARCH64_FL_V8_1): Renumber.
23874 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23875 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23876 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23877 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23878 rdma to feature modifiers list.
23880 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23882 PR middle-end/56727
23883 * ipa-visibility (function_and_variable_visibility): Convert
23884 recursive PLT call to direct call if appropriate.
23886 2017-07-21 Andrew Pinski <apinski@cavium.com>
23888 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23889 operand 1 to see if the types precision matches.
23890 * fold-const.c (operand_equal_p): Likewise.
23892 2017-07-21 Richard Biener <rguenther@suse.de>
23894 PR tree-optimization/81303
23895 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23896 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23897 (vect_peeling_hash_get_lowest_cost): Adjust.
23898 (vect_enhance_data_refs_alignment): Likewise. Use
23899 vect_get_peeling_costs_all_drs to compute the penalty for no
23900 peeling to match up costs.
23902 2017-07-21 Richard Biener <rguenther@suse.de>
23904 PR tree-optimization/81500
23905 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23906 we didn't identify a reduction path.
23908 2017-07-21 Tom de Vries <tom@codesourcery.com>
23909 Cesar Philippidis <cesar@codesourcery.com>
23911 PR gcov-profile/81442
23912 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23915 2017-07-21 Tom de Vries <tom@codesourcery.com>
23918 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23920 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23921 nvptx_override_options_after_change.
23923 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23925 * dwarf2out.c (output_file_names): Avoid double testing for
23926 dwarf_version >= 5.
23928 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23930 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23932 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23934 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23936 (try_crossjump_to_edge): Do not punt on partitioned functions.
23938 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23940 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23941 Put all BBs reachable only via paths crossing cold region to cold
23943 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23945 2016-07-21 Richard Biener <rguenther@suse.de>
23947 PR tree-optimization/81303
23948 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23949 into account prologue and epilogue iterations when raising
23950 min_profitable_iters to sth at least covering one vector iteration.
23952 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23954 * config/arm/arm.c (arm_test_cpu_arch_dat):
23957 2017-07-20 Nathan Sidwell <nathan@acm.org>
23959 Remove TYPE_METHODS.
23960 * tree.h (TYPE_METHODS): Delete.
23961 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23962 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23963 (dbxout_type_methods): Scan TYPE_FIELDS.
23964 (dbxout_type): Don't check TYPE_METHODS here.
23965 * function.c (use_register_for_decl): Always ignore register for
23966 class types when not optimizing.
23967 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23968 * tree.c (free_lang_data_in_type): Stitch out member functions and
23969 templates from TYPE_FIELDS.
23970 (build_distinct_type_copy, verify_type_variant,
23971 verify_type): Member fns are on TYPE_FIELDS.
23972 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23973 * tree-pretty-print.c (dump_generic_node): Likewise.
23975 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23978 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23979 V2TImode and V4TImode.
23980 (ix86_expand_vector_extract): Likewise.
23981 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23982 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23983 (ssescalarmode): Handle V4TImode and V2TImode.
23984 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23985 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23986 (VEXTRACTI128_MODE): New mode iterator.
23987 (splitter for *vec_extractv?ti first element): New.
23988 (VEC_INIT_MODE): New mode iterator.
23989 (vec_init<mode>): Consolidate 3 expanders into one using
23990 VEC_INIT_MODE mode iterator.
23992 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23994 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23995 non_spilled_static_chain_regno_p.
23997 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23999 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
24001 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
24003 * bb-reorder.c (connect_traces): Allow copying of blocks within
24006 2017-07-20 Richard Biener <rguenther@suse.de>
24008 * gimple.h (gimple_phi_result): Add gphi * overload.
24009 (gimple_phi_result_ptr): Likewise.
24010 (gimple_phi_arg): Likewise. Adjust index assert to only
24011 allow actual argument accesses rather than all slots available
24013 (gimple_phi_arg_def): Add gphi * overload.
24014 * tree-phinodes.c (make_phi_node): Initialize only actual
24016 (resize_phi_node): Clear memory not covered by old node,
24017 do not initialize excess argument slots.
24018 (reserve_phi_args_for_new_edge): Initialize new argument slot
24021 2017-07-20 Bin Cheng <bin.cheng@arm.com>
24023 PR tree-optimization/81388
24025 2016-07-21 Bin Cheng <bin.cheng@arm.com>
24027 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
24028 by removing computation of may_be_zero.
24030 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24031 Tom de Vries <tom@codesourcery.com>
24033 PR middle-end/81030
24034 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
24035 when gimple level profile disagrees with what RTL expander did.
24037 2017-07-20 Richard Biener <rguenther@suse.de>
24039 PR tree-optimization/61171
24040 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
24041 (vect_analyze_stmt): Add slp instance parameter.
24042 (vectorizable_reduction): Likewise.
24043 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
24044 (vect_is_simple_reduction): Deal with chains not detected
24045 as SLP reduction chain, specifically not properly associated
24046 chains containing a mix of plus/minus.
24047 (get_reduction_op): Remove.
24048 (get_initial_defs_for_reduction): Simplify, pass in whether
24049 this is a reduction chain, pass in the SLP node for the PHIs.
24050 (vect_create_epilog_for_reduction): Get the SLP instance as
24052 (vectorizable_reduction): Get the SLP instance as arg.
24053 During analysis remember the SLP node with the PHIs in the
24054 instance. Simplify getting at the vectorized reduction PHIs.
24055 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
24056 through SLP instance.
24057 (vect_slp_analyze_operations): Likewise.
24058 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
24059 (vect_transform_stmt): Likewise.
24061 2017-07-20 Tom de Vries <tom@codesourcery.com>
24063 PR tree-optimization/81489
24064 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
24065 read of phi arg location to before loop that modifies phi.
24067 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24069 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
24072 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24074 PR middle-end/81331
24075 * except.c (execute): Fix ordering issue.
24077 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24079 PR rtl-optimization/81423
24080 * combine.c (make_compound_operation_int): Don't try to optimize
24081 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
24083 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24085 PR rtl-optimization/81423
24086 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
24087 with a constant that is -1 in the truncated to mode.
24089 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24091 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
24092 (determine_unlikely_bbs): ... here.
24093 * predict.h (propagate_unlikely_bbs_forward): Declare.
24094 * cfgexpand.c (pass_expand::execute): Use it.
24095 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
24097 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
24098 propagate_unlikely_bbs_forward.
24100 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24102 PR middle-end/81331
24103 * except.c (maybe_add_nop_after_section_switch): New function.
24106 2017-07-19 Tom de Vries <tom@codesourcery.com>
24108 * gimple.h (gimple_phi_set_arg): Make assert more strict.
24110 2017-07-19 Tom de Vries <tom@codesourcery.com>
24112 * gimple.h (gimple_phi_arg): Make assert more strict.
24114 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
24116 * config.gcc (powerpc*-*-*): Add mmintrin.h.
24117 * config/rs6000/mmintrin.h: New file.
24118 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
24120 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24122 PR tree-optimization/81346
24123 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
24125 2017-07-19 Tom de Vries <tom@codesourcery.com>
24127 * config/nvptx/nvptx.md (VECIM): Add V2DI.
24129 2017-07-19 Tom de Vries <tom@codesourcery.com>
24131 * config/nvptx/nvptx-modes.def: Add V2DImode.
24132 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
24133 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
24134 (nvptx_output_mov_insn): Handle lack of mov.b128.
24135 (nvptx_print_operand): Handle 'H' and 'L' codes.
24136 (nvptx_vector_mode_supported): Allow V2DImode.
24137 (nvptx_preferred_simd_mode): New function.
24138 (nvptx_data_alignment): New function.
24139 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
24140 nvptx_preferred_simd_mode.
24141 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
24143 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
24145 2017-07-19 Tom de Vries <tom@codesourcery.com>
24147 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
24148 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
24149 (nvptx_vector_mode_supported): New function. Allow V2SImode.
24150 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
24151 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
24152 (mov<VECIM>_insn): New define_insn.
24153 (define_expand "mov<VECIM>): New define_expand.
24155 2017-07-19 Tom de Vries <tom@codesourcery.com>
24157 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
24159 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24161 PR tree-optimization/81346
24162 * fold-const.h (fold_div_compare, range_check_type): Declare.
24163 * fold-const.c (range_check_type): New function.
24164 (build_range_check): Use range_check_type.
24165 (fold_div_compare): No longer static, rewritten into
24166 a match.pd helper function.
24167 (fold_comparison): Don't call fold_div_compare here.
24168 * match.pd (X / C1 op C2): New optimization using fold_div_compare
24169 as helper function.
24171 2017-07-19 Nathan Sidwell <nathan@acm.org>
24173 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
24174 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
24175 * tree.c (find_decls_types_r, verify_type): Use
24176 TYPE_{MIN,MAX}_VALUE_RAW.
24177 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
24178 (hash_tree): Likewise.
24179 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
24181 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
24184 2017-07-18 Tom de Vries <tom@codesourcery.com>
24186 PR middle-end/81464
24187 * omp-expand.c (expand_omp_for_static_chunk): Handle
24188 equal-argument loop exit phi.
24190 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
24193 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
24194 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
24195 operand 2 predicate.
24196 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
24197 operand 2 predicate.
24198 (ror,rol -> rorx splitters): Use const_int_operand as
24199 operand 2 predicate.
24201 2017-06-18 Richard Biener <rguenther@suse.de>
24203 PR tree-optimization/81410
24204 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
24205 the gap in the ! slp_perm SLP case after each group.
24207 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24209 PR middle-end/81463
24210 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
24213 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24215 PR middle-end/81462
24216 * predict.c (set_even_probabilities): Cleanup; do not affect
24217 probabilities that are already known.
24218 (combine_predictions_for_bb): Call even when count is set.
24220 2017-07-18 Nathan Sidwell <nathan@acm.org>
24222 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24225 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24228 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24229 optimization for loop niter analysis.
24231 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24234 * config/avr/avr.c (avr_optimize_casesi): Don't use
24235 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24237 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24239 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24240 body_cost_vec from _vect_peel_extended_info.
24241 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24242 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24245 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24247 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24249 2017-07-18 Richard Biener <rguenther@suse.de>
24251 PR tree-optimization/80620
24252 PR tree-optimization/81403
24253 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24254 info when re-using a VN table entry.
24256 2017-07-18 Richard Biener <rguenther@suse.de>
24258 PR tree-optimization/81418
24259 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24260 vectype_in. Verify that with lane-reducing reduction operations
24261 we have a single def-use cycle.
24263 2017-07-17 Carl Love <cel@us.ibm.com>
24265 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24267 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24268 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24269 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24270 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24271 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24272 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24273 VMULOSW): New enum "unspec" values.
24274 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24275 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24276 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24277 altivec_vmulosw): New patterns.
24278 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24279 VMULOSW): Add definitions.
24281 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24283 * config/alpha/alpha.c: Include predict.h.
24285 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24287 * tree-vrp.c (compare_assert_loc): Fix comparison function
24288 to return predictable results.
24290 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24292 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24294 (subdi3): Likewise.
24295 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24296 * doc/invoke.texi (mexpand-adddi): Update text.
24298 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24300 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24301 that also clobbers the CC register. The old expand code is moved
24303 (*arc_clzsi2): ... here.
24304 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24305 the CC register. The old expand code is moved to ...
24306 (arc_ctzsi2): ... here.
24308 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24310 * config/arc/arc.opt (mindexed-loads): Use initial value
24311 TARGET_INDEXED_LOADS_DEFAULT.
24312 (mauto-modify-reg): Use initial value
24313 TARGET_AUTO_MODIFY_REG_DEFAULT.
24314 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24315 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24316 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24317 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24319 2017-07-17 Martin Liska <mliska@suse.cz>
24322 * opts.c (finish_options): Do not allow -fgnu-tm
24323 w/ -fsanitize={kernel-,}address. Say sorry.
24325 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24328 * tree-loop-distribution.c (classify_partition): Only assert on
24329 numer of iterations.
24330 (merge_dep_scc_partitions): Delete prameter. Update function call.
24331 (distribute_loop): Remove code handling loop with unknown niters.
24332 (pass_loop_distribution::execute): Skip loop with unknown niters.
24334 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24337 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24338 function sort_partitions_by_post_order.
24340 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24342 PR tree-optimization/81374
24343 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24344 the max index of basic blocks, rather than number of basic blocks.
24346 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24348 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24350 (arc_legitimate_pic_operand_p): Likewise.
24351 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24353 (arc_needs_pcl_p): Likewise.
24354 (arc_legitimate_pc_offset_p): Likewise.
24355 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24356 function is also used in constrains.md.
24357 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24358 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24359 PLUS. Only return true/false in known cases, otherwise assert.
24360 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24361 is already called in arc_legitimate_constant_p.
24362 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24364 (LEGITIMATE_PIC_OPERAND_P): Use
24365 arc_raw_symbolic_reference_mentioned_p function.
24366 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24371 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24372 Andrew Burgess <andrew.burgess@embecosm.com>
24374 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24375 (arc_return_address_register): New function.
24376 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24377 (arc_handle_fndecl_attribute): Add naked attribute.
24378 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24379 (TARGET_WARN_FUNC_RETURN): Likewise.
24380 (arc_allocate_stack_slots_for_args): New function.
24381 (arc_warn_func_return): Likewise.
24382 (machine_function): Change type fn_type.
24383 (arc_compute_function_type): Consider new naked function type,
24384 change function return type.
24385 (arc_must_save_register): Adapt to handle new
24386 arc_compute_function_type's return type.
24387 (arc_expand_prologue): Likewise.
24388 (arc_expand_epilogue): Likewise.
24389 (arc_return_address_regs): Delete.
24390 (arc_return_address_register): New function.
24391 (arc_epilogue_uses): Use above function.
24392 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24393 (arc_function_type): Change encoding, add naked type.
24394 (ARC_INTERRUPT_P): Change to handle the new encoding.
24395 (ARC_FAST_INTERRUPT_P): Likewise.
24396 (ARC_NORMAL_P): Define.
24397 (ARC_NAKED_P): Likewise.
24398 (arc_compute_function_type): Delete prototype.
24399 * config/arc/arc.md (in_ret_delay_slot): Use
24400 arc_return_address_register function.
24401 (simple_return): Likewise.
24402 (p_return_i): Likewise.
24404 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24406 PR tree-optimization/81428
24407 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24408 can't be built for those types.
24410 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24412 Remove stuff dead since r239246.
24414 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24415 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24416 (avr_inform_devices): Remove dead stuff.
24418 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24420 * config/arm/arm_neon.h: Fix softp typo.
24422 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24424 PR tree-optimization/81365
24425 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24426 aggregate moves onto bb predecessor edges, make sure there are no
24427 loads that could alias the lhs in between the start of bb and the
24430 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24433 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24434 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24435 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24437 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24439 PR tree-optimization/81396
24440 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24441 (init_symbolic_number): Initialize it to 1.
24442 (perform_symbolic_merge): Add n_ops from both operands into the new
24444 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24445 without base_addr as useless if they need more than one operation.
24446 (bswap_replace): Handle !bswap case for NULL base_addr.
24448 2017-07-17 Tom de Vries <tom@codesourcery.com>
24451 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24454 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24456 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24457 conditional builtin define __FIX_LEON3FT_B2BST.
24459 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24461 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24462 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24465 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24467 PR rtl-optimization/81424
24468 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24469 to remove potential trapping from operands if -fnon-call-exceptions.
24471 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24473 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24474 profile_proability for scalling.
24475 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24477 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24479 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24481 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24483 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24484 fixpoint arithmetics.
24486 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24488 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24489 fixpoint arithmetics.
24491 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24493 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24494 fixpoint arithmetics.
24496 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24498 * profile-count.h (profile_probability::from_reg_br_prob_note,
24499 profile_probability::to_reg_br_prob_note): New functions.
24500 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24501 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24502 * predict.c (probability_reliable_p): Update.
24503 (edge_probability_reliable_p): Update.
24504 (br_prob_note_reliable_p): Update.
24505 (invert_br_probabilities): Update.
24506 (add_reg_br_prob_note): New function.
24507 (combine_predictions_for_insn): Update.
24508 * asan.c (asan_clear_shadow): Update.
24509 * cfgbuild.c (compute_outgoing_frequencies): Update.
24510 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24511 (update_br_prob_note): Update.
24512 (rtl_verify_edges): Update.
24513 (purge_dead_edges): Update.
24514 (fixup_reorder_chain): Update.
24515 * emit-rtl.c (try_split): Update.
24516 * ifcvt.c (cond_exec_process_insns): Update.
24517 (cond_exec_process_if_block): Update.
24518 (dead_or_predicable): Update.
24519 * internal-fn.c (expand_addsub_overflow): Update.
24520 (expand_neg_overflow): Update.
24521 (expand_mul_overflow): Update.
24522 * loop-doloop.c (doloop_modify): Update.
24523 * loop-unroll.c (compare_and_jump_seq): Update.
24524 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24525 * predict.h: Update.
24526 * reorg.c (mostly_true_jump): Update.
24528 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24529 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24530 * config/arc/arc.c: (emit_unlikely_jump): Update.
24531 * config/arm/arm.c: (emit_unlikely_jump): Update.
24532 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24533 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24534 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24535 (ix86_print_operand): Update.
24536 (ix86_split_fp_branch): Update.
24537 (predict_jump): Update.
24538 * config/ia64/ia64.c (ia64_print_operand): Update.
24539 * config/mmix/mmix.c (mmix_print_operand): Update.
24540 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24541 (rs6000_expand_split_stack_prologue): Update.
24542 * config/rs6000/rs6000.c: Update.
24543 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24544 (s390_expand_vec_movstr): Update.
24545 (s390_expand_cs_tdsi): Update.
24546 (s390_expand_split_stack_prologue): Update.
24547 * config/sh/sh.c (sh_print_operand): Update.
24548 (expand_cbranchsi4): Update.
24549 (expand_cbranchdi4): Update.
24550 * config/sparc/sparc.c (output_v9branch): Update.
24551 * config/spu/spu.c (get_branch_target): Update.
24552 (ea_load_store_inline): Update.
24553 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24554 * config/tilepro/tilepro.c: Update.
24556 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24558 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24559 (gimplify_save_expr): Likewise.
24561 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24563 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24565 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24567 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24569 * timevar.def (TV_IPA_FNSUMMARY): Define.
24571 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24573 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24574 to back store errata sensitive sequence from being generated.
24575 (sqrtdf2_fix): Likewise.
24577 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24579 * tree-ssa-threadupdate.c (compute_path_counts,
24580 update_joiner_offpath_counts): Use profile_probability.
24582 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24585 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24587 * config/arm/arm-c.c (arm_cpu_builtins): Define
24588 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24590 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24592 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24593 array entries to represent __ieee128 versions of the
24594 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24595 scalar_extract_sig, and scalar_insert_exp built-in functions.
24596 (altivec_resolve_overloaded_builtin): Add special case handling
24597 for the __builtin_scalar_insert_exp function, as represented by
24598 the P9V_BUILTIN_VEC_VSIEDP constant.
24599 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24600 exponent support for __ieee128 argument.
24601 (VSESQP): Add scalar extract signature support for __ieee128
24603 (VSTDCNQP): Add scalar test negative support for __ieee128
24605 (VSIEQP): Add scalar insert exponent support for __int128 argument
24606 with __ieee128 result.
24607 (VSIEQPF): Add scalar insert exponent support for __ieee128
24608 argument with __ieee128 result.
24609 (VSTDCQP): Add scalar test data class support for __ieee128
24611 (VSTDCNQP): Add overload support for scalar test negative with
24612 __ieee128 argument.
24613 (VSTDCQP): Add overload support for scalar test data class
24614 __ieee128 argument.
24615 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24616 UNSPEC_VSX_SXSIGDP.
24617 (UNSPEC_VSX_SIEXPQP): New constant.
24618 (xsxexpqp): New insn for VSX scalar extract exponent quad
24620 (xsxsigqp): New insn for VSX scalar extract significand quad
24622 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24623 precision with floating point argument.
24624 (xststdcqp): New expand for VSX scalar test data class quad
24626 (xststdcnegqp): New expand for VSX scalar test negative quad
24628 (xststdcqp): New insn to match expansions for VSX scalar test data
24629 class quad precision and VSX scalar test negative quad precision.
24630 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24631 special case operand checking to enforce that second operand of
24632 VSX scalar test data class with quad precision argument is a 7-bit
24634 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24635 prototypes and descriptions of __ieee128 versions of
24636 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24637 scalar_test_data_class, and scalar_test_neg built-in functions.
24639 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24641 PR tree-optimization/81162
24642 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24643 replace a negate with an add.
24645 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24647 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24649 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24651 * config/arm/arm-c.c (arm_cpu_builtins): Define
24652 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24654 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24656 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24657 (armv8-r): Set ARM Cortex-R52 as default CPU.
24658 * config/arm/arm-tables.opt: Regenerate.
24659 * config/arm/arm-tune.md: Regenerate.
24660 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24662 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24663 extension for -mcpu=cortex-r52.
24665 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24667 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24668 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24669 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24670 (fp-armv8): Define it as FP_ARMv8 only.
24671 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24672 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24674 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24675 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24676 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24677 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24678 than TARGET_FPU_ARMV8.
24679 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24680 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24681 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24683 * config/arm/neon.md (neon_vrint): Likewise.
24684 (neon_vcvt): Likewise.
24685 (neon_<fmaxmin_op><mode>): Likewise.
24686 (<fmaxmin><mode>3): Likewise.
24687 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24688 * config/arm/predicates.md (arm_cond_move_operator): Check against
24689 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24691 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24693 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24694 to top of function.
24696 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24698 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24699 loop in comment with memset.
24701 2017-07-14 Martin Liska <mliska@suse.cz>
24703 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24704 * dwarf2out.c (is_java): Remove the function.
24705 (output_pubname): Remove usage of the function.
24706 (lower_bound_default): Remove usage of DW_LANG_Java.
24707 (gen_compile_unit_die): Likewise.
24708 * gcc.c: Remove compiler defaults for .java and .zip files.
24709 * gimple-expr.c (remove_suffix): Change as there's no longer
24710 extension than 4-letter one.
24711 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24712 (gimplify_save_expr): Likewise.
24713 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24714 as it's possible even for other languages than Java.
24715 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24716 * lto-opts.c (lto_write_options): Remove reference to Java.
24717 * opts.c (strip_off_ending): Update file extension handling.
24718 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24719 * tree-eh.c (lower_resx): Likewise.
24720 * tree.c (free_lang_data_in_type): Remove dead code.
24721 (find_decls_types_r): Likewise.
24722 (build_common_builtin_nodes): Remove Java from a comment.
24723 (verify_type): Remove dead code.
24724 * varasm.c (assemble_external): Remove Java from a comment.
24726 2017-07-14 Martin Liska <mliska@suse.cz>
24728 * opts.c (finish_options): Add quotes.
24729 (common_handle_option): Likewise.
24731 2017-07-14 Martin Liska <mliska@suse.cz>
24733 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24734 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24735 Remove N_SO_PASCAL.
24736 * dwarf2out.c (lower_bound_default): Do not handle
24738 (gen_compile_unit_die): Likewise.
24739 * gcc.c: Remove default extension binding for GNU Pascal.
24740 * stmt.c: Remove Pascal language from a comment.
24741 * xcoffout.c: Likewise.
24743 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24746 * diagnostic-show-locus.c (fixit_cmp): New function.
24747 (layout::layout): Sort m_fixit_hints.
24748 (column_range::column_range): Assert that the values are valid.
24749 (struct char_span): New struct.
24750 (correction::overwrite): New method.
24751 (struct source_line): New struct.
24752 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24753 calls in terms of classes source_line and char_span, and
24754 correction::overwrite.
24755 (selftest::test_overlapped_fixit_printing_2): New function.
24756 (selftest::diagnostic_show_locus_c_tests): Call it.
24758 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24760 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24761 early if there is no lhs.
24763 2017-07-13 Martin Liska <mliska@suse.cz>
24765 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24766 (gen_reference_type_die): Likewise.
24767 * stor-layout.c: Remove Pascal-related comment.
24769 2017-07-13 Martin Liska <mliska@suse.cz>
24771 * opts.c (finish_options): Add quotes to error messages.
24772 (parse_sanitizer_options): Likewise.
24774 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24776 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24778 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24780 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24782 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24784 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24786 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24788 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24791 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24792 provides the hardware capability bits, define the macro
24793 __BUILTIN_CPU_SUPPORTS__.
24794 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24795 if GLIBC does not provide the hardware capability bits. Add a
24796 gcc_unreachable call if the built-in cpu function is neither
24797 __builtin_cpu_is nor __builtin_cpu_supports.
24798 (rs6000_get_function_versions_dispatcher): Change the warning
24799 that an old GLIBC is used which does not export the capability
24800 bits to be an error.
24801 * doc/extend.texi (target_clones attribute): Document the
24802 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24803 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24804 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24805 the macros defined by GCC if the newer GLIBC is available.
24807 2017-07-12 Jeff Law <law@redhat.com>
24809 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24810 remaining includes slightly.
24811 * config/riscv/riscv-builtins.c: Include profile-count.h.
24813 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24816 * config/avr/avr.c (avr_set_current_function): In diagnostic
24817 messages: Quote keywords and (parts of) identifiers.
24818 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24821 2017-07-12 Carl Love <cel@us.ibm.com>
24823 * config/rs6000/rs6000-c.c: Add support for built-in functions
24824 vector bool char vec_revb (vector bool char);
24825 vector bool short vec_revb (vector short char);
24826 vector bool int vec_revb (vector bool int);
24827 vector bool long long vec_revb (vector bool long long);
24828 * doc/extend.texi: Update the built-in documentation file for the
24829 new built-in functions.
24831 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24833 * config/s390/s390.md: Remove movcc splitter.
24835 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24837 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24838 load/store on condition.
24840 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24843 * config/avr/avr.c (avr_encode_section_info)
24844 [progmem && !TREE_READONLY]: Error if progmem object needs
24847 2017-07-11 Michael Collison <michael.collison@arm.com>
24849 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24852 2017-07-11 Carl Love <cel@us.ibm.com>
24854 * config/rs6000/rs6000-c.c: Add support for builtins
24855 vector unsigned int vec_parity_lsbb (vector signed int);
24856 vector unsigned int vec_parity_lsbb (vector unsigned int);
24857 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24858 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24859 vector unsigned long long vec_parity_lsbb (vector signed long long);
24860 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24861 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24862 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24863 * doc/extend.texi: Update the built-in documentation file for the
24864 new built-in functions.
24866 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24868 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24869 (layout::m_primary_loc): New field.
24870 (layout::layout): Initialize new field. Move location filtering
24871 logic from here to...
24872 (layout::maybe_add_location_range): ...this new method. Add
24873 support for filtering to just the lines already specified by other
24875 (layout::will_show_line_p): New method.
24876 (gcc_rich_location::add_location_if_nearby): New method.
24877 (selftest::test_add_location_if_nearby): New test function.
24878 (selftest::diagnostic_show_locus_c_tests): Call it.
24879 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24882 2017-07-11 Tom de Vries <tom@codesourcery.com>
24884 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24885 (bb_first_real_insn): New function.
24886 (nvptx_single): Add extra initialization of broadcasted condition
24889 2017-07-11 Nathan Sidwell <nathan@acm.org>
24891 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24893 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24895 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24896 Binutils doc as TEXI will mess them up.
24897 * doc/invoke.texi (AVR Options): Same here.
24899 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24901 * config/sparc/sparc.opt (mfix-ut700): New option.
24902 (mfix-gr712rc): Likewise.
24903 (sparc_fix_b2bst): New variable.
24904 * doc/invoke.texi (SPARC options): Document them.
24905 (ARM options): Fix warnings.
24906 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24907 instructions to prevent sequences that can trigger the store-store
24908 errata for certain LEON3FT processors.
24909 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24910 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24911 * config/sparc/sparc.md (fix_b2bst): New attribute.
24912 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24914 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24917 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24919 (*rsqrtsf2_sse): Ditto.
24921 (div<mode>3): Macroize insn from divdf3 and divsf3
24922 using MODEF mode iterator.
24924 2017-07-10 Martin Sebor <msebor@redhat.com>
24926 PR tree-optimization/80397
24927 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24928 instead of testing for equality to INTEGER_TYPE.
24930 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24932 * config.gcc: Remove uclibc from arc target spec.
24934 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24936 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24938 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24941 * lto-wrapper.c (remove_option): New function.
24942 (merge_and_complain): Merge PIC/PIE options more realistically.
24944 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24946 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24950 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24951 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24952 * config.in: Regenerate.
24953 * configure: Regenerate.
24954 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24955 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24956 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24957 (TARGET_GASISR_PROLOGUES): ...target mask.
24958 * common/config/avr/avr-common.c
24959 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24960 Set -mgas-isr-prologues.
24961 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24962 INSERT_PASS_BEFORE for it.
24963 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24964 * config/avr/avr.c (avr_option_override)
24965 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24966 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24967 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24968 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24969 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24971 (make_avr_pass_pre_proep): New function.
24972 (emit_push_sfr) <treg>: Add argument to function and use it
24973 instead of TMP_REG.
24974 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24975 and set machine->gasisr.yes.
24976 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24977 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24978 __gcc_isr.n_pushed to .L__stack_usage.
24979 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24980 (avr_asm_final_postscan_insn): ...this new static function.
24981 * config/avr/avr.h (machine_function)
24982 <is_no_gccisr, use_L__stack_usage>: New fields.
24983 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24984 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24985 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24986 (gasisr, *gasisr): New expander and insn.
24987 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24988 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24989 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24991 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24993 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24996 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24998 Move jump-tables out of .text again.
25001 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
25002 (ASM_OUTPUT_ADDR_VEC): New function.
25003 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
25004 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
25005 INSN_ADDRESSes as asm comment.
25006 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
25007 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
25008 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
25009 * config/avr/avr.md (*tablejump): Adjust comment.
25010 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
25011 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
25013 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
25014 (avr_output_addr_vec): New proto.
25015 (avr_log_t) <insn_addresses>: New field.
25017 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
25020 * config/i386/i386.c (ix86_function_arg_advance): Set
25021 outgoing_args_on_stack to true if there are outgoing arguments
25023 (ix86_function_arg): Likewise.
25024 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
25025 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
25026 * config/i386/i386.h (machine_function): Add
25027 outgoing_args_on_stack.
25029 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
25031 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
25032 supporting pthreds.
25033 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
25035 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
25037 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
25038 (REAL_H): Remove $(MACHMODE_H).
25039 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
25041 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
25042 $(MACHMODE_H) and double-int.h.
25043 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
25045 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
25048 2017-07-07 Andrew Pinski <apinski@cavium.com>
25050 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
25051 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
25053 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
25055 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
25056 Add warning if GCC was not configured to link against a GLIBC that
25057 exports the hardware capability bits.
25058 (make_resolver_func): Make resolver function private and not a
25059 COMDAT function. Create the name with clone_function_name instead
25060 of make_unique_name.
25063 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
25064 correct operand in doing the split.
25066 2017-07-07 Carl Love <cel@us.ibm.com>
25068 * config/rs6000/rs6000-c: Add support for built-in function
25069 vector unsigned short vec_pack_to_short_fp32 (vector float,
25071 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
25072 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
25073 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
25074 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
25075 (convert_4f32_8i16): Add define_expand.
25076 * doc/extend.texi: Update the built-in documentation file for the
25077 new built-in function.
25079 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25081 * config/sparc/m8.md: New file.
25082 * config/sparc/sparc.md: Include m8.md.
25084 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25086 * config/sparc/sparc.opt: New option -mvis4b.
25087 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
25088 (sparc_option_override): Handle VIS4B.
25089 (enum sparc_builtins): Define
25090 SPARC_BUILTIN_DICTUNPACK{8,16,32},
25091 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
25092 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
25093 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
25094 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
25095 (check_constant_argument): New function.
25096 (sparc_vis_init_builtins): Define builtins
25097 __builtin_vis_dictunpack{8,16,32},
25098 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
25099 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
25100 __builtin_vis_fpcmpde{8,16,32}shl and
25101 __builtin_vis_fpcmpur{8,16,32}shl.
25102 (sparc_expand_builtin): Check that the constant operands to
25103 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
25104 constant and in range.
25105 * config/sparc/sparc-c.c (sparc_target_macros): Handle
25107 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
25108 (SPARC_IMM5_P): Likewise.
25109 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
25110 (enabled): Handle vis4b.
25111 (UNSPEC_DICTUNPACK): New unspec.
25112 (UNSPEC_FPCMPSHL): Likewise.
25113 (UNSPEC_FPUCMPSHL): Likewise.
25114 (UNSPEC_FPCMPDESHL): Likewise.
25115 (UNSPEC_FPCMPURSHL): Likewise.
25116 (cpu_feature): New CPU feature `vis4b'.
25117 (dictunpack{8,16,32}): New insns.
25118 (FPCSMODE): New mode iterator.
25119 (fpcscond): New code iterator.
25120 (fpcsucond): Likewise.
25121 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
25122 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
25123 (fpcmpde{8,16,32}{si,di}shl): Likewise.
25124 (fpcmpur{8,16,32}{si,di}shl): Likewise.
25125 * config/sparc/constraints.md: Define constraints `q' for unsigned
25126 2-bit integer constants and `t' for unsigned 5-bit integer
25128 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
25130 (imm5_operand_dictunpack16): Likewise.
25131 (imm5_operand_dictunpack32): Likewise.
25132 (imm2_operand): Likewise.
25133 * doc/invoke.texi (SPARC Options): Document -mvis4b.
25134 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
25135 ditunpack* and fpcmp*shl builtins.
25137 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25139 * config.gcc: Handle m8 in --with-{cpu,tune} options.
25140 * config.in: Add HAVE_AS_SPARC6 define.
25141 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
25143 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
25145 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
25146 (CPP_CPU_SPEC): Handle m8.
25147 (ASM_CPU_SPEC): Likewise.
25148 * config/sparc/sparc-opts.h (enum processor_type): Add
25150 * config/sparc/sparc.c (m8_costs): New struct.
25151 (sparc_option_override): Handle TARGET_CPU_m8.
25152 (sparc32_initialize_trampoline): Likewise.
25153 (sparc64_initialize_trampoline): Likewise.
25154 (sparc_issue_rate): Likewise.
25155 (sparc_register_move_cost): Likewise.
25156 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
25157 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
25158 (ASM_CPU64_DEFAULT_SPEC): Likewise.
25159 (CPP_CPU_SPEC): Handle M8.
25160 (ASM_CPU_SPEC): Likewise.
25161 (AS_M8_FLAG): Define.
25162 * config/sparc/sparc.md: Add m8 to the cpu attribute.
25163 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
25164 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
25166 * configure: Regenerate.
25167 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
25170 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25172 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
25174 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
25175 ("*movdi_insn_sp32"): Do not set v3pipe.
25176 ("*movsi_insn"): Likewise.
25177 ("*movdi_insn_sp64"): Likewise.
25178 ("*movsf_insn"): Likewise.
25179 ("*movdf_insn_sp32"): Likewise.
25180 ("*movdf_insn_sp64"): Likewise.
25181 ("*zero_extendsidi2_insn_sp64"): Likewise.
25182 ("*sign_extendsidi2_insn"): Likewise.
25183 ("*mov<VM32:mode>_insn"): Likewise.
25184 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25185 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25186 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25187 ("<vlop:code><VL:mode>3"): Likewise.
25188 ("*not_<vlop:code><VL:mode>3"): Likewise.
25189 ("*nand<VL:mode>_vis"): Likewise.
25190 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
25191 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
25192 ("one_cmpl<VL:mode>2"): Likewise.
25193 ("faligndata<VM64:mode>_vis"): Likewise.
25194 ("alignaddrsi_vis"): Likewise.
25195 ("alignaddrdi_vis"): Likweise.
25196 ("alignaddrlsi_vis"): Likewise.
25197 ("alignaddrldi_vis"): Likewise.
25198 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25199 ("bmaskdi_vis"): Likewise.
25200 ("bmasksi_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 ("pdistn<P:mode>_vis"): Likewise.
25206 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25208 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25210 * config/sparc/sparc.md ("subtype"): New insn attribute.
25211 ("*wrgsr_sp64"): Set insn subtype.
25212 ("*rdgsr_sp64"): Likewise.
25213 ("alignaddrsi_vis"): Likewise.
25214 ("alignaddrdi_vis"): Likewise.
25215 ("alignaddrlsi_vis"): Likewise.
25216 ("alignaddrldi_vis"): Likewise.
25217 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25218 ("fexpand_vis"): Likewise.
25219 ("fpmerge_vis"): Likewise.
25220 ("faligndata<VM64:mode>_vis"): Likewise.
25221 ("bshuffle<VM64:mode>_vis"): Likewise.
25222 ("cmask8<P:mode>_vis"): Likewise.
25223 ("cmask16<P:mode>_vis"): Likewise.
25224 ("cmask32<P:mode>_vis"): Likewise.
25225 ("fchksm16_vis"): Likewise.
25226 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25227 ("fmean16_vis"): Likewise.
25228 ("fp<plusminus_insn>64_vis"): Likewise.
25229 ("<plusminus_insn>v8qi3"): Likewise.
25230 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25231 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25232 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25233 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25234 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25235 ("*movqi_insn"): Likewise.
25236 ("*movhi_insn"): Likewise.
25237 ("*movsi_insn"): Likewise.
25238 ("movsi_pic_gotdata_op"): Likewise.
25239 ("*movdi_insn_sp32"): Likewise.
25240 ("*movdi_insn_sp64"): Likewise.
25241 ("movdi_pic_gotdata_op"): Likewise.
25242 ("*movsf_insn"): Likewise.
25243 ("*movdf_insn_sp32"): Likewise.
25244 ("*movdf_insn_sp64"): Likewise.
25245 ("*zero_extendhisi2_insn"): Likewise.
25246 ("*zero_extendqihi2_insn"): Likewise.
25247 ("*zero_extendqisi2_insn"): Likewise.
25248 ("*zero_extendqidi2_insn"): Likewise.
25249 ("*zero_extendhidi2_insn"): Likewise.
25250 ("*zero_extendsidi2_insn_sp64"): Likewise.
25251 ("ldfsr"): Likewise.
25252 ("prefetch_64"): Likewise.
25253 ("prefetch_32"): Likewise.
25254 ("tie_ld32"): Likewise.
25255 ("tie_ld64"): Likewise.
25256 ("*tldo_ldub_sp32"): Likewise.
25257 ("*tldo_ldub1_sp32"): Likewise.
25258 ("*tldo_ldub2_sp32"): Likewise.
25259 ("*tldo_ldub_sp64"): Likewise.
25260 ("*tldo_ldub1_sp64"): Likewise.
25261 ("*tldo_ldub2_sp64"): Likewise.
25262 ("*tldo_ldub3_sp64"): Likewise.
25263 ("*tldo_lduh_sp32"): Likewise.
25264 ("*tldo_lduh1_sp32"): Likewise.
25265 ("*tldo_lduh_sp64"): Likewise.
25266 ("*tldo_lduh1_sp64"): Likewise.
25267 ("*tldo_lduh2_sp64"): Likewise.
25268 ("*tldo_lduw_sp32"): Likewise.
25269 ("*tldo_lduw_sp64"): Likewise.
25270 ("*tldo_lduw1_sp64"): Likewise.
25271 ("*tldo_ldx_sp64"): Likewise.
25272 ("*mov<VM32:mode>_insn"): Likewise.
25273 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25274 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25276 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25278 * config/sparc/sparc.md ("type"): New insn type viscmp.
25279 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25281 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25282 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25283 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25284 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25286 ("n7_vis_logical_11cycle"): Likewise.
25287 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25288 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25289 * config/sparc/niagara.md ("niag_vis"): Likewise.
25290 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25291 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25293 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25295 * config/sparc/sparc.md: New instruction type `bmask'.
25296 (bmaskdi_vis): Use the `bmask' type.
25297 (bmasksi_vis): Likewise.
25298 * config/sparc/ultra3.md (us3_array): Likewise.
25299 * config/sparc/niagara7.md (n7_array): Likewise.
25300 * config/sparc/niagara4.md (n4_array): Likewise.
25301 * config/sparc/niagara2.md (niag2_vis): Likewise.
25302 (niag3_vis): Likewise.
25303 * config/sparc/niagara.md (niag_vis): Likewise.
25305 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25307 * ipa-comdats.c: Remove optimize check from gate.
25308 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25309 for functions not optimized.
25310 (ipa_fn_summary_read): Skip optimize check.
25311 (ipa_fn_summary_write): Likewise.
25312 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25314 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25316 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25317 (check_callers): Give up on caller which is not optimized.
25318 (inline_small_functions): Likewise.
25319 (ipa_inline): Do not give up when not optimizing.
25320 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25321 away unoptimizes cdtors.
25322 (whole_program_function_and_variable_visibility): Do
25323 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25324 * ipa.c (process_references): Do not check optimize.
25325 (symbol_table::remove_unreachable_nodes): Update optimize check.
25326 (set_writeonly_bit): Update optimize check.
25327 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25328 (pass_ipa_single_use::gate): Remove.
25330 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25332 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25333 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25334 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25335 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25336 permute_load, permute_store, adjust_extract, adjust_splat,
25337 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25338 replace_swap_with_copy, dump_swap_insn_table,
25339 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25340 recombine_lvx_pattern, recombine_stvx_pattern,
25341 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25342 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25343 to file rs6000-p8swap.c.
25344 * config/rs6000/rs6000-p8swap.c: New file.
25345 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25346 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25347 and rs6000*-*-* targets.
25349 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25351 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25353 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25355 * lto-wrapper.c (merge_and_complain): Do not merge
25356 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25357 fsigned_zeros, ftrapping_math, fwrapv.
25358 (append_compiler_options): Do not track these options.
25359 (append_linker_options): Likewie
25361 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25363 * cgraphunit.c (cgraph_node::finalize_function): When
25364 !flag_toplevel_reorde set no_reorder flag.
25365 (varpool_node::finalize_decl): Likewise.
25366 (symbol_table::compile): Drop no toplevel reorder path.
25368 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25370 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25371 edges; zero probability is not better than uninitialized.
25373 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25375 * asan.h (asan_sanitize_allocas_p): Declare.
25376 * asan.c (asan_sanitize_allocas_p): New function.
25377 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25378 (handle_builtin_alloca): Likewise.
25379 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25380 if !asan_sanitize_allocas_p.
25381 * params.def (asan-instrument-allocas): Add new option.
25382 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25383 * opts.c (common_handle_option): Disable allocas sanitization for
25386 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25388 * asan.c: Include gimple-fold.h.
25389 (get_last_alloca_addr): New function.
25390 (handle_builtin_stackrestore): Likewise.
25391 (handle_builtin_alloca): Likewise.
25392 (asan_emit_allocas_unpoison): Likewise.
25393 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25394 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25395 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25396 (instrument_builtin_call): Pass gimple iterator to
25397 get_mem_refs_of_builtin_call.
25398 (last_alloca_addr): New global.
25399 * asan.h (asan_emit_allocas_unpoison): Declare.
25400 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25401 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25402 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25403 if function calls alloca.
25404 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25405 * gimple-fold.h (replace_call_with_value): Declare.
25406 * internal-fn.c: Include asan.h.
25407 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25408 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25410 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25412 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25413 (C_SELFTEST_FLAGS): New.
25414 (CPP_SELFTEST_FLAGS): New.
25415 (SELFTEST_DEPS): New, from deps of s-selftest.
25416 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25417 (CPP_SELFTEST_DEPS): New.
25418 (selftest): Add dependency on s-selftest-c++.
25419 (s-selftest): Rename to...
25420 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25421 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25422 than SELFTEST_FLAGS.
25423 (selftest-gdb): Rename to...
25424 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25426 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25427 (selftest-valgrind): Rename to...
25428 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25430 (selftest-valgrind): Reintroduce as an alias for
25431 selftest-c-valgrind.
25432 (s-selftest-c++): New.
25433 (selftest-c++-gdb): New.
25434 (selftest-c++-valgrind): New.
25436 2017-07-06 Olivier Hainque <hainque@adacore.com>
25438 * gcc.c (process_command): When deciding if undefined variables
25439 should be ignored when processing specs, accept "gcc -v" as well.
25441 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25443 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25444 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25446 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25448 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25449 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25450 * config/arm/arm-tables.opt: Regenerate.
25451 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25453 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25455 2017-07-06 Carl Love <cel@us.ibm.com>
25457 * ChangeLog: Clean up from mid air collision
25459 2017-07-06 Carl Love <cel@us.ibm.com>
25461 * config/rs6000/rs6000-c.c: Add support for built-in functions
25462 vector signed int vec_subc (vector signed int, vector signed int);
25463 vector signed __int128 vec_subc (vector signed __int128,
25464 vector signed __int128);
25465 vector unsigned __int128 vec_subc (vector unsigned __int128,
25466 vector unsigned __int128);
25467 vector signed int vec_sube (vector signed int, vector signed int,
25468 vector signed int);
25469 vector unsigned int vec_sube (vector unsigned int,
25470 vector unsigned int,
25471 vector unsigned int);
25472 vector signed __int128 vec_sube (vector signed __int128,
25473 vector signed __int128,
25474 vector signed__int128);
25475 vector unsigned __int128 vec_sube (vector unsigned __int128,
25476 vector unsigned __int128,
25477 vector unsigned __int128);
25478 vector signed int vec_subec (vector signed int, vector signed int,
25479 vector signed int);
25480 vector unsigned int vec_subec (vector unsigned int,
25481 vector unsigned int,
25482 vector unsigned int);
25483 vector signed __int128 vec_subec (vector signed __int128,
25484 vector signed __int128,
25485 vector signed__int128);
25486 vector unsigned __int128 vec_subec (vector unsigned __int128,
25487 vector unsigned __int128,
25488 vector unsigned __int128);
25489 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25490 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25491 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25492 BU_ALTIVEC_OVERLOAD_X definitions.
25493 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25494 * doc/extend.texi: Update the built-in documentation file for the new
25495 built-in functions.
25497 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25500 * diagnostic-show-locus.c (layout::layout): Use start and finish
25501 spelling location for the start and finish of each range.
25502 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25503 Add unused aspect param.
25504 * input.c (expand_location_1): Add "aspect" param, and use it
25505 to access the correct part of the location.
25506 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25508 (expand_location_to_spelling_point): Likewise.
25509 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25510 param, and pass it to expand_location_1.
25512 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25514 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25515 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25516 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25517 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25518 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25519 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25520 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25521 _mm_maskz_getmant_ss): New intrinsics.
25522 (__builtin_ia32_getexpss128_mask): Changed to ...
25523 __builtin_ia32_getexpss128_round ... this.
25524 (__builtin_ia32_getexpsd128_mask): Changed to ...
25525 __builtin_ia32_getexpsd128_round ... this.
25526 * config/i386/i386-builtin-types.def
25527 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25528 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25529 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25530 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25531 __builtin_ia32_getmantss_mask_round): New builtins.
25532 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25533 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25534 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25535 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25536 * config/i386/sse.md
25537 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25538 avx512f_sgetexp<mode><mask_scalar_name>
25539 <round_saeonly_scalar_name> ... this.
25540 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25541 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25542 vgetexp<ssescalarmodesuffix>
25543 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25544 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25545 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25546 avx512f_vgetmant<mode><mask_scalar_name>
25547 <round_saeonly_scalar_name> ... this.
25548 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25549 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25550 vgetmant<ssescalarmodesuffix>
25551 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25552 %0<mask_scalar_operand4>, %1, %2
25553 <round_saeonly_scalar_mask_op4>, %3} ... this.
25554 * config/i386/subst.md (mask_scalar_operand4,
25555 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25556 round_saeonly_scalar_nimm_predicate): New subst attributes.
25558 2017-07-06 Julia Koval <julia.koval@intel.com>
25560 * config/i386/i386.c (ix86_erase_embedded_rounding):
25561 Remove code for old rounding pattern.
25563 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25565 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25567 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25569 * doc/sourcebuild.texi (Test Directives, Variants of
25570 dg-require-support): Add documentation for dg-require-stack-check.
25572 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25574 * config/i386/subst.md (mask_scalar, round_scalar,
25575 round_saeonly_scalar): New meta-templates.
25576 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25577 round_scalar_mask_operand3, round_scalar_mask_op3,
25578 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25579 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25580 round_saeonly_scalar_constraint,
25581 round_saeonly_scalar_prefix): New subst attribute.
25582 * config/i386/sse.md
25583 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25584 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25585 <round_scalar_name> ... this.
25586 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25587 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25588 <round_scalar_name> ... this.
25589 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25590 <sse>_vm<code><mode>3<mask_scalar_name>
25591 <round_saeonly_scalar_name> ... this.
25592 (v<plusminus_mnemonic><ssescalarmodesuffix>
25593 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25594 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25595 v<plusminus_mnemonic><ssescalarmodesuffix>
25596 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25597 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25598 (v<multdiv_mnemonic><ssescalarmodesuffix>
25599 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25600 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25601 v<multdiv_mnemonic><ssescalarmodesuffix>
25602 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25603 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25604 (v<maxmin_float><ssescalarmodesuffix>
25605 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25606 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25607 v<maxmin_float><ssescalarmodesuffix>
25608 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25609 %0<mask_scalar_operand3>, %1, %<iptr>2
25610 <round_saeonly_scalar_mask_op3>} ... this.
25612 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25614 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25615 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25617 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25618 Alan Hayward <alan.hayward@arm.com>
25619 David Sherwood <david.sherwood@arm.com>
25621 * combine.c (simplify_if_then_else): Remove "enum" before
25623 * compare-elim.c (can_eliminate_compare): Likewise.
25624 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25626 (aarch64_lookup_simd_builtin_type): Likewise.
25627 (aarch64_simd_builtin_type): Likewise.
25628 (aarch64_init_simd_builtin_types): Likewise.
25629 (aarch64_simd_expand_args): Likewise.
25630 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25632 (aarch64_reverse_mask): Likewise.
25633 (aarch64_simd_emit_reg_reg_move): Likewise.
25634 (aarch64_gen_adjusted_ldpstp): Likewise.
25635 (aarch64_ccmp_mode_to_code): Likewise.
25636 (aarch64_operands_ok_for_ldpstp): Likewise.
25637 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25638 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25640 (aarch64_min_divisions_for_recip_mul): Likewise.
25641 (aarch64_reassociation_width): Likewise.
25642 (aarch64_get_condition_code_1): Likewise.
25643 (aarch64_simd_emit_reg_reg_move): Likewise.
25644 (aarch64_simd_attr_length_rglist): Likewise.
25645 (aarch64_reverse_mask): Likewise.
25646 (aarch64_operands_ok_for_ldpstp): Likewise.
25647 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25648 (aarch64_gen_adjusted_ldpstp): Likewise.
25649 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25651 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25652 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25653 (arm_lookup_simd_builtin_type): Likewise.
25654 (arm_simd_builtin_type): Likewise.
25655 (arm_init_simd_builtin_types): Likewise.
25656 (arm_expand_builtin_args): Likewise.
25657 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25658 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25659 (ft32_setup_incoming_varargs): Likewise.
25660 (ft32_function_arg): Likewise.
25661 (ft32_function_arg_advance): Likewise.
25662 (ft32_pass_by_reference): Likewise.
25663 (ft32_arg_partial_bytes): Likewise.
25664 (ft32_valid_pointer_mode): Likewise.
25665 (ft32_addr_space_pointer_mode): Likewise.
25666 (ft32_addr_space_legitimate_address_p): Likewise.
25667 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25669 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25670 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25671 (iamcu_alignment): Likewise.
25672 (canonicalize_vector_int_perm): Likewise.
25673 (ix86_noce_conversion_profitable_p): Likewise.
25674 (ix86_mpx_bound_mode): Likewise.
25675 (ix86_operands_ok_for_move_multiple): Likewise.
25676 * config/microblaze/microblaze-protos.h
25677 (microblaze_expand_conditional_branch_reg): Likewise.
25678 * config/microblaze/microblaze.c
25679 (microblaze_expand_conditional_branch_reg): Likewise.
25680 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25682 (rs6000_reassociation_width): Likewise.
25683 (rs6000_invalid_binary_op): Likewise.
25684 (fusion_p9_p): Likewise.
25685 (emit_fusion_p9_load): Likewise.
25686 (emit_fusion_p9_store): Likewise.
25687 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25689 (riscv_hard_regno_mode_ok_p): Likewise.
25690 (riscv_address_insns): Likewise.
25691 (riscv_split_symbol): Likewise.
25692 (riscv_legitimize_move): Likewise.
25693 (riscv_function_value): Likewise.
25694 (riscv_hard_regno_nregs): Likewise.
25695 (riscv_expand_builtin): Likewise.
25696 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25697 (riscv_build_integer): Likewise.
25698 (riscv_split_integer): Likewise.
25699 (riscv_legitimate_constant_p): Likewise.
25700 (riscv_cannot_force_const_mem): Likewise.
25701 (riscv_regno_mode_ok_for_base_p): Likewise.
25702 (riscv_valid_base_register_p): Likewise.
25703 (riscv_valid_offset_p): Likewise.
25704 (riscv_valid_lo_sum_p): Likewise.
25705 (riscv_classify_address): Likewise.
25706 (riscv_legitimate_address_p): Likewise.
25707 (riscv_address_insns): Likewise.
25708 (riscv_load_store_insns): Likewise.
25709 (riscv_force_binary): Likewise.
25710 (riscv_split_symbol): Likewise.
25711 (riscv_force_address): Likewise.
25712 (riscv_legitimize_address): Likewise.
25713 (riscv_move_integer): Likewise.
25714 (riscv_legitimize_const_move): Likewise.
25715 (riscv_legitimize_move): Likewise.
25716 (riscv_address_cost): Likewise.
25717 (riscv_subword): Likewise.
25718 (riscv_output_move): Likewise.
25719 (riscv_canonicalize_int_order_test): Likewise.
25720 (riscv_emit_int_order_test): Likewise.
25721 (riscv_function_arg_boundary): Likewise.
25722 (riscv_pass_mode_in_fpr_p): Likewise.
25723 (riscv_pass_fpr_single): Likewise.
25724 (riscv_pass_fpr_pair): Likewise.
25725 (riscv_get_arg_info): Likewise.
25726 (riscv_function_arg): Likewise.
25727 (riscv_function_arg_advance): Likewise.
25728 (riscv_arg_partial_bytes): Likewise.
25729 (riscv_function_value): Likewise.
25730 (riscv_pass_by_reference): Likewise.
25731 (riscv_setup_incoming_varargs): Likewise.
25732 (riscv_print_operand): Likewise.
25733 (riscv_elf_select_rtx_section): Likewise.
25734 (riscv_save_restore_reg): Likewise.
25735 (riscv_for_each_saved_reg): Likewise.
25736 (riscv_register_move_cost): Likewise.
25737 (riscv_hard_regno_mode_ok_p): Likewise.
25738 (riscv_hard_regno_nregs): Likewise.
25739 (riscv_class_max_nregs): Likewise.
25740 (riscv_memory_move_cost): Likewise.
25741 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25742 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25743 (rl78_addr_space_address_mode): Likewise.
25744 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25746 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25747 (rs6000_reassociation_width): Likewise.
25748 (rs6000_invalid_binary_op): Likewise.
25749 (fusion_p9_p): Likewise.
25750 (emit_fusion_p9_load): Likewise.
25751 (emit_fusion_p9_store): Likewise.
25752 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25753 (ok_for_simple_move_operands): Likewise.
25754 (ok_for_simple_move_strict_operands): Likewise.
25755 (ok_for_simple_arith_logic_operands): Likewise.
25756 (visium_legitimize_reload_address): Likewise.
25757 (visium_select_cc_mode): Likewise.
25758 (output_cbranch): Likewise.
25759 (visium_split_double_move): Likewise.
25760 (visium_expand_copysign): Likewise.
25761 (visium_expand_int_cstore): Likewise.
25762 (visium_expand_fp_cstore): Likewise.
25763 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25764 (visium_function_arg): Likewise.
25765 (visium_function_arg_advance): Likewise.
25766 (visium_libcall_value): Likewise.
25767 (visium_setup_incoming_varargs): Likewise.
25768 (visium_legitimate_constant_p): Likewise.
25769 (visium_legitimate_address_p): Likewise.
25770 (visium_legitimize_address): Likewise.
25771 (visium_secondary_reload): Likewise.
25772 (visium_register_move_cost): Likewise.
25773 (visium_memory_move_cost): Likewise.
25774 (prepare_move_operands): Likewise.
25775 (ok_for_simple_move_operands): Likewise.
25776 (ok_for_simple_move_strict_operands): Likewise.
25777 (ok_for_simple_arith_logic_operands): Likewise.
25778 (visium_function_value_1): Likewise.
25779 (rtx_ok_for_offset_p): Likewise.
25780 (visium_legitimize_reload_address): Likewise.
25781 (visium_split_double_move): Likewise.
25782 (visium_expand_copysign): Likewise.
25783 (visium_expand_int_cstore): Likewise.
25784 (visium_expand_fp_cstore): Likewise.
25785 (visium_split_cstore): Likewise.
25786 (visium_select_cc_mode): Likewise.
25787 (visium_split_cbranch): Likewise.
25788 (output_cbranch): Likewise.
25789 (visium_print_operand_address): Likewise.
25790 * expmed.c (flip_storage_order): Likewise.
25791 * expmed.h (emit_cstore): Likewise.
25792 (flip_storage_order): Likewise.
25793 * genrecog.c (validate_pattern): Likewise.
25794 * hsa-gen.c (gen_hsa_addr): Likewise.
25795 * internal-fn.c (expand_arith_overflow): Likewise.
25796 * ira-color.c (allocno_copy_cost_saving): Likewise.
25797 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25798 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25799 (process_invariant_for_inheritance): Likewise.
25800 * lra-eliminations.c (move_plus_up): Likewise.
25801 * omp-low.c (lower_oacc_reductions): Likewise.
25802 * simplify-rtx.c (simplify_subreg): Likewise.
25803 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25804 (TARGET_CHKP_BOUND_MODE): Likewise..
25805 * targhooks.c (default_chkp_bound_mode): Likewise.
25806 (default_setup_incoming_vararg_bounds): Likewise.
25807 * targhooks.h (default_chkp_bound_mode): Likewise.
25808 (default_setup_incoming_vararg_bounds): Likewise.
25809 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25810 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25811 (have_whole_vector_shift): Likewise.
25812 * tree-vect-stmts.c (vectorizable_load): Likewise.
25813 * doc/tm.texi: Regenerate.
25815 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25817 Graceful degrade if Binutils PR21472 is not available.
25820 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25821 .rodata in flash test fails.
25822 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25823 * confgure: Regenerate.
25824 * config.in: Regenerate.
25825 * config/avr/avr.c (avr_asm_named_section)
25826 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25827 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25828 (avr_asm_init_sections): Same.
25830 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25832 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25833 (fma<VH:mode>4_intrinsic): Likewise.
25834 (*fmsub<VCVTF:mode>4): Likewise.
25835 (*fmsub<VH:mode>4_intrinsic): Likewise.
25837 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25840 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25841 Don't depend on "optimize > 0".
25842 (out_movhi_r_mr, out_movqi_mr_r): Same.
25843 (out_movhi_mr_r, out_movqi_r_mr): Same.
25844 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25845 io_address_operand on "optimize > 0".
25847 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25849 * tree-loop-distribution.c: Add general explanantion on the pass.
25850 (generate_loops_for_partition): Mark distributed loop.
25851 (pg_add_dependence_edges): New parameter. Handle alias data
25852 dependence specially and record it in the parameter if asked.
25853 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25854 (init_partition_graph_vertices, add_partition_graph_edge): New.
25855 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25856 (free_partition_graph_vdata, build_partition_graph): New.
25857 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25858 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25859 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25860 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25861 (version_for_distribution_p, finalize_partitions): New.
25862 (distribute_loop): Handle alias data dependence specially. Factor
25863 out loop fusion code as functions and call these functions.
25865 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25867 * tree-loop-distribution.c (classify_partition): New parameter and
25868 better handle reduction statement.
25869 (rdg_build_partitions): Revise comment.
25870 (distribute_loop): Compute statements in all partitions and pass it
25871 to classify_partition.
25873 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25875 * tree-loop-distribution.c (enum partition_type): New.
25876 (struct partition): New field type.
25877 (partition_merge_into): Add parameter. Update partition type.
25878 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25879 (build_rdg_partition_for_vertex): Compute partition type.
25880 (rdg_build_partitions): Dump partition type.
25881 (distribute_loop): Update calls to partition_merge_into.
25883 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25885 * tree-loop-distribution.c (struct ddr_hasher): New.
25886 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25888 (classify_partition): Call get_data_dependence.
25889 (pg_add_dependence_edges): Ditto.
25890 (distribute_loop): Release data dependence hash table.
25892 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25894 * tree-loop-distribution.c (ref_base_address): Delete.
25895 (similar_memory_accesses): Rename ...
25896 (share_memory_accesses): ... to this. Check if partitions access
25897 the same memory reference.
25898 (distribute_loop): Call share_memory_accesses.
25900 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25902 * tree-loop-distribution.c (struct partition): New field recording
25903 its data reference.
25904 (partition_alloc, partition_free): Init and release data refs.
25905 (partition_merge_into): Merge data refs.
25906 (build_rdg_partition_for_vertex): Collect data refs for partition.
25907 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25909 (distribute_loop): Remve data refs from vertice data of partition
25912 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25914 * tree-loop-distribution.c (params.h): Include header file.
25915 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25916 (datarefs_vec): New global var.
25917 (create_rdg_vertices): Use datarefs_vec directly.
25918 (free_rdg): Don't free data references.
25919 (build_rdg): Update use. Don't free data references.
25920 (distribute_loop): Compute global variable for data references.
25921 Bail out if there are too many data references.
25923 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25925 * tree-loop-distribution.c (loop_nest): New global var.
25926 (build_rdg): Use loop directly, rather than loop nest.
25927 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25929 (distribute_loop): Compute global variable loop nest. Update use.
25931 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25933 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25934 (partition_merge_into): New parameter. Dump reason for fusion.
25935 (distribute_loop): Update use of partition_merge_into.
25937 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25939 * tree-loop-distribution.c (bb_top_order_index): New.
25940 (bb_top_order_index_size, bb_top_order_cmp): New.
25941 (stmts_from_loop): Use topological order.
25942 (pass_loop_distribution::execute): Compute and release topological
25943 order for basic blocks.
25945 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25947 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25950 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25952 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25953 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25954 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25955 * internal-fn.def (LOOP_DIST_ALIAS): New.
25956 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25957 (fold_loop_internal_call): ... this.
25958 (vect_loop_dist_alias_call): New function.
25959 (set_uid_loop_bbs): Call fold_loop_internal_call.
25960 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25963 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25966 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25967 Require dead FLAGS_REG at the beginning of a peephole.
25969 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25972 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25973 arguments in the call to __builtin_ia32_sbb_u32.
25974 (_subborrow_u64): Swap _X and _Y arguments in the call to
25975 __builtin_ia32_sbb_u64.
25977 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25980 * tree-vrp.c (compare_assert_loc): Turn into a function template
25981 with stable template parameter. Only test if a->e is NULL,
25982 !a->e == !b->e has been verified already. Use e == NULL or
25983 e != NULL instead of e or ! e tests. If stable is true, don't use
25984 iterative_hash_expr, on the other side allow a or b or both NULL
25985 and sort the NULLs last.
25986 (process_assert_insertions): Sort using compare_assert_loc<false>
25987 instead of compare_assert_loc, later sort using
25988 compare_assert_loc<true> before calling process_assert_insertions_for
25989 in a loop. Use break instead of continue once seen NULL pointer.
25991 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25993 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25994 Cortex-R7 and Cortex-R8 processors.
25996 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25998 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25999 uninitialized while src is not.
26001 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
26003 * common/config/arm/arm-common.c: Adjust include path for
26005 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
26006 (arm-cpu.h): Create in build directory. Adjust dependency rules.
26007 (arm-cpu-data.h): Likewise.
26008 (arm-cpu-cdata.h): Likewise.
26009 * config/arm/arm-cpu.h: Delete.
26010 * config/arm/arm-cpu-cdata.h: Delete.
26011 * config/arm/arm-cpu-data.h: Delete.
26013 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
26015 * config/arm/arm-cpus.in (cortex-a55): New.
26016 (cortex-a75): Likewise.
26017 (cortex-a75.cortex-a55): Likewise.
26018 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
26020 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
26021 * config/arm/arm-cpu-cdata.h: Regenerate.
26022 * config/arm/arm-cpu-data.h: Regenerate.
26023 * config/arm/arm-cpu.h: Regenerate.
26024 * config/arm/arm-tables.opt: Regenerate.
26025 * config/arm/arm-tune.md: Regenerate.
26027 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
26029 * haifa-sched.c (sched_create_recovery_edges): Update profile.
26031 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
26033 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
26036 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
26038 PR tree-optimization/81292
26039 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
26040 full_string_p, also call adjust_related_strinfos if the adjustment
26041 is simple, otherwise invalidate related strinfos.
26043 2017-07-04 Martin Liska <mliska@suse.cz>
26046 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
26047 newly created variable as DECL_IGNORED_P.
26049 2017-07-04 Martin Liska <mliska@suse.cz>
26052 * ipa-inline.c (inline_small_functions):
26053 Use xstrdup_for_dump.
26055 2017-07-04 Tom de Vries <tom@codesourcery.com>
26057 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
26059 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
26062 * config/darwin.c (darwin_function_switched_text_sections):
26065 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
26067 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
26069 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26071 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
26073 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26075 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
26076 min_profitable_iters, and th as inclusive lower bounds.
26077 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
26078 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
26079 for min_profitable_iters and min_profitable_estimate.
26080 (vect_transform_loop): Treat th as an inclusive lower bound.
26081 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26083 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
26086 * config/darwin.c (darwin_function_switched_text_sections):
26087 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
26088 in two pieces, and suppress the use of buf.
26090 2017-07-03 Nathan Sidwell <nathan@acm.org>
26092 * hash-table.h (hash_table_mod1): Fix indentation.
26094 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26096 PR middle-end/81290
26097 * predict.c (force_edge_cold): Be more careful about propagation
26099 * profile-count.h (profile_probability::guessed,
26100 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
26102 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
26104 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
26106 * doc/invoke.texi (rcpc architecture extension): Document it.
26108 2017-07-03 Richard Biener <rguenther@suse.de>
26110 PR tree-optimization/60510
26111 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
26112 the scalar reduction PHI and use it.
26113 (vectorizable_reduction): Properly guard the single_defuse_cycle
26114 path for non-SLP reduction chains where we cannot use it.
26115 Rework reduc_def/index and vector type deduction. Rework
26116 vector operand gathering during reduction op code-gen.
26117 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
26118 chains dissolve the chain and leave it to non-SLP reduction
26121 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26123 * tree-data-ref.h (dr_alignment): Declare.
26124 * tree-data-ref.c (dr_alignment): New function.
26125 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
26126 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
26128 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
26130 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26132 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
26133 and base_misalignment fields.
26134 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
26135 * tree-data-ref.c: Include builtins.h.
26136 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
26137 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
26138 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
26139 * tree-vect-data-refs.c: Include tree-cfg.h.
26140 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
26141 fields instead of calculating an alignment here.
26142 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
26143 innermost_loop_behavior fields.
26145 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26147 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
26149 (DR_STEP_ALIGNMENT): New macro.
26150 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
26151 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
26152 (create_data_ref): Print it.
26153 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
26154 to tell whether the step preserves vector (mis)alignment.
26155 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26156 Move the check for an integer step and generalise to all INTEGER_CST.
26157 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
26158 Print the outer step alignment.
26160 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26162 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
26163 with offset_alignment.
26164 (DR_ALIGNED_TO): Delete.
26165 (DR_OFFSET_ALIGNMENT): New macro.
26166 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
26167 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
26168 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
26169 (create_data_ref): Likewise.
26170 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26171 (vect_analyze_data_refs): Likewise.
26172 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
26173 creating dummy innermost behavior.
26175 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26177 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
26178 with a "innermost_loop_behavior *" and refeence tree.
26179 * tree-data-ref.c (dr_analyze_innermost): Likewise.
26180 (create_data_ref): Update call accordingly.
26181 * tree-predcom.c (find_looparound_phi): Likewise.
26183 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26185 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
26186 fields with dr_wrt_vec_loop.
26187 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
26188 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
26189 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
26190 (vect_dr_behavior): New function.
26191 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26192 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
26193 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
26194 track whether the step preserves the misalignment.
26195 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26196 Use vect_dr_behavior.
26197 (vect_setup_realignment): Update call accordingly.
26198 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
26199 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
26200 call to vect_create_addr_base_for_vector_ref.
26201 (vect_create_cond_for_align_checks): Likewise.
26202 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
26203 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
26204 (vect_recog_mask_conversion_pattern): Likewise.
26205 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
26206 (new_stmt_vec_info): Remove redundant zeroing.
26208 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26210 * common/config/arm/arm-common.c (arm_be8_option): New function.
26211 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
26212 (ISA_ARMv6): Add isa_bit_be8.
26213 * config/arm/arm.h (arm_be8_option): Add prototype.
26214 (BE8_SPEC_FUNCTION): New define.
26215 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
26216 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
26217 (mlittle-endian): Similarly.
26218 (mbe8, mbe32): New options.
26219 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
26220 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
26222 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26224 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26226 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26228 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26229 (cleanup_tree_cfg_bb): Use it.
26230 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26232 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26234 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26237 * loop-doloop.c (add_test): Update profile.
26239 2017-07-03 Martin Liska <mliska@suse.cz>
26242 * sanopt.c (rewrite_usage_of_param): New function.
26243 (sanitize_rewrite_addressable_params): Likewise.
26244 (pass_sanopt::execute): Call rewrite_usage_of_param.
26246 2017-07-03 Richard Biener <rguenther@suse.de>
26248 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26249 back to using VIEW_CONVERT_EXPR.
26251 2017-07-03 Martin Liska <mliska@suse.cz>
26254 * doc/extend.texi: Document when a resolver function is
26255 generated for target_clones.
26257 2017-07-03 Martin Liska <mliska@suse.cz>
26259 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26260 and shadow memory of auto variables which are subject of
26261 use-after-scope sanitization.
26262 (asan_expand_mark_ifn): Add do set only when is_poison.
26264 2016-07-03 Richard Biener <rguenther@suse.de>
26266 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26268 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26269 (vectorizable_reduction): Perform reduction PHI creation when
26270 visiting a reduction PHI and adjust and simplify code generation
26271 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26272 (vect_transform_loop): Visit reduction PHIs.
26273 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26274 defs into the SLP tree.
26275 (vect_build_slp_tree): Reduction defs terminate the recursion.
26276 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26278 (vect_get_vec_defs_for_stmt_copy): Export.
26279 (vect_get_vec_defs): Likewise.
26280 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26282 (vect_get_vec_defs_for_stmt_copy): Declare.
26283 (vect_get_vec_defs): Likewise.
26285 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26287 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26288 parameter with a "loop" parameter and use it instead of the
26289 loop containing DR_STMT. Don't check simple_iv when doing
26290 BB analysis. Describe the two analysis modes in the comment.
26292 2017-07-03 Tom de Vries <tom@codesourcery.com>
26294 PR tree-optimization/69468
26295 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26296 (find_same_succ_bb): Handle ignore_edge_flags.
26298 2017-07-03 Tom de Vries <tom@codesourcery.com>
26300 PR tree-optimization/81192
26301 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26303 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26305 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26307 2017-07-03 Tom de Vries <tom@codesourcery.com>
26309 PR tree-optimization/81192
26310 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26311 BB_SAME_SUCC (bb) == NULL.
26313 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26315 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26318 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26320 * dumpfile.c: Include profile-count.h
26321 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26323 (insert_cond_bb): Update profile.
26324 * tree-cfg.h (insert_cond_bb): Update prototype.
26325 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26326 * tree-dump.c: Do not include tree-cfg.
26328 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26330 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26332 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26334 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26337 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26339 * tree-complex.c (expand_complex_div_wide): update profile.
26341 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26342 Alan Hayward <alan.hayward@arm.com>
26343 David Sherwood <david.sherwood@arm.com>
26345 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26346 (CORETYPES_H): New define.
26347 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26348 (insn-modes-inline.h, s-modes-inline-h): New rules.
26349 (generated_files): Add insn-modes-inline.h.
26350 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26351 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26352 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26353 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26354 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26355 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26356 (build/gencodes.o, build/genconditions.o): Likewise.
26357 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26358 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26359 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26360 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26361 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26362 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26363 * coretypes.h: Include everything up to real.h for generators.
26364 Include insn-modes.h first. Include wide-int-print.h after
26365 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26366 * machmode.h: Don't include insn-modes.h here.
26367 * function-tests.c: Remove includes of signop.h, machmode.h,
26368 double-int.h and wide-int.h.
26370 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26372 * optc-save-gen.awk: Likewise.
26373 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26374 * godump.c: Remove include of wide-int-print.h.
26375 * pretty-print.h: Likewise.
26376 * wide-int-print.cc: Likewise.
26377 * wide-int.cc: Likewise.
26378 * hash-map-tests.c: Remove include of signop.h.
26379 * hash-set-tests.c: Likewise.
26380 * rtl-tests.c: Likewise.
26381 * mkconfig.sh: Remove include of machmode.h.
26382 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26384 (emit_insn_modes_inline_h): ...this new function. Emit the code
26385 into an insn-modes-inline.h header file, adding appropriate
26386 include guards and end comments.
26387 (emit_insn_modes_c_header): Remove include of machmode.h.
26388 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26390 (main): Handle -i flag and call emit_insn_modes_inline_h when
26393 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26395 * tree-ssa-strlen.c (strinfo): Rename the length field to
26396 nonzero_chars. Add a full_string_p field.
26397 (compare_nonzero_chars, zero_length_string_p): New functions.
26398 (get_addr_stridx): Add an offset_out parameter.
26399 Use compare_nonzero_chars.
26400 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26401 (new_strinfo): Update after above changes to strinfo.
26402 (set_endptr_and_length): Set full_string_p.
26403 (get_string_length): Update after above changes to strinfo.
26404 (unshare_strinfo): Update call to new_strinfo.
26405 (maybe_invalidate): Likewise.
26406 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26407 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26408 as a uhwi instead of an shwi. Update after above changes to
26409 strinfo and new_strinfo.
26410 (zero_length_string): Assert that chainsi contains full strings.
26411 Use zero_length_string_p. Update call to new_strinfo.
26412 (adjust_related_strinfos): Update after above changes to strinfo.
26413 Copy full_string_p from origsi.
26414 (adjust_last_stmt): Use zero_length_string_p.
26415 (handle_builtin_strlen): Update after above changes to strinfo and
26416 new_strinfo. Install the lhs as the string length if the previous
26417 entry didn't describe a full string.
26418 (handle_builtin_strchr): Update after above changes to strinfo
26420 (handle_builtin_strcpy): Likewise.
26421 (handle_builtin_strcat): Likewise.
26422 (handle_builtin_malloc): Likewise.
26423 (handle_pointer_plus): Likewise.
26424 (handle_builtin_memcpy): Likewise. Track nonzero characters
26425 that aren't necessarily followed by a nul terminator.
26426 (handle_char_store): Likewise.
26428 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26430 PR tree-optimization/80769
26431 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26432 for malloc and calloc. Document the new invariant that all related
26433 strinfos have delayed lengths or none do.
26434 (verify_related_strinfos): Move earlier in file.
26435 (set_endptr_and_length): New function, split out from...
26436 (get_string_length): ...here. Also set the lengths of related
26438 (zero_length_string): Assert that chainsi has known (rather than
26440 (adjust_related_strinfos): Likewise.
26442 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26444 PR tree-optimization/81136
26445 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26446 assert that two references with the same misalignment have the same
26447 compile-time misalignment if those compile-time misalignments
26450 2017-07-01 Andi Kleen <ak@linux.intel.com>
26452 * print-tree.c (print_node): Print all attributes.
26454 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26456 * cfg.c (scale_bbs_frequencies): New function.
26457 * cfg.h (scale_bbs_frequencies): Declare it.
26458 * cfgloopanal.c (single_likely_exit): Cleanup.
26459 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26461 (scale_loop_profile): Likewise.
26462 (loop_version): Likewise.
26463 (create_empty_loop_on_edge): Update.
26464 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26465 scale_loop_frequencies, scale_loop_profile, loopify,
26466 loop_version): Update prototypes.
26467 * modulo-sched.c (sms_schedule): Update.
26468 * predict.c (unlikely_executed_edge_p): Also check probability.
26469 (probably_never_executed_edge_p): Fix typo.
26470 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26471 * tree-parloops.c (gen_parallel_loop): Update.
26472 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26473 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26474 * tree-ssa-loop-split.c (split_loop): Update.
26475 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26476 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26477 (vect_loop_versioning): Update.
26478 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26480 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26482 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26484 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26486 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26487 to keep profile consistent.
26489 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26491 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26492 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26493 * profile-count.h (max_safe_multiplier): Make unsigned.
26494 (profile_count::guessed_zero): New.
26496 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26498 * bb-reorder.c (fix_up_crossing_landing_pad,
26499 fix_crossing_conditional_branches): Use make_single_succ_edge
26500 to keep profile consistent.
26502 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26504 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26507 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26510 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26511 the right scopes, make sure cond_jump isn't preserved between multiple
26512 iterations. Search for fallthru edge whenever there are 3+ edges and
26513 use find_fallthru_edge for it.
26515 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26517 Patch by Alexander Monakov <amonakov@ispras.ru>
26518 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26519 probabilities consistently.
26521 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26523 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26524 emit_cmp_and_jump_insns.
26526 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26529 * tree-inline.c (expand_call_inline): Combine profile statuses.
26531 2017-06-30 Andrew Pinski <apinski@cavium.com>
26533 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26534 fold_stmt returned true.
26536 2017-06-30 Nathan Sidwell <nathan@acm.org>
26538 * ggc.h (empty_string): Delete.
26539 * cfgexpand.c (expand_asm_stmt): Use plain "".
26540 * optabs.c (expand_asm_memory_barrier): Likewise.
26541 * stringpool.c (empty_string): Delete.
26542 (digit_vector, digit_string): Delete.
26543 (ggc_alloc_string): Use plain "", don't optimize single digit
26544 strings. Use ggc_alloc_atomic.
26546 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26548 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26549 comparison set and one other set, use the cost of the non-comparison
26552 2017-06-30 Nathan Sidwell <nathan@acm.org>
26554 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26557 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26559 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26560 loops. Remove now unneeded calls to gimple_switch_set_label() that
26561 just set removed labels to NULL_TREE.
26563 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26565 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26566 (set_range_info): ...here. Only call set_range_info_raw if domain
26568 (set_nonzero_bits): Call set_range_info_raw.
26569 * tree-ssanames.h (set_range_info_raw): New.
26571 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26574 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26575 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26576 of nonimmediate_operand and <store_mask_constraint> instead of m
26577 for the input operand. For V8FI iterator, always split if input
26578 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26579 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26580 <store_mask_predicate> instead of register_operand and
26581 <store_mask_constraint> instead of v for the input operand. Make
26582 sure both operands aren't MEMs for if not <mask_applied>.
26584 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26586 * lto-wrapper.c (copy_file) Close both file descriptors before
26589 2017-06-30 Martin Liska <mliska@suse.cz>
26592 * multiple_target.c (create_dispatcher_calls): Make ifunc
26593 also for function that don't have calls or are not referenced.
26595 2017-06-30 Richard Biener <rguenther@suse.de>
26597 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26598 analyze the first scalar stmt. Move vector type computation
26599 for the BB case here from ...
26600 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26601 live operation processing in the SLP case properly.
26603 2017-06-30 Richard Biener <rguenther@suse.de>
26605 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26607 2017-06-30 Martin Liska <mliska@suse.cz>
26610 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26611 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26613 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26615 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26618 2017-06-30 Richard Biener <rguenther@suse.de>
26620 PR tree-optimization/81249
26621 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26622 condition reduction result to original scalar type.
26624 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26626 * profile-count.h (enum profile_quality): Fix typos and whitespace
26629 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26631 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26632 type for branch probabilities.
26634 2017-06-29 Julian Brown <julian@codesourcery.com>
26635 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26637 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26638 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26639 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26640 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26642 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26644 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26645 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26646 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26647 CC usage from generic code to here.
26648 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26649 CC usage into the target macros.
26651 2017-06-29 Maya Rashish <coypu@sdf.org>
26653 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26656 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26658 * arm/arm-builtins.c: Include profile-count.h
26659 * except.c (sjlj_emit_function_enter): Use
26660 profile_probability::unlikely.
26662 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26664 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26665 and tocrel_offset be pointer args rather than implicitly using
26667 (legitimate_constant_pool_address_p, rs6000_emit_move,
26668 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26669 tocrel_offset and use in toc_relative_expr_p call.
26670 (print_operand, print_operand_address): Use static tocrel_base_oac
26671 and tocrel_offset_oac.
26672 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26675 2017-06-29 Maya Rashish <coypu@sdf.org>
26677 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26679 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26681 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26682 objects, take into account only the alignment of 'op0' and 'mode1' if
26685 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26687 * ccmp.c (ccmp_tree_comparison_p): New function.
26688 (ccmp_candidate_p): Update to use above function.
26689 (get_compare_parts): New function.
26690 (expand_ccmp_next): Update to use new functions.
26691 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26693 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26694 take mode as argument.
26695 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26696 * expr.c (expand_expr_real_1): Pass mode as argument.
26698 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26700 * combine.c (combine_instructions): Print insns to dump_file, together
26703 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26705 * asan.c (asan_emit_stack_protection): Update.
26706 (create_cond_insert_point): Update.
26707 * auto-profile.c (afdo_propagate_circuit): Update.
26708 * basic-block.h (struct edge_def): Turn probability to
26709 profile_probability.
26710 (EDGE_FREQUENCY): Update.
26711 * bb-reorder.c (find_traces_1_round): Update.
26712 (better_edge_p): Update.
26713 (sanitize_hot_paths): Update.
26714 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26715 (make_single_succ_edge): Update.
26716 (check_bb_profile): Update.
26717 (dump_edge_info): Update.
26718 (update_bb_profile_for_threading): Update.
26719 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26720 probabilitycount to 0.
26721 * cfgbuild.c (compute_outgoing_frequencies): Update.
26722 * cfgcleanup.c (try_forward_edges): Update.
26723 (outgoing_edges_match): Update.
26724 (try_crossjump_to_edge): Update.
26725 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26726 (expand_gimple_tailcall): Update.
26727 (construct_init_block): Use make_single_succ_edge.
26728 (construct_exit_block): Use make_single_succ_edge.
26729 * cfghooks.c (verify_flow_info): Update.
26730 (redirect_edge_succ_nodup): Update.
26731 (split_edge): Update.
26732 (account_profile_record): Update.
26733 * cfgloopanal.c (single_likely_exit): Update.
26734 * cfgloopmanip.c (scale_loop_profile): Update.
26735 (set_zero_probability): Remove.
26736 (duplicate_loop_to_header_edge): Update.
26737 * cfgloopmanip.h (loop_version): Update prototype.
26738 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26739 (force_nonfallthru_and_redirect): Update.
26740 (update_br_prob_note): Update.
26741 (rtl_verify_edges): Update.
26742 (purge_dead_edges): Update.
26743 (rtl_lv_add_condition_to_bb): Update.
26744 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26745 * cgraphunit.c (init_lowered_empty_function): Update.
26746 (cgraph_node::expand_thunk): Update.
26747 * cilk-common.c: Include profile-count.h
26748 * dojump.c (inv): Remove.
26749 (jumpifnot): Update.
26750 (jumpifnot_1): Update.
26751 (do_jump_1): Update.
26753 (do_jump_by_parts_greater_rtx): Update.
26754 (do_compare_rtx_and_jump): Update.
26755 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26756 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26757 * dwarf2cfi.c: Include profile-count.h
26758 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26759 (sjlj_emit_dispatch_table): Likewise.
26760 * explow.c: Include profile-count.h
26761 * expmed.c (emit_store_flag_force): Update.
26762 (do_cmp_and_jump): Update.
26763 * expr.c (compare_by_pieces_d::generate): Update.
26764 (compare_by_pieces_d::finish_mode): Update.
26765 (emit_block_move_via_loop): Update.
26766 (store_expr_with_bounds): Update.
26767 (store_constructor): Update.
26768 (expand_expr_real_2): Update.
26769 (expand_expr_real_1): Update.
26770 * expr.h (try_casesi, try_tablejump): Update prototypes.
26771 * gimple-pretty-print.c (dump_probability): Update.
26772 (dump_profile): New.
26773 (dump_gimple_label): Update.
26774 (dump_gimple_bb_header): Update.
26775 * graph.c (draw_cfg_node_succ_edges): Update.
26776 * hsa-gen.c (convert_switch_statements): Update.
26777 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26778 (find_if_case_1): Update.
26779 (find_if_case_2): Update.
26780 * internal-fn.c (expand_arith_overflow_result_store): Update.
26781 (expand_addsub_overflow): Update.
26782 (expand_neg_overflow): Update.
26783 (expand_mul_overflow): Update.
26784 (expand_vector_ubsan_overflow): Update.
26785 * ipa-cp.c (good_cloning_opportunity_p): Update.
26786 * ipa-split.c (split_function): Use make_single_succ_edge.
26787 * ipa-utils.c (ipa_merge_profiles): Update.
26788 * loop-doloop.c (add_test): Update.
26789 (doloop_modify): Update.
26790 * loop-unroll.c (compare_and_jump_seq): Update.
26791 (unroll_loop_runtime_iterations): Update.
26792 * lra-constraints.c (lra_inheritance): Update.
26793 * lto-streamer-in.c (input_cfg): Update.
26794 * lto-streamer-out.c (output_cfg): Update.
26795 * mcf.c (adjust_cfg_counts): Update.
26796 * modulo-sched.c (sms_schedule): Update.
26797 * omp-expand.c (expand_omp_for_init_counts): Update.
26798 (extract_omp_for_update_vars): Update.
26799 (expand_omp_ordered_sink): Update.
26800 (expand_omp_for_ordered_loops): Update.
26801 (expand_omp_for_generic): Update.
26802 (expand_omp_for_static_nochunk): Update.
26803 (expand_omp_for_static_chunk): Update.
26804 (expand_cilk_for): Update.
26805 (expand_omp_simd): Update.
26806 (expand_omp_taskloop_for_outer): Update.
26807 (expand_omp_taskloop_for_inner): Update.
26808 * omp-simd-clone.c (simd_clone_adjust): Update.
26809 * optabs.c (expand_doubleword_shift): Update.
26810 (expand_abs): Update.
26811 (emit_cmp_and_jump_insn_1): Update.
26812 (expand_compare_and_swap_loop): Update.
26813 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26814 * predict.c (predictable_edge_p): Update.
26815 (edge_probability_reliable_p): Update.
26816 (set_even_probabilities): Update.
26817 (combine_predictions_for_insn): Update.
26818 (combine_predictions_for_bb): Update.
26819 (propagate_freq): Update.
26820 (estimate_bb_frequencies): Update.
26821 (force_edge_cold): Update.
26822 * profile-count.c (profile_count::dump): Add missing space into dump.
26823 (profile_count::debug): Add newline.
26824 (profile_count::differs_from_p): Explicitly convert to unsigned.
26825 (profile_count::stream_in): Update.
26826 (profile_probability::dump): New member function.
26827 (profile_probability::debug): New member function.
26828 (profile_probability::differs_from_p): New member function.
26829 (profile_probability::differs_lot_from_p): New member function.
26830 (profile_probability::stream_in): New member function.
26831 (profile_probability::stream_out): New member function.
26832 * profile-count.h (profile_count_quality): Rename to ...
26833 (profile_quality): ... this one.
26834 (profile_probability): New.
26835 (profile_count): Update.
26836 * profile.c (compute_branch_probabilities): Update.
26837 * recog.c (peep2_attempt): Update.
26838 * sched-ebb.c (schedule_ebbs): Update.
26839 * sched-rgn.c (find_single_block_region): Update.
26840 (compute_dom_prob_ps): Update.
26841 (schedule_region): Update.
26842 * sel-sched-ir.c (compute_succs_info): Update.
26843 * stmt.c (struct case_node): Update.
26844 (do_jump_if_equal): Update.
26845 (get_outgoing_edge_probs): Update.
26846 (conditional_probability): Update.
26847 (emit_case_dispatch_table): Update.
26848 (expand_case): Update.
26849 (expand_sjlj_dispatch_table): Update.
26850 (emit_case_nodes): Update.
26851 * targhooks.c: Update.
26852 * tracer.c (better_p): Update.
26853 (find_best_successor): Update.
26854 * trans-mem.c (expand_transaction): Update.
26855 * tree-call-cdce.c: Update.
26856 * tree-cfg.c (gimple_split_edge): Upate.
26857 (move_sese_region_to_fn): Upate.
26858 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26859 * tree-eh.c (lower_resx): Upate.
26860 (cleanup_empty_eh_move_lp): Upate.
26861 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26862 * tree-inline.c (copy_edges_for_bb): Update.
26863 (copy_cfg_body): Update.
26864 * tree-parloops.c (gen_parallel_loop): Update.
26865 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26866 (gimple_gen_time_profiler): Update.
26867 * tree-ssa-dce.c (remove_dead_stmt): Update.
26868 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26869 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26870 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26871 (unloop_loops): Update.
26872 (try_peel_loop): Update.
26873 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26874 * tree-ssa-loop-split.c (connect_loops): Update.
26875 (split_loop): Update.
26876 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26877 (hoist_guard): Update.
26878 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26879 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26880 (value_replacement): Update.
26881 * tree-ssa-reassoc.c (branch_fixup): Update.
26882 * tree-ssa-tail-merge.c (replace_block_by): Update.
26883 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26884 (create_edge_and_update_destination_phis): Update.
26885 (compute_path_counts): Update.
26886 (recompute_probabilities): Update.
26887 (update_joiner_offpath_counts): Update.
26888 (freqs_to_counts_path): Update.
26889 (duplicate_thread_path): Update.
26890 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26891 (struct switch_conv_info): Update.
26892 (gen_inbound_check): Update.
26893 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26894 (vect_do_peeling): Update.
26895 (vect_loop_versioning): Update.
26896 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26897 (optimize_mask_stores): Update.
26898 * ubsan.c (ubsan_expand_null_ifn): Update.
26899 * value-prof.c (gimple_divmod_fixed_value): Update.
26900 (gimple_divmod_fixed_value_transform): Update.
26901 (gimple_mod_pow2): Update.
26902 (gimple_mod_pow2_value_transform): Update.
26903 (gimple_mod_subtract): Update.
26904 (gimple_mod_subtract_transform): Update.
26905 (gimple_ic): Update.
26906 (gimple_stringop_fixed_value): Update.
26907 (gimple_stringops_transform): Update.
26908 * value-prof.h: Update.
26910 2017-06-29 Carl Love <cel@us.ibm.com>
26912 * config/rs6000/rs6000-c.c: Add support for built-in functions
26913 vector signed int vec_signed (vector float);
26914 vector signed long long vec_signed (vector double);
26915 vector signed int vec_signed2 (vector double, vector double);
26916 vector signed int vec_signede (vector double);
26917 vector signed int vec_signedo (vector double);
26918 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26919 instruction generator.
26920 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26921 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26922 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26924 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26925 vunsignede_v2df): Add define_expands.
26926 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26927 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26928 VEC_UNSIGNEDO): Add definitions.
26929 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26930 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26931 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26932 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26933 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26934 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26935 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26936 vec_unsignede, vec_unsignedo): Add builtin defines.
26937 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26939 * doc/extend.texi: Update the built-in documentation file for the
26940 new built-in functions.
26942 2017-06-29 Richard Biener <rguenther@suse.de>
26944 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26945 reduction chains to LOOP_VINFO_REDUCTIONS.
26946 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26947 SLP reductions after processing reduction chains.
26949 2017-06-29 Nathan Sidwell <nathan@acm.org>
26951 * builtins.c (fold_builtin_FUNCTION): Use
26952 lang_hooks.decl_printable_name.
26954 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26956 PR middle-end/81194
26957 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26958 with only one label.
26959 * stmt.c (expand_case): Assert NCASES is greater than one.
26961 2017-06-29 Richard Biener <rguenther@suse.de>
26963 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26965 (group_case_labels): Likewise.
26966 (find_taken_edge): Push sanity checking on val to workers...
26967 (find_taken_edge_cond_expr): ... here
26968 (find_taken_edge_switch_expr): ... and here, handle cases
26969 with just a default label.
26970 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26971 (group_case_labels): Likewise.
26972 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26973 group_case_labels does anything cleanup the CFG again.
26975 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26977 PR tree-optimization/81196
26978 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26979 exit condition comparing two IVs.
26981 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26983 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26984 profile to the dummy entry at the end of the list of architectures.
26985 * config/arm/arm-cpu-cdata.h: Regenerated.
26987 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26988 Michael Collison <michael.collison@arm.com>
26991 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26993 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26994 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26995 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26996 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26997 of shift when the shift amount is masked with constant equal to
26998 the size of the mode.
26999 * config/aarch64/predicates.md (subreg_lowpart_operator): New
27002 2017-06-29 Martin Liska <mliska@suse.cz>
27004 * config/i386/i386.opt: Change range from [1,5] to [0,5].
27006 2017-06-29 Yury Gribov <tetra2005@gmail.com>
27009 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
27012 (edge_growth_cache_entry::edge_growth_cache_entry): New
27014 (reset_edge_growth_cache): Update to use constructor.
27016 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27018 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
27019 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
27020 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
27022 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
27024 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
27025 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
27027 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
27029 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
27030 (*-linux-uclibc*): Add t-uclibc tmake_file.
27031 * config/t-musl: New.
27032 * config/t-uclibc: New.
27034 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
27036 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
27038 (gen_comm_data): Emit architectural setting of arch_prof.
27039 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
27041 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
27042 (armv8-m.base, armv8-m.main): Likewise.
27043 * arm-protos.h (arm_build_target): Add profile field.
27044 (arch_option): Likewise.
27045 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
27047 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
27048 arm_active_target.profile.
27050 2017-06-28 Richard Biener <rguenther@suse.de>
27052 PR middle-end/81227
27053 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
27054 TYPE_OVERFLOW_WRAPS.
27055 * match.pd (negate_expr_p): Likewise.
27056 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
27057 fold_build2, not fold_binary.
27059 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
27061 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
27062 Convert memory address to Pmode.
27063 (aarch64_print_operand): Assert MEM operands are always Pmode.
27065 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
27068 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
27069 Remove redundant if.
27070 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
27071 * config/arm/aarch-common-protos.h
27072 (aarch_forward_to_shift_is_not_shifted_re): Remove.
27073 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
27075 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
27078 * multiple_target.c (create_dispatcher_calls): Set the default
27079 clone to be static, not public.
27081 2017-06-28 Richard Biener <rguenther@suse.de>
27083 * tree-vect-loop.c (vectorizable_reduction): Move special
27084 cond reduction IV var creation ...
27085 (vect_create_epilog_for_reduction): ... here. Remove induction_index
27086 parameter. Use STMT_VINFO_VECTYPE.
27087 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
27090 2017-06-28 Martin Liska <mliska@suse.cz>
27093 * ipa-visibility.c (non_local_p): Handle visibility.
27095 2017-06-28 Martin Liska <mliska@suse.cz>
27098 * common.opt: Add IntegerRange to various options.
27099 * opt-functions.awk (integer_range_info): New function.
27100 * optc-gen.awk: Add integer_range_info to cl_options struct.
27101 * opts-common.c (decode_cmdline_option): Handle
27102 CL_ERR_INT_RANGE_ARG.
27103 (cmdline_handle_error): Likewise.
27104 * opts.c (print_filtered_help): Show valid interval in
27105 when --help is provided.
27106 * opts.h (struct cl_option): Add range_min and range_max fields.
27107 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
27109 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
27111 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
27112 (x * C EQ/NE y * C): New transformation.
27114 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
27116 * genmultilib (combination_space): Accept '+' in option names.
27118 2017-06-28 Martin Liska <mliska@suse.cz>
27121 * asan.c (instrument_derefs): Bail out inner references
27122 that are hard register variables.
27124 2017-06-28 Jakub Jelinek <jakub@redhat.com>
27127 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
27128 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
27130 2017-06-28 Richard Biener <rguenther@suse.de>
27132 * tree-vectorizer.h (vect_get_vec_defs): Remove.
27133 (vect_get_slp_defs): Adjust.
27134 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
27136 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
27138 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
27139 get_initial_defs_for_reduction instead of vect_get_vec_defs.
27140 (vectorizable_reduction): Adjust.
27141 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
27143 (vect_get_slp_defs): Likewise.
27144 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
27145 (vectorizable_bswap): Adjust.
27146 (vectorizable_call): Likewise.
27147 (vectorizable_conversion): Likewise.
27148 (vectorizable_assignment): Likewise.
27149 (vectorizable_shift): Likewise.
27150 (vectorizable_operation): Likewise.
27151 (vectorizable_store): Likewise.
27152 (vectorizable_condition): Likewise.
27153 (vectorizable_comparison): Likewise.
27155 2017-06-28 Michael Collison <michael.collison@arm.com>
27158 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
27160 (arm_gen_movmemqi): Removed unused variable 'i'.
27161 Convert 'for' loop into 'while' loop.
27162 (arm_expand_prologue): Remove last unnecessary set of insn.
27163 (thumb_pop): Remove unused variable 'pushed_words'.
27164 (thumb_exit): Remove last unnecessary set of regs_to_pop.
27166 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27168 * config/s390/predicates.md: Use s390_rel_address_ok_p.
27169 * config/s390/s390-protos.h: Add prototype of
27170 s390_rel_address_ok_p.
27171 * config/s390/s390.c (s390_got_symbol): New function.
27172 (s390_rel_address_ok_p): New function.
27173 (legitimize_pic_address): Use s390_rel_address_ok_p.
27174 (s390_load_got): Use s390_got_symbol.
27175 (s390_option_override): Issue error if
27176 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
27177 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
27179 * config/s390/s390.opt: New option mpic-data-is-text-relative.
27181 2017-06-27 Andrew Pinski <apinski@cavium.com>
27183 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
27184 (X * copysign (1.0, X)): New pattern.
27185 (X * copysign (1.0, -X)): New pattern.
27186 (copysign (-1.0, CST)): New pattern.
27188 2017-06-27 Joseph Myers <joseph@codesourcery.com>
27190 * genmultilib (combination_space): Remove variable.
27191 Validate reuse rules against regular expression for any sequence
27192 of multilib options in any order.
27194 2017-06-27 Michael Collison <michael.collison@arm.com>
27196 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
27197 call aarch64_split_simd_combine.
27198 * (aarch64_combine_internal<mode>): Delete pattern.
27199 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
27200 Allow register and subreg operands.
27202 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27204 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
27205 specific need, just fallback on defaults.
27206 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
27208 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27209 Olivier Hainque <hainque@adacore.com>
27211 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
27213 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
27214 targets. Pick a default if no particular attempt applied.
27215 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
27218 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27220 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
27221 (x86_64-wrs-vxworks7): Likewise.
27223 2017-06-27 Marek Polacek <polacek@redhat.com>
27226 * ubsan.c (instrument_null): Check get_base_address's result for null.
27228 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27230 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27232 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27234 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27235 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27236 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27237 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27238 New function types.
27239 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27240 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27241 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27242 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27243 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27244 BUILT_IN_FEUPDATEENV): New builtins.
27245 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27246 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27247 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27248 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27250 (builtin_structptr_types): Adjust size.
27251 * tree.c (builtin_structptr_types): Add four entries.
27253 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27254 Olivier Hainque <hainque@adacore.com>
27256 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27257 (TLS_SYM): New local macro, forcing reference to __tls__ on
27258 link command lines for VxWorks 7 RTPs, triggering initialization
27260 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27261 OS features TLS support, true for RTPs on VxWorks 7.
27262 * config/vxworks.c (vxworks_override_options): Setup emutls
27265 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27267 * predict.c (test_prediction_value_range): Use -1U instead of -1
27268 to avoid narrowing conversion warning.
27269 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27270 to avoid narrowing conversion warning.
27271 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27273 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27275 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27277 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27278 64bit configurations.
27279 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27280 (SIZE_TYPE): Likewise.
27281 * config/vxworks.c (vxworks_emutls_var_fields): Use
27282 long_unsigned_type_node instead of unsigned_type_node as the offset
27283 field type, which is "pointer" mode in emutls.c.
27285 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27288 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27290 PR middle-end/81207
27291 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27292 gimple_vuse copying separately from gimple_vdef copying.
27294 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27296 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27299 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27300 Olivier Hainque <hainque@adacore.com>
27302 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27303 for all vxworks7 targets.
27304 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27305 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27306 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27307 variations for VX6/VX7 and 32/64bits later on in ...
27308 (VXWORKS_LIB_SPEC): Leverage new macros.
27309 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27310 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27312 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27314 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27315 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27317 2017-06-26 Carl Love <cel@us.ibm.com>
27319 * config/rs6000/rs6000-c.c: Add support for built-in functions
27320 vector bool char vec_reve (vector bool char);
27321 vector signed char vec_reve (vector signed char);
27322 vector unsigned char vec_reve (vector unsigned char);
27323 vector bool int vec_reve (vector bool int);
27324 vector signed int vec_reve (vector signed int);
27325 vector unsigned int vec_reve (vector unsigned int);
27326 vector bool long long vec_reve (vector bool long long);
27327 vector signed long long vec_reve (vector signed long long);
27328 vector unsigned long long vec_reve (vector unsigned long long);
27329 vector bool short vec_reve (vector bool short);
27330 vector signed short vec_reve (vector signed short);
27331 vector double vec_reve (vector double);
27332 vector float vec_reve (vector float);
27333 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27334 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27335 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27336 (altivec_vreve): New pattern.
27337 * config/rs6000/altivec.h (vec_reve): New define.
27338 * doc/extend.texi (vec_rev): Update the built-in documentation file
27339 for the new built-in functions.
27341 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27343 PR tree-optimization/71815
27344 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27346 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27348 (slsr_process_phi): Likewise.
27349 (replace_uncond_cands_and_profitable_phis): Don't replace a
27350 multiply candidate with a stride of 1 (copy or cast).
27351 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27353 (lowest_cost_path): Likewise.
27354 (total_savings): Likewise.
27356 2017-06-26 Richard Biener <rguenther@suse.de>
27359 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27360 Use def_builtin_pure for all gather builtins.
27362 2017-06-26 Richard Biener <rguenther@suse.de>
27364 PR tree-optimization/81203
27365 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27366 non-dominating BBs.
27368 2017-06-26 Marek Polacek <polacek@redhat.com>
27371 * doc/invoke.texi: Document -Wmultistatement-macros.
27373 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27375 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27376 arm_neon_ok_no_float_abi effective target.
27378 2017-06-26 Richard Biener <rguenther@suse.de>
27380 PR tree-optimization/80928
27381 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27382 (copy_bbs): Set BB_DUPLICATED flag early.
27383 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27385 (execute_on_shrinking_pred): Likewise.
27386 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27387 BB_DUPLICATED blocks.
27388 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27389 iterate over all PHIs considering removal of *gsi.
27391 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27393 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27396 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27398 * config/rs6000/rs6000-string.c: (expand_block_clear,
27399 do_load_for_compare, select_block_compare_mode,
27400 compute_current_alignment, expand_block_compare,
27401 expand_strncmp_align_check, expand_strn_compare,
27402 expand_block_move, rs6000_output_load_multiple)
27403 Move functions related to string/block move/compare
27404 to a separate file.
27405 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27406 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27407 for this function which is now used in two files.
27408 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27409 * config.gcc: Add rs6000-string.o to extra_objs for
27410 targets powerpc*-*-* and rs6000*-*-*.
27412 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27415 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27416 32-bit, since indexed is not valid for DImode.
27417 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27418 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27419 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27420 (define_peephole2 for Altivec d-form store): Likewise.
27423 * multiple_target.c (create_dispatcher_calls): Only create the
27424 dispatcher call if the function is the default clone of a
27425 versioned function.
27427 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27429 PR middle-end/80902
27430 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27431 a call, force the call to not be a tail call.
27433 2017-06-23 Jeff Law <law@redhat.com>
27435 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27438 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27440 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27441 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27442 handling for early expansion of vector shifts (sl,sr,sra,rl).
27443 (builtin_function_type): Add vector shift right instructions
27444 to the unsigned argument list.
27446 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27448 rtl-optimizatoin/79286
27449 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27450 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27451 trap. PIC register plus a const unspec without offset can never trap.
27453 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27455 * tree.h (builtin_structptr_type): New type.
27456 (builtin_structptr_types): Declare new array.
27457 * tree.c (builtin_structptr_types): New array.
27458 (free_lang_data, build_common_tree_nodes): Use it.
27460 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27463 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27466 2017-06-22 Matt Turner <mattst88@gmail.com>
27468 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27469 Lake models to skylake case. Assume skylake for unknown
27470 models with clflushopt.
27472 2017-06-22 Jeff Law <law@redhat.com>
27474 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27475 frame sizes that do not satisfy aarch64_uimm12_shift.
27477 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27479 * profile-count.h (apply_probability,
27480 apply_scale, probability_in): Fix checks for zero.
27482 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27484 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27485 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27487 2016-06-22 Richard Biener <rguenther@suse.de>
27489 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27490 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27491 REDUC_MAX_EXPR support.
27492 (vectorizable_reduction): Likewise.
27493 (vect_create_epilog_for_reduction): Likewise.
27495 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27497 * match.pd (A / (1 << B) -> A >> B): New.
27498 * generic-match-head.c: Include optabs-tree.h.
27499 * gimple-match-head.c: Likewise.
27500 * optabs-tree.h (target_supports_op_p): New.
27501 * optabs-tree.c (target_supports_op_p): New.
27503 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27505 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27506 $gcc_cv_ld --help output.
27507 (gcc_cv_ld_demangle): Likewise.
27508 (gcc_cv_ld_eh_frame_hdr): Likewise.
27509 (gcc_cv_ld_pie): Likewise.
27510 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27511 (gcc_cv_ld_buildid): Likewise.
27512 (gcc_cv_ld_sysroot): Likewise.
27513 (ld_bndplt_support): Likewise.
27514 (ld_pushpopstate_support): Likewise.
27515 * configure: Regenerate.
27516 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27518 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27521 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27522 operands indexes to avoid gap between operands and match_dups.
27524 2017-06-21 Andrew Pinski <apinski@cavium.com>
27526 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27527 Increment Arith_shift and Arith_shift_reg by 1.
27528 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27530 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27531 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27532 (aarch64_strip_extend): Add new argument and test for it.
27533 (aarch64_cheap_mult_shift_p): New function.
27534 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27535 add a cost if it is true.
27536 Update calls to aarch64_strip_extend.
27537 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27539 2017-06-21 Andrew Pinski <apinski@cavium.com>
27541 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27543 (thunderxt88): Likewise.
27544 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27545 (thunderx_prefetch_tune): New variable.
27546 (thunderx2t99_prefetch_tune): Update for the correct values.
27547 (thunderxt88_tunings): New variable.
27548 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27549 generic_prefetch_tune.
27550 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27552 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27554 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27555 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27556 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27557 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27558 (aarch64_atomic_cas<mode>, GPI): Likewise.
27560 2017-06-21 Martin Liska <mliska@suse.cz>
27562 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27563 statements on cold and hot labels.
27564 * predict.c (tree_estimate_probability_bb): Remove the
27565 prediction from this place.
27567 2017-06-21 Martin Liska <mliska@suse.cz>
27569 PR tree-optimization/79489
27570 * gimplify.c (maybe_add_early_return_predict_stmt): New
27572 (gimplify_return_expr): Call the function.
27573 * predict.c (tree_estimate_probability_bb): Remove handling
27575 * predict.def: Update comment about early return predictor.
27576 * gimple-predict.h (is_gimple_predict): New function.
27577 * predict.def: Change default value of early return to 66.
27578 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27580 * passes.def: Put pass_strip_predict_hints to the beginning of
27583 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27585 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27586 FUNCTION_DECL declarations.
27587 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27589 (dwaf2out_decl): Likewise.
27590 * godump.c (go_early_global_decl): Skip call to the real debug hook
27591 for FUNCTION_DECL declarations.
27592 * passes.c (rest_of_decl_compilation): Skip call to the
27593 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27594 -fdump-go-spec is passed.
27596 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27598 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27599 Reorder for compactness.
27600 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27601 (def_builtin_pure, def_builtin_pure2): New functions.
27602 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27604 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27606 * match.pd (nop_convert): New predicate.
27607 ((A +- CST1) +- CST2): Allow some NOP conversions.
27609 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27612 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27613 with ctors/dtors if GOVD_SHARED is set.
27615 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27617 * config/aarch64/aarch64.md (movti_aarch64):
27618 Emit mov rather than orr.
27619 (movtf_aarch64): Likewise.
27620 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27621 Emit mov rather than orr.
27623 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27625 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27626 Swap alternatives, make integer dup more expensive.
27628 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27630 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27631 Return true for non-tls symbols.
27633 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27635 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27636 (cortex-a75): Likewise.
27637 (cortex-a75.cortex-a55): Likewise.
27638 * config/aarch64/aarch64-tune.md: Regenerate.
27639 * doc/invoke.texi (-mtune): Document new values for -mtune.
27641 2017-06-21 Tom de Vries <tom@codesourcery.com>
27643 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27644 stack_size feature.
27645 (Effective-Target Keywords, Other attributes): Suggest using
27646 dg-add-options stack_size feature to get stack limit in stack_size
27647 effective target documentation.
27649 2017-06-21 Julian Brown <julian@codesourcery.com>
27650 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27652 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27653 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27654 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27656 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27657 attribute type list for neon_multiply.
27658 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27659 attribute type list for neon_multiply.
27660 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27661 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27662 attribute type list for neon_multiply.
27663 * config/arm/types.md (crypto_pmull): Add.
27664 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27665 attribute type list.
27667 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27669 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27672 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27674 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27675 to make sure not to dereference a NULL cost_classes_ptr pointer.
27677 2017-06-20 Carl Love <cel@us.ibm.com>
27679 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27680 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27681 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27682 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27683 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27684 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27685 VMULOSW): New enum "unspec" values.
27686 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27687 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27688 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27689 altivec_vmulosw): New patterns.
27690 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27691 VMULOSW): Add definitions.
27693 2017-06-20 Julia Koval <julia.koval@intel.com>
27695 * config/i386/i386.c: Fix rounding expand for new pattern.
27696 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27698 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27700 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27701 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27703 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27705 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27708 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27710 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27711 architecture, then by implementer ID.
27712 * config/aarch64/aarch64-tune.md: Regenerate.
27714 2017-06-20 Richard Biener <rguenther@suse.de>
27716 PR middle-end/81097
27717 * fold-const.c (split_tree): Fold to type before negating.
27719 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27721 * diagnostic-show-locus.c
27722 (selftest::test_fixit_deletion_affecting_newline): New function.
27723 (selftest::diagnostic_show_locus_c_tests): Call it.
27725 2017-06-20 Andreas Schwab <schwab@suse.de>
27728 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27731 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27733 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27734 __ARM_FEATURE_COPROC according to support.
27736 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27738 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27739 Rewritten to avoid overflow for > 32-bit pointers.
27742 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27743 by removing enum keyword.
27744 (ubsan_type_descriptor): Likewise. Formatting fix.
27747 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27748 splitter): Require TARGET_SSE2 in the condition.
27750 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27753 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27754 for doing vector set of SFmode on ISA 3.0.
27755 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27756 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27758 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27759 SFmode value into a V4SF variable that was extracted from another
27760 V4SF variable without converting the element to double precision
27761 and back to single precision vector format.
27762 (vsx_insert_extract_v4sf_p9_2): Likewise.
27764 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27766 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27767 in UWHI to avoid undefined overflow.
27770 * ubsan.h (enum ubsan_encode_value_phase): New.
27771 (ubsan_encode_value): Change second argument to
27772 enum ubsan_encode_value_phase with default value of
27773 UBSAN_ENCODE_VALUE_GENERIC.
27774 * ubsan.c (ubsan_encode_value): Change second argument to
27775 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27776 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27777 create_tmp_var_raw instead of create_tmp_var and use a
27779 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27780 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27781 ubsan_encode_value callers.
27784 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27785 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27786 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27788 2017-06-19 Richard Biener <rguenther@suse.de>
27790 PR middle-end/81118
27791 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27792 estimates if we changed anything.
27794 2017-06-19 Richard Biener <rguenther@suse.de>
27796 PR tree-optimization/80887
27797 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27798 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27799 simplified lookups, then reset mprts_hook.
27800 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27802 (try_to_simplify): Likewise.
27804 2017-06-19 Martin Liska <mliska@suse.cz>
27807 * gimplify.c (gimplify_switch_expr):
27808 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27810 2017-06-19 Martin Liska <mliska@suse.cz>
27812 * doc/install.texi: Document that PGO runs in 4 stages.
27814 2017-06-19 Martin Liska <mliska@suse.cz>
27817 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27818 to dispatcher function name.
27819 * multiple_target.c (replace_function_decl): New function.
27820 (create_dispatcher_calls): Redirect both edges and references.
27822 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27824 * profile-count.c (profile_count::dump): Dump quality.
27825 (profile_count::differs_from_p): Update for unsigned val.
27826 * profile-count.h (profile_count_quality): New enum.
27827 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27829 2017-06-19 Richard Biener <rguenther@suse.de>
27831 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27832 struct function as arg.
27833 (estimate_numbers_of_iterations): Export overload with loop arg.
27834 (free_numbers_of_iterations_estimates_loop): Use an overload of
27835 free_numbers_of_iterations_estimates instead.
27836 * tree-cfg.c (remove_bb): Adjust.
27837 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27838 * tree-parloops.c (gen_parallel_loop): Likewise.
27839 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27841 (tree_unroll_loops_completely): Likewise.
27842 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27843 Use an overload instead and export.
27844 (estimated_loop_iterations): Adjust.
27845 (max_loop_iterations): Likewise.
27846 (likely_max_loop_iterations): Likewise.
27847 (estimate_numbers_of_iterations): Take struct function as arg
27849 (loop_exits_before_overflow): Adjust.
27850 (free_numbers_of_iterations_estimates_loop): Use an overload.
27851 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27852 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27854 2017-06-19 Richard Biener <rguenther@suse.de>
27857 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27858 RANGE_EXPR conservatively.
27860 2017-06-16 Carl Love <cel@us.ibm.com>
27862 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27863 definitions for vec_float, vec_float2, vec_floato,
27864 vec_floate built-ins.
27865 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27866 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27868 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27869 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27870 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27871 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27872 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27873 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27874 vec_floato): Add builtin defines.
27875 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27876 Update the built-in documentation file for the new built-in
27879 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27881 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27882 (mthumb): Mark as the negative of -marm.
27884 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27886 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27888 (ARM Options, -mtune): Document that this accepts the same
27889 extension options as -mcpu.
27890 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27892 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27894 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27895 permitted extensions.
27897 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27899 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27900 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27901 (armv8-m.main): Add option +nodsp.
27902 * config/arm/arm-cpu-cdata.h: Regenerated.
27904 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27906 * config/arm/t-fuchsia: New file.
27907 * config.gcc (arm*-*-fuchsia*): Use it.
27909 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27911 * config/arm/t-symbian: Rewrite for new option infrastructure.
27913 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27915 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27916 (MULTILIB_REQUIRED): Likewise.
27918 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27920 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27921 (MULTILIB_RESUE): Likewise.
27922 (MULTILIB_MATCHES): Likewise.
27923 (MULTLIB_REQUIRED): Likewise.
27925 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27927 * config/arm/t-rtems: Rewrite for new option framework.
27929 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27931 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27932 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27933 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27934 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27935 * config/arm/t-multilib: ... here.
27936 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27937 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27938 armv7-a and armv8*-a when A-profile libraries have not been built.
27939 * config/arm/t-rmprofile: Rewrite.
27941 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27943 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27944 with a backslash. Remove the backslash after substituting unescaped
27946 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27948 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27950 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27951 the driver to pass through the default setting of -mfloat-abi.
27952 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27954 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27955 (all_feat_combs): New rule.
27956 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27958 * config/arm/t-aprofile: Rewrite.
27960 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27962 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27963 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27964 fpu is not specified by the user/command-line.
27965 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27966 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27967 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27968 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27969 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27970 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27972 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27974 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27975 * config/arm/t-arm-elf: Rewritten.
27977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27979 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27980 have some floating-point instructions.
27981 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27982 (TARGET_MAYBE_HARD_FLOAT): New macro.
27983 * config/arm/arm-builtins.c (arm_init_builtins): Use
27984 TARGET_MAYBE_HARD_FLOAT.
27985 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27987 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27989 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27990 (configargs.h): Include it.
27991 (arm_print_hint_for_fpu_option): New function.
27992 (arm_parse_fpu_option): New function.
27993 (candidate_extension): New class.
27994 (arm_canon_for_multilib): New function.
27995 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27996 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27997 (ARCH_CANONICAL_SPECS): New macro.
27998 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
28000 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28002 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
28003 are set after handling multilib fragments. Set target_cpu_default2
28006 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28008 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
28010 (arm*-*-*): Set target_cpu_default2 to a quoted string.
28011 * config/arm/parsecpu.awk (check_cpu): Validate any extension
28013 (check_arch): Likewise.
28014 * config/arm/arm.c (arm_configure_build_target): Handle
28015 TARGET_CPU_DEFAULT being a string constant. Scan any feature
28016 options in the default.
28018 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28020 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
28021 when an option is an alias of another.
28022 * config/arm/parsecpu.awk (optalias): New parser token.
28023 (gen_comm_data): Mark non-alias options as such. Emit entries
28024 for extension aliases.
28025 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
28026 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
28027 (armv6kz, armv6zk, armv6t2): Likewise.
28028 (armv7): Make vfpv3-d16 an alias.
28029 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
28031 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
28032 Sort in canonical order.
28033 (armv8-a): Sort in canonical order.
28034 (armv8.1-a, armv8.2-a): Likewise.
28035 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
28037 (cortex-a9): Sort in canonical order.
28038 * config/arm/arm.c (selftests.h): Include it.
28039 (arm_test_cpu_arch_data): New function.
28040 (arm_run_self_tests): New function.
28041 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
28042 (targetm): Move declaration to the end of the file.
28043 * arm-cpu-cdata.h: Regenerated.
28045 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28047 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
28048 call to target_mode_check describing the type of option passed.
28049 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
28050 (arm_target_thumb_only): Use arm_parse_arch_option_name or
28051 arm_parse_cpu_option_name to match parameters against list of
28053 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
28054 arm_arch_core_flags data structure.
28055 * config/arm/arm-cpu_cdata.h: Regenerated.
28057 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28059 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
28061 (arm_print_hint_for_cpu_option): Likewise.
28062 (arm_print_hint_for_arch_option): Likewise.
28063 (arm_parse_cpu_option_name): Likewise.
28064 (arm_parse_arch_option_name): Likewise.
28065 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
28066 of entries in the all_fpus list.
28067 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
28068 (arm_parse_cpu_option_name): Declare.
28069 (arm_parse_arch_option_name): Declare.
28070 (arm_parse_option_features): Declare.
28071 (arm_intialize_isa): Declare.
28072 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
28074 (gen_comm_data): ... here. Make definitions non-static.
28075 * config/arm/arm-cpu-data.h: Regenerated.
28076 * config/arm/arm-cpu-cdata.h: Regenerated.
28078 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28080 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
28081 (cpu_arch_extension): New structure.
28082 (cpu_arch_option, arch_option, cpu_option): New structures.
28083 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
28084 architecture types.
28085 (gen_data): Generate new format data tables.
28086 * config/arm/arm.c (cpu_tune): New structure.
28087 (cpu_option, processors): Delete.
28088 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
28089 (arm_print_hint_for_cpu_option): ... this and ...
28090 (arm_print_hint_for_arch_option): ... this.
28091 (arm_parse_arch_cpu_name): Delete. Replace with ...
28092 (arm_parse_cpu_option_name): ... this and ...
28093 (arm_parse_arch_option_name): ... this.
28094 (arm_unrecognized_feature): Change type of target parameter to
28096 (arm_parse_arch_cpu_features): Delete. Replace with ...
28097 (arm_parse_option_features): ... this.
28098 (arm_configure_build_target): Rework to use new configuration data
28100 (arm_print_tune_info): Rework for new configuration data tables.
28101 * config/arm/arm-cpu-data.h: Regenerated.
28102 * config/arm/arm-cpu.h: Regenerated.
28104 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28106 * Makefile.in (OBJS): Move sbitmap.o from here ...
28107 (OBJS-libcommon): ... to here.
28109 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28111 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
28112 (ISA_ALL_CRYPTO): New macro.
28113 (ISA_ALL_SIMD): New macro
28114 (ISA_ALL_FP): New macro.
28115 * config/arm/arm.c (fpu_bitlist): Update initializer.
28116 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
28118 (arm9e): Add fpu. Add option for nofp
28119 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
28120 (arm926ej-s, arm1026ej-s): Likewise.
28121 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
28122 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
28123 neon-fp16, neon-vfpv4, nofp and nosimd.
28124 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
28125 (cortex-a8): Add fpu. Add option for nofp.
28126 (cortex-a9): Add fpu. Add options for nosimd and nofp.
28127 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
28128 (cortex-r4f): Add fpu.
28129 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
28130 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
28132 (cortex-r8): Likewise.
28133 (cortex-m4): Add fpu. Add option for nofp.
28134 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
28135 (cortex-a17.cortex-a7): Likewise.
28136 (cortex-a32): Add fpu. Add options for crypto and nofp.
28137 (cortex-a35, cortex-a53): Likewise.
28138 (cortex-a57): Add fpu. Add option for crypto.
28139 (cortex-a72, cortex-a73): Likewise.
28140 (exynos-m1): Likewise.
28141 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
28142 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
28143 (cortex-m33): Add fpu. Add option for nofp.
28144 * config/arm/arm-cpu-cdata.h: Regenerated
28145 * config/arm/arm-cpu-data.h: Regenerated.
28147 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28149 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
28150 (armv5te, armv5tej): Likewise.
28151 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
28152 (armv7): Add options fp and vfpv3-d16.
28153 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
28154 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
28156 (armv7ve): Likewise.
28157 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
28158 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
28159 (armv8-a): Add nocrypto option.
28160 (armv8.1-a, armv8.2-a): Likewise.
28161 (armv8-m.main): add options fp, fp.dp and nofp.
28163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28165 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
28167 (armv8-a+crc): Delete.
28168 (armv8.1-a): Add options simd, crypto and nofp.
28169 (armv8.2-a): Add options fp16, simd, crypto and nofp.
28170 (armv8.2-a+fp16): Delete.
28171 (armv8-m.main): Add option dsp.
28172 (armv8-m.main+dsp): Delete.
28173 (cortex-a8): Add fpu. Add nofp option.
28174 (cortex-a9): Add fpu. Add nofp and nosimd options.
28175 * config/arm/parsecpu.awk (gen_data): Generate option tables and
28176 link to main cpu and architecture data structures.
28177 (gen_comm_data): Only put isa attributes from the main architecture
28179 (option): New statement for architecture and CPU entries.
28180 * arm.c (struct cpu_option): New structure.
28181 (struct processors): Add entry for options.
28182 (arm_unrecognized_feature): New function.
28183 (arm_parse_arch_cpu_name): Ignore any characters after the first
28185 (arm_parse_arch_cpu_feature): New function.
28186 (arm_configure_build_target): Separate out any CPU and architecture
28187 features and parse separately. Don't error out if -mfpu=auto is
28188 used with only an architecture string.
28189 (arm_print_asm_arch_directives): New function.
28190 (arm_file_start): Call it.
28191 * config/arm/arm-cpu-cdata.h: Regenerated.
28192 * config/arm/arm-cpu-data.h: Likewise.
28193 * config/arm/arm-tables.opt: Likewise.
28195 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28197 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
28198 assembler when it is not -mfpu=auto.
28200 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28202 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
28203 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
28204 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
28205 (ASM_CPU_SPEC): Rewrite.
28206 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
28207 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
28208 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
28209 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
28210 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
28211 copied string is NUL-terminated. Also strip any characters prefixed
28213 (arm_rewrite_selected_arch): New function.
28214 (arm_rewrite_march): New function.
28216 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28218 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
28219 (x_arm_cpu_string, x_arm_tune_string): Likewise.
28220 (march, mcpu, mtune): Convert to string-based options.
28221 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
28222 (arm_parse_arch_cpu_name): New function.
28223 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28224 identify selected architecture or CPU.
28225 (arm_option_save): New function.
28226 (TARGET_OPTION_SAVE): Redefine.
28227 (arm_option_restore): Restore string options.
28228 (arm_option_print): Print string options.
28230 2017-06-16 Martin Sebor <msebor@redhat.com>
28232 PR tree-optimization/80933
28233 PR tree-optimization/80934
28234 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28235 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28236 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28237 (gimple_fold_builtin): Call them.
28239 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28241 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28242 as unlikely; update profile.
28244 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28246 * predict.c (force_edge_cold): Handle declaring edges impossible
28249 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28251 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28253 (try_unroll_loop_completely): Fix reporting.
28255 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28257 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28259 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28262 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28263 if given a non-constant argument for an intrinsic which requires a
28266 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28268 * profile.c (compare_freqs): New function.
28269 (branch_prob): Sort edge list.
28270 (find_spanning_tree): Assume that the list is priority sorted.
28272 2017-06-16 Richard Biener <rguenther@suse.de>
28274 PR tree-optimization/81090
28275 * passes.def (pass_record_bounds): Remove.
28276 * tree-pass.h (make_pass_record_bounds): Likewise.
28277 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28278 make_pass_record_bounds): Likewise.
28279 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28280 not free niter estimates at the beginning but at the end.
28281 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28283 2017-06-16 Richard Biener <rguenther@suse.de>
28285 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28286 initializer to workaround ICE in host GCC 4.8.
28288 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28290 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28292 (clone_inlined_nodes): Update.
28294 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28296 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28297 prefetch settings, and enable prefetching by default at -O3.
28299 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28301 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28302 Set flag_prefetch_loop_arrays according to tuning data.
28304 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28306 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28307 New tune structure.
28308 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28309 [Unrelated to main purpose of the patch] Place the pointer field last
28310 to enable type checking errors when tune structure are wrongly merged.
28311 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28312 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28313 (thunderx2t99_prefetch_tune): New tune constants.
28314 (tune_params *_tunings): Update all tunings (no functional change).
28315 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28316 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28317 from tunings structures.
28319 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28322 * ubsan.c (instrument_null): Add T argument, use it instead
28323 of computing it based on IS_LHS.
28324 (instrument_object_size): Likewise.
28325 (pass_ubsan::execute): Adjust instrument_null and
28326 instrument_object_size callers to pass gimple_get_lhs or
28327 gimple_assign_rhs1 result to it. Use instrument_null instead of
28328 calling get_base_address and instrument_mem_ref. Handle
28329 aggregate call arguments for object-size sanitization.
28331 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28333 PR tree-optimization/81089
28334 * tree-vrp.c (is_masked_range_test): Validate operands of
28337 2017-06-15 Martin Sebor <msebor@redhat.com>
28340 * dumpfile.c (dump_register): Avoid calling memset to initialize
28341 a class with a default ctor.
28342 * gcc.c (struct compiler): Remove const qualification.
28343 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28344 * hash-table.h: Ditto.
28345 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28347 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28348 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28350 * params.h (struct param_info): Make struct members non-const.
28351 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28352 with default initialization.
28353 * vec.h (vec_copy_construct, vec_default_construct): New helper
28355 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28356 with vec_copy_construct.
28357 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28358 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28359 * doc/invoke.texi (-Wclass-memaccess): Document.
28361 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28363 * emit-rtl.h (is_leaf): Update comment about local
28364 register allocator.
28366 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28369 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28370 for a variable to have a section before checking if the section has a
28372 Set section to.persistent if persistent attribute is set.
28373 Warn if .persistent attribute is used on an automatic variable.
28375 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28377 PR rtl-optimization/80474
28378 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28380 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28382 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28385 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28387 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28388 (casesi): Emit bounds checking as RTL.
28389 (casesi_internal_mips16_<mode>): Remove bounds checking.
28391 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28393 * config/xtensa/xtensa.c (xtensa_option_override): Append
28394 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28395 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28396 xtensa_doloop_hooks): Define unconditionally.
28397 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28399 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28400 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28401 for it in xtensa_option_override.
28402 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28403 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28405 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28407 * doc/cppopts.texi: Document '-' special value to -MF.
28409 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28411 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28412 (cortex_a53_fconst): Likewise.
28413 (cortex_a53_fpmul): Likewise.
28414 (cortex_a53_f_load_64): Likewise.
28415 (cortex_a53_f_load_many): Likewise.
28416 (cortex_a53_advsimd_alu): Likewise.
28417 (cortex_a53_advsimd_alu_q): Likewise.
28418 (cortex_a53_advsimd_mul): Likewise.
28419 (cortex_a53_advsimd_mul_q): Likewise.
28420 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28421 Add missing fmul, r2f_cvt and fconst cases.
28423 2017-06-14 Richard Biener <rguenther@suse.de>
28425 PR middle-end/81088
28426 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28428 (fold_binary_loc): When associating do not treat pre-existing
28429 TREE_OVERFLOW on literal constants as a reason to allow
28430 TREE_OVERFLOW on associated literal constants.
28432 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28434 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28435 (MASK_FEATURES): New macro.
28436 * config/sparc/sparc.c (sparc_option_override): Remove the special
28437 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28439 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28441 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28442 a division of 0 if non-call exceptions are enabled.
28444 2017-06-14 Andrew Pinski <apinski@cavium.com>
28445 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28448 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28449 Improve vector initialization code gen for only variable case.
28451 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28453 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28455 2017-06-14 Richard Biener <rguenther@suse.de>
28457 PR tree-optimization/81083
28458 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28461 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28463 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28464 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28465 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28466 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28467 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28468 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28470 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28472 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28473 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28475 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28477 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28479 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28481 * config/rs6000/t-rtems: Don't handle SPE.
28483 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28485 * config/rs6000/t-linux: Don't handle SPE.
28487 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28489 * config/rs6000/eabispe.h: Delete file.
28491 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28493 * config/rs6000/t-spe: Delete file.
28495 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28497 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28498 (rs6000_legitimate_offset_address_p): Return false for anything in
28499 V2SImode or V2SFmode.
28501 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28503 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28504 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28505 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28507 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28508 (rs6000_legitimate_offset_address_p): Ditto.
28509 (rs6000_emit_move): Ditto.
28510 (rs6000_init_builtins): Remove V4HI_type_node.
28512 2017-06-13 Martin Liska <mliska@suse.cz>
28515 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28516 (gate_asan): Likewise.
28517 * asan.h (asan_no_sanitize_address_p): Remove the function.
28518 (sanitize_flags_p): New function.
28519 * builtins.def: Fix coding style.
28520 * common.opt: Use renamed enum value.
28521 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28522 * doc/extend.texi: Document no_sanitize attribute.
28523 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28524 to SANITIZE_UNDEFINED_NONDEFAULT.
28525 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28526 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28527 Use sanitize_flags_p.
28528 * gimplify.c (gimplify_function_tree): Likewise.
28529 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28530 * opts.c (parse_no_sanitize_attribute): New function.
28531 (common_handle_option): Use renamed enum value.
28532 * opts.h (parse_no_sanitize_attribute): Declare.
28533 * tree.c (sanitize_flags_p): New function.
28534 * tree.h: Declared here.
28535 * tsan.c: Use sanitize_flags_p.
28536 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28537 (instrument_mem_ref): Likewise.
28538 (instrument_bool_enum_load): Likewise.
28539 (do_ubsan_in_current_function): Remove the function.
28540 (pass_ubsan::execute): Use sanitize_flags_p.
28541 * ubsan.h: Remove do_ubsan_in_current_function
28542 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28543 (dump_function_to_file): Use it here.
28545 2017-06-13 Martin Jambor <mjambor@suse.cz>
28547 PR tree-optimization/80803
28548 PR tree-optimization/81063
28549 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28550 (propagate_subaccesses_across_link): Enqueue subtree whenever
28551 necessary instead of relying on the caller.
28553 2017-06-13 Martin Jambor <mjambor@suse.cz>
28555 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28556 that have a first_link.
28557 (sort_and_splice_var_accesses): Do not check first_link before
28559 (subtree_mark_written_and_enqueue): Likewise.
28560 (propagate_all_subaccesses): Likewise and do not stop at first
28561 parent with a first_link.
28563 2017-06-13 Martin Jambor <mjambor@suse.cz>
28565 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28568 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28570 * match.pd: New pattern.
28572 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28574 * tree-vrp.c (is_masked_range_test): New function.
28575 (register_edge_assert_for): Determine ranges for
28578 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28580 PR tree-optimization/67328
28581 * fold-const.c (maskable_range_p): New function.
28582 (build_range_check): Generate bittests if possible.
28584 2017-06-13 Martin Liska <mliska@suse.cz>
28586 * gimple-pretty-print.c (dump_probability): Add new argument.
28587 (dump_edge_probability): Dump both probability and count.
28588 (dump_gimple_label): Likewise.
28589 (dump_gimple_bb_header): Likewise.
28591 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28594 * config/avr/avr-devices.c: Fix indentation.
28595 * config/avr/gen-avr-mmcu-specs.c: Dito.
28597 2017-06-13 Richard Biener <rguenther@suse.de>
28599 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28600 instead get vector type from stmt_info.
28601 (vectorizable_reduction): Adjust. Remove dead code.
28603 2017-06-13 Richard Biener <rguenther@suse.de>
28605 PR middle-end/81065
28606 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28607 case of C * (x * C2 + C3).
28608 (fold_addr_of_array_ref_difference): Properly fold index difference.
28610 2017-06-12 David S. Miller <davem@davemloft.net>
28613 * config/sparc/sparc.md (return expander): Emit frame blockage if
28614 function uses alloca.
28616 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28618 * combine.c (make_field_assignment): Check len rather than the mode
28619 precision when calling force_to_mode.
28621 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28623 Support multilibs and devices that see flash in RAM address range.
28626 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28627 (avr_mcu_t) <flash_pm_offset>: New field.
28628 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28629 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28630 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28631 (AVR_TINY_PM_OFFSET): Remove macro.
28632 * config/avr/avr.opt (-mshort-calls): New option.
28633 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28634 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28635 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28636 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28637 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28638 instead of avr_arch->have_jmp_call.
28639 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28640 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28641 avr_arch->flash_pm_offset to define.
28642 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28643 new field flash_pm_offset. Add entry for avrxmega3.
28644 (avr_texinfo): Add entry for avrxmega3.
28645 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28646 attiny212, attiny214,
28647 attiny412, attiny414, attiny416, attiny417,
28648 attiny814, attiny816, attiny817,
28649 attiny1614, attiny1616, attiny1617,
28650 attiny3214, attiny3216, attiny3217.
28651 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28652 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28653 (avr_print_operand_address) [AVR_TINY]: Same.
28654 (avr_asm_init_sections) <readonly_data_section>: Only patch
28655 callback if avr_arch->flash_pm_offset = 0.
28656 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28657 for rodata if avr_arch->flash_pm_offset != 0.
28658 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28659 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28660 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28661 (m_options): Append opt_rcall.
28662 (m_dirnames): Append dir_rcall.
28663 * config/avr/t-multilib: Regenerate.
28665 * configure.ac [target=avr]: Check whether avrxmega3 default
28666 linker description file works as needed.
28667 * configure: Regenerate.
28668 * doc/avr-mmcu.texi: Regenerate.
28669 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28670 <__AVR_ARCH__>: Document avrxmega3 and 103.
28671 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28672 <__AVR_SHORT_CALLS__>: Document it.
28673 <__AVR_PM_BASE_ADDRESS__>: Document it.
28674 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28675 (AVR Variable Attributes) <progmem>: Document this is
28676 not needed for avrxmega3.
28677 (AVR Named Address Spaces) <__flash>: Dito.
28679 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28681 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28683 2017-06-12 Doug Rupp <rupp@adacore.com>
28685 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28686 Append vxworks-stdint.h to the tm_file list.
28687 * config/vxworks-stdint.h: New file.
28689 2017-06-12 Martin Liska <mliska@suse.cz>
28691 PR tree-optimization/81041
28692 * tree-profile.c (gimple_gen_ic_func_profiler):
28693 Create an extra BB in profile-generate
28694 (gimple_gen_time_profiler): Likewise.
28696 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28698 PR tree-optimization/81003
28699 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28700 (update_range_test): Use it instead of force_gimple_operand_gsi.
28702 2017-06-12 Richard Biener <rguenther@suse.de>
28704 PR tree-optimization/81053
28705 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28706 with backedge value not defined in loop. Simplify def stmt
28709 2017-06-11 Tom de Vries <tom@codesourcery.com>
28712 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28714 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28715 nvptx_cannot_force_const_mem.
28717 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28719 * opts.c (finish_options): Move test for flag_split_stack after
28720 it has been initialized.
28722 2017-06-11 Jason Merrill <jason@redhat.com>
28724 * tree.h (id_equal): New.
28725 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28726 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28727 instead of strcmp of IDENTIFIER_POINTER.
28729 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28731 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28732 (mark_all_inlined_calls_cdtor): Fix formating.
28733 (inline_transform): Rescale profile before inlining.
28735 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28737 * cgraph.h (cgraph_edge::clone): Update prototype.
28738 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28739 (cgraph_node::create_clone): Update.
28740 (cgraph_node::create_version_clone): Update.
28741 * tree-inline.c (copy_bb): Update.
28742 (expand_call_inline): Update.
28744 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28746 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28747 factored out from ...
28748 (rs6000_emit_prologue): ... here.
28750 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28752 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28753 factored out from ...
28754 (rs6000_emit_prologue): ... here.
28756 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28758 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28760 (handle_missing_profiles): Fix computation of tp_first_run.
28761 (counts_to_freqs): Do not touch freqs when count is 0.
28763 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28765 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28768 2017-06-10 Tom de Vries <tom@codesourcery.com>
28770 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28771 attributes): Document signal effective target.
28773 2017-06-10 Tom de Vries <tom@codesourcery.com>
28775 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28776 Document effective target stack_size.
28778 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28780 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28781 to the edit_context if they can be auto-applied.
28783 2017-06-9 Ian Lance Taylor <iant@golang.org>
28785 * opts.c (finish_options): If -fsplit-stack, disable implicit
28786 -forder-blocks-and-partition.
28787 * doc/invoke.texi (Optimize Options): Document that when using
28788 -fsplit-stack -forder-blocks-and-partition is not implicitly
28791 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28793 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28794 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28795 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28796 * builtins.def (abort, trap, unreachable): Declare cold.
28797 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28798 * tree-core.h (ECF_COLD): New.
28799 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28800 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28802 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28804 * predict.c (unlikely_executed_stmt_p): Cleanup.
28806 2017-06-09 Richard Biener <rguenther@suse.de>
28808 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28809 model if the ref is always written to.
28811 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28813 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28815 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28817 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28820 2017-06-09 Tom de Vries <tom@codesourcery.com>
28823 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28824 "target cannot support label values" when encountering LABEL_REF.
28826 2017-06-09 Martin Liska <mliska@suse.cz>
28828 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28829 (gimple_gen_ic_func_profiler): Emit direct comparison
28830 of __gcov_indirect_call_callee with NULL.
28831 (gimple_gen_time_profiler): Change probability from
28832 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28834 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28836 * profile.c (edge_gcov_counts): Turn to pointer.
28837 (compute_branch_probabilities, compute_branch_probabilities): Update.
28838 (branch_prob): Do not clear edge_gcov_count.
28839 * profile.h (edge_gcov_counts): Turn to pointer.
28840 (edge_gcov_count): Update.
28842 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28844 * gimple.h (gimple_check_failed): Mark cold.
28846 2017-06-09 Richard Biener <rguenther@suse.de>
28848 PR tree-optimization/66623
28849 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28850 refactor check_reduction into two parts, properly computing
28851 whether we have to check reduction validity for outer loop
28854 2017-06-09 Richard Biener <rguenther@suse.de>
28856 PR tree-optimization/79483
28857 * graphite-scop-detection.c (order): New global.
28858 (get_order): Compute bb to order mapping that satisfies code
28859 generation constraints.
28860 (cmp_pbbs): New helper.
28861 (build_scops): Start domwalk at entry block, sort generated
28864 2017-06-09 Richard Biener <rguenther@suse.de>
28866 PR middle-end/81007
28867 * ipa-polymorphic-call.c
28868 (ipa_polymorphic_call_context::restrict_to_inner_class):
28869 Skip FIELD_DECLs with error_mark_node type.
28870 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28873 2017-06-09 Martin Liska <mliska@suse.cz>
28875 * predict.c (struct branch_predictor): New struct.
28876 (test_prediction_value_range): New test.
28877 (predict_c_tests): New function.
28878 * selftest-run-tests.c (selftest::run_tests): Run the function.
28879 * selftest.h: Declare new tests.
28881 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28884 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28885 gen_add3_insn did not fail.
28886 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28887 r0, construct that number in a temporary reg and add that reg to r0.
28888 If asked to put the result in r0 as well, fail.
28890 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28892 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28893 for early expansion of vec_eqv.
28895 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28897 PR middle-end/81005
28898 * ubsan.c (instrument_null): Avoid pointless code temporary.
28899 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28901 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28905 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28908 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28911 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28913 * predict.c (unlikely_executed_edge_p): Move ahead.
28914 (probably_never_executed_edge_p): Use it.
28916 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28918 PR middle-end/79988
28919 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28920 gimple_call_builtin_p call.
28922 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28924 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28925 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28926 rtl_check_failed_type2, rtl_check_failed_code1,
28927 rtl_check_failed_code2, rtl_check_failed_code_mode,
28928 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28929 rtvec_check_failed_bounds, rtl_check_failed_flag,
28930 _fatal_insn_not_found, _fatal_insn): Likewise.
28931 * tree.h (tree_contains_struct_check_failed,
28932 tree_check_failed, tree_not_check_failed,
28933 tree_class_check_failed, tree_range_check_failed,
28934 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28935 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28936 tree_operand_check_failed, omp_clause_check_failed,
28937 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28940 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28942 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28943 flag_branch_probabilities.
28944 * ipa-inline.c (edge_badness): Likewise.
28945 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28946 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28947 * predict.c (maybe_hot_frequency_p): Likewise.
28948 (probably_never_executed): Likewise.
28949 * sched-ebb.c (schedule_ebbs): Likewise.
28950 * sched-rgn.c (find_single_block_region): Likewise.
28951 * tracer.c (tail_duplicate): Likewise.
28953 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28955 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28956 longer requires x_flag_profile_use.
28958 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28960 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28961 instead of flag_reorder_blocks_and_partition.
28962 * dbxout.c (dbxout_function_end): Likewise.
28963 * dwarf2out.c (gen_subprogram_die): Likewise.
28964 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28965 * hw-doloop.c (reorg_loops): Likewise.
28966 * varasm.c (assemble_start_function,
28967 assemble_end_function): Likewise.
28968 (decide_function_section): Do not check for
28969 flag_reorder_blocks_and_partition.
28971 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28973 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28975 (chkp_get_hard_register_fake_addr_expr): Ditto.
28976 (chkp_build_addr_expr): Add check for hard reg case.
28977 (chkp_parse_array_and_component_ref): Ditto.
28978 (chkp_find_bounds_1): Ditto.
28979 (chkp_process_stmt): Don't generate bounds store for
28982 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28984 * predict.c (maybe_hot_bb_p): Do not check profile status.
28985 (maybe_hot_edge_p): Likewise.
28986 (probably_never_executed): Check for zero counts even if profile
28988 (unlikely_executed_edge_p): New function.
28989 (unlikely_executed_stmt_p): New function.
28990 (unlikely_executed_bb_p): New function.
28991 (set_even_probabilities): Use unlikely predicates.
28992 (combine_predictions_for_bb): Likewise.
28993 (predict_paths_for_bb): Likewise.
28994 (predict_paths_leading_to_edge): Likewise.
28995 (determine_unlikely_bbs): New function.
28996 (estimate_bb_frequencies): Use it.
28997 (compute_function_frequency): Use zero counts even if profile is
28999 * profile-count.h: Fix typo.
29001 2017-08-08 Julia Koval <julia.koval@intel.com>
29003 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
29004 _mm512_mask_cvtsepi16_storeu_epi8,
29005 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
29006 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
29007 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
29008 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
29009 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
29010 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
29011 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
29012 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
29013 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
29014 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
29015 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
29016 __builtin_ia32_pmovuswb256mem_mask,
29017 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
29018 __builtin_ia32_pmovwb512mem_mask): New builtins.
29020 2017-08-08 Julia Koval <julia.koval@intel.com>
29022 PR target/73350,80862
29023 * config/i386/subst.md (round): Fix round pattern.
29024 * config/i386/i386.c (ix86_erase_embedded_rounding):
29025 Fix erasing rounding for the fixed pattern.
29027 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
29029 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
29031 2017-06-08 Martin Liska <mliska@suse.cz>
29033 PR gcov-profile/80911
29034 * gcov.c (block_info::block_info): New constructor.
29036 2017-06-07 Carl Love <cel@us.ibm.com>
29038 * config/rs6000/rs6000-c: The return type of the following
29039 built-in functions was implemented as int not long long. Fix sign
29040 of return value for the unsigned version of vec_mulo and vec_mule.
29041 vector unsigned long long vec_bperm (vector unsigned long long,
29042 vector unsigned char)
29043 vector signed long long vec_mule (vector signed int,
29045 vector unsigned long long vec_mule (vector unsigned int,
29046 vector unsigned int)
29047 vector signed long long vec_mulo (vector signed int,
29049 vector unsigned long long vec_mulo (vector unsigned int,
29050 vector unsigned int)
29051 * doc/extend.texi: Fix the documentation for the built-in
29054 2017-06-07 Carl Love <cel@us.ibm.com>
29057 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
29060 2017-06-07 Carl Love <cel@us.ibm.com>
29062 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
29063 support, Generate doublehv for signed int/float for BE case only.
29065 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
29067 * doc/invoke.texi (mcx16): Rewrite.
29069 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29071 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
29072 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
29073 *mov<mode>_softfloat, and an anonymous splitter): Use
29074 nonimmediate_operand instead of rs6000_nonimmediate_operand.
29076 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29078 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
29080 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29081 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
29082 (rs6000_debug_reg_global): Adjust.
29083 (rs6000_init_hard_regno_mode_ok): Adjust.
29084 (rs6000_dbx_register_number): Adjust.
29085 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
29086 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29087 Remove SPE_ACC and SPEFSCR.
29088 (REG_ALLOC_ORDER): Ditto.
29089 (FRAME_POINTER_REGNUM): Change to 111.
29090 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
29091 (REG_CLASS_NAMES): Ditto.
29092 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
29093 (REGISTER_NAMES): Ditto.
29094 (ADDITIONAL_REG_NAMES): Ditto.
29095 (rs6000_reg_names): Ditto.
29096 * config/rs6000/rs6000.md: Renumber some register number
29099 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29101 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
29103 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29104 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
29106 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
29107 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29108 Delete the SPE high registers.
29109 (REG_ALLOC_ORDER): Ditto.
29110 (enum reg_class): Remove SPE_HIGH_REGS.
29111 (REG_CLASS_NAMES): Ditto.
29112 (REG_CLASS_CONTENTS): Delete the SPE high registers.
29113 (REGISTER_NAMES): Ditto.
29114 (rs6000_reg_names): Ditto.
29115 * doc/tm.texi.in: Remove SPE as example.
29116 * doc/tm.texi: Regenerate.
29118 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29120 * config/rs6000/8540.md (ppc8540_brinc): Delete.
29121 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
29122 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
29123 * config/rs6000/rs6000.md (type): Remove "brinc".
29125 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29127 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
29128 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
29129 * config/rs6000/linuxspe.h: Delete file.
29130 * config/rs6000/rs6000.md: Don't include spe.md.
29131 * config/rs6000/spe.h: Delete file.
29132 * config/rs6000/spe.md: Delete file.
29133 * config/rs6000/t-rs6000: Remove spe.md.
29135 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29137 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
29138 (reg_or_none500mem_operand): Delete.
29139 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
29140 instead of reg_or_none500mem_operand.
29142 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29144 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29145 handling of SPE flags.
29146 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
29148 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29150 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
29152 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
29153 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
29154 paired_divv2sf3): Similar.
29155 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
29156 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
29157 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
29159 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
29160 Rename the paired_* instruction patterns.
29161 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
29163 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
29164 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
29165 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
29166 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
29167 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
29168 PAIRED_VECTOR_MODE.
29169 (struct machine_function): Delete field spe_insn_chain_scanned_p.
29170 (spe_func_has_64bit_regs_p): Delete.
29171 (spe_expand_predicate_builtin): Delete.
29172 (spe_expand_evsel_builtin): Delete.
29173 (TARGET_DWARF_REGISTER_SPAN): Do not define.
29174 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
29175 (invalid_e500_subreg): Delete.
29176 (rs6000_legitimize_address): Always force_reg op2 as well, for
29177 paired single memory accesses.
29178 (rs6000_member_type_forces_blk): Delete.
29179 (rs6000_spe_function_arg): Delete.
29180 (rs6000_expand_unop_builtin): Delete SPE handling.
29181 (rs6000_expand_binop_builtin): Ditto.
29182 (spe_expand_stv_builtin): Delete.
29183 (bdesc_2arg_spe): Delete.
29184 (spe_expand_builtin): Delete.
29185 (spe_expand_predicate_builtin): Delete.
29186 (spe_expand_evsel_builtin): Delete.
29187 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
29188 (spe_init_builtins): Delete.
29189 (spe_func_has_64bit_regs_p): Delete.
29190 (savres_routine_name): Delete "info" parameter. Adjust callers.
29191 (rs6000_emit_stack_reset): Ditto.
29192 (rs6000_dwarf_register_span): Delete.
29193 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
29194 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
29195 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
29197 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
29199 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
29200 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
29201 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
29202 mulv2sf3, divv2sf3): Delete expanders.
29204 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29206 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
29208 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29210 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
29211 * config/rs6000/rs6000.c: Ditto.
29213 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29215 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
29216 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
29217 comparison_operator.
29219 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29221 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
29222 * config/rs6000/rs6000.opt: Ditto.
29223 * config/rs6000/t-rtems: Ditto.
29225 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29227 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29228 TARGET_E500_SINGLE by 0, simplify.
29229 * config/rs6000/rs6000.c: Ditto.
29230 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29231 (spe_build_register_parallel): Delete.
29232 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29233 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29234 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29235 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29236 (E500_CONVERT): Delete.
29237 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29239 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29241 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29242 * config/rs6000/dfp.md: Ditto.
29243 (negdd2, *negdd2_fpr): Merge.
29244 (absdd2, *absdd2_fpr): Merge.
29245 (negtd2, *negtd2_fpr): Merge.
29246 (abstd2, *abstd2_fpr): Merge.
29247 * config/rs6000/e500.h: Delete file.
29248 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29249 TARGET_FPRS by 1 and simplify.
29250 * config/rs6000/rs6000-c.c: Ditto.
29251 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29252 TARGET_DF_SPE by 0.
29253 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29255 * config/rs6000/rs6000.md: Ditto.
29256 (floatdidf2, *floatdidf2_fpr): Merge.
29257 (move_from_CR_gt_bit): Delete.
29258 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29259 (E500_CR_IOR_COMPARE): Delete.
29260 (All patterns that require !TARGET_FPRS): Delete.
29261 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29263 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29265 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29267 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29269 * graphds.c (add_edge): Intitialize edge's attached data.
29270 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29271 pointer parameter. Call pointed function on each edge during
29272 graph traversing. Skip traversing the edge when the function
29274 (graphds_dfs, graphds_scc): Ditto.
29275 (for_each_edge): New parameter. Pass the new parameter to callback
29277 * graphds.h (skip_edge_callback): New function pointer type.
29278 (graphds_dfs, graphds_scc): New function pointer parameter.
29279 (graphds_edge_callback, for_each_edge): New parameter.
29281 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29283 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29284 out code checking if runtime alias check is possible to below ...
29285 Call the new function.
29286 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29287 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29289 2017-06-07 Marek Polacek <polacek@redhat.com>
29292 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29293 TYPE_OVERFLOW_WRAPS check.
29295 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29297 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29298 if versioning is required.
29299 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29300 peeling with the check for versioning.
29302 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29304 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29305 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29306 Set true to new parameter if new ssa variable is defined.
29307 (vect_gen_vector_loop_niters): Refactor. Set range information
29308 for the new vector loop bound variable.
29309 (vect_do_peeling): Ditto.
29311 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29313 * tree-affine.c (ssa.h): Include header file.
29314 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29315 has wrapping overflow behavior.
29317 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29319 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29321 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29323 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29324 (tree_to_aff_combination): ... here.
29326 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29328 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29329 reg_pressure model function.
29330 (ivopts_global_cost_for_size): Delete.
29331 (determine_set_costs, iv_ca_recount_cost): Call new model function
29332 ivopts_estimate_reg_pressure.
29334 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29336 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29337 expensive than udiv. Remove floating point cases from mod.
29339 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29341 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29342 Increase idiv cost.
29344 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29346 * config/aarch64/aarch64.md
29347 (copysignsf3): Fix mask generation.
29349 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29351 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29353 (class dump_manager): Add register_dumps method.
29354 * dumpfile.c: Include langhooks.h.
29355 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29356 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29357 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29358 (dump_manager::dump_register): Start with 512 entries instead of 32.
29359 (dump_manager::register_dumps): New method.
29360 * toplev.c (general_init): Instead of invoking register_dumps
29361 langhook, invoke register_dumps method on the dump manager.
29362 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29365 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29367 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29368 Say that # requires an associated define_split to exist, and that
29369 the define_split must be suitable for use after register allocation.
29371 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29373 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29374 (compute_outgoing_frequencies): Also initialize zero counts.
29375 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29376 around loops; preserve more of profile when nothing changes.
29378 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29380 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29382 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29383 * config/arm/arm-cpu-cdata.h: Regenerate.
29384 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29385 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29386 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29387 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29388 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29390 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29392 * config/arm/t-rmprofile: Likewise.
29393 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29395 2017-06-06 David S. Miller <davem@davemloft.net>
29398 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29399 blockage if function uses alloca.
29401 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29403 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29404 New "uid" fields to hold pretty-print IDs of group and ref.
29405 Memory references are now identified as <group_id>:<ref_id>
29406 instead of using [random] addresses.
29407 (dump_mem_details): Simplify, no functional change.
29408 (dump_mem_ref): Simplify and make output more concise.
29409 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29410 (find_or_create_group): Initialize group uid.
29411 (record_ref): Initialize ref uid. Improve debug output.
29412 (prune_group_by_reuse, should_issue_prefetch_p,)
29413 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29414 (mark_nontemporal_store, determine_loop_nest_reuse):
29415 Improve debug output.
29417 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29419 * dbgcnt.def (prefetch): New debug counter.
29420 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29421 (schedule_prefetches): Stop issueing prefetches if debug counter
29424 2017-06-06 Tom de Vries <tom@codesourcery.com>
29426 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29427 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29429 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29431 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29432 Use aarch64_reg_or_zero predicate for operand 4.
29433 (aarch64_compare_and_swap<mode> define_insn_and_split):
29434 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29435 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29437 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29439 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29440 (arm_compute_save_core_reg_mask): This.
29441 (thumb1_compute_save_reg_mask): Rename into ...
29442 (thumb1_compute_save_core_reg_mask): This.
29443 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29444 (arm_compute_frame_layout): Likewise.
29446 2017-06-06 Richard Biener <rguenther@suse.de>
29448 PR tree-optimization/80974
29449 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29450 keep or clear leaders SSA info.
29452 2017-06-06 Tom de Vries <tom@codesourcery.com>
29454 * config/nvptx/nvptx.c (split_mode_p): New function.
29455 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29457 2017-06-06 Tom de Vries <tom@codesourcery.com>
29459 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29461 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29464 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29467 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29469 * shrink-wrap.c (handle_simple_exit): Update profile.
29470 (try_shrink_wrapping): Upate profile.
29472 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29474 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29475 (tree_guess_outgoing_edge_probabilities): New.
29476 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29477 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29479 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29481 * ipa-split.c (split_function): Initialize return bb profile.
29483 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29485 * profile.c (compute_branch_probabilities): Also initialize
29486 EXIT_BLOCK profile.
29488 2017-06-06 Richard Biener <rguenther@suse.de>
29490 PR tree-optimization/80928
29491 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29492 (vect_analyze_loop_operations): Properly guard analysis for
29494 (vect_transform_loop): Likewise.
29495 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29496 (vect_model_induction_cost): Do not cost for pure SLP.
29497 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29498 of induction in inner loop vectorization.
29499 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29500 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29501 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29503 (vect_analyze_slp_cost_1): Cost induction.
29504 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29505 (vect_get_slp_vect_defs): Likewise.
29506 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29507 (vect_transform_stmt): Handle SLP reductions.
29508 * tree-vectorizer.h (vectorizable_induction): Adjust.
29510 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29512 * config/rs6000/rs6000.c (make_resolver_func): Update
29513 init_lowered_empty_function call.
29515 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29517 * doc/invoke.texi: Document the -fprofile-abs-path option.
29518 * common.opt (fprofile-abs-path): New option.
29519 * gcov-io.h (gcov_write_filename): Declare.
29520 * gcov-io.c (gcov_write_filename): New function.
29521 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29522 * profile.c (output_location): Likewise.
29524 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29526 * shring-wrap.c: Revert accidental commit.
29528 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29530 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29532 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29534 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29536 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29537 profile in callgraph edge.
29538 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29539 (apply_scale): Likewise.
29540 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29541 Also scale profile when inlining function with zero profile.
29542 (initialize_cfun): Update exit block profile even when it is zero.
29543 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29544 when profile is read.
29546 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29548 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29549 (CLONE_*): New constants to define the processors we can generate
29550 code for with the target_clone attribute.
29551 (rs6000_clone_map): New array to identify which clone processors
29552 the current program is running on.
29553 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29554 target_clone attribute.
29555 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29556 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29557 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29558 (cpu_expand_builtin): Add support for target_clone attribute.
29559 (rs6000_valid_attribute_p): Allow "default" attribute.
29560 (get_decl_name): New debug function to simplify printing the
29561 current function name in debugging statements.
29562 (rs6000_clone_priority): New functions to support the target_clone
29563 attribute, and be able to generate code to switch between ISA 2.05
29564 through ISA 3.0 (power6 through power9).
29565 (rs6000_compare_version_priority): Likewise.
29566 (rs6000_get_function_versions_dispatcher): Likewise.
29567 (make_resolver_func): Likewise.
29568 (add_condition_to_bb): Likewise.
29569 (dispatch_function_versions): Likewise.
29570 (rs6000_generate_version_dispatcher_body): Likewise.
29571 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29572 (fusion_gpr_load_p): Fix a spacing issue.
29573 * doc/extend.texi (Common Function Attributes): Document that the
29574 PowerPC supports the target_clone attribute.
29576 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29578 * config/arm/arm.h: explain F symbol found in description of ARM
29579 register allocation in its legend.
29581 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29583 * config/mips/frame-header-opt.c: Include profile-count.h.
29584 * config/riscv/riscv.c: Include profile-count.h
29586 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29588 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29590 (sm_set_flag_if_changed): Add bbs field.
29591 (execute_sm_if_changed_flag_set): Pass BBS.
29592 (execute_sm): Update.
29594 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29596 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29599 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29601 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29602 (peephole2): New peephole2 to emit the above.
29603 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29605 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29607 * config/aarch64/aarch64.c (define_peephole2 above
29608 *sub_<shift>_<mode>): New peephole.
29610 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29612 * config/i386/i386.c (make_resolver_func): Update.
29613 * Makefile.in: Add profile-count.h and profile-count.o
29614 * auto-profile.c (afdo_indirect_call): Update to new API.
29615 (afdo_set_bb_count): Update.
29616 (afdo_propagate_edge): Update.
29617 (afdo_propagate_circuit): Update.
29618 (afdo_calculate_branch_prob): Update.
29619 (afdo_annotate_cfg): Update.
29620 * basic-block.h: Include profile-count.h
29621 (struct edge_def): Turn count to profile_count.
29622 (struct basic_block_def): Likewie.
29623 (REG_BR_PROB_BASE): Move to profile-count.h
29624 (RDIV): Move to profile-count.h
29625 * bb-reorder.c (max_entry_count): Turn to profile_count.
29626 (find_traces): Update.
29627 (rotate_loop):Update.
29628 (connect_traces):Update.
29629 (sanitize_hot_paths):Update.
29630 * bt-load.c (migrate_btr_defs): Update.
29631 * cfg.c (RDIV): Remove.
29632 (init_flow): Use alloc_block.
29633 (alloc_block): Uninitialize count.
29634 (unchecked_make_edge): Uninitialize count.
29635 (check_bb_profile): Update.
29636 (dump_edge_info): Update.
29637 (dump_bb_info): Update.
29638 (update_bb_profile_for_threading): Update.
29639 (scale_bbs_frequencies_int): Update.
29640 (scale_bbs_frequencies_gcov_type): Update.
29641 (scale_bbs_frequencies_profile_count): New.
29642 * cfg.h (update_bb_profile_for_threading): Update.
29643 (scale_bbs_frequencies_profile_count): Declare.
29644 * cfgbuild.c (compute_outgoing_frequencies): Update.
29645 (find_many_sub_basic_blocks): Update.
29646 * cfgcleanup.c (try_forward_edges): Update.
29647 (try_crossjump_to_edge): Update.
29648 * cfgexpand.c (expand_gimple_tailcall): Update.
29649 (construct_exit_block): Update.
29650 * cfghooks.c (verify_flow_info): Update.
29651 (dump_bb_for_graph): Update.
29652 (split_edge): Update.
29653 (make_forwarder_block): Update.
29654 (duplicate_block): Update.
29655 (account_profile_record): Update.
29656 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29657 (get_estimated_loop_iterations): Update.
29658 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29659 (single_likely_exit): Update.
29660 * cfgloopmanip.c (scale_loop_profile): Update.
29662 (set_zero_probability): Update.
29663 (lv_adjust_loop_entry_edge): Update.
29664 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29665 (purge_dead_edges): Update.
29666 (rtl_account_profile_record): Update.
29667 * cgraph.c (cgraph_node::create): Uninitialize count.
29668 (symbol_table::create_edge): Uninitialize count.
29669 (cgraph_update_edges_for_call_stmt_node): Update.
29670 (cgraph_edge::dump_edge_flags): Update.
29671 (cgraph_node::dump): Update.
29672 (cgraph_edge::maybe_hot_p): Update.
29673 * cgraph.h: Include profile-count.h
29674 (create_clone), create_edge, create_indirect_edge): Update.
29675 (cgraph_node): Turn count to profile_count.
29676 (cgraph_edge0: Likewise.
29677 (make_speculative, clone): Update.
29678 (create_edge): Update.
29679 (init_lowered_empty_function): Update.
29680 * cgraphclones.c (cgraph_edge::clone): Update.
29681 (duplicate_thunk_for_node): Update.
29682 (cgraph_node::create_clone): Update.
29683 * cgraphunit.c (cgraph_node::analyze): Update.
29684 (cgraph_node::expand_thunk): Update.
29685 * final.c (dump_basic_block_info): Update.
29686 * gimple-streamer-in.c (input_bb): Update.
29687 * gimple-streamer-out.c (output_bb): Update.
29688 * graphite.c (print_global_statistics): Update.
29689 (print_graphite_scop_statistics): Update.
29690 * hsa-brig.c: Include basic-block.h.
29691 * hsa-dump.c: Include basic-block.h.
29692 * hsa-gen.c (T sum_slice): Update.
29693 (convert_switch_statements):Update.
29694 * hsa-regalloc.c: Include basic-block.h.
29695 * ipa-chkp.c (chkp_produce_thunks): Update.
29696 * ipa-cp.c (struct caller_statistics): Update.
29697 (init_caller_stats): Update.
29698 (gather_caller_stats): Update.
29699 (ipcp_cloning_candidate_p): Update.
29700 (good_cloning_opportunity_p): Update.
29701 (get_info_about_necessary_edges): Update.
29702 (dump_profile_updates): Update.
29703 (update_profiling_info): Update.
29704 (update_specialized_profile): Update.
29705 (perhaps_add_new_callers): Update.
29706 (decide_about_value): Update.
29707 (ipa_cp_c_finalize): Update.
29708 * ipa-devirt.c (struct odr_type_warn_count): Update.
29709 (struct decl_warn_count): Update.
29710 (struct final_warning_record): Update.
29711 (possible_polymorphic_call_targets): Update.
29712 (ipa_devirt): Update.
29713 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29714 * ipa-icf.c (sem_function::merge): Update.
29715 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29716 * ipa-inline.c (compute_uninlined_call_time): Update.
29717 (compute_inlined_call_time): Update.
29718 (want_inline_small_function_p): Update.
29719 (want_inline_self_recursive_call_p): Update.
29720 (edge_badness): Update.
29721 (lookup_recursive_calls): Update.
29722 (recursive_inlining): Update.
29723 (inline_small_functions): Update.
29724 (dump_overall_stats): Update.
29725 (dump_inline_stats): Update.
29726 * ipa-profile.c (ipa_profile_generate_summary): Update.
29727 (ipa_propagate_frequency): Update.
29728 (ipa_profile): Update.
29729 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29730 * ipa-utils.c (ipa_merge_profiles): Update.
29731 * loop-doloop.c (doloop_modify): Update.
29732 * loop-unroll.c (report_unroll): Update.
29733 (unroll_loop_runtime_iterations): Update.
29734 * lto-cgraph.c (lto_output_edge): Update.
29735 (lto_output_node): Update.
29736 (input_node): Update.
29737 (input_edge): Update.
29738 (merge_profile_summaries): Update.
29739 * lto-streamer-in.c (input_cfg): Update.
29740 * lto-streamer-out.c (output_cfg): Update.
29741 * mcf.c (create_fixup_graph): Update.
29742 (adjust_cfg_counts): Update.
29743 (sum_edge_counts): Update.
29744 * modulo-sched.c (sms_schedule): Update.
29745 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29746 * predict.c (maybe_hot_count_p): Update.
29747 (probably_never_executed): Update.
29748 (dump_prediction): Update.
29749 (combine_predictions_for_bb): Update.
29750 (propagate_freq): Update.
29751 (handle_missing_profiles): Update.
29752 (counts_to_freqs): Update.
29753 (rebuild_frequencies): Update.
29754 (force_edge_cold): Update.
29755 * predict.h: Include profile-count.h
29756 (maybe_hot_count_p, counts_to_freqs): UPdate.
29757 * print-rtl-function.c: Do not include cfg.h
29758 * print-rtl.c: Include basic-block.h
29759 * profile-count.c: New file.
29760 * profile-count.h: New file.
29761 * profile.c (is_edge_inconsistent): Update.
29762 (correct_negative_edge_counts): Update.
29763 (is_inconsistent): Update.
29764 (set_bb_counts): Update.
29765 (read_profile_edge_counts): Update.
29766 (compute_frequency_overlap): Update.
29767 (compute_branch_probabilities): Update; Initialize and deinitialize
29769 (branch_prob): Update.
29770 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29771 (edge_gcov_count): New.
29772 (bb_gcov_count): New.
29773 * shrink-wrap.c (try_shrink_wrapping): Update.
29774 * tracer.c (better_p): Update.
29775 * trans-mem.c (expand_transaction): Update.
29776 (ipa_tm_insert_irr_call): Update.
29777 (ipa_tm_insert_gettmclone_call): Update.
29778 * tree-call-cdce.c: Update.
29779 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29780 (gimple_duplicate_sese_tail): Update.
29781 (gimple_account_profile_record): Update.
29782 (execute_fixup_cfg): Update.
29783 * tree-inline.c (copy_bb): Update.
29784 (copy_edges_for_bb): Update.
29785 (initialize_cfun): Update.
29786 (freqs_to_counts): Update.
29787 (copy_cfg_body): Update.
29788 (expand_call_inline): Update.
29789 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29790 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29791 (try_unroll_loop_completely): Update.
29792 (try_peel_loop): Update.
29793 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29794 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29795 * tree-ssa-loop-split.c (connect_loops): Update.
29796 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29797 * tree-ssa-reassoc.c (branch_fixup): Update.
29798 * tree-ssa-tail-merge.c (replace_block_by): Update.
29799 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29800 (compute_path_counts): Update.
29801 (update_profile): Update.
29802 (recompute_probabilities): Update.
29803 (update_joiner_offpath_counts): Update.
29804 (estimated_freqs_path): Update.
29805 (freqs_to_counts_path): Update.
29806 (clear_counts_path): Update.
29807 (ssa_fix_duplicate_block_edges): Update.
29808 (duplicate_thread_path): Update.
29809 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29810 (struct switch_conv_info): Update.
29811 * tree-tailcall.c (decrease_profile): Update.
29812 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29813 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29814 * value-prof.c (check_counter): Update.
29815 (gimple_divmod_fixed_value): Update.
29816 (gimple_mod_pow2): Update.
29817 (gimple_mod_subtract): Update.
29818 (gimple_ic_transform): Update.
29819 (gimple_stringop_fixed_value): Update.
29820 * value-prof.h (gimple_ic): Update.
29822 2017-06-02 Carl Love <cel@us.ibm.com>
29824 * config/rs6000/rs6000-c: Add support for built-in functions
29825 vector double vec_doublee (vector signed int);
29826 vector double vec_doublee (vector unsigned int);
29827 vector double vec_doublee (vector float);
29828 vector double vec_doubleh (vector signed int);
29829 vector double vec_doubleh (vector unsigned int);
29830 vector double vec_doubleh (vector float);
29831 vector double vec_doublel (vector signed int);
29832 vector double vec_doublel (vector unsigned int);
29833 vector double vec_doublel (vector float);
29834 vector double vec_doubleo (vector signed int);
29835 vector double vec_doubleo (vector unsigned int);
29836 vector double vec_doubleo (vector float);.
29837 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29838 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29840 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29841 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29842 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29844 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29845 vec_doublel, vec_doubleh.
29846 * doc/extend.texi: Update the built-in documentation file for the
29847 new built-in functions.
29849 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29852 * ipa-inline-analysis.c (free_growth_caches): Set
29853 edge_removal_hook_holder to NULL after removing it.
29855 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29857 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29858 comparision with zero.
29860 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29861 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29862 for early expansion of vec_min and vec_max builtins.
29863 (builtin_function_type): Add min/max unsigned variants to those
29864 identified as having unsigned arguments.
29866 2017-06-02 Olivier Hainque <hainque@adacore.com>
29868 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29870 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29872 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29873 Use VALL_F16 iterator rather than VALL.
29875 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29877 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29878 Emit CBNZ inside loop when doing a strong exchange and comparing
29879 against zero. Generate the CC flags after the loop.
29881 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29883 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29884 (dl_section_ref): New.
29885 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29886 On AIX, append an expression to subtract the size of the
29887 section length to dl_section_ref.
29889 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29891 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29892 for early expansion of vector absolute builtins.
29894 2017-06-02 Richard Biener <rguenther@suse.de>
29896 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29897 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29899 2017-06-02 Richard Biener <rguenther@suse.de>
29901 PR tree-optimization/80948
29902 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29903 stmt order as well.
29905 2017-06-02 Richard Biener <rguenther@suse.de>
29907 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29909 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29910 for inductions as relevant.
29912 2017-06-02 Richard Biener <rguenther@suse.de>
29914 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29915 (vectorizable_induction): ... this. Remove dead code.
29917 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29919 * builtins. (expand_builtin_alloca): Remove second parameter and
29920 infer its value from the first parameter instead.
29921 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29923 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29925 PR rtl-optimization/80903
29926 * loop-doloop.c (add_test): Unshare sequence.
29928 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29930 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29932 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29934 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29936 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29937 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29938 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29939 (xlogue_layout::get_stub_rtx): Make static.
29940 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29941 (xlogue_layout::compute_stub_managed_regs): Rename to...
29942 (xlogue_layout::count_stub_managed_regs): ...this.
29943 (xlogue_layout::is_stub_managed_reg): New function.
29944 (xlogue_layout::m_stub_names): Rename to...
29945 (xlogue_layout::s_stub_names): ...this, make static.
29946 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29947 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29948 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29949 xlogue_layout::s_stub_names): Instantiate statics.
29950 (stub_managed_regs): Remove.
29951 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29952 (disable_call_ms2sysv_xlogues): Rename to...
29953 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29954 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29956 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29957 change after reload_completed.
29958 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29960 (ix86_expand_prologue): Likewise.
29961 (ix86_expand_epilogue): Likewise.
29962 (ix86_expand_split_stack_prologue): Likewise.
29963 (ix86_compute_frame_layout): Remove frame parameter ...
29964 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29965 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29967 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29968 (ix86_frame): Move from here ...
29969 * config/i386/i386.h (ix86_frame): ... to here.
29970 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29971 complete ix86_frame data structure instead. Remove some_ld_name.
29973 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29975 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29976 symbols that hold a DECL_VALUE_EXPR.
29978 2017-06-01 Martin Jambor <mjambor@suse.cz>
29980 PR tree-optimization/80898
29981 * tree-sra.c (process_subtree_disqualification): Removed.
29982 (disqualify_candidate): Do not acll
29983 process_subtree_disqualification.
29984 (subtree_mark_written_and_enqueue): New function.
29985 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29986 RHS has been disqualified and re-queue LHS if necessary. Apart
29987 from that, ignore disqualified RHS.
29989 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29991 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29992 address fetch for z10 or later.
29994 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29996 * config/arc/arc.md (tst_movb): Add guard when splitting.
29998 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30000 * config/arc/arc.c (arc_can_eliminate): Test against
30001 arc_frame_pointer_needed.
30003 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30005 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
30006 to prevent store reordering.
30007 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
30008 (type): Add block type.
30009 (stack_tie): Define special instruction to be used in
30012 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30014 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
30015 constraint. It is not valid for the pattern.
30016 (noncommutative_binary_comparison): Likewise.
30018 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30020 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
30023 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30025 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
30026 be used by the reg-alloc.
30028 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30030 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
30031 reg-alloc when having mul64 or mul32x16 instructions.
30032 (mulsidi3): Likewise.
30033 (umulsidi3): Likewise.
30034 (mulsi32x16): New pattern.
30035 (mulsi64): Likewise.
30036 (mulsidi64): Likewise.
30037 (umulsidi64): Likewise.
30038 (MUL32x16_REG): Define.
30039 (mul64_600): Use MUL32x16_REG.
30040 (mac64_600): Likewise.
30041 (umul64_600): Likewise.
30042 (umac64_600): Likewise.
30044 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30046 * config/arc/arc.md (mulsi3_700): Make it commutative.
30048 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
30050 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
30052 (*sign_extendsidi2_insn): Likewise for movstosw.
30054 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
30056 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
30057 the type of the input discriminant value. Convert the
30058 discriminant value of signedness vary.
30060 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
30062 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
30063 Add to -Wall section.
30065 2017-06-01 Richard Biener <rguenther@suse.de>
30067 PR middle-end/66313
30068 * fold-const.c (fold_plusminus_mult_expr): If the factored
30069 factor may be zero use a wrapping type for the inner operation.
30070 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
30071 and handle moved defs.
30072 (process_assignment): Properly guard the unary op case. Return a
30073 tri-state indicating that moving the stmt before the call may allow
30074 to continue. Pass through to_move.
30075 (find_tail_calls): Handle moving unrelated defs before
30078 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
30081 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
30082 splitter result in the canonical way.
30084 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
30086 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
30087 also for 32bit target. Update insn attributes.
30088 (zero-extendsidi2 splitter): Allow all registers for operand 1.
30090 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
30092 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
30093 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
30094 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
30095 (_mm_maskz_min_ss): New intrinsics.
30097 2017-05-31 Martin Liska <mliska@suse.cz>
30099 * tree-vect-loop.c (vect_create_epilog_for_reduction):
30100 Change comment style to one we normally use.
30101 (vectorizable_reduction): Likewise.
30102 (vectorizable_induction): Likewise.
30103 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
30104 (vectorizable_call): Likewise.
30105 (vectorizable_simd_clone_call): Likewise.
30106 (vectorizable_conversion): Likewise.
30107 (vectorizable_assignment): Likewise.
30108 (vectorizable_shift): Likewise.
30109 (vectorizable_operation): Likewise.
30110 (vectorizable_store): Likewise.
30111 (vectorizable_load): Likewise.
30112 * tree-vectorizer.h: Likewise.
30114 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
30116 * passes.c (emergency_dump_function): New.
30117 * tree-pass.h (emergency_dump_function): Declare.
30118 * plugin.c (plugins_internal_error_function): Remove.
30119 * plugin.h (plugins_internal_error_function): Remove declaration.
30120 * toplev.c (internal_error_function): New static function. Use it...
30121 (general_init): ...here.
30123 2017-05-31 Graham Markall <graham.markall@embecosm.com>
30125 * config/arc/arc.c (arc_print_operand): Handle constant operands.
30126 (arc_rtx_costs): Add costs for new patterns.
30127 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
30128 * config/arc/predicates.md: Add _1_2_3_operand predicate.
30130 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30132 * tree-ssa-strlen.c (get_next_strinfo): New function.
30133 (get_stridx_plus_constant): Use it.
30134 (zero_length_string): Likewise.
30135 (adjust_related_strinfos): Likewise.
30136 (adjust_last_stmt): Likewise.
30138 2017-05-31 Richard Biener <rguenther@suse.de>
30141 * config/i386/i386.c (ix86_expand_builtin): Remove assert
30142 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
30144 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30146 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
30147 loop_vinfo argument and use of dependence distance vectors.
30148 Check instead whether the two references differ only in their
30149 initial value and assume that they have the same alignment if the
30150 difference is a multiple of the vector alignment.
30151 (vect_analyze_data_refs_alignment): Update call accordingly.
30153 2017-05-31 Martin Liska <mliska@suse.cz>
30156 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
30158 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30160 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
30161 (create_intersect_range_checks): Move from ...
30162 * tree-data-ref.c (create_intersect_range_checks_index)
30163 (create_intersect_range_checks): ... to here.
30164 (create_runtime_alias_checks): New function factored from ...
30165 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
30166 here. Call above function.
30167 * tree-data-ref.h (create_runtime_alias_checks): New function.
30169 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30171 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
30172 segment length for dr_b and compute it in wide_int.
30174 2017-05-31 Richard Biener <rguenther@suse.de>
30176 PR tree-optimization/80906
30177 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
30178 and pass through iv_map.
30179 (copy_bb_and_scalar_dependences): Adjust.
30180 (translate_pending_phi_nodes): Likewise.
30181 (copy_loop_close_phi_args): Handle code-generating IVs instead
30184 2017-05-30 David Malcolm <dmalcolm@redhat.com>
30186 * diagnostic-color.c (color_dict): Add "type-diff".
30187 (parse_gcc_colors): Update comment.
30188 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
30189 -fdiagnostics-show-template-tree and -fno-elide-type.
30190 (GCC_COLORS): Add type-diff to example.
30192 (-fdiagnostics-show-template-tree): New.
30193 (-fno-elide-type): New.
30194 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
30195 the pp_format_decoder callback. Call any m_format_postprocessor's
30197 (pretty_printer::pretty_printer): Initialize
30198 m_format_postprocessor.
30199 (pretty_printer::~pretty_printer): Delete any
30200 m_format_postprocessor.
30201 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
30202 (class format_postprocessor): New class.
30203 (struct pretty_printer::format_decoder): Document the new parameters.
30204 (struct pretty_printer::m_format_postprocessor): New field.
30205 * tree-diagnostic.c (default_tree_printer): Update for new
30206 bool and const char ** params.
30207 * tree-diagnostic.h (default_tree_printer): Likewise.
30209 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
30211 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
30212 (lwa_operand): Delete rs6000_gen_cell_microcode test.
30213 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30214 rs6000_gen_cell_microcode code.
30215 (rs6000_final_prescan_insn): Delete.
30216 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
30217 "warn-cell-microcode" entries.
30218 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
30219 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
30220 throughout. Change cc_reg_not_micro_cr0_operand to
30221 cc_reg_not_cr0_operand throughout.
30222 (*extendhi<mode>2_noload): Delete.
30223 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30224 (mwarn-cell-microcode): Delete.
30225 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30226 -mgen-cell-microcode and -mwarn-cell-microcode.
30228 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30231 * config/i386/constraints.md (Yd): New constraint.
30233 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30234 and (?Yd, r) alternatives. Update insn attributes.
30235 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30236 and (?*Yd, r) alternatives. Update insn attributes.
30237 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30239 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30241 * gimplify.c (gimplify_modify_expr): Don't create a
30242 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30245 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30247 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30249 2017-05-30 Richard Biener <rguenther@suse.de>
30251 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30252 and reduc_def fields.
30253 (STMT_VINFO_REDUC_TYPE): New define.
30254 (STMT_VINFO_REDUC_DEF): Likewise.
30255 (vect_force_simple_reduction): Adjust prototype.
30256 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30257 (vect_is_simple_reduction): Remove check_reduction argument.
30258 (vect_force_simple_reduction): Adjust and set
30259 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30260 (vectorizable_reduction): Do not re-do reduction analysis
30261 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30262 * tree-parloops.c (gather_scalar_reductions): Adjust.
30264 2017-05-30 Richard Biener <rguenther@suse.de>
30266 PR middle-end/80901
30267 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30270 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30272 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30273 Introduce unknown_misalignment parameter and remove vf.
30274 (vect_peeling_hash_get_lowest_cost):
30275 Pass unknown_misalignment parameter.
30276 (vect_enhance_data_refs_alignment):
30277 Fix unsupportable data ref treatment.
30279 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30281 * tree-vect-data-refs.c (vect_get_data_access_cost):
30282 Workaround for SLP handling.
30283 (vect_enhance_data_refs_alignment):
30284 Compute costs for doing no peeling at all, compare to the best
30285 peeling costs so far and avoid peeling if cheaper.
30287 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30289 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30290 Return peeling info and set costs to zero for unlimited cost
30292 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30293 with unknown misalignment. Compute and costs for unknown
30294 misalignment, compare them to the costs for known misalignment
30295 and choose the cheapest for peeling.
30297 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30299 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30300 (vect_get_peeling_costs_all_drs): Create function.
30301 (vect_peeling_hash_get_lowest_cost):
30302 Use vect_get_peeling_costs_all_drs.
30303 (vect_peeling_supportable): Create function.
30304 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30306 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30308 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30309 DR_HAS_NEGATIVE_STEP.
30310 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30311 (vect_enhance_data_refs_alignment): Use.
30312 (vect_duplicate_ssa_name_ptr_info): Use.
30313 * tree-vectorizer.h (dr_misalignment): Use.
30314 (known_alignment_for_access_p): Use.
30316 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30319 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30321 (has_section_name): New function.
30323 2017-05-30 Martin Liska <mliska@suse.cz>
30326 * auto-profile.c (get_function_decl_from_block): Fix
30329 2017-05-30 Richard Biener <rguenther@suse.de>
30331 PR middle-end/80876
30332 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30334 2017-05-30 Martin Liska <mliska@suse.cz>
30336 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30337 * dumpfile.h (struct dump_file_info): Remove ctors.
30339 2017-05-30 Martin Liska <mliska@suse.cz>
30341 * predict.def: Fix GNU coding style.
30343 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30345 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30346 Mark 'to' argument with ATTRIBUTE_UNUSED.
30348 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30350 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30351 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30352 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30355 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30357 * doc/install.texi (Options specification): Restore entry of
30358 --enable-sjlj-exceptions.
30360 2017-05-27 Michael Eager <eager@eagercon.com>
30363 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30365 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30367 * config/microblaze/microblaze.h
30368 (FIXED_REGISTERS): Update in macro.
30369 (CALL_USED_REGISTERS): Update in macro.
30371 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30373 * doc/install.texi: Add links to macOS binary distributions.
30375 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30379 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30381 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30383 2017-05-26 Martin Liska <mliska@suse.cz>
30385 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30387 2017-05-26 Martin Liska <mliska@suse.cz>
30389 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30390 always leading ';; '.
30391 (dump_bb_info): Likewise.
30392 (brief_dump_cfg): Likewise.
30393 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30394 * dumpfile.c: Remove usage of TDF_VERBOSE.
30395 * dumpfile.h (enum dump_kind): Likewise.
30396 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30397 * print-tree.c (debug_verbose): Remove.
30398 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30399 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30400 * tree-diagnostic.c (default_tree_printer): Replace
30401 TDF_DIAGNOSTIC with TDF_SLIM.
30403 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30405 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30406 in parameter loop, rather than loop_vinfo.
30407 (create_intersect_range_checks): Ditto.
30408 (vect_create_cond_for_alias_checks): Update call to above functions.
30410 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30412 PR tree-optimization/80815
30413 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30414 for merging runtime alias checks. Handle negative DR_STEPs.
30416 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30418 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30420 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30421 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30422 out code pruning runtime alias checks.
30423 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30424 factored out from above.
30425 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30427 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30429 (prune_runtime_alias_test_list): New decalaration.
30431 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30433 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30434 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30435 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30436 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30437 (operator==, comp_dr_with_seg_len_pair): Ditto.
30438 (vect_prune_runtime_alias_test_list): Ditto.
30440 2017-05-26 Martin Liska <mliska@suse.cz>
30443 * params.def: Bound partial-inlining-entry-probability param.
30445 2017-05-26 Marek Polacek <polacek@redhat.com>
30448 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30451 2017-05-26 Richard Biener <rguenther@suse.de>
30453 PR tree-optimization/80842
30454 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30457 2017-05-26 Richard Biener <rguenther@suse.de>
30459 PR tree-optimization/80844
30460 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30462 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30464 * doc/md.texi (Machine Constraints): Update x86 family
30465 machine constraints section to match 'config/i386/constraints.md'.
30467 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30469 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30471 2017-05-25 Nathan Sidwell <nathan@acm.org>
30473 * doc/invoke.texi (--enable-languages): Update documentation.
30475 2017-05-25 Martin Liska <mliska@suse.cz>
30477 * dumpfile.c: Add TDF_FOLDING.
30478 * dumpfile.h (enum dump_kind): Likewise.
30479 * genmatch.c (dt_simplify::gen_1): Use it.
30481 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30483 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30485 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30487 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30488 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30490 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30492 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30493 * match.pd (X == C): Rewrite it here.
30494 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30495 with_certain_nonzero_bits2): New predicates.
30496 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30498 2017-05-24 Nathan Sidwell <nathan@acm.org>
30500 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30503 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30506 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30508 * config/powerpcspe: New port. Files are copied from the rs6000
30509 port, with "rs6000" in filenames replaced by "powerpcspe".
30511 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30513 PR rtl-optimization/80754
30514 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30516 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30518 * config/sparc/sparc.md (length): Return the correct value for -mflat
30519 sibcalls to match output_sibcall.
30521 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30525 * config/rs6000/rs6000.c (struct machine_function): Add new field
30527 (rs6000_get_separate_components): Init that field, use it.
30528 (rs6000_components_for_bb): Use the field.
30530 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30532 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30534 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30536 PR middle-end/80823
30537 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30539 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30542 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30543 address against address_operand predicate.
30544 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30546 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30548 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30549 parameters passed indirectly.
30551 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30553 * config/i386/i386.md (*movdi_internal): Remove SSE4
30554 alternative 18 (?r, *v). Update insn attributes.
30555 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30556 Update insn attributes.
30557 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30558 Update insn attributes.
30559 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30560 alternative 1 (r, v). Remove isa attribute.
30561 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30562 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30563 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30565 2017-05-23 Tom de Vries <tom@codesourcery.com>
30567 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30570 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30572 * cgraphunit.c (symbol_table::process_new_functions): Update.
30573 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30574 (inline_generate_summary): Rename to ...
30575 (ipa_fn_summary_generate): ... this one.
30576 (inline_read_summary): Rename to ...
30577 (ipa_fn_summary_read): ... this one.
30578 (inline_write_summary): Rename to ...
30579 (ipa_fn_summary_write): ... this one.
30580 (inline_free_summary): Rename to ...
30581 (ipa_free_fn_summary): ... this one.
30582 (pass_data_local_fn_summary, pass_local_fn_summary,
30583 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30584 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30585 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30586 make_pass_ipa_fn_summary): New.
30587 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30588 inline_write_summary, inline_free_summary): Remove.
30589 (ipa_free_fn_summary) : New.
30590 * ipa-inline.c (ipa_inline): Update.
30591 (pass_ipa_inline): Do not generate summaries.
30592 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30594 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30595 and add pass_ipa_fn_summary.
30596 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30598 (make_pass_inline_parameters): Remove.
30600 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30602 * omp-low.c (struct omp_context): Remove "default_kind" member.
30605 * omp-offload.c (execute_oacc_device_lower): Remove the
30606 parallelism dimensions function attributes for unparallelized
30607 OpenACC kernels constructs.
30609 2017-05-23 Martin Liska <mliska@suse.cz>
30611 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30613 (cgraph_edge::make_speculative): Likewise.
30614 (cgraph_edge::resolve_speculation): Likewise.
30615 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30616 (cgraph_node::dump): Likewise.
30617 * cgraph.h: Likewise.
30618 * cgraphunit.c (analyze_functions): Likewise.
30619 (symbol_table::compile): Likewise.
30620 * ipa-cp.c (print_all_lattices): Likewise.
30621 (determine_versionability): Likewise.
30622 (initialize_node_lattices): Likewise.
30623 (ipcp_verify_propagated_values): Likewise.
30624 (estimate_local_effects): Likewise.
30625 (update_profiling_info): Likewise.
30626 (create_specialized_node): Likewise.
30627 (perhaps_add_new_callers): Likewise.
30628 (decide_about_value): Likewise.
30629 (decide_whether_version_node): Likewise.
30630 (identify_dead_nodes): Likewise.
30631 (ipcp_store_bits_results): Likewise.
30632 * ipa-devirt.c (dump_targets): Likewise.
30633 (ipa_devirt): Likewise.
30634 * ipa-icf.c (sem_item::dump): Likewise.
30635 (sem_function::equals): Likewise.
30636 (sem_variable::equals): Likewise.
30637 (sem_item_optimizer::read_section): Likewise.
30638 (sem_item_optimizer::execute): Likewise.
30639 (congruence_class::dump): Likewise.
30640 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30641 (dump_inline_summary): Likewise.
30642 (estimate_node_size_and_time): Likewise.
30643 (inline_analyze_function): Likewise.
30644 * ipa-inline-transform.c (inline_call): Likewise.
30645 * ipa-inline.c (report_inline_failed_reason): Likewise.
30646 (want_early_inline_function_p): Likewise.
30647 (edge_badness): Likewise.
30648 (update_edge_key): Likewise.
30649 (inline_small_functions): Likewise.
30650 * ipa-profile.c (ipa_profile): Likewise.
30651 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30652 (ipa_make_edge_direct_to_target): Likewise.
30653 (remove_described_reference): Likewise.
30654 (ipa_impossible_devirt_target): Likewise.
30655 (propagate_controlled_uses): Likewise.
30656 (ipa_print_node_params): Likewise.
30657 (ipcp_transform_function): Likewise.
30658 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30659 (propagate_pure_const): Likewise.
30660 * ipa-reference.c (generate_summary): Likewise.
30661 (read_write_all_from_decl): Likewise.
30662 (propagate): Likewise.
30663 (ipa_reference_read_optimization_summary): Likewise.
30664 * ipa-utils.c (ipa_merge_profiles): Likewise.
30665 * ipa.c (walk_polymorphic_call_targets): Likewise.
30666 (symbol_table::remove_unreachable_nodes): Likewise.
30667 (ipa_single_use): Likewise.
30668 * passes.c (execute_todo): Likewise.
30669 * predict.c (drop_profile): Likewise.
30670 * symtab.c (symtab_node::get_dump_name): New function.
30671 (symtab_node::dump_name): Likewise.
30672 (symtab_node::dump_asm_name): Likewise.
30673 (symtab_node::dump_references): Likewise.
30674 (symtab_node::dump_referring): Likewise.
30675 (symtab_node::dump_base): Likewise.
30676 (symtab_node::debug_symtab): Likewise.
30677 * tree-sra.c (convert_callers_for_node): Likewise.
30678 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30679 * value-prof.c (init_node_map): Likewise.
30681 2017-05-23 Martin Liska <mliska@suse.cz>
30683 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30684 and symtab_node::debug_symtab to symbol_table::debug.
30685 * cgraphunit.c (analyze_functions): Use the renamed function.
30686 (symbol_table::compile): Likewise.
30687 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30688 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30689 * passes.c (execute_todo): Likewise.
30690 * symtab.c (symbol_table::dump): New function.
30691 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30693 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30695 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30696 that nonconst implies exec.
30698 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30700 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30701 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30702 (inline_edge_summary_vec): Turn into ...
30703 (ipa_call_summaries): ... this one.
30704 (redirect_to_unreachable, edge_set_predicate,
30705 evaluate_properties_for_edge, inline_summary_alloc,
30706 reset_ipa_call_summary, reset_inline_summary,
30707 inline_summary_t::duplicate): Update.
30708 (inline_edge_duplication_hook): Turn to ...
30709 (ipa_call_summary_t::duplicate): ... this one.
30710 (inline_edge_removal_hook): Turn to ...
30711 (ipa_call_summary_t::remove): ... this one.
30712 (dump_inline_edge_summary): Turn to ...
30713 (dump_ipa_call_summary): ... this one.
30714 (estimate_function_body_sizes): Update.
30715 (inline_update_callee_summaries): Update.
30716 (remap_edge_change_prob): Update.
30717 (remap_edge_summaries): Update.
30718 (inline_merge_summary): Update.
30719 (do_estimate_edge_time): Update.
30720 (inline_generate_summary): Update.
30721 (inline_read_section): Update.
30722 (inline_read_summary): Update.
30723 (inline_free_summary): Update.
30724 * ipa-inline.c (can_inline_edge_p): Update.
30725 (compute_inlined_call_time): Update.
30726 (want_inline_small_function_p): Update.
30727 (edge_badness): Update.
30728 (early_inliner): Update.
30729 * ipa-inline.h (inline_edge_summary): Turn to ...
30730 (ipa_call_summary): ... this one.
30731 (ipa_call_summary_t): New class.
30732 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30733 (ipa_call_summaries): New.
30734 (inline_edge_summary): Remove.
30735 (estimate_edge_growth): Update.
30736 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30737 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30738 * ipa-split.c (execute_split_functions): Update.
30739 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30741 2017-05-23 Tom de Vries <tom@codesourcery.com>
30743 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30744 attributes): Document rdrand effective target.
30746 2017-05-23 Tom de Vries <tom@codesourcery.com>
30748 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30749 attributes): Sort alphabetically.
30751 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30753 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30755 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30758 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30759 V2DF/V2DI splat into two separate patterns, one that handles
30760 registers, and the other that only handles memory. Drop support
30761 for splatting from a GPR on ISA 2.07 and then splitting the
30762 splat into direct move and splat.
30763 (vsx_splat_<mode>_reg): Likewise.
30764 (vsx_splat_<mode>_mem): Likewise.
30766 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30768 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30770 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30772 PR middle-end/80809
30773 * omp-low.c (finish_taskreg_remap): New function.
30774 (finish_taskreg_scan): If unit size of ctx->record_type
30775 is non-constant, unshare the size expression and replace
30776 decls in it with possible outer var refs.
30778 PR middle-end/80809
30779 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30780 GOVD_SHARED rather than GOVD_PRIVATE with it.
30781 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30782 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30784 PR middle-end/80853
30785 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30786 as last argument to build_outer_var_ref for pointer bases of array
30787 section reductions.
30789 2017-05-19 Martin Sebor <msebor@redhat.com>
30791 * print-tree.c (print_node): Print DECL_READ_P flag.
30793 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30795 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30796 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30797 * cgraph.c: Likewise.
30798 * cgraphunit.c: Likewise.
30799 * gengtype.c: Likewise.
30800 * ipa-cp.c: Likewise.
30801 * ipa-devirt.c: Likewise.
30802 * ipa-icf.c: Likewise.
30803 * ipa-predicate.c: Likewise.
30804 * ipa-profile.c: Likewise.
30805 * ipa-prop.c: Likewise.
30806 * ipa-split.c: Likewise.
30808 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30809 edge_predicate_pool, dump_inline_hints,
30810 inline_summary::account_size_time, redirect_to_unreachable,
30811 edge_set_predicate, set_hint_predicate,
30812 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30813 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30814 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30815 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30816 ipa_call_summary_t::remove, initialize_growth_caches,
30817 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30818 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30819 mark_modified, unmodified_parm_1, unmodified_parm,
30820 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30821 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30822 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30823 will_be_nonconstant_predicate, record_modified_bb_info,
30824 get_minimal_bb, record_modified, param_change_prob,
30825 phi_result_unknown_predicate, predicate_for_phi_result,
30826 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30827 estimate_function_body_sizes, compute_inline_parameters,
30828 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30829 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30830 inline_update_callee_summaries, remap_edge_change_prob,
30831 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30832 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30833 inline_analyze_function, inline_summary_t::insert,
30834 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30835 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30836 inline_free_summary): Move to ipa-fnsummary.h
30837 (predicate_t): Remove.
30838 * ipa-fnsummary.c: New file.
30839 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30840 (enum inline_hints_vals, inline_hints, agg_position_info,
30841 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30842 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30843 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30844 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30845 inline_read_summary, inline_write_summary, inline_free_summary,
30846 inline_analyze_function, initialize_inline_failed,
30847 inline_merge_summary, inline_update_overall_summary,
30848 compute_inline_parameters): Move to ipa-fnsummary.h
30849 * ipa-fnsummary.h: New file.
30850 * ipa-inline-transform.h: Include ipa-inline.h.
30851 * ipa-inline.c: LIkewise.
30853 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30855 * ipa-inline.c (edge_badness): Use inlined_time instead of
30856 inline_summaries->get.
30858 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30860 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30862 2017-05-22 Nathan Sidwell <nathan@acm.org>
30864 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30865 (fdump-lang): Document 'raw' option.
30866 * dumpfile.h (TDI_tu): Delete.
30867 * dumpfile.c (dump_files): Remove translation-unit.
30868 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30870 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30872 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30873 command option from $(AWK) call.
30874 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30875 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30876 [FORMAT]: Remove handling of variable.
30877 * config/avr/t-multilib: Regenerate.
30879 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30881 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30883 (dump_inline_summary): Do not print self_time.
30884 (estimate_function_body_sizes): Do not set self_time.
30885 (compute_inline_parameters): Likewise.
30886 (inline_read_section, inline_write_summary): Do not stream self_time.
30887 * ipa-inline.h (inline_summary): Drop self_time.
30889 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30891 * ipa-inline-analysis.c (account_size_time): Rename to ...
30892 (inline_summary::account_size_time): ... this one.
30893 (reset_ipa_call_summary): Turn to ...
30894 (ipa_call_summary::reset): ... this one.
30895 (reset_inline_summary): Turn to ...
30896 (inline_summary::reset): ... this one.
30897 (inline_summary_t::remove): Update.
30898 (inline_summary_t::duplicate): Update.
30899 (ipa_call_summary_t::remove): Update.
30900 (dump_inline_summary): Update.
30901 (estimate_function_body_sizes): Update.
30902 (compute_inline_parameters): Update.
30903 (estimate_node_size_and_time): Update.
30904 (inline_merge_summary): Update.
30905 (inline_update_overall_summary): Update.
30906 (inline_read_section): Update.
30907 (inline_write_summary): Update.
30908 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30909 add account_size_time and reset member functions.
30910 (ipa_call_summary): Add reset function.
30911 * ipa-predicate.h (predicate::operator &): Constify.
30913 2017-05-22 Richard Biener <rguenther@suse.de>
30915 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30917 2017-05-19 Jason Merrill <jason@redhat.com>
30919 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30921 2017-05-19 Marek Polacek <polacek@redhat.com>
30924 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30925 TYPE_OVERFLOW_WRAPS checks.
30927 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30929 * tree-core.h (enum omp_clause_default_kind): Add
30930 "OMP_CLAUSE_DEFAULT_PRESENT".
30931 * tree-pretty-print.c (dump_omp_clause): Handle it.
30932 * gimplify.c (enum gimplify_omp_var_data): Add
30933 "GOVD_MAP_FORCE_PRESENT".
30934 (gimplify_adjust_omp_clauses_1): Map it to
30935 "GOMP_MAP_FORCE_PRESENT".
30936 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30938 * gimplify.c (oacc_default_clause): Clarify.
30940 2017-05-19 Nathan Sidwell <nathan@acm.org>
30942 LANG_HOOK_REGISTER_DUMPS
30943 * toplev.c (general_init): Call register dump lang hook.
30944 * doc/invoke.texi: Document -fdump-lang option family.
30945 * dumpfile.c (dump_files): Remove class dump here.
30946 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30947 * dumpfile.h (tree_dump_index): Remove TDI_class.
30948 * langhooks-def.h (lhd_register_dumps): Declare.
30949 (LANG_HOOKS_REGISTER_DUMPS): Define.
30950 (LANG_HOOKS_INITIALIZER): Add it.
30951 * langhooks.c (lhd_register_dumps): Define.
30952 * langhooks.h (struct lang_hooks): Add register_dumps.
30954 2017-05-19 Nathan Sidwell <nathan@acm.org>
30956 * context.h (context::set_passes): New.
30957 * context.c (context::context): Do not create pass manager.
30958 * toplev.c (general_init): Create pass manager here.
30960 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30962 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30963 use this splitter if two add or or instructions would also work for
30964 the constant we want to generate.
30966 2017-05-19 Richard Biener <rguenther@suse.de>
30969 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30970 predicate evaluation.
30972 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30974 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30976 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30979 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30981 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30982 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30983 (inline_edge_summary_vec): Turn into ...
30984 (ipa_call_summaries): ... this one.
30985 (redirect_to_unreachable, edge_set_predicate,
30986 evaluate_properties_for_edge, inline_summary_alloc,
30987 reset_ipa_call_summary, reset_inline_summary,
30988 inline_summary_t::duplicate): Update.
30989 (inline_edge_duplication_hook): Turn to ...
30990 (ipa_call_summary_t::duplicate): ... this one.
30991 (inline_edge_removal_hook): Turn to ...
30992 (ipa_call_summary_t::remove): ... this one.
30993 (dump_inline_edge_summary): Turn to ...
30994 (dump_ipa_call_summary): ... this one.
30995 (estimate_function_body_sizes): Update.
30996 (inline_update_callee_summaries): Update.
30997 (remap_edge_change_prob): Update.
30998 (remap_edge_summaries): Update.
30999 (inline_merge_summary): Update.
31000 (do_estimate_edge_time): Update.
31001 (inline_generate_summary): Update.
31002 (inline_read_section): Update.
31003 (inline_read_summary): Update.
31004 (inline_free_summary): Update.
31005 * ipa-inline.c (can_inline_edge_p): Update.
31006 (compute_inlined_call_time): Update.
31007 (want_inline_small_function_p): Update.
31008 (edge_badness): Update.
31009 (early_inliner): Update.
31010 * ipa-inline.h (inline_edge_summary): Turn to ...
31011 (ipa_call_summary): ... this one.
31012 (ipa_call_summary_t): New class.
31013 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
31014 (ipa_call_summaries): New.
31015 (inline_edge_summary): Remove.
31016 (estimate_edge_growth): Update.
31017 * ipa-profile.c (ipa_propagate_frequency_1): Update.
31018 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
31019 * ipa-split.c (execute_split_functions): Update.
31020 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
31022 2017-05-19 Richard Biener <rguenther@suse.de>
31024 PR middle-end/80764
31025 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
31027 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
31029 * config/rs6000/rs6000.c (struct machine_function): Add field
31030 fpr_is_wrapped_separately.
31031 (rs6000_get_separate_components): Use 64 components. Handle the
31032 new FPR components.
31033 (rs6000_components_for_bb): Handle the FPR components.
31034 (rs6000_emit_prologue_components): Handle the FPR components.
31035 (rs6000_emit_epilogue_components): Handle the FPR components.
31036 (rs6000_set_handled_components): Handle the FPR components.
31037 (rs6000_emit_prologue): Don't output prologue code for those FPRs
31038 that are already separately shrink-wrapped.
31039 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
31040 that are already separately shrink-wrapped.
31042 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31045 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
31048 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
31049 (define_peephole2 for Altivec d-form load): Add peepholes to catch
31050 cases where the register allocator uses a move and an offsettable
31051 memory operation to/from a FPR register on ISA 2.06/2.07.
31052 (define_peephole2 for Altivec d-form store): Likewise.
31054 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
31057 * config/i386/mmx.md (*mov<mode>_internal): Enable
31058 alternatives 11, 12, 13 and 14 also for 32bit targets.
31059 Remove alternatives 15, 16, 17 and 18.
31060 * config/i386/sse.md (vec_concatv2di): Change
31061 alternative (!x, *y) to (x, ?!*Yn).
31063 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
31065 * dumpfile.h (enum dump_kind): Remove stray comma.
31067 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31069 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
31070 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
31071 predicate::num_conditions
31072 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
31073 (CHANGED): turn into predicate::changed.
31074 (agg_position_info): Move to ipa-predicate.h
31075 (add_condition, predicate::add_clause, predicate::operator &=,
31076 predicate::or_with, predicate::evaluate, predicate::probability,
31077 dump_condition, dump_clause, predicate::dump,
31078 predicate::remap_after_duplication, predicate::remap_after_inlining,
31079 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
31080 (evaluate_conditions_for_known_args): Update.
31081 (set_cond_stmt_execution_predicate): Update.
31082 * ipa-inline.h: Include ipa-predicate.h
31083 (condition, inline_param_summary, conditions, agg_position_info,
31084 predicate): Move to ipa-predicate.h
31085 * ipa-predicate.c: New file.
31086 * ipa-predicate.h: New file.
31088 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
31090 * final.c (leaf_function_p): Check we are not in a sequence.
31092 2017-05-18 Martin Liska <mliska@suse.cz>
31094 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
31095 * dumpfile.c (dump_register): Use new enum dump_kind.
31096 (get_dump_file_name): Likewise.
31097 (dump_enable_all): Likewise.
31098 (dump_switch_p_1): Likewise.
31099 (enable_rtl_dump_file): Remove usage of TDF_RTL.
31100 * dumpfile.h (enum dump_kind): New enum type.
31101 (struct dump_file_info): Create constructor and
31102 format fields and comments.
31103 * passes.c (pass_manager::register_one_dump_file):
31105 * statistics.c (statistics_early_init): Likewise.
31106 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
31107 TDF_TREE with TDF_SLIM.
31108 (gather_memory_references_ref): Likewise.
31110 2017-05-18 Martin Liska <mliska@suse.cz>
31112 * vec.h (struct vnull): Use it.
31114 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31116 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
31117 (true_predicate, false_predicate, true_predicate_p,
31118 false_predicate_p): Remove.
31119 (single_cond_predicate, not_inlined_predicate): Turn to member function
31121 (add_condition): Update.
31122 (add_clause): Turn to...
31123 (predicate::add_clause): ... this one; update; allow passing NULL
31125 (and_predicates): Turn to ...
31126 (predicate::operator &=): ... this one.
31127 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
31128 (or_predicates): Turn to ...
31129 (predicate::or_with): ... this one.
31130 (evaluate_predicate): Turn to ...
31131 (predicate::evaluate): ... this one.
31132 (predicate_probability): Turn to ...
31133 (predicate::probability): ... this one.
31134 (dump_condition): Update.
31135 (dump_predicate): Turn to ...
31136 (predicate::dump): ... this one.
31137 (account_size_time): Update.
31138 (edge_set_predicate): Update.
31139 (set_hint_predicate): UPdate.
31140 (evaluate_conditions_for_known_args): Update.
31141 (evaluate_properties_for_edge): Update.
31142 (remap_predicate_after_duplication): Turn to...
31143 (predicate::remap_after_duplication): ... this one.
31144 (remap_hint_predicate_after_duplication): Update.
31145 (inline_summary_t::duplicate): UPdate.
31146 (dump_inline_edge_summary): Update.
31147 (dump_inline_summary): Update.
31148 (set_cond_stmt_execution_predicate): Update.
31149 (set_switch_stmt_execution_predicate): Update.
31150 (compute_bb_predicates): Update.
31151 (will_be_nonconstant_expr_predicate): Update.
31152 (will_be_nonconstant_predicate): Update.
31153 (phi_result_unknown_predicate): Update.
31154 (predicate_for_phi_result): Update.
31155 (array_index_predicate): Update.
31156 (estimate_function_body_sizes): Update.
31157 (estimate_node_size_and_time): Update.
31158 (estimate_ipcp_clone_size_and_time): Update.
31159 (remap_predicate): Rename to ...
31160 (predicate::remap_after_inlining): ... this one.
31161 (remap_hint_predicate): Update.
31162 (inline_merge_summary): Update.
31163 (inline_update_overall_summary): Update.
31164 (estimate_size_after_inlining): Update.
31165 (read_predicate): Rename to ...
31166 (predicate::stream_in): ... this one.
31167 (read_inline_edge_summary): Update.
31168 (write_predicate): Rename to ...
31169 (predicate::stream_out): ... this one.
31170 (write_inline_edge_summary): Update.
31171 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
31172 (clause_t): Turn to uint32_t
31173 (predicate): Turn to class; implement constructor and operators
31175 (size_time_entry): Update.
31176 (inline_summary): Update.
31177 (inline_edge_summary): Update.
31179 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
31181 * fold-const.c (fold_binary_loc): Move transformation...
31182 * match.pd (C - X CMP X): ... here.
31184 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
31186 * config/sparc/sparc.c (sparc_option_override): Set function
31187 alignment for -mcpu=niagara7 to 64 to match the I$ line.
31188 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
31190 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
31192 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
31194 2017-05-18 Marek Polacek <polacek@redhat.com>
31197 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
31198 (pass_ubsan::execute): Call gimple_assign_single_p instead of
31199 gimple_assign_load_p.
31201 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
31203 PR middle-end/80692
31204 * real.c (do_compare): Give decimal_do_compare preference over
31205 comparing just the signs.
31207 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
31209 * doc/md.texi (Canonicalization of Instructions): Describe the
31210 canonical form of instructions that inherently set a condition
31213 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
31215 PR middle-end/80775
31216 * tree-cfg.c: Move deletion of unreachable case statements to after
31217 the merging of consecutive case labels.
31219 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
31221 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
31222 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31223 restoring of callee-saved registers.
31225 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31227 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31228 * config/visium/visium.c (single_set_and_flags): Likewise.
31229 * config/visium/visium.md (Substitutions): Likewise.
31231 2017-05-17 Martin Liska <mliska@suse.cz>
31233 * cfg.c: Introduce dump_flags_t type and
31234 use it instead of int type.
31236 * cfghooks.c: Likewise.
31237 * cfghooks.h (struct cfg_hooks): Likewise.
31238 * cfgrtl.c: Likewise.
31239 * cfgrtl.h: Likewise.
31240 * cgraph.c (cgraph_node::get_body): Likewise.
31241 * coretypes.h: Likewise.
31242 * domwalk.c: Likewise.
31243 * domwalk.h: Likewise.
31244 * dumpfile.c (struct dump_option_value_info): Likewise.
31245 (dump_enable_all): Likewise.
31246 (dump_switch_p_1): Likewise.
31247 (opt_info_switch_p): Likewise.
31248 * dumpfile.h (enum tree_dump_index): Likewise.
31249 (struct dump_file_info): Likewise.
31250 * genemit.c: Likewise.
31251 * generic-match-head.c: Likewise.
31252 * gengtype.c (open_base_files): Likewise.
31253 * gimple-pretty-print.c: Likewise.
31254 * gimple-pretty-print.h: Likewise.
31255 * graph.c (print_graph_cfg): Likewise.
31256 * graphite-scop-detection.c (dot_all_sese): Likewise.
31257 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31258 * loop-unroll.c (report_unroll): Likewise.
31259 * passes.c (pass_manager::register_one_dump_file): Likewise.
31260 * print-tree.c: Likewise.
31261 * statistics.c: Likewise.
31262 * tree-cfg.c: Likewise.
31263 * tree-cfg.h: Likewise.
31264 * tree-dfa.c: Likewise.
31265 * tree-dfa.h: Likewise.
31266 * tree-dump.c (dump_function): Likewise.
31267 * tree-dump.h (struct dump_info): Likewise.
31268 * tree-pretty-print.c: Likewise.
31269 * tree-pretty-print.h: Likewise.
31270 * tree-ssa-live.c: Likewise.
31271 * tree-ssa-live.h: Likewise.
31272 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31273 * tree-vect-loop.c: Likewise.
31274 * tree-vect-slp.c: Likewise.
31276 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31277 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31279 PR tree-optimization/80457
31280 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31281 of all arguments to a statement as scalar_to_vec operations.
31282 (vectorizable_call): Adjust call to vect_model_simple_cost for
31284 (vectorizable_conversion): Likewise.
31285 (vectorizable_assignment): Likewise.
31286 (vectorizable_shift): Likewise.
31287 (vectorizable_operation): Likewise.
31288 (vectorizable_comparison): Likewise.
31289 (vect_is_simple_cond): Record the def types for operands.
31290 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31291 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31292 for statement argument count.
31294 2017-05-16 Carl Love <cel@us.ibm.com>
31296 * config/rs6000/rs6000-c: Add support for built-in functions
31297 vector unsigned long long vec_bperm (vector unsigned long long,
31298 vector unsigned char)
31299 vector signed long long vec_mule (vector signed int,
31301 vector unsigned long long vec_mule (vector unsigned int,
31302 vector unsigned int)
31303 vector signed long long vec_mulo (vector signed int,
31305 vector unsigned long long vec_mulo (vector unsigned int,
31306 vector unsigned int)
31307 vector signed char vec_sldw (vector signed char,
31308 vector signed char,
31310 vector unsigned char vec_sldw (vector unsigned char,
31311 vector unsigned char,
31313 vector signed short vec_sldw (vector signed short,
31314 vector signed short,
31316 vector unsigned short vec_sldw (vector unsigned short,
31317 vector unsigned short,
31319 vector signed int vec_sldw (vector signed int,
31322 vector unsigned int vec_sldw (vector unsigned int,
31323 vector unsigned int,
31325 vector signed long long vec_sldw (vector signed long long,
31326 vector signed long long,
31328 vector unsigned long long vec_sldw (vector unsigned long long,
31329 vector unsigned long long,
31331 * config/rs6000/rs6000-c: Add support for built-in functions
31332 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31333 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31334 * doc/extend.texi: Update the built-in documentation for the
31335 new built-in functions.
31337 2017-05-16 Marek Polacek <polacek@redhat.com>
31341 * tree.c (save_expr): Don't fold the expression.
31343 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31345 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31346 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31347 and (?*y,m). Update insn attributes.
31349 2017-05-16 Martin Liska <mliska@suse.cz>
31351 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31352 flags argument of print_gimple_stmt, print_gimple_expr,
31353 print_generic_stmt and print_generic_expr.
31354 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31355 * coretypes.h: Likewise.
31356 * except.c (dump_eh_tree): Likewise.
31357 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31358 * gimple-pretty-print.h: Likewise.
31359 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31360 (backprop::push_to_worklist): Likewise.
31361 (backprop::pop_from_worklist): Likewise.
31362 (backprop::process_use): Likewise.
31363 (backprop::intersect_uses): Likewise.
31364 (note_replacement): Likewise.
31365 * gimple-ssa-store-merging.c
31366 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31367 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31368 (pass_store_merging::execute): Likewise.
31369 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31370 (ssa_base_cand_dump_callback): Likewise.
31371 (dump_incr_vec): Likewise.
31372 (replace_refs): Likewise.
31373 (replace_mult_candidate): Likewise.
31374 (create_add_on_incoming_edge): Likewise.
31375 (create_phi_basis): Likewise.
31376 (insert_initializers): Likewise.
31377 (all_phi_incrs_profitable): Likewise.
31378 (introduce_cast_before_cand): Likewise.
31379 (replace_one_candidate): Likewise.
31380 * gimplify.c (gimplify_expr): Likewise.
31381 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31382 (set_rename): Likewise.
31383 (rename_uses): Likewise.
31384 (copy_loop_phi_nodes): Likewise.
31385 (add_close_phis_to_merge_points): Likewise.
31386 (copy_loop_close_phi_args): Likewise.
31387 (copy_cond_phi_args): Likewise.
31388 (graphite_copy_stmts_from_block): Likewise.
31389 (translate_pending_phi_nodes): Likewise.
31390 * graphite-poly.c (print_pdr): Likewise.
31391 (dump_gbb_cases): Likewise.
31392 (dump_gbb_conditions): Likewise.
31393 (print_scop_params): Likewise.
31394 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31395 (build_cross_bb_scalars_use): Likewise.
31396 (gather_bbs::before_dom_children): Likewise.
31397 * hsa-dump.c (dump_hsa_immed): Likewise.
31398 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31399 (get_replacement_map): Likewise.
31400 * ipa-inline-analysis.c (dump_condition): Likewise.
31401 (estimate_function_body_sizes): Likewise.
31402 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31403 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31404 * ipa-prop.c (ipa_dump_param): Likewise.
31405 (ipa_print_node_jump_functions_for_edge): Likewise.
31406 (ipa_modify_call_arguments): Likewise.
31407 (ipa_modify_expr): Likewise.
31408 (ipa_dump_param_adjustments): Likewise.
31409 (ipa_dump_agg_replacement_values): Likewise.
31410 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31411 * ipa-pure-const.c (check_stmt): Likewise.
31412 (pass_nothrow::execute): Likewise.
31413 * ipa-split.c (execute_split_functions): Likewise.
31414 * omp-offload.c (dump_oacc_loop_part): Likewise.
31415 (dump_oacc_loop): Likewise.
31416 * trans-mem.c (tm_log_emit): Likewise.
31417 (tm_memopt_accumulate_memops): Likewise.
31418 (dump_tm_memopt_set): Likewise.
31419 (dump_tm_memopt_transform): Likewise.
31420 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31421 (print_loop): Likewise.
31422 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31423 (chkp_gather_checks_info): Likewise.
31424 (chkp_get_check_result): Likewise.
31425 (chkp_remove_check_if_pass): Likewise.
31426 (chkp_use_outer_bounds_if_possible): Likewise.
31427 (chkp_reduce_bounds_lifetime): Likewise.
31428 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31429 (chkp_mark_completed_bounds): Likewise.
31430 (chkp_register_incomplete_bounds): Likewise.
31431 (chkp_mark_invalid_bounds): Likewise.
31432 (chkp_maybe_copy_and_register_bounds): Likewise.
31433 (chkp_build_returned_bound): Likewise.
31434 (chkp_get_bound_for_parm): Likewise.
31435 (chkp_build_bndldx): Likewise.
31436 (chkp_get_bounds_by_definition): Likewise.
31437 (chkp_generate_extern_var_bounds): Likewise.
31438 (chkp_get_bounds_for_decl_addr): Likewise.
31439 * tree-chrec.c (chrec_apply): Likewise.
31440 * tree-data-ref.c (dump_data_reference): Likewise.
31441 (dump_subscript): Likewise.
31442 (dump_data_dependence_relation): Likewise.
31443 (analyze_overlapping_iterations): Likewise.
31444 * tree-inline.c (expand_call_inline): Likewise.
31445 (tree_function_versioning): Likewise.
31446 * tree-into-ssa.c (dump_defs_stack): Likewise.
31447 (dump_currdefs): Likewise.
31448 (dump_names_replaced_by): Likewise.
31449 (dump_update_ssa): Likewise.
31450 (update_ssa): Likewise.
31451 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31452 * tree-parloops.c (build_new_reduction): Likewise.
31453 (try_create_reduction_list): Likewise.
31454 (ref_conflicts_with_region): Likewise.
31455 (oacc_entry_exit_ok_1): Likewise.
31456 (oacc_entry_exit_single_gang): Likewise.
31457 * tree-pretty-print.h: Likewise.
31458 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31459 (get_scalar_evolution): Likewise.
31460 (add_to_evolution): Likewise.
31461 (get_loop_exit_condition): Likewise.
31462 (analyze_evolution_in_loop): Likewise.
31463 (analyze_initial_condition): Likewise.
31464 (analyze_scalar_evolution): Likewise.
31465 (instantiate_scev): Likewise.
31466 (number_of_latch_executions): Likewise.
31467 (gather_chrec_stats): Likewise.
31468 (final_value_replacement_loop): Likewise.
31469 (scev_const_prop): Likewise.
31470 * tree-sra.c (dump_access): Likewise.
31471 (disqualify_candidate): Likewise.
31472 (create_access): Likewise.
31473 (reject): Likewise.
31474 (maybe_add_sra_candidate): Likewise.
31475 (create_access_replacement): Likewise.
31476 (analyze_access_subtree): Likewise.
31477 (analyze_all_variable_accesses): Likewise.
31478 (sra_modify_assign): Likewise.
31479 (initialize_constant_pool_replacements): Likewise.
31480 (find_param_candidates): Likewise.
31481 (decide_one_param_reduction): Likewise.
31482 (replace_removed_params_ssa_names): Likewise.
31483 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31484 * tree-ssa-copy.c (dump_copy_of): Likewise.
31485 (copy_prop_visit_cond_stmt): Likewise.
31486 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31487 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31488 (record_equivalences_from_stmt): Likewise.
31489 * tree-ssa-dse.c (compute_trims): Likewise.
31490 (delete_dead_call): Likewise.
31491 (delete_dead_assignment): Likewise.
31492 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31493 (forward_propagate_into_cond): Likewise.
31494 (pass_forwprop::execute): Likewise.
31495 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31496 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31498 (move_computations_worker): Likewise.
31499 (execute_sm): Likewise.
31500 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31501 (remove_exits_and_undefined_stmts): Likewise.
31502 (remove_redundant_iv_tests): Likewise.
31503 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31504 (adjust_iv_update_pos): Likewise.
31505 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31506 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31507 (value_replacement): Likewise.
31508 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31509 * tree-ssa-pre.c (print_pre_expr): Likewise.
31510 (get_representative_for): Likewise.
31511 (create_expression_by_pieces): Likewise.
31512 (insert_into_preds_of_block): Likewise.
31513 (eliminate_insert): Likewise.
31514 (eliminate_dom_walker::before_dom_children): Likewise.
31515 (eliminate): Likewise.
31516 (remove_dead_inserted_code): Likewise.
31517 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31518 * tree-ssa-reassoc.c (get_rank): Likewise.
31519 (eliminate_duplicate_pair): Likewise.
31520 (eliminate_plus_minus_pair): Likewise.
31521 (eliminate_not_pairs): Likewise.
31522 (undistribute_ops_list): Likewise.
31523 (eliminate_redundant_comparison): Likewise.
31524 (update_range_test): Likewise.
31525 (optimize_range_tests_var_bound): Likewise.
31526 (optimize_vec_cond_expr): Likewise.
31527 (rewrite_expr_tree): Likewise.
31528 (rewrite_expr_tree_parallel): Likewise.
31529 (linearize_expr): Likewise.
31530 (break_up_subtract): Likewise.
31531 (linearize_expr_tree): Likewise.
31532 (attempt_builtin_powi): Likewise.
31533 (attempt_builtin_copysign): Likewise.
31534 (transform_stmt_to_copy): Likewise.
31535 (transform_stmt_to_multiply): Likewise.
31536 (dump_ops_vector): Likewise.
31537 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31538 (print_scc): Likewise.
31539 (set_ssa_val_to): Likewise.
31540 (visit_reference_op_store): Likewise.
31541 (visit_use): Likewise.
31542 (sccvn_dom_walker::before_dom_children): Likewise.
31543 (run_scc_vn): Likewise.
31544 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31546 (expr_hash_elt::print): Likewise.
31547 (const_and_copies::pop_to_marker): Likewise.
31548 (const_and_copies::record_const_or_copy_raw): Likewise.
31549 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31550 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31551 (dump_predicates): Likewise.
31552 (find_uninit_use): Likewise.
31553 (warn_uninitialized_phi): Likewise.
31554 (pass_late_warn_uninitialized::execute): Likewise.
31555 * tree-ssa.c (verify_vssa): Likewise.
31556 (verify_ssa): Likewise.
31557 (maybe_optimize_var): Likewise.
31558 * tree-vrp.c (dump_value_range): Likewise.
31559 (dump_all_value_ranges): Likewise.
31560 (dump_asserts_for): Likewise.
31561 (register_edge_assert_for_2): Likewise.
31562 (vrp_visit_cond_stmt): Likewise.
31563 (vrp_visit_switch_stmt): Likewise.
31564 (vrp_visit_stmt): Likewise.
31565 (vrp_visit_phi_node): Likewise.
31566 (simplify_cond_using_ranges_1): Likewise.
31567 (fold_predicate_in): Likewise.
31568 (evrp_dom_walker::before_dom_children): Likewise.
31569 (evrp_dom_walker::push_value_range): Likewise.
31570 (evrp_dom_walker::pop_value_range): Likewise.
31571 (execute_early_vrp): Likewise.
31573 2017-05-16 Richard Biener <rguenther@suse.de>
31575 * dwarf2out.c (loc_list_from_tree_1): Do not create
31576 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31578 2017-05-16 Richard Biener <rguenther@suse.de>
31580 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31582 (note_variable_value_in_expr): If we resolved the decl ref
31583 do not push to the stack.
31585 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31587 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31588 operations in fast-math mode.
31589 (vaddq_f16): Likewise.
31590 (vmul_f16): Likewise.
31591 (vmulq_f16): Likewise.
31592 (vsub_f16): Likewise.
31593 (vsubq_f16): Likewise.
31594 * config/arm/neon.md (add<mode>3): New.
31596 (fma:<VH:mode>3): New. Also remove outdated comment.
31599 2017-05-16 Martin Liska <mliska@suse.cz>
31603 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31604 (odr_types_equivalent_p): Likewise.
31606 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31608 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31610 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31613 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31614 non-interunit SSE move alternatives with '?'.
31615 (zero-extendsidi peephole2): New peephole to skip intermediate
31616 general register in SSE zero-extend sequence.
31618 2017-05-15 Jeff Law <law@redhat.com>
31620 * reorg.c (relax_delay_slots): Create a new variable to hold
31621 the temporary target rather than clobbering TARGET_LABEL.
31623 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31624 missing argument to extract_bit_field call.
31625 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31627 2017-05-15 Martin Liska <mliska@suse.cz>
31630 * gcc.c (process_command): Do not allow empty argument of -o option.
31632 2017-05-15 Renlin Li <renlin.li@arm.com>
31634 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31635 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31636 * config/aarch64/constraints.md (Usf): Add long call check.
31637 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31638 (call_value): Likewise.
31639 (sibcall): Likewise.
31640 (sibcall_value): Likewise.
31642 (call_value_insn): New.
31643 (sibcall_insn): Update rtx pattern.
31644 (sibcall_value_insn): Likewise.
31645 (call_internal): Remove.
31646 (call_value_internal): Likewise.
31647 (sibcall_internal): Likewise.
31648 (sibcall_value_internal): Likewise.
31649 (call_reg): Likewise.
31650 (call_symbol): Likewise.
31651 (call_value_reg): Likewise.
31652 (call_value_symbol): Likewise.
31654 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31657 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31659 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31661 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31662 compatible with CCGOCmode and with CCZmode.
31664 2017-05-14 Martin Sebor <msebor@redhat.com>
31666 PR middle-end/77671
31667 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31668 (gimple_fold_builtin_snprintf): Same.
31669 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31670 (gimple_fold_builtin_snprintf): Same.
31671 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31672 of character types.
31673 (is_call_safe): New function.
31674 (try_substitute_return_value): Call it.
31675 (try_simplify_call): New function.
31676 (pass_sprintf_length::handle_gimple_call): Call it.
31678 2017-05-14 Martin Sebor <msebor@redhat.com>
31680 PR middle-end/80669
31681 * builtins.c (expand_builtin_stpncpy): Simplify.
31683 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31685 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31686 * config/i386/i386.h
31687 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31688 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31689 (struct machine_function): Add new members call_ms2sysv,
31690 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31691 (struct machine_frame_state): New fields sp_realigned and
31692 sp_realigned_offset.
31693 * config/i386/i386.c
31694 (enum xlogue_stub): New enum.
31695 (enum xlogue_stub_sets): New enum.
31696 (class xlogue_layout): New class.
31697 (struct ix86_frame): New fields stack_realign_allocate_offset,
31698 stack_realign_offset and outlined_save_offset. Modify comments to
31699 detail stack layout when using out-of-line stubs.
31700 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31701 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31702 -mcall-ms2sysv-xlogues.
31703 (stub_managed_regs): New static variable.
31704 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31705 registers managed by out-of-line stub.
31706 (disable_call_ms2sysv_xlogues): New function.
31707 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31708 m->call_ms2sysv when appropriate and compute frame layout for
31710 (sp_valid_at, fp_valid_at): New inline functions.
31711 (choose_basereg): New function.
31712 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31714 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31715 Use align parameter of choose_baseaddr to generated aligned SSE movs
31717 (pro_epilogue_adjust_stack): Modify to track
31718 machine_frame_state::sp_realigned.
31719 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31720 (ix86_nsaved_sseregs): Likewise.
31721 (ix86_emit_save_regs): Likewise.
31722 (ix86_emit_save_regs_using_mov): Likewise.
31723 (ix86_emit_save_sse_regs_using_mov): Likewise.
31724 (get_scratch_register_on_entry): Likewise.
31725 (gen_frame_set): New function.
31726 (gen_frame_load): Likewise.
31727 (gen_frame_store): Likewise.
31728 (emit_outlined_ms2sysv_save): Likewise.
31729 (emit_outlined_ms2sysv_restore): Likewise.
31730 (ix86_expand_prologue): Modify stack re-alignment code and call
31731 emit_outlined_ms2sysv_save when appropriate.
31732 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31733 parameter rtx_insn *insn, which allows the function to be used to only
31734 generate the notes.
31735 (ix86_expand_epilogue): Modify validity checks of frame and stack
31736 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31737 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31738 * config/i386/predicates.md
31739 (save_multiple): New predicate.
31740 (restore_multiple): Likewise.
31741 * config/i386/sse.md
31742 (save_multiple<mode>): New pattern.
31743 (save_multiple_realign<mode>): Likewise.
31744 (restore_multiple<mode>): Likewise.
31745 (restore_multiple_and_return<mode>): Likewise.
31746 (restore_multiple_leave_return<mode>): Likewise.
31747 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31749 2017-05-14 Julia Koval <julia.koval@intel.com>
31751 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31752 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31753 (__builtin_ia32_xsetbv): New builtins.
31754 * config/i386/i386.c (ix86_expand_special_args_builtin):
31756 (ix86_expand_builtin): Special expand for new intrinsics.
31757 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31758 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31759 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31761 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31763 * cfganal.c (inverted_post_order_compute): Change argument type
31765 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31766 * df-core.c (rest_of_handle_df_initialize): Adjust.
31767 (rest_of_handle_df_finish): Likewise.
31768 (df_analyze_1): Likewise.
31769 (df_analyze): Likewise.
31770 (loop_inverted_post_order_compute): Change argument to be a vec *.
31771 (df_analyze_loop): Adjust.
31772 (df_get_n_blocks): Likewise.
31773 (df_get_postorder): Likewise.
31774 * df.h (struct df_d): Change field to be a vec.
31775 * lcm.c (compute_laterin): Adjust.
31776 (compute_available): Likewise.
31777 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31778 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31779 * tree-ssa-pre.c (compute_antic): Likewise.
31781 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31783 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31784 (depth_first_search::depth_first_search): Change structure init
31785 function to this constructor.
31786 (depth_first_search::add_bb): Rename function to this member.
31787 (depth_first_search::execute): Likewise.
31788 (flow_dfs_compute_reverse_finish): Adjust.
31790 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31792 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31793 (longest_simple_path): Likewise.
31794 * shrink-wrap.c (spread_components): Likewise.
31795 (disqualify_problematic_components): Likewise.
31796 (emit_common_heads_for_components): Likewise.
31797 (emit_common_tails_for_components): Likewise.
31798 (insert_prologue_epilogue_for_components): Likewise.
31800 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31802 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31805 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31807 * df-core.c (df_set_blocks): Start using auto_bitmap.
31808 (df_compact_blocks): Likewise.
31809 * df-problems.c (df_rd_confluence_n): Likewise.
31810 * df-scan.c (df_insn_rescan_all): Likewise.
31811 (df_process_deferred_rescans): Likewise.
31812 (df_update_entry_block_defs): Likewise.
31813 (df_update_exit_block_uses): Likewise.
31814 (df_entry_block_bitmap_verify): Likewise.
31815 (df_exit_block_bitmap_verify): Likewise.
31816 (df_scan_verify): Likewise.
31817 * lra-constraints.c (lra_constraints): Likewise.
31818 (undo_optional_reloads): Likewise.
31819 (lra_undo_inheritance): Likewise.
31820 * lra-remat.c (calculate_gen_cands): Likewise.
31821 (do_remat): Likewise.
31822 * lra-spills.c (assign_spill_hard_regs): Likewise.
31823 (spill_pseudos): Likewise.
31824 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31825 (bitmap_set_subtract_values): Likewise.
31827 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31829 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31830 management with auto_bitmap.
31831 (fix_inter_tick): Likewise.
31832 (fix_recovery_deps): Likewise.
31833 * ira.c (add_store_equivs): Likewise.
31834 (find_moveable_pseudos): Likewise.
31835 (split_live_ranges_for_shrink_wrap): Likewise.
31836 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31837 (rtx_reuse_manager::seen_def_p): Likewise.
31838 (rtx_reuse_manager::set_seen_def): Likewise.
31839 * print-rtl.h (class rtx_reuse_manager): Likewise.
31841 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31843 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31845 (migrate_btr_def): Likewise.
31846 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31847 * df-core.c (loop_post_order_compute): Likewise.
31848 (loop_inverted_post_order_compute): Likewise.
31849 * hsa-common.h: Likewise.
31850 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31851 * init-regs.c (initialize_uninitialized_regs): Likewise.
31852 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31853 (inline_small_functions): Likewise.
31854 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31855 * ira.c (combine_and_move_insns): Likewise.
31856 (build_insn_chain): Likewise.
31857 * loop-invariant.c (find_invariants): Likewise.
31858 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31859 * predict.c (tree_predict_by_opcode): Likewise.
31860 (predict_paths_leading_to): Likewise.
31861 (predict_paths_leading_to_edge): Likewise.
31862 (estimate_loops_at_level): Likewise.
31863 (estimate_loops): Likewise.
31864 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31865 (spread_components): Likewise.
31866 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31867 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31868 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31869 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31870 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31871 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31872 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31873 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31874 (mark_threaded_blocks): Likewise.
31875 (thread_through_all_blocks): Likewise.
31876 * tree-ssa.c (verify_ssa): Likewise.
31877 (execute_update_addresses_taken): Likewise.
31878 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31880 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31882 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31884 (post_order_compute): Likewise.
31885 (inverted_post_order_compute): Likewise.
31886 (pre_and_rev_post_order_compute_fn): Likewise.
31888 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31890 * genrecog.c (int_set::int_set): Explicitly construct our
31891 auto_vec base class.
31892 * vec.h (auto_vec::auto_vec): New constructor.
31894 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31896 * bitmap.h (class auto_bitmap): New constructor taking
31897 bitmap_obstack * argument.
31899 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31901 * bitmap.h (class auto_bitmap): Change type of m_bits to
31902 bitmap_head, and adjust ctor / dtor and member operators.
31904 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31906 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31907 when returned register mode doesn't match original mode.
31909 2017-05-12 Jeff Law <law@redhat.com>
31910 Jakub Jelinek <jakub@redhat.com>
31912 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31913 we look for cc setter after the compare-elim changes.
31914 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31915 within the vector to match what compare-elim now expects.
31916 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31917 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31919 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31920 after the compare-elim changes.
31921 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31922 the vector to match what compare-elim now expects.
31923 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31924 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31925 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31926 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31927 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31929 * config/visium/visium.c (single_set_and_flags): Fix where
31930 we look for cc setter after the compare-elim changes.
31931 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31932 with the vector to match what compare-elim now expects.
31933 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31934 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31935 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31936 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31937 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31938 (neg<mode>2_insn_set_overflow): Likewise.
31940 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31942 PR middle-end/79794
31943 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31944 maybe_expand_insn call, set ops[0].target. If still set after call,
31945 set alt_rtl. Add extra arg to recursive calls.
31946 (extract_bit_field): Add alt_rtl argument. Pass to
31948 * expmed.h (extract_bit_field): Fix prototype.
31949 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31950 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31951 to extract_bit_field_calls.
31952 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31953 Pass alt_rtl to extract_bit_field calls.
31954 * calls.c (store_unaligned_arguments_into_psuedos)
31955 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31956 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31958 * optabs.h (struct expand_operand): Add target bitfield.
31960 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31962 * compare-elim.c (try_eliminate_compare): Canonicalize
31963 operation with embedded compare to
31964 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31965 (set (reg) (operation)].
31967 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31969 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31972 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31973 cost of adding a carry flag for ADC instruction.
31974 [case MINUS]: Ignore the cost of subtracting a carry flag
31975 for SBB instruction.
31977 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31979 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31981 * config/rs6000/bmiintrin.h: New file.
31982 * config/rs6000/bmi2intrin.h: New file.
31983 * config/rs6000/x86intrin.h: New file.
31985 2017-05-12 Jeff Law <law@redhat.com>
31987 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31990 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31992 PR middle-end/80707
31993 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31995 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31997 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31998 early expansion of vector divide builtins.
31999 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
32000 builtins identified as having unsigned arguments.
32002 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
32004 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
32005 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
32006 expansion of vector logical operations (and, andc, or, xor,
32009 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
32011 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
32012 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
32014 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
32016 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
32017 early GIMPLE expansion of vector multiplies.
32019 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32021 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
32022 TARGET_HAVE_MOVT conditional.
32023 (movt splitter): Likewise.
32025 2017-05-12 Richard Biener <rguenther@suse.de>
32027 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
32028 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32029 Fold all stmts not inplace.
32031 2017-05-12 Richard Biener <rguenther@suse.de>
32033 PR tree-optimization/80713
32034 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
32035 inserted_exprs bit for not removed stmts.
32037 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
32039 PR middle-end/69921
32040 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
32041 parallelized" attribute for parallelized OpenACC kernels.
32042 * omp-offload.c (execute_oacc_device_lower): Use it.
32044 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
32045 Set "oacc kernels" attribute.
32046 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
32047 parameter. Adjust all users.
32048 (oacc_fn_attrib_kernels_p): Remove function.
32049 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
32050 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
32051 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
32052 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
32053 assert "oacc kernels" attribute is set.
32055 2017-05-11 Carl Love <cel@us.ibm.com>
32057 * config/rs6000/rs6000-c: Add support for built-in functions
32058 vector unsigned char vec_popcnt (vector signed char)
32059 vector unsigned char vec_popcnt (vector unsigned char)
32060 vector unsigned short vec_popcnt (vector signed short)
32061 vector unsigned short vec_popcnt (vector unsigned short)
32062 vector unsigned int vec_popcnt (vector signed int)
32063 vector unsigned int vec_popcnt (vector unsigned int)
32064 vector unsigned long long vec_popcnt (vector signed long long)
32065 vector unsigned long long vec_popcnt (vector unsigned long long)
32066 vector signed long long vec_slo (vector signed long long,
32067 vector signed char)
32068 vector signed long long vec_slo (vector signed long long,
32069 vector unsigned char)
32070 vector unsigned long long vec_slo (vector unsigned long long,
32071 vector signed char)
32072 vector unsigned long long vec_slo (vector unsigned long long,
32073 vector unsigned char)
32074 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
32075 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
32076 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
32077 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
32078 * doc/extend.texi: Update the built-in documentation file for the
32079 new built-in functions.
32081 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
32083 * attribs.h (sorted_attr_string): Move machine independent
32084 functions for target clone support from the i386 port to common
32085 code. Rename ix86_function_versions to common_function_versions.
32086 Rename make_name to make_unique_name.
32087 (common_function_versions): Likewise.
32088 (make_unique_name): Likewise.
32089 (make_dispatcher_decl): Likewise.
32090 (is_function_default_version): Likewise.
32091 * attribs.c (attr_strcmp): Likewise.
32092 (sorted_attr_string): Likewise.
32093 (common_function_versions): Likewise.
32094 (make_unique_name): Likewise.
32095 (make_dispatcher_decl): Likewise.
32096 (is_function_default_version): Likewise.
32097 * config/i386/i386.c (attr_strcmp): Likewise.
32098 (sorted_attr_string): Likewise.
32099 (ix86_function_versions): Likewise.
32100 (make_name): Likewise.
32101 (make_dispatcher_decl): Likewise.
32102 (is_function_default_version): Likewise.
32103 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
32105 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32108 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
32109 Account for direct move costs for vec_construct of integer
32112 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
32115 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
32116 (UNSPEC_STX_ATOMIC): Ditto.
32117 (loaddi_via_sse): New insn.
32118 (storedi_via_sse): Ditto.
32119 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
32120 Update corresponding peephole2 patterns.
32121 (atomic_storedi_fpu): Ditto.
32123 2017-05-11 Julia Koval <julia.koval@intel.com>
32125 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
32126 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
32128 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
32129 (__builtin_ia32_rsqrt14ss_mask): New builtins.
32130 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
32132 2017-05-11 Nathan Sidwell <nathan@acm.org>
32134 * graphite-poly.c: Include dumpfile.h.
32136 * dumpfle.h (dump_function): Declare here ...
32137 * tree-dump.h (dump_function): ... not here.
32138 * dumpfile.c: #include tree-cfg.h.
32139 (dump_function): Move here from ...
32140 * tree-dump.c (dump_function): ... here.
32141 * gimplify.c: #include splay-tree.h, not tree-dump.h.
32142 * graphite-poly.c: Don't include tree-dump.h.
32143 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
32144 * print-tree.c: Likewise.
32145 * stor-layout.c: Likewise.
32146 * tree-nested.c: Likewise.
32148 * dumpfile.c (dump_start): Use TDF_FLAGS.
32149 (dump_enable_all): Fix TDF_KIND check thinko.
32151 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
32153 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32154 array entries to represent two legal parameterizations of the
32155 overloaded __builtin_cmpb function, as represented by the
32156 P6_OV_BUILTIN_CMPB constant.
32157 (altivec_resolve_overloaded_builtin): Add special case handling
32158 for the __builtin_cmpb function, as represented by the
32159 P6_OV_BUILTIN_CMPB constant.
32160 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
32161 (BU_P6_64BIT_2): New macro.
32162 (BU_P6_OVERLOAD_2): New macro
32163 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
32164 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
32165 (CMPB): Add overload support to represent both 32-bit and 64-bit
32166 compare-bytes function.
32167 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
32168 support for TARGET_CMPB.
32169 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
32170 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32171 documentation of the __builtin_cmpb overloaded built-in function.
32173 2017-05-11 Richard Biener <rguenther@suse.de>
32175 PR tree-optimization/80705
32176 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
32177 bases are not vectorizable.
32179 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32181 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
32182 when counting register pressure.
32184 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32186 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
32187 (iv_ca_more_deps): Renamed to ...
32188 (iv_ca_compare_deps): ... this.
32189 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
32191 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32193 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
32195 (determine_group_iv_costs): ... here.
32196 (find_inv_vars_cb): Record inv var if it's not recorded before.
32198 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32200 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
32201 (get_shiftadd_cost): Ditto.
32203 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32205 * tree-ssa-address.c: Include header file.
32206 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
32208 (add_to_parts): Refactor.
32209 (addr_to_parts): New parameter. Update use of move_hint_to_base.
32210 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
32213 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32215 PR tree-optimization/53090
32216 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
32218 (extract_cond_operands): Detect condition with IV on both sides
32219 and return COMP_IV_EXPR_2.
32220 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
32221 (rewrite_use_compare): Simplify by removing call to function
32222 extract_cond_operands.
32224 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32226 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32227 (extract_cond_operands): Detect condition comparing against non-
32228 invariant bound and return appropriate enum value.
32229 (find_interesting_uses_cond): Update use of extract_cond_operands.
32230 Handle its return value accordingly.
32231 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32233 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32235 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32236 nonlinear iv_use computation in loop invariant sensitive way.
32238 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32240 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32241 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32243 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32245 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32246 (dump_cand): Support iv_cand.inv_exprs.
32247 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32249 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32252 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32254 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32256 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32257 as local function. Include necessary header files.
32258 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32260 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32262 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32264 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32266 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32267 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32268 RSHIFT_EXPR and BIT_NOT_EXPR.
32270 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32272 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32273 (adjust_setup_cost): New parameter supporting round up adjustment.
32274 (struct address_cost_data): Delete.
32275 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32276 (split_address_cost, ptr_difference_cost): Delete.
32277 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32278 (struct ainc_cost_data): New struct.
32279 (get_address_cost_ainc): New function.
32280 (get_address_cost, get_computation_cost): Reimplement.
32281 (determine_group_iv_cost_address): Record inv_expr for all uses of
32283 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32284 (iv_ca_has_deps): Reimplemented to ...
32285 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32287 (iv_ca_extend): Call iv_ca_more_deps.
32289 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32291 * tree-ssa-address.c (struct mem_address): Move to header file.
32292 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32293 * tree-ssa-address.h (struct mem_address): Move from C file.
32294 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32296 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32298 * tree-affine.h (aff_combination_type): New interface.
32299 (aff_combination_zero_p): Remove static.
32300 (aff_combination_const_p): New interface.
32301 (aff_combination_singleton_var_p): New interfaces.
32303 2017-05-11 Richard Biener <rguenther@suse.de>
32305 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32306 Skip unreachable blocks and destinations.
32307 (eliminate): Move stmt removal and fixup ...
32308 (fini_eliminate): ... here. Skip inserted exprs.
32309 (pass_pre::execute): Move fini_pre after fini_eliminate.
32310 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32311 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32312 PRE to get rid of dead code that has invalid SSA form and
32313 split critical edges again.
32315 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32317 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32319 2017-05-11 Richard Biener <rguenther@suse.de>
32321 * passes.c (execute_function_todo): Verify loops if they are
32322 said to be up-to-date.
32323 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32324 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32326 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32329 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32330 handle calling assemble_external ourself.
32333 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32334 modes with zero size. Enhance comment.
32336 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32338 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32339 built-ins for vec_xl and vec_xst with short and char pointer
32342 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32344 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32345 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32346 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32347 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32348 (_mm_maskz_min_round_ss): New intrinsics.
32349 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32350 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32351 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32352 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32353 (__builtin_ia32_minss_mask_round): New builtins.
32354 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32355 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32356 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32358 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32359 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32361 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32364 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32366 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32367 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32368 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32369 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32370 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32371 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32372 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32373 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32374 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32375 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32376 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32377 (__builtin_ia32_mulss_mask_round): New builtins.
32378 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32379 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32380 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32382 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32383 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32385 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32388 2017-05-10 Julia Koval <julia.koval@intel.com>
32390 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32391 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32392 (_mm256_setr_m128i): New intrinsics.
32394 2017-05-10 Julia Koval <julia.koval@intel.com>
32396 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32397 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32398 (_mm_maskz_rcp14_ss): New intrinsics.
32399 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32400 (__builtin_ia32_rcp14ss_mask): New builtins.
32401 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32403 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32405 PR tree-optimization/51513
32406 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32407 (assert_unreachable_fallthru_edge_p): Use it.
32408 (group_case_labels_stmt): Likewise.
32409 * tree-cfg.h: Prototype it.
32410 * stmt.c: Include cfghooks.h and tree-cfg.h.
32411 (emit_case_dispatch_table) <gap_label>: New local variable.
32412 Use it to fill dispatch table gaps.
32413 Test for default_label before updating probabilities.
32414 (expand_case) <default_label>: Remove unneeded initialization.
32415 Test for unreachable default case statement and remove its edge.
32416 Set default_label accordingly.
32417 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32419 2017-05-10 Carl Love <cel@us.ibm.com>
32421 * config/rs6000/rs6000-c: Add support for built-in functions
32422 vector signed char vec_neg (vector signed char)
32423 vector signed short int vec_neg (vector short int)
32424 vector signed int vec_neg (vector signed int)
32425 vector signed long long vec_neg (vector signed long long)
32426 vector float vec_neg (vector float)
32427 vector double vec_neg (vector double)
32428 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32430 * config/rs6000/altivec.h: Add define for vec_neg
32431 * doc/extend.texi: Update the built-in documentation for the
32432 new built-in functions.
32434 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32436 PR tree-optimization/77644
32437 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32439 2017-05-10 Nathan Sidwell <nathan@acm.org>
32441 * dumpfile.h (TDI_lang_all): New.
32442 (TDF_KIND): New. Renumber others
32443 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32445 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32447 (get_dump_file_name): Adjust suffix generation.
32448 (dump_enable_all): Use TDF_KIND.
32449 * doc/invoke.texi (-fdump-lang-all): Document.
32451 * dumpfile.h: Tabify.
32453 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32456 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32457 Move member access before delete.
32459 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32461 * tree-inline.c (expand_call_inline): Split block at stmt
32464 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32467 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32468 are now unused after splitting mov{sf,sd}_hardfloat.
32469 (f32_lr2): Likewise.
32470 (f32_lm): Likewise.
32471 (f32_lm2): Likewise.
32472 (f32_li): Likewise.
32473 (f32_li2): Likewise.
32474 (f32_lv): Likewise.
32475 (f32_sr): Likewise.
32476 (f32_sr2): Likewise.
32477 (f32_sm): Likewise.
32478 (f32_sm2): Likewise.
32479 (f32_si): Likewise.
32480 (f32_si2): Likewise.
32481 (f32_sv): Likewise.
32482 (f32_dm): Likewise.
32483 (f32_vsx): Likewise.
32484 (f32_av): Likewise.
32485 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32486 For movsf, order stores so the VSX stores occur before the GPR
32487 store which encourages the register allocator to use a traditional
32488 FPR instead of a GPR. For movsd, order the stores so that the GPR
32489 store comes before the VSX stores to allow the power6 to work.
32490 This is due to the power6 not having a 32-bit integer store
32491 instruction from a FPR.
32492 (movsf_hardfloat): Likewise.
32493 (movsd_hardfloat): Likewise.
32495 2017-05-09 Martin Sebor <msebor@redhat.com>
32497 PR translation/80280
32498 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32501 PR translation/80280
32502 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32503 data member added in r247778.
32504 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32506 2017-05-09 Nathan Sidwell <nathan@acm.org>
32508 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32510 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32513 2017-05-09 Marek Polacek <polacek@redhat.com>
32515 * doc/invoke.texi: Fix typo.
32517 2017-05-09 Richard Biener <rguenther@suse.de>
32519 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32520 (vrp_val_is_min): Likewise.
32521 (set_value_range_to_value): Likewise.
32522 (set_value_range_to_nonnegative): Likewise.
32523 (gimple_assign_nonzero_p): Likewise.
32524 (gimple_stmt_nonzero_p): Likewise.
32525 (vrp_int_const_binop): Likewise. Remove unreachable case.
32526 (adjust_range_with_scev): Adjust comments.
32527 (compare_range_with_value): Likewise.
32528 (extract_range_from_phi_node): Likewise.
32529 (test_for_singularity): Likewise.
32531 2017-05-09 Richard Biener <rguenther@suse.de>
32533 * tree-vrp.c (get_single_symbol): Add assert that we don't
32534 get overflowed constants as invariant part.
32535 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32536 checks. Use wi::cmp instead of recursing for integer constants.
32537 (compare_values): Just ignore whether we assumed undefined
32538 overflow instead of failing the compare.
32539 (extract_range_for_var_from_comparison_expr): Add comment before the
32540 TREE_NO_WARNING sets.
32541 (test_for_singularity): Likewise.
32542 (extract_range_from_comparison): Do not disable optimization
32543 when we assumed undefined overflow.
32544 (extract_range_basic): Remove init of unused var.
32546 2017-05-09 Richard Biener <rguenther@suse.de>
32548 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32549 (extract_range_from_multiplicative_op_1): Adjust.
32550 (extract_range_from_binary_expr_1): Use int_const_binop.
32552 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32555 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32556 rs6000_store_data_bypass_p in seven define_bypass directives and
32557 in several comments.
32558 * config/rs6000/rs6000-protos.h: Add prototype for
32559 rs6000_store_data_bypass_p function.
32560 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32561 function implements slightly different (rs6000-specific) semantics
32562 than store_data_bypass_p, returning false rather than aborting
32563 with assertion error when arguments do not satisfy the
32564 requirements of store data bypass.
32565 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32566 rs6000_store_data_bypass_p.
32568 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32570 * config/xtensa/xtensa-protos.h
32571 (xtensa_initial_elimination_offset): New declaration.
32572 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32573 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32574 macro definition, add case for FRAME_POINTER_REGNUM when
32575 FRAME_GROWS_DOWNWARD.
32576 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32577 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32578 xtensa_initial_elimination_offset.
32580 2017-05-08 Nathan Sidwell <nathan@acm.org>
32582 * doc/invoke.texi: Alphabetize -fdump options.
32584 2017-05-08 Martin Sebor <msebor@redhat.com>
32586 PR translation/80280
32587 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32589 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32591 * target.def (compute_frame_layout): New optional target hook.
32592 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32593 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32594 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32596 * reload1.c (verify_initial_elim_offsets): Likewise.
32597 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32598 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32599 (arm_get_frame_offsets): Split up into this ...
32600 (arm_compute_frame_layout): ... and this function.
32602 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32604 * config/aarch64/constraints.md (Usa): New constraint.
32605 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32607 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32609 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32610 with_multilib_list after it has been checked.
32612 2017-05-08 Richard Biener <rguenther@suse.de>
32614 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32615 (bitmap_set_subtract_values): Likewise.
32617 2017-05-08 Richard Biener <rguenther@suse.de>
32619 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32620 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32622 (gimple_stmt_nonzero_warnv_p): Rename to ...
32623 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32625 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32626 (extract_range_basic): Adjust, do not disable propagation on
32627 strict overflow sensitive simplification.
32628 (vrp_visit_cond_stmt): Likewise.
32630 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32632 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32633 body size unconditionally.
32635 2017-05-07 Jeff Law <law@redhat.com>
32638 2017-05-06 Jeff Law <law@redhat.com>
32639 PR tree-optimization/78496
32640 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32643 PR tree-optimization/78496
32644 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32645 (simplify_stmt_using_ranges): Call it.
32646 (vrp_dom_walker::before_dom_children): Extract equivalences
32647 from an ASSERT_EXPR with an equality comparison against a
32650 2017-05-06 Jeff Law <law@redhat.com>
32652 PR tree-optimization/78496
32653 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32656 PR tree-optimization/78496
32657 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32658 (simplify_stmt_using_ranges): Call it.
32659 (vrp_dom_walker::before_dom_children): Extract equivalences
32660 from an ASSERT_EXPR with an equality comparison against a
32663 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32665 * lra-constraints.c (lra_copy_reg_equiv): New function.
32666 (split_reg): Use it to copy equivalence information from the
32667 original register to the spill register.
32669 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32671 PR rtl-optimization/75964
32672 * simplify-rtx.c (simplify_const_relational_operation): Remove
32673 invalid handling of comparisons of integer ABS.
32675 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32677 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32678 initialize to zero.
32679 (init_regs): Remove declaration.
32680 (function_arg_advance_32): Initialize error_p as boolean variable.
32682 2017-05-05 Nathan Sidwell <nathan@acm.org>
32684 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32685 lines. Use for (;;).
32687 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32689 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32690 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32691 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32692 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32693 VF=2 that require versioning.
32695 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32697 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32700 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32702 * diagnostic.h (diagnostic_override_option_index): Convert from
32703 macro to inline function.
32705 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32707 * diagnostic.c (last_module_changed_p): New function.
32708 (set_last_module): New function.
32709 (diagnostic_report_current_module): Convert macro usage to
32710 the above functions.
32711 * diagnostic.h (diagnostic_context::last_module): Strengthen
32712 from const line_map * to const line_map_ordinary *.
32713 (diagnostic_last_module_changed): Delete macro.
32714 (diagnostic_set_last_module): Delete macro.
32716 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32718 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32719 with diagnostic_report_diagnostic.
32720 (diagnostic_n_impl_richloc): Likewise.
32721 * diagnostic.h (report_diagnostic): Delete macro.
32722 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32723 with diagnostic_report_diagnostic.
32724 * substring-locations.c (format_warning_va): Likewise.
32726 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32728 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32729 save/restor of format_spec. Move option-printing code to...
32730 (print_option_information): ...this new function, and
32731 reimplement by simply printing to the pretty_printer,
32732 rather than appending to the format string.
32734 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32736 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32737 handling logic into...
32738 (update_effective_level_from_pragmas): ...this new function.
32740 2017-05-04 Andrew Waterman <andrew@sifive.com>
32742 * config/riscv/riscv.opt (mstrict-align): New option.
32743 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32744 (SLOW_UNALIGNED_ACCESS): Define.
32745 (riscv_slow_unaligned_access): Declare.
32746 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32748 (riscv_slow_unaligned_access): New variable.
32749 (rocket_tune_info): Set slow_unaligned_access to true.
32750 (optimize_size_tune_info): Set slow_unaligned_access to false.
32751 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32752 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32753 (riscv_option_override): Set riscv_slow_unaligned_access.
32754 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32756 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32758 * config/riscv/riscv.md: Unify indentation.
32760 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32765 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32767 (extendsi<mode>2): Add support for doing sign extension via
32768 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32769 don't have ISA 3.0 instructions.
32770 (extendsi<mode>2 splitter): Likewise.
32771 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32772 generate the normal insns since SImode can now go in vector
32773 registers. Disallow the special UNSPECs needed for previous
32774 machines to hide SImode being used. Add new insns
32775 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32776 (fix_trunc<mode>si2_stfiwx): Likewise.
32777 (fix_trunc<mode>si2_internal): Likewise.
32778 (fixuns_trunc<mode>si2): Likewise.
32779 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32780 (fctiw<u>z_<mode>_smallint): Likewise.
32781 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32782 of floating point to 32-bit integer from doing a direct move to
32783 the GPR registers to do a store.
32784 (fctiwz_<mode>): Break long line.
32786 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32788 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32789 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32790 (addr_list, addr_offset_valid_p): New.
32791 (split_address_groups): Check offset validity with above function.
32792 (gt-tree-ssa-loop-ivopts.h): Include header file.
32794 2017-05-05 Nathan Sidwell <nathan@acm.org>
32796 * config.gcc (arm*-*-*): Add missing 'fi'.
32798 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32800 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32801 included in -fopt-info does not matter.
32802 * doc/optinfo.texi (-fopt-info): Fix description of default
32803 behavour. Explicitly say order of options included in -fopt-info
32806 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32808 * config.gcc: Allow combinations of aprofile and rmprofile values for
32809 --with-multilib-list.
32810 * config/arm/t-multilib: New file.
32811 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32812 variables. Remove setting of ISA and floating-point ABI in
32813 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32814 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32815 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32816 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32818 * config/arm/t-rmprofile: Likewise except for the matches changes.
32819 * doc/install.texi (--with-multilib-list): Document the combination of
32820 aprofile and rmprofile values and warn about pitfalls in doing that.
32822 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32824 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32825 (movdi_aarch64): Likewise.
32827 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32829 PR tree-optimization/80632
32830 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32832 (build_arrays): Initialize it for virtual phis.
32833 (fix_phi_nodes): Use it for virtual phis.
32835 PR tree-optimization/80558
32836 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32837 [x, y] op z into [x op, y op z] for op & or | if conditions
32840 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32841 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32844 * config/arm/arm.md (use_literal_pool): Remove.
32845 (64-bit immediate split): No longer takes cost into consideration
32846 if arm_disable_literal_pool is enabled.
32847 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32848 used when arm_disable_literal_pool is enabled.
32849 (arm_max_const_double_inline_cost): Remove use of
32850 arm_disable_literal_pool.
32851 (push_minipool_fix): Add assert.
32852 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32853 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32854 (no_literal_pool_sf_immediate): New.
32856 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32858 PR tree-optimization/80613
32859 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32860 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32862 2017-05-05 Richard Biener <rguenther@suse.de>
32864 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32866 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32868 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32869 of this flag from insn conditions due to removal from r247495.
32871 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32873 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32875 (arm_early_store_addr_dep_ptr): Likewise.
32876 * config/arm/aarch-common-protos.h
32877 (arm_early_load_addr_dep_ptr): Add prototype.
32878 (arm_early_store_addr_dep_ptr): Likewise.
32879 * config/arm/cortex-a53.md: Add new bypasses.
32881 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32883 * tree.c (next_type_uid): Change type to unsigned.
32884 (type_hash_canon): Decrement back next_type_uid if
32885 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32886 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32889 2017-05-04 Martin Sebor <msebor@redhat.com>
32891 * builtins.c: Fix a trivial typo in a comment.
32893 PR middle-end/79234
32894 * builtins.c (check_sizes): Adjust to handle reading past the end.
32895 Avoid printing excessive upper bound of ranges. Use %E to print
32896 tree nodes instead of converting them to %wu.
32897 (expand_builtin_memchr): New function.
32898 (compute_dest_size): Rename...
32899 (compute_objsize): ...to this.
32900 (expand_builtin_memcpy): Adjust.
32901 (expand_builtin_mempcpy): Adjust.
32902 (expand_builtin_strcat): Adjust.
32903 (expand_builtin_strcpy): Adjust.
32904 (check_strncat_sizes): Adjust.
32905 (expand_builtin_strncat): Adjust.
32906 (expand_builtin_strncpy): Adjust and simplify.
32907 (expand_builtin_memset): Adjust.
32908 (expand_builtin_bzero): Adjust.
32909 (expand_builtin_memcmp): Adjust.
32910 (expand_builtin): Handle memcmp.
32911 (maybe_emit_chk_warning): Check strncat just once.
32913 2017-05-04 Martin Sebor <msebor@redhat.com>
32915 PR preprocessor/79214
32916 PR middle-end/79222
32917 PR middle-end/79223
32918 * builtins.c (check_sizes): Add inlining context and issue
32919 warnings even when -Wno-system-headers is set.
32920 (check_strncat_sizes): Same.
32921 (expand_builtin_strncat): Same.
32922 (expand_builtin_memmove): New function.
32923 (expand_builtin_stpncpy): Same.
32924 (expand_builtin): Handle memmove and stpncpy.
32926 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32928 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32929 which is not used any more.
32931 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32933 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32935 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32937 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32938 (cortexa53_tunings): Likewise.
32939 (cortexa57_tunings): Likewise.
32940 (cortexa72_tunings): Likewise.
32941 (cortexa73_tunings): Likewise.
32943 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32945 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32946 Set loop alignment to 8.
32948 2017-05-04 Martin Sebor <msebor@redhat.com>
32950 PR translation/80280
32951 * builtins.c (expand_builtin_object_size): Add missing quoting to
32952 %D and like directives.
32953 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32954 (hsa_type_for_tree_type): Same.
32955 (verify_function_arguments): Same.
32956 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32957 * varasm.c (get_section): Same.
32960 2017-05-04 Martin Sebor <msebor@redhat.com>
32962 PR translation/80280
32963 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32965 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32967 * config/aarch64/aarch64.c (generic_addrcost_table):
32968 Change HI/TI mode setting.
32970 2017-05-04 Martin Jambor <mjambor@suse.cz>
32972 PR tree-optimization/80622
32973 * tree-sra.c (comes_initialized_p): New function.
32974 (build_accesses_from_assign): Only set write lazily when
32975 comes_initialized_p is false.
32976 (analyze_access_subtree): Use comes_initialized_p.
32977 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32978 instead of testing for PARM_DECL.
32980 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32982 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32983 constraint on operand 0 to allow more general addressing modes.
32984 Adjust output template.
32985 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32987 * config/aarch64/aarch64-protos.h
32988 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32989 * config/aarch64/constraints.md (Dp): New address constraint.
32990 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32993 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32995 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32996 update use of estimate_ipcp_clone_size_and_time.
32997 (estimate_local_effects): Update use of
32998 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32999 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
33000 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
33001 Return nonspecialized time.
33003 2017-05-04 Richard Biener <rguenther@suse.de>
33005 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
33006 for the last VUSE which def dominates the PHI. Directly call
33008 (get_continuation_for_phi_1): Remove.
33010 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
33012 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
33013 to explain the use of truncating division. Cap the number of
33014 iterations to the maximum given by nb_iterations_upper_bound,
33017 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33019 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
33020 * configure: Regenerate.
33021 * config.in: Regenerate.
33022 * config/i386/driver-mingw32.c: new file.
33023 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
33024 * config.host: Link driver-mingw32.o on MinGW host.
33025 * doc/install.texi: Document new --enable-mingw-wildcard configure
33028 2017-05-04 Marek Polacek <polacek@redhat.com>
33030 PR tree-optimization/80612
33031 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
33033 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
33034 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
33036 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
33037 (movt splitter): Likewise.
33038 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
33039 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
33040 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
33041 block for Thumb-1 with MOVT.
33042 (thumb2_legitimate_address_p): Move code block ...
33043 (can_avoid_literal_pool_for_label_p): ... into this new function.
33044 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
33046 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
33047 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
33048 "M-profile targets with the MOVT instruction".
33050 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
33052 * config/arm/arm-builtins.c (arm_init_builtins): Rename
33053 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
33054 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
33056 2017-05-04 Martin Liska <mliska@suse.cz>
33058 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
33059 variable cond_code.
33061 2017-05-04 Richard Biener <rguenther@suse.de>
33063 * tree.c (array_at_struct_end_p): Handle arrays at struct
33064 end with flexarrays more conservatively. Refactor and treat
33065 arrays of arrays or aggregates more strict. Fix
33066 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
33067 * tree.c (array_at_struct_end_p): Adjust prototype.
33068 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
33069 * gimple-fold.c (get_range_strlen): Likewise.
33070 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
33072 2017-05-04 Richard Biener <rguenther@suse.de>
33074 PR tree-optimization/31130
33075 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
33077 (supports_overflow_infinity): Likewise.
33078 (is_negative_overflow_infinity): Likewise.
33079 (is_positive_overflow_infinity): Likewise.
33080 (is_overflow_infinity): Likewise.
33081 (stmt_overflow_infinity): Likewise.
33082 (overflow_infinity_range_p): Likewise.
33083 (usable_range_p): Remove as always returning true.
33084 (make_overflow_infinity): Remove.
33085 (negative_overflow_infinity): Likewise.
33086 (positive_overflow_infinity): Likewise.
33087 (avoid_overflow_infinity): Likewise.
33088 (set_value_range): Adjust accordingly.
33089 (set_value_range_to_nonnegative): Likewise, remove now unused
33090 overflow_infinity arg.
33091 (vrp_operand_equal_p): Adjust.
33092 (update_value_range): Likewise.
33093 (range_int_cst_singleton_p): Likewise.
33094 (operand_less_p): Likewise.
33095 (compare_values_warnv): Likewise.
33096 (extract_range_for_var_from_comparison_expr): Likewise.
33097 (vrp_int_const_binop): Likewise.
33098 (zero_nonzero_bits_from_vr): Likewise.
33099 (extract_range_from_multiplicative_op_1): Likewise.
33100 (extract_range_from_binary_expr_1): Likewise.
33101 (extract_range_from_unary_expr): Likewise.
33102 (extract_range_from_comparison): Likewise.
33103 (extract_range_basic): Likewise.
33104 (adjust_range_with_scev): Likewise.
33105 (compare_ranges): Likewise.
33106 (compare_range_with_value): Likewise.
33107 (dump_value_range): Likewise.
33108 (test_for_singularity): Likewise, remove strict_overflow_p parameter
33110 (simplify_cond_using_ranges): Adjust.
33112 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33114 * brig-builtins.def: Added a builtin for class_f64.
33115 * builtin-types.def: Added a builtin type needed by class_f64.
33117 2017-05-03 Jason Merrill <jason@redhat.com>
33119 * timevar.def: Add TV_CONSTEXPR.
33121 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33123 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
33125 2017-05-03 Martin Jambor <mjambor@suse.cz>
33127 * ipa-prop.c (ipa_update_after_lto_read): Removed.
33128 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
33129 * ipa-cp.c (ipcp_propagate_stage): Do not call
33130 ipa_update_after_lto_read.
33131 * ipa-inline.c (ipa_inline): Likewise.
33133 2017-05-03 Martin Jambor <mjambor@suse.cz>
33135 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
33136 tag. Added a default constructor and a destructor.
33137 (ipa_edge_args_sum_t): New class;
33138 (ipa_edge_args_sum): Declare.
33139 (ipa_edge_args_vector): Remove declaration.
33140 (IPA_EDGE_REF): Use ipa_edge_args_sum.
33141 (ipa_free_edge_args_substructures): Remove declaration.
33142 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
33143 (ipa_edge_args_info_available_for_edge_p): Likewise.
33144 * ipa-prop.c (ipa_edge_args_vector): Removed.
33145 (edge_removal_hook_holder): Likewise.
33146 (edge_duplication_hook_holder): Likewise.
33147 (ipa_edge_args_sum): New variable.
33148 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
33149 ipa_edge_args_vector.
33150 (ipa_free_edge_args_substructures): Likewise.
33151 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
33152 ipa_edge_args_vector.
33153 (ipa_edge_removal_hook): Turned into method
33154 ipa_edge_args_sum_t::remove.
33155 (ipa_edge_duplication_hook): Turned into method
33156 ipa_edge_args_sum_t::duplicate.
33157 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
33158 registering edge hooks.
33159 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
33160 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
33161 ipa_edge_args_sum instead of ipa_edge_args_vector.
33162 * ipa-profile.c (ipa_profile): Likewise.
33164 2017-05-03 Martin Jambor <mjambor@suse.cz>
33166 * symbol-summary.h (function_summary): New method exists.
33167 (function_summary::symtab_removal): Deallocate through release.
33168 (call_summary): New class.
33169 (gt_ggc_mx): New overload.
33170 (gt_pch_nx): Likewise.
33171 (gt_pch_nx): Likewise.
33173 2017-05-03 Jeff Law <law@redhat.com>
33175 PR tree-optimization/78496
33176 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
33177 from simplify_cond_using_ranges. Split off code to walk
33178 backwards through casts into ...
33179 (simplify_cond_using_ranges_2): New function.
33180 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
33181 (execute_vrp): After identifying jump threads, call
33182 simplify_cond_using_ranges_2.
33184 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
33187 * ipa-inline.h (inline_summary): Add ctor.
33188 (create_ggc): Do not use ggc_cleared_alloc.
33190 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
33191 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33193 * gcc.c (handle_braces): Support escaping in switch matching
33195 * doc/invoke.texi (Spec Files): Document it.
33196 Remove superfluous @code markup in items.
33198 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33200 * diagnostic-show-locus.c (struct column_range): New struct.
33201 (get_affected_columns): New function.
33202 (get_printed_columns): New function.
33203 (struct correction): New struct.
33204 (correction::ensure_capacity): New function.
33205 (correction::ensure_terminated): New function.
33206 (struct line_corrections): New struct.
33207 (line_corrections::~line_corrections): New dtor.
33208 (line_corrections::add_hint): New function.
33209 (layout::print_trailing_fixits): Reimplement in terms of the new
33211 (selftest::test_overlapped_fixit_printing): New function.
33212 (selftest::diagnostic_show_locus_c_tests): Call it.
33214 2017-05-03 Nathan Sidwell <nathan@acm.org>
33216 Canonicalize canonical type hashing
33217 * tree.h (type_hash_canon_hash): Declare.
33218 * tree.c (type_hash_list, attribute_hash_list): Move into
33219 type_hash_canon_hash.
33220 (build_type_attribute_qual_variant): Break out hash code calc into
33221 type_hash_canon_hash.
33222 (type_hash_canon_hash): New. Generic type hash computation.
33223 (build_range_type_1, build_array_type_1, build_function_type,
33224 build_method_type_directly, build_offset_type, build_complex_type,
33225 make_vector_type): Call it.
33227 2017-05-03 Richard Biener <rguenther@suse.de>
33229 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33230 When all DRs have unknown misaligned do not always peel
33231 when there is a store but apply the same costing model as if
33232 there were only loads.
33234 2017-05-03 Richard Biener <rguenther@suse.de>
33237 PR tree-optimization/80492
33238 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33239 compare_base_decls returning dont-know properly.
33241 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33243 * config/arm/iterators.md (CCSI): New mode iterator.
33244 (arch): New mode attribute.
33245 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33246 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33247 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33248 code iterator for success result mode.
33249 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33250 the corresponding new insn generators.
33252 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33255 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33256 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33258 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33260 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33261 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33262 (DDR_A): Wrap DDR argument in brackets.
33263 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33264 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33265 (DDR_REVERSED_P): Likewise.
33267 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33269 PR tree-optimization/79472
33270 * tree-switch-conversion.c (struct switch_conv_info): Add
33271 contiguous_range and default_case_nonstandard fields.
33272 (collect_switch_conv_info): Compute contiguous_range and
33273 default_case_nonstandard fields, don't clear final_bb if
33274 contiguous_range and only the default case doesn't have the required
33276 (check_all_empty_except_final): Set default_case_nonstandard instead
33277 of failing if contiguous_range and the default case doesn't have empty
33279 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33280 and only the default case doesn't have the required constants. Skip
33282 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33283 if default_case_nonstandard.
33284 (build_constructors): Build constant 1 just once. Assert that default
33285 values aren't inserted in between cases if contiguous_range. Skip
33287 (build_arrays): Skip virtual phis.
33288 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33289 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33290 Handle virtual phis.
33291 (gen_inbound_check): Handle default_case_nonstandard case.
33292 (process_switch): Adjust check_final_bb caller. Call
33293 gather_default_values with the first non-default case instead of
33294 default case if default_case_nonstandard.
33296 2017-05-02 Nathan Sidwell <nathan@acm.org>
33298 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33299 check. Fix formatting.
33301 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33303 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33304 errors when comparing specialized and unspecialized times.
33306 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33308 * diagnostic-show-locus.c
33309 (layout::should_print_annotation_line_p): Make private.
33310 (layout::print_annotation_line): Make private.
33311 (layout::annotation_line_showed_range_p): Make private.
33312 (layout::show_ruler): Make private.
33313 (layout::print_source_line): Make private. Pass in line and
33314 line_width, rather than calling location_get_source_line. Drop
33316 (layout::print_leading_fixits): New method.
33317 (layout::print_any_fixits): Rename to...
33318 (layout::print_trailing_fixits): ...this, and make private.
33319 Don't print newline fixits.
33320 (diagnostic_show_locus): Move logic for printing one row into...
33321 (layout::print_line): ...this new function. Move the
33322 location_get_source_line call and error-handling from
33323 print_source_line to here. Call print_leading_fixits, and rename
33324 print_any_fixits to print_trailing_fixits.
33325 (selftest::test_fixit_insert_containing_newline): Update now that
33326 newlines are partially supported.
33327 (selftest::test_fixit_insert_containing_newline_2): New test.
33328 (selftest::test_fixit_replace_containing_newline): Update comments.
33329 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33330 * edit-context.c (class added_line): New class.
33331 (class edited_line): Describe newline handling in comment.
33332 (edited_line::actually_edited_p): New method.
33333 (edited_line::print_content): Delete redundant decl.
33334 (edited_line::m_predecessors): New field.
33335 (edited_file::print_content): Call edited_line::print_content.
33336 (edited_file::print_diff): Update to support newlines.
33337 (edited_file::print_diff_hunk): Likewise.
33338 (edited_file::print_run_of_changed_lines): New function.
33339 (edited_file::print_diff_line): Convert to...
33340 (print_diff_line): ...this.
33341 (edited_file::get_effective_line_count): New function.
33342 (edited_line::edited_line): Initialize new field m_predecessors.
33343 (edited_line::~edited_line): Clean up m_predecessors.
33344 (edited_line::apply_fixit): Handle newlines.
33345 (edited_line::get_effective_line_count): New function.
33346 (edited_line::print_content): New function.
33347 (edited_line::print_diff_lines): New function.
33348 (selftest::test_applying_fixits_insert_containing_newline): New
33350 (selftest::test_applying_fixits_replace_containing_newline): New
33352 (selftest::insert_line): New function.
33353 (selftest::test_applying_fixits_multiple_lines): Add example of
33355 (selftest::edit_context_c_tests): Call the new tests.
33357 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33359 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33360 parameter cand. Update dump information.
33361 (get_computation_cost): Update uses.
33363 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33365 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33366 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33367 (get_computation_at, rewrite_use_address): Update use of
33368 get_computation_aff.
33370 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33372 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33373 (get_computation): Delete.
33374 (get_computation_cost): Implement like get_computation_cost_at.
33375 Use get_computation_at.
33376 (get_computation_cost_at): Delete.
33377 (rewrite_use_nonlinear_expr): Use get_computation_at.
33378 (rewrite_use_compare, remove_unused_ivs): Ditto.
33380 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33382 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33384 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33386 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33387 (ivopts_global_cost_for_size): Rename parameter and update uses.
33388 (iv_ca_recount_cost): Update uses.
33389 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33390 candidates seperately in n_invs and n_cands.
33391 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33393 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33395 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33396 (find_inv_vars_cb): New.
33397 (find_depends): Renamed to ...
33398 (find_inv_vars): ... this.
33399 (add_candidate_1, force_var_cost): Call find_inv_vars.
33400 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33402 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33404 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33405 inv_vars. Add inv_exprs.
33406 (struct iv_cand): Rename depends_on to inv_vars.
33407 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33408 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33409 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33410 (dump_cand): Dump inv_vars.
33411 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33412 (record_invariant, find_depends, add_candidate_1): Ditto.
33413 (set_group_iv_cost, force_var_cost): Ditto.
33414 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33415 (get_computation_cost_at, get_computation_cost): Ditto.
33416 (determine_group_iv_cost_generic): Ditto.
33417 (determine_group_iv_cost_address): Ditto.
33418 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33419 (determine_group_iv_costs): Ditto.
33420 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33421 (iv_ca_set_remove_invariants): Renamed to ...
33422 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33423 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33424 (iv_ca_set_add_invariants): Renamed to ...
33425 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33426 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33427 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33428 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33429 (create_new_ivs): Remove useless dump.
33431 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33433 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33435 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33436 (iv_ca_set_no_cp, create_new_iv): Ditto.
33438 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33440 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33442 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33444 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33445 function tree_check2.
33447 2017-05-02 Martin Liska <mliska@suse.cz>
33449 * doc/gcov.texi: Add missing preposition.
33450 * gcov.c (function_info::function_info): Properly fill up
33451 all member variables.
33453 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33455 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33457 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33459 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33461 2017-05-02 Martin Liska <mliska@suse.cz>
33464 * lto-streamer-in.c (lto_read_tree_1): Remove
33465 LTO_STREAMER_DEBUG.
33466 * lto-streamer.c (struct tree_hash_entry): Likewise.
33467 (struct tree_entry_hasher): Likewise.
33468 (tree_entry_hasher::hash): Likewise.
33469 (tree_entry_hasher::equal): Likewise.
33470 (lto_streamer_init): Likewise.
33471 (lto_orig_address_map): Likewise.
33472 (lto_orig_address_get): Likewise.
33473 (lto_orig_address_remove): Likewise.
33474 * lto-streamer.h: Likewise.
33475 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33476 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33478 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33480 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33481 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33482 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33483 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33484 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33485 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33486 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33487 (mm_maskz_sub_ss): New intrinsics.
33488 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33489 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33490 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33491 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33492 (__builtin_ia32_subss_mask_round): New builtins.
33493 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33494 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33495 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33497 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33498 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33500 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33503 2017-05-02 Martin Jambor <mjambor@suse.cz>
33505 PR tree-optimization/78687
33506 * tree-sra.c (access): New field parent.
33507 (process_subtree_disqualification): New function.
33508 (disqualify_candidate): Call it.
33509 (build_accesses_from_assign): Reset write flag if creating an
33511 (build_access_subtree): Fill in parent field and also prpagate
33512 down grp_write flag.
33513 (create_artificial_child_access): New parameter set_grp_write, set
33514 grp_write to its value.
33515 (propagate_subaccesses_across_link): Also propagate grp_write flag
33517 (propagate_all_subaccesses): Push the closest parent back to work
33518 queue if add_access_to_work_queue returned true.
33520 2017-05-02 Richard Biener <rguenther@suse.de>
33522 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33523 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33524 -fstrict-overflow documentation.
33525 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33526 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33527 flag_strict_overflow.
33528 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33529 * lto-opts.c (lto_write_options): Do not stream it.
33530 * lto-wrapper.c (merge_and_complain): Do not handle it.
33531 * opts.c (default_options_table): Do not set -fstrict-overflow.
33532 (finish_options): Likewise do not clear it when sanitizing.
33533 * simplify-rtx.c (simplify_const_relational_operation): Do not
33534 test flag_strict_overflow.
33536 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33538 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33539 using enabled attribute.
33540 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33541 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33542 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33543 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33544 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33545 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33546 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33547 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33548 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33549 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33551 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33553 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33555 2017-05-02 Richard Biener <rguenther@suse.de>
33557 PR tree-optimization/80591
33559 2017-04-10 Richard Biener <rguenther@suse.de>
33561 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33564 2017-05-02 Richard Biener <rguenther@suse.de>
33566 PR tree-optimization/80549
33567 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33568 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33569 headers if they do not have a preheader.
33571 2017-05-02 Martin Liska <mliska@suse.cz>
33574 * common.opt: Fix typo.
33575 * doc/invoke.texi: Likewise.
33577 2017-05-01 Jan Beulich <jbeulich@suse.com>
33579 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33580 swapping, add (x,x,m,x,n) alternative.
33582 2017-05-01 Nathan Sidwell <nathan@acm.org>
33584 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33585 unnecessary unadjusted_alignment check.
33587 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33590 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33592 * gimplify.c (gimplify_cilk_detach): New function.
33593 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33594 * tree-core.h: Document EXPR_CILK_SPAWN.
33595 * tree.h (EXPR_CILK_SPAWN): Define.
33597 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33599 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33600 to use new fixit_hint representation, using the "replace" logic.
33601 (get_line_span_for_fixit_hint): Likewise.
33602 (layout::print_any_fixits): Likewise.
33603 (selftest::test_one_liner_many_fixits): Rename to...
33604 (selftest::test_one_liner_many_fixits_1): ...this, and update
33605 comment and expected output to reflect that the multiple fix-it
33606 hints are now consolidated into one insertion.
33607 (selftest::test_one_liner_many_fixits_2): New test.
33608 (selftest::test_diagnostic_show_locus_one_liner): Update for
33610 (selftest::test_fixit_consolidation): Update for fix-it API
33612 * diagnostic.c (print_parseable_fixits): Likewise.
33613 * edit-context.c (edited_line::m_line_events): Convert from
33614 auto_vec <line_event *> to auto_vec <line_event>.
33615 (class line_event): Convert from abstract base class to a concrete
33616 class, taking over the role of replace_event.
33617 (class insert_event): Delete.
33618 (class replace_event): Rename to class line_event. Convert to
33620 (edit_context::add_fixits): Reimplement.
33621 (edit_context::apply_insert): Delete.
33622 (edit_context::apply_replace): Rename to...
33623 (edit_context::apply_fixit): ...this. Convert to half-open range.
33624 (edited_file::apply_insert): Delete.
33625 (edited_file::apply_replace): Rename to...
33626 (edited_file::apply_fixit): ...this.
33627 (edited_line::~edited_line): Drop deletion of events.
33628 (edited_line::apply_insert): Delete.
33629 (edited_line::apply_replace): Rename to...
33630 (edited_line::apply_fixit): ...this. Convert to half-open range.
33631 Update for change to type of m_line_events.
33632 * edit-context.h (edit_context::apply_insert): Delete.
33633 (edit_context::apply_replace): Rename to...
33634 (edit_context::apply_fixit): ...this.
33636 2017-05-01 Martin Sebor <msebor@redhat.com>
33638 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33641 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33644 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33645 __get_cpuid_max returns 0.
33646 (__get_cpuid_count): Ditto.
33648 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33650 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33651 replacement expression is another instance of one of its arguments.
33653 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33656 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33657 check for stack push/pop autoinc.
33658 * config/i386/i386.c (ix86_agi_dependent): Return false
33659 if the only reason why modified_in_p returned true is that
33660 addr is SP based and set_insn is a push or pop.
33662 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33664 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33667 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33670 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33671 (account_size_time): Use two predicates - exec_pred and
33673 (evaluate_conditions_for_known_args): Compute both clause and
33675 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33676 (inline_summary_t::duplicate): Update.
33677 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33679 (compute_inline_parameters): Likewise.
33680 (estimate_edge_size_and_time): Update caluclation of time.
33681 (estimate_node_size_and_time): Compute both time and nonspecialized
33683 (estimate_ipcp_clone_size_and_time): Update.
33684 (inline_merge_summary): Update.
33685 (do_estimate_edge_time): Update.
33686 (do_estimate_edge_size): Update.
33687 (do_estimate_edge_hints): Update.
33688 (inline_read_section, inline_write_summary): Stream both new predicates.
33689 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33691 (compute_inlined_call_time): Cleanup.
33692 (big_speedup_p): Update.
33693 (edge_badness): Update.
33694 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33695 (size_time_entry): Replace predicate by exec_predicate and
33696 nonconst_predicate.
33697 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33698 (estimate_edge_time): Return also nonspec_time.
33699 (reset_edge_growth_cache): Update.
33701 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33703 PR rtl-optimization/80491
33704 * ifcvt.c (noce_process_if_block): When looking for x setter
33705 with missing else_bb, don't check only the insn right before
33706 cond_earliest, but look for the last insn that x is modified in
33707 within the same bb.
33709 PR rtl-optimization/80491
33710 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33712 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33714 PR tree-optimization/80487
33715 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33717 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33719 PR tree-optimization/79697
33720 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33721 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33722 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33724 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33725 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33727 2017-04-28 Martin Sebor <msebor@redhat.com>
33729 PR tree-optimization/80523
33730 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33731 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33733 (maybe_warn, format_directive, parse_directive): Use new functions.
33734 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33736 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33738 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33740 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33742 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33743 target_header_dir): Set correctly.
33744 * configure: Regenerated.
33745 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33746 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33747 instead of SYSTEM_HEADER_DIR.
33749 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33751 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33752 (estimate_local_effects): Likewise.
33753 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33754 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33755 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33756 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33757 do_estimate_edge_time, estimate_edge_time): Likewise.
33758 * ipa-inline-analysis.c (estimate_node_size_and_time,
33759 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33760 (estimate_time_after_inlining): Remove.
33762 2017-04-28 Martin Liska <mliska@suse.cz>
33764 * doc/gcov.texi: Enhance documentation of gcov.
33766 2017-04-28 Martin Liska <mliska@suse.cz>
33768 * doc/gcov.texi: Sort options in alphabetic order.
33769 * doc/gcov-dump.texi: Likewise.
33770 * doc/gcov-tool.texi: Likewise.
33771 * gcov.c (print_usage): Likewise.
33772 * gcov-dump.c (print_usage): Likewise.
33773 * gcov-tool.c (print_merge_usage_message): Likewise.
33774 (print_rewrite_usage_message): Likewise.
33775 (print_overlap_usage_message): Likewise.
33777 2017-04-28 Martin Liska <mliska@suse.cz>
33779 PR gcov-profile/53915
33780 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33782 2017-04-28 Martin Liska <mliska@suse.cz>
33784 PR gcov-profile/79891
33785 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33786 is marked by compiler as living on a line.
33787 (get_cycles_count): Remove usage of the union.
33788 (output_intermediate_file): Likewise.
33789 (find_source): Fix GNU coding style.
33790 (accumulate_line_counts): Remove old non-all block mode.
33791 (output_lines): Remove usage of the union.
33792 * profile.c (output_location): Include all BBs, even if
33793 belonging to a same line (and file) as a previous BB.
33795 2017-04-28 Martin Liska <mliska@suse.cz>
33797 * gcov.c (process_args): Handle new argument 'w'.
33798 (read_graph_file): Assign ID to BBs.
33799 (output_branch_count): Display BB # if verbose flag is set.
33800 (output_lines): Likewise for arcs.
33801 (print_usage): Add '--verbose' option help.
33802 * doc/gcov.texi: Document --verbose (-w) option.
33804 2017-04-28 Martin Liska <mliska@suse.cz>
33806 * gcov.c (struct block_location_info): New struct.
33807 (process_file): Fill up the new structure.
33808 (read_graph_file): Replace usage of encoding by the newly added
33810 (add_line_counts): Likewise.
33811 (accumulate_line_counts): Remove usage of the union.
33812 (function_info::function_info): New function.
33813 (function_info::~function_info): Likewise.
33814 (process_file): Call delete instead of release_function.
33815 (release_function): Release the function.
33816 (release_structures): Call delete instead of release_function.
33817 (solve_flow_graph): Replace usage of num_blocks.
33818 (find_exception_blocks): Likewise.
33819 (output_lines): Fix GNU coding style.
33821 2017-04-28 Martin Liska <mliska@suse.cz>
33824 * coverage.c (coverage_remove_note_file): New function.
33825 * coverage.h: Declare the function.
33826 * toplev.c (finalize): Clean if an error has been seen.
33828 2017-04-28 Martin Liska <mliska@suse.cz>
33830 PR gcov-profile/80031
33831 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33832 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33833 * gcov.c (read_graph_file): Read just number of blocks.
33834 * profile.c (branch_prob): Do not stream 0 flags per a basic
33837 2017-04-28 Martin Liska <mliska@suse.cz>
33839 * gcov-dump.c (tag_*): Add new argument to declarations.
33840 (dump_gcov_file): Likewise.
33841 (tag_blocks): Add and use new argument depth.
33842 (tag_arcs): Likewise.
33843 (tag_lines): Likewise.
33844 (tag_counters): Likewise.
33845 (tag_summary): Likewise.
33846 (dump_working_sets): Use depth to do a proper indentation.
33848 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33851 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33852 * symtab.c (symtab_node::debug_symtab): Move definition here.
33854 2017-04-28 Richard Biener <rguenther@suse.de>
33856 * lto-streamer.h (LTO_major_version): Bump to 7.
33858 2017-04-28 Richard Biener <rguenther@suse.de>
33860 * tree-vrp.c (assert_info): New struct.
33861 (add_assert_info): New helper.
33862 (register_edge_assert_for_2): Refactor to add asserts to a vector
33864 (register_edge_assert_for_1): Likewise.
33865 (register_edge_assert_for): Likewise.
33866 (finish_register_edge_assert_for): New helper actually registering
33867 asserts where live on edge.
33868 (find_conditional_asserts): Adjust.
33869 (find_switch_asserts): Likewise.
33870 (evrp_dom_walker::try_find_new_range): Generalize.
33871 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33873 2017-04-27 Marek Polacek <polacek@redhat.com>
33876 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33877 arg10 and arg11 to itype.
33879 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33881 * doc/extend.texi (Object Size Checking): Improve grammar.
33883 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33886 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33887 that the logic for permitting reciprocal estimates matches that
33890 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33893 * tree.c (type_cache_hasher::equal): Only compare
33894 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33895 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33896 non-aggregate element types.
33897 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33898 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33900 2017-04-27 Richard Biener <rguenther@suse.de>
33902 PR middle-end/80533
33903 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33904 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33905 keeping a reference to a trailing array.
33907 2017-04-27 Richard Biener <rguenther@suse.de>
33909 PR middle-end/80539
33910 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33911 being in loop-closed SSA form conservatively.
33912 (chrec_fold_multiply_poly_poly): Likewise.
33914 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33916 PR middle-end/79665
33917 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33918 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33920 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33923 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33924 (aarch64_function_arg_alignment): Return unsigned int again, but still
33925 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33926 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33927 Don't emit -Wpsabi note.
33928 (aarch64_function_arg_boundary): Likewise.
33929 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33932 2017-04-26 Nathan Sidwell <nathan@acm.org>
33934 * tree.h (crc32_unsigned_n): Declare.
33935 (crc32_unsigned, crc32_unsigned): Make inline.
33936 * tree.c (crc32_unsigned_bits): Replace with ...
33937 (crc32_unsigned_n): ... this.
33938 (crc32_unsigned, crc32_byte): Remove.
33939 (crc32_string): Remove unnecessary braces.
33941 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33943 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33944 * ipa-inline-analysis.c (MAX_TIME): Remove.
33945 (account_size_time): Use sreal for time.
33946 (dump_inline_summary): Update.
33947 (estimate_function_body_sizes): Update.
33948 (estimate_edge_size_and_time): Update.
33949 (estimate_calls_size_and_time): Update.
33950 (estimate_node_size_and_time): Update.
33951 (inline_merge_summary): Update.
33952 (inline_update_overall_summary): Update.
33953 (estimate_time_after_inlining): Update.
33954 (inline_read_section): Update.
33955 (inline_write_summary): Update.
33956 * ipa-inline.c (compute_uninlined_call_time): Update.
33957 (compute_inlined_call_time): Update.
33958 (recursive_inlining): Update.
33959 (inline_small_functions): Update.
33960 (dump_overall_stats): Update.
33961 * ipa-inline.h: Include sreal.h.
33962 (size_time_entry): Turn time to sreal.
33963 (inline_summary): Turn self_time nad time to sreal.
33965 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33967 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33969 (sreal::stream_out, sreal::stream_in): New.
33970 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33972 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33974 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33977 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33980 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33982 (dimode_scalar_chain::compute_convert_gain): Ditto.
33983 (dimode_scalar_chain::make_vector_copies): Ditto.
33984 (dimode_scalar_chain::convert_reg): Ditto.
33985 (dimode_scalar_chain::convert_insn): Ditto.
33986 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33987 (VI248_AVX512BW_1): New mode iterator.
33988 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33989 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33992 2017-04-25 Martin Sebor <msebor@redhat.com>
33994 PR tree-optimization/80497
33995 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33996 constants are representable in HOST_WIDE_INT.
33997 (parse_directive): Ditto.
33999 2017-04-25 Martin Sebor <msebor@redhat.com>
34002 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
34003 (new_zero_array): Adjust signature.
34004 (dom_info::dom_init): Used unsigned rather that size_t.
34005 (dom_info::dom_info): Same.
34007 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
34008 Jakub Jelinek <jakub@redhat.com>
34011 * config/arm/arm.c: Include gimple.h.
34012 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
34013 returns negative, increment ncrn only if it returned positive.
34014 (arm_needs_doubleword_align): Return int instead of bool,
34015 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
34016 members, but if there is any such non-FIELD_DECL
34017 > PARM_BOUNDARY aligned decl, return -1 instead of false.
34018 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
34019 returns negative, increment nregs only if it returned positive.
34020 (arm_setup_incoming_varargs): Likewise.
34021 (arm_function_arg_boundary): Emit -Wpsabi note if
34022 arm_needs_doubleword_align returns negative, return
34023 DOUBLEWORD_ALIGNMENT only if it returned positive.
34025 2017-04-25 Marek Polacek <polacek@redhat.com>
34028 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34029 first argument to type.
34031 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
34034 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
34035 type checks to test for compatibility instead of equality.
34037 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
34038 Jakub Jelinek <jakub@redhat.com>
34041 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
34043 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
34044 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
34045 the alignment computation, but return their maximum in warn_alignment.
34046 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
34047 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
34049 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
34050 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
34053 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34055 * config/arc/simdext.md (dmpyh): Fix typo.
34057 2017-04-25 Richard Biener <rguenther@suse.de>
34059 PR tree-optimization/80492
34060 * alias.c (compare_base_decls): Handle registers with asm
34061 specification conservatively.
34062 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
34063 compare_base_decls returning dont-know properly.
34065 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34067 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
34068 (legitimate_offset_address_p): New function.
34069 (arc_legitimate_address_p): Use above function.
34071 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34073 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
34075 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34077 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
34078 ACCH registers whenever they are available.
34080 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34082 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
34083 double regs fix when not used.
34085 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34087 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
34089 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
34090 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
34092 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34094 * config/arc/arc.c (arc_output_addsi): Check for h-register class
34095 when emitting short ADD instructions.
34097 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34099 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
34101 (cmpsi_cc_c_insn): Likewise.
34102 (cbranchsi4_scratch): Compute proper instruction length using
34103 compact_hreg_operand.
34104 * config/arc/predicates.md (compact_hreg_operand): New predicate.
34106 2017-04-25 Richard Biener <rguenther@suse.de>
34108 PR middle-end/80509
34109 * passes.c (pass_manager::pass_manager): Initialize
34110 m_name_to_pass_map.
34112 2017-04-25 Richard Biener <rguenther@suse.de>
34114 PR tree-optimization/79201
34115 * tree-ssa-sink.c (statement_sink_location): Handle calls.
34117 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34120 * config/s390/vector.md: Split MEM->GPR vector moves for
34121 non-s_operand addresses.
34123 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34126 * config/s390/predicates.md (reload_const_wide_int_operand): New
34128 * config/s390/s390.md ("movti"): Remove d/P alternative.
34129 ("movti_bigconst"): New pattern definition.
34131 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34134 * s390-protos.h (s390_expand_cs_hqi): Removed.
34135 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
34136 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
34137 modes as well as CCZ1mode and CCZmode.
34138 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
34139 signature of s390_emit_compare_and_swap.
34140 (s390_expand_cs_hqi): Likewise, make static.
34141 (s390_expand_cs_tdsi): Generate an explicit compare before trying
34142 compare-and-swap, in some cases.
34143 (s390_expand_cs): Wrapper function.
34144 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
34146 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
34147 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
34148 patterns for small and large integers. Forbid symref memory operands.
34149 Move expander to s390.c. Require cc register.
34150 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
34151 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
34152 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
34153 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
34154 symref memory operands. Remove CC mode and call s390_match_ccmode
34156 ("atomic_exchange<mode>"): Allow and implement all integer modes.
34158 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34160 * config/s390/s390.md (define_peephole2): New peephole to help
34161 combining the load-and-test pattern with volatile memory.
34163 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34165 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
34166 with CCZmode for TARGET_Z196.
34168 2017-04-25 Jakub Jelinek <jakub@redhat.com>
34170 PR rtl-optimization/80501
34171 * combine.c (make_compound_operation_int): Set subreg_code to SET
34172 even for AND with mask of the sign bit of mode.
34174 PR rtl-optimization/80500
34175 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
34176 sum's initial value.
34178 2017-04-25 Julian Brown <julian@codesourcery.com>
34179 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34181 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
34183 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
34185 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
34187 2017-04-25 Julian Brown <julian@codesourcery.com>
34188 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34190 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
34191 (thunderx2t99_sha): New Reservation.
34193 2017-04-25 Julian Brown <julian@codesourcery.com>
34194 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34196 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
34197 type for 1-element load.
34199 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
34201 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
34203 2017-04-24 Martin Jambor <mjambor@suse.cz>
34205 PR tree-optimization/80293
34206 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
34207 char arrays not totally scalarizable if it is false.
34208 (analyze_all_variable_accesses): Pass correct value in the new
34209 parameter. Add a statistics counter.
34211 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
34213 PR middle-end/79931
34214 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
34216 2017-04-24 Richard Biener <rguenther@suse.de>
34218 PR tree-optimization/80494
34219 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
34220 out for complex types.
34222 2017-04-24 Richard Biener <rguenther@suse.de>
34224 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34225 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34226 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34227 (DFS): Adjust and never fail.
34228 (sccvn_dom_walker::fail): Remove.
34229 (sccvn_dom_walker::before_dom_children): Adjust.
34230 (run_scc_vn): Likewise and never fail.
34231 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34232 (pass_fre::execute): Likewise.
34234 2017-04-24 Richard Biener <rguenther@suse.de>
34236 PR tree-optimization/79725
34237 * tree-ssa-sink.c (statement_sink_location): Return whether
34238 failure reason was zero uses. Move that check later.
34239 (sink_code_in_bb): Deal with zero uses by removing the stmt
34242 2017-04-24 Richard Biener <rguenther@suse.de>
34245 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34246 pointer-based references.
34248 2017-04-24 Richard Biener <rguenther@suse.de>
34251 * pass_manager.h (pass_manager::operator new): Remove.
34252 (pass_manager::operator delete): Likewise.
34253 * passes.c (pass_manager::operator new): Remove.
34254 (pass_manager::operator delete): Likewise.
34255 (pass_manager::pass_manager): Zero individual pass members.
34257 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34260 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34261 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34262 Check "XEXP (src, 1)" operand here.
34263 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34264 Check "XEXP (src, 1)" operand here.
34265 (dimode_scalar_chain::make_vector_copies): Detect count register
34266 of a shift instruction. Zero extend count register from QImode
34267 to DImode to satisfy vector shift pattern count operand predicate.
34268 Substitute vector shift count operand with a DImode copy.
34269 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34272 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34274 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34275 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34276 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34277 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34278 (UNSPEC_NOREX_MEM): Remove definition.
34280 2017-04-21 Richard Biener <rguenther@suse.de>
34282 PR tree-optimization/79547
34283 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34284 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34285 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34286 without any constraints.
34288 2017-04-21 Richard Biener <rguenther@suse.de>
34290 PR tree-optimization/78847
34291 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34293 2017-04-21 Richard Biener <rguenther@suse.de>
34295 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34296 (build_distinct_type_copy): Likewise.
34297 (build_variant_type_copy): Likewise.
34298 * tree.c (build_qualified_type): Pass down mem-stat info.
34299 (build_distinct_type_copy): Likewise.
34300 (build_variant_type_copy): Likewise.
34302 2017-04-21 Richard Biener <rguenther@suse.de>
34304 PR tree-optimization/80237
34305 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34307 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34308 for a simplified result.
34310 2016-04-21 Richard Biener <rguenther@suse.de>
34312 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34313 sth as strict as a simple_iv but a chrec without symbols and an
34314 operand defined in the loop we are peeling (and not some subloop).
34315 (propagate_constants_for_unrolling): Propagate all constants.
34317 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34320 * config/i386/i386.c (print_reg): Remove assert for disalowed
34321 regno values, call output_operand_lossage instead.
34323 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34326 * config/i386/constraints.md (Yc): New register constraint.
34327 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34328 Use Yc constraint for alternative 2 of operand 0. Remove
34329 preferred_for_speed attribute.
34331 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34333 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34334 lastprivate clauses in SIMT case.
34336 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34338 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34340 2017-04-20 Richard Biener <rguenther@suse.de>
34342 PR tree-optimization/57796
34343 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34344 as N scalar stores.
34345 (vect_model_load_cost): Cost gathers as N scalar loads.
34347 2017-04-20 Richard Biener <rguenther@suse.de>
34349 * ggc-page.c (ggc_allocated_p): Rename to ...
34350 (safe_lookup_page_table_entry): ... this and return the lookup
34352 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34354 2017-04-20 Richard Biener <rguenther@suse.de>
34356 PR tree-optimization/80453
34357 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34358 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34359 from the conditions.
34360 (vn_phi_eq): Pass them down.
34361 (vn_phi_lookup): Record them.
34362 (vn_phi_insert): Likewise.
34364 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34366 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34367 uninitialized variable warning to avoid buffer overrun.
34369 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34372 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34373 is suppressed for '{ 0 }' in C.
34375 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34377 * BASE-VER: Set to 8.0.0.
34379 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34381 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34382 priority .init_array and .fini_array section with SECTION_NOTYPE
34385 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34387 PR middle-end/80423
34388 * tree.h (build_array_type): Add typeless_storage default argument.
34389 * tree.c (type_cache_hasher::equal): Also compare
34390 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34391 (build_array_type): Add typeless_storage argument, set
34392 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34394 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34395 (build_array_type): Likewise. Add typeless_storage argument.
34397 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34398 Jakub Jelinek <jakub@redhat.com>
34400 PR tree-optimization/80426
34401 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34402 operation on symbolic operands, also compute the overflow for the
34403 invariant part when the operation degenerates into a negation.
34405 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34408 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34409 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34412 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34414 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34417 * config/avr/avr.c (tree.h): Include it.
34418 (cgraph.h): Include it.
34419 (avr_encode_section_info): Don't warn for uninitialized progmem
34420 variable if it's just an alias.
34422 2017-04-19 Richard Biener <rguenther@suse.de>
34425 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34426 when needed by AutoPGO.
34428 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34431 * doc/lto.texi: Remove an extra 'that'.
34433 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34435 PR rtl-optimization/80429
34436 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34437 are only used in debug insns.
34439 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34440 Vladimir Makarov <vmakarov@redhat.com>
34442 * config/sparc/predicates.md (input_operand): Add comment. Return
34443 true for any memory operand when LRA is in progress.
34444 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34446 2017-04-18 Jeff Law <law@redhat.com>
34449 * mips.md ({return,simple_return}_internal): Do not overwrite
34452 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34454 PR tree-optimization/80443
34455 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34456 instead of adding 1, subtract -1 and similarly instead of subtracting
34459 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34461 PR rtl-optimization/80357
34462 * haifa-sched.c (tmp_bitmap): New variable.
34463 (model_recompute): Handle duplicate use records.
34464 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34465 (free_global_sched_pressure_data): Free it.
34467 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34470 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34471 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34472 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34473 instead of SYSTEM_HEADER_DIR.
34475 2017-04-18 Jeff Law <law@redhat.com>
34477 PR middle-end/80422
34478 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34479 predecessors after walking up the insn chain.
34481 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34484 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34485 sizetype type into debug info.
34487 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34490 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34491 unneeded test for TARGET_UPPER_REGS_SF.
34492 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34494 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34497 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34498 instead of gsi_after_labels.
34500 2017-04-18 Jeff Law <law@redhat.com>
34502 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34506 2017-04-13 Jeff Law <law@redhat.com>
34507 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34508 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34510 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34513 * config/avr/avr.c (intl.h): Include it.
34514 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34516 2017-04-18 Martin Liska <mliska@suse.cz>
34518 PR gcov-profile/78783
34519 * gcov-tool.c (gcov_output_files): Validate that destination
34520 file is either removed by the tool or by a user.
34522 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34523 Guy Benyei <guybe@mellanox.com>
34525 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34526 block, and do not negate it, the stored id is already negative.
34528 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34530 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34532 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34535 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34536 masks of options that should be turned off if the VSX vector
34537 options are turned off.
34538 (OTHER_P8_VECTOR_MASKS): Likewise.
34539 (OTHER_VSX_VECTOR_MASKS): Likewise.
34540 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34541 rs6000_disable_incompatible_switches to validate no type switches
34543 (rs6000_incompatible_switch): New function to disallow turning on
34544 other vector options if -mno-vsx, -mno-power8-vector, or
34545 -mno-power9-vector are specified.
34547 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34549 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34551 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34553 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34554 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34555 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34556 (ARG_POINTER_CFA_OFFSET): Likewise.
34558 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34560 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34561 conditions to take advantage of various optimizations.
34563 2017-04-13 Jeff Law <law@redhat.com>
34565 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34566 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34567 (zero_extendsidi2_dext): Likewise.
34569 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34572 * fold-const.c (fold_ternary_loc): Revert
34573 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34576 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34578 PR rtl-optimization/80343
34579 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34580 new scratch pseudo.
34582 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34585 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34586 to ubsan_encode_value.
34588 2017-04-13 Jeff Law <law@redhat.com>
34590 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34591 appearing in DEBUG_INSNs.
34593 2017-04-13 Martin Liska <mliska@suse.cz>
34595 PR gcov-profile/80413
34596 * gcov-io.c (gcov_write_string): Copy to buffer just when
34597 allocated size is greater than zero.
34599 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34602 * dwarf2out.c (decls_for_scope): Ignore declarations of
34603 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34605 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34608 * ipa-visibility.c (non_local_p): Fix typos.
34609 (localize_node): When localizing symbol in same comdat group,
34610 dissolve the group only when we know external symbols are going
34612 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34614 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34616 PR tree-optimization/79390
34617 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34618 order does not result in usable sequence, retry with reversed operand
34624 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34625 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34626 op0 instead of fold_convert_loc (loc, type, arg0).
34628 2017-04-12 Jeff Law <law@redhat.com>
34630 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34631 has a delay slot in the generated code.
34633 * config/cris/cris.md (cris_preferred_reload_class): Return
34634 GENNONACR_REGS rather than GENERAL_REGS.
34636 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34639 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34640 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34641 signedness of the result type.
34643 2017-04-12 Richard Biener <rguenther@suse.de>
34644 Jeff Law <law@redhat.com>
34646 PR tree-optimization/80359
34647 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34648 trim stores to TARGET_MEM_REFs.
34650 2017-04-12 Richard Biener <rguenther@suse.de>
34652 PR tree-optimization/79390
34653 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34654 threading case even more.
34656 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34659 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34660 for quad_address_p for TImode, instead of just not indexed_address.
34662 2017-04-12 Richard Biener <rguenther@suse.de>
34663 Bernd Edlinger <bernd.edlinger@hotmail.de>
34665 PR middle-end/79671
34666 * alias.c (component_uses_parent_alias_set_from): Handle
34667 TYPE_TYPELESS_STORAGE.
34668 (get_alias_set): Likewise.
34669 * tree-core.h (tree_type_common): Add typeless_storage flag.
34670 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34671 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34672 for types containing members with TYPE_TYPELESS_STORAGE.
34673 (place_field): Likewise.
34674 (layout_type): Likewise for ARRAY_TYPE.
34675 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34676 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34677 TYPE_TYPELESS_STORAGE.
34678 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34680 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34683 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34684 first argument to type.
34686 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34690 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34691 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34692 (rs6000_expand_binop_builtin): Likewise.
34693 (rs6000_expand_ternop_builtin): Likewise; also add missing
34694 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34695 vshasigma built-ins.
34696 * doc/extend.texi: Document that vec_xxpermdi's third argument
34697 must be a constant.
34699 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34701 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34702 Use shift_const cost parameter when calculating gain of STV shifts.
34704 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34706 PR rtl-optimization/70478
34707 * lra-constraints.c (process_alt_operands): Check memory for
34708 disfavoring memory insn operand.
34710 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34712 PR middle-end/80100
34713 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34714 left shift in unsigned HOST_WIDE_INT type.
34716 PR rtl-optimization/80385
34717 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34718 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34721 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34722 if they have any depend clauses.
34724 2017-04-11 Martin Liska <mliska@suse.cz>
34727 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34728 * ipa-split.c (split_function): Create a local comdat symbol
34729 if caller is in a comdat group.
34731 2017-04-11 Martin Liska <mliska@suse.cz>
34734 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34737 2017-04-11 Martin Sebor <msebor@redhat.com>
34739 PR middle-end/80364
34740 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34741 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34743 (directive::set_width, directive::set_precision, format_character):
34745 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34748 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34751 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34752 conflict, set target->arch_name instead of target->cpu_name.
34754 2017-04-11 Richard Biener <rguenther@suse.de>
34756 PR tree-optimization/80374
34757 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34758 build_zero_cst, remove fold_convertible_p check again.
34760 2017-04-11 Martin Liska <mliska@suse.cz>
34763 * ubsan.c (instrument_object_size): Do not instrument register
34766 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34769 * config/i386/i386-builtin-types.def
34770 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34771 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34772 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34773 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34774 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34775 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34776 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34777 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34778 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34779 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34780 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34781 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34782 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34783 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34784 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34785 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34786 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34787 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34788 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34789 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34790 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34791 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34792 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34793 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34794 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34795 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34796 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34797 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34798 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34799 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34800 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34801 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34802 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34803 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34804 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34805 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34806 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34807 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34808 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34809 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34810 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34811 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34812 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34813 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34814 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34815 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34817 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34818 flag to second_arg_count, handle 4 argument function type _COUNT
34819 aliases, handle second_arg_count on second argument rather than last.
34821 2017-04-10 Jeff Law <law@redhat.com>
34823 PR tree-optimization/80374
34824 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34825 record anything if we can not convert integer_zero_node to the
34828 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34831 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34832 Enhance special handling given to the TARGET_P9_MINMAX option in
34833 relation to certain other options.
34835 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34837 PR tree-optimization/80153
34838 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34839 remove POINTER_PLUS_EXPR's base part directly, rather than through
34842 2017-04-10 Richard Biener <rguenther@suse.de>
34843 Bin Cheng <bin.cheng@arm.com>
34845 PR tree-optimization/80153
34846 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34847 the first element of pointer type aff_tree. Build result expr in
34849 (add_elt_to_tree): Convert to type unconditionally. Remove other
34850 fold_convert calls.
34851 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34852 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34854 2017-04-10 Richard Biener <rguenther@suse.de>
34856 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34859 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34861 PR rtl-optimization/70478
34862 * lra-constraints.c (curr_small_class_check): New.
34863 (update_and_check_small_class_inputs): New.
34864 (process_alt_operands): Update curr_small_class_check. Disfavor
34865 alternative insn memory operands. Check available regs for small
34868 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34871 * config/mips/mips.opt (-mvirt): Update description.
34872 * doc/invoke.texi (-mvirt): Likewise.
34874 2017-04-10 Richard Biener <rguenther@suse.de>
34876 PR middle-end/80362
34877 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34878 looking for NEGATE_EXPR in -A / -B to A / B folding.
34880 2017-04-10 Martin Liska <mliska@suse.cz>
34882 PR gcov-profile/80224
34883 * gcov.c (print_usage): Fix usage string.
34884 (get_gcov_intermediate_filename): Remove.
34885 (output_gcov_file): Use both for normal and intermediate format.
34886 (generate_results): Do not initialize special file for
34887 intermediate format.
34889 2017-04-10 Richard Biener <rguenther@suse.de>
34891 PR tree-optimization/80304
34892 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34895 2017-04-10 Nathan Sidwell <nathan@acm.org>
34898 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34899 (rs6000_init_builtins): Use it.
34901 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34903 * config/arm/arm.md (<mrc>): Add mode to SET source.
34904 (<mrrc>): Likewise.
34906 2017-04-10 Richard Biener <rguenther@suse.de>
34908 PR middle-end/80344
34909 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34911 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34914 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34915 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34916 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34917 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34918 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34919 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34920 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34921 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34922 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34923 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34924 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34925 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34926 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34927 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34928 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34929 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34930 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34931 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34932 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34933 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34934 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34935 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34936 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34938 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34940 PR rtl-optimization/70478
34941 * lra-constraints.c: Reverse the last patch.
34943 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34945 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34946 Add comment for WCHAR_T.
34948 2017-04-08 Martin Liska <mliska@suse.cz>
34951 2017-04-07 Martin Liska <mliska@suse.cz>
34954 * ipa-split.c (split_function): Add function part to a same comdat
34957 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34960 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34962 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34964 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34965 (vec_pairing): ...to this.
34966 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34967 (rs6000_sched_init): Adjust for name change.
34968 (struct rs6000_sched_context): Likewise.
34969 (rs6000_init_sched_context): Likewise.
34970 (rs6000_set_sched_context): Likewise.
34972 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34978 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34980 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34981 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34982 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34984 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34986 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34988 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34990 PR rtl-optimization/70703
34991 * ira-color.c (update_conflict_hard_regno_costs): Use
34992 int64_t instead of HOST_WIDE_INT.
34994 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34996 PR rtl-optimization/70478
34997 * lra-constraints.c (process_alt_operands): Disfavor alternative
34998 insn memory operands.
35000 2017-04-07 Jeff Law <law@redhat.com>
35002 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
35003 CALL and NOTE_INSN_CALL_ARG_LOCATION.
35005 2017-04-07 Martin Liska <mliska@suse.cz>
35008 * config/aarch64/aarch64.c (aarch64_process_target_attr):
35009 Show error message instead of an ICE.
35011 2017-04-07 Martin Liska <mliska@suse.cz>
35014 * ipa-split.c (split_function): Add function part to a same comdat
35017 2017-04-07 Richard Biener <rguenther@suse.de>
35019 PR middle-end/80341
35020 * tree.c (get_unwidened): Also handle ! for_type case for
35022 * convert.c (do_narrow): Split out from ...
35023 (convert_to_integer_1): ... here. Do not pass final truncation
35024 type to get_unwidened for TRUNC_DIV_EXPR.
35026 2017-04-07 Richard Biener <rguenther@suse.de>
35028 * tree-affine.c (wide_int_ext_for_comb): Take type rather
35030 (aff_combination_const): Adjust.
35031 (aff_combination_scale): Likewise.
35032 (aff_combination_add_elt): Likewise.
35033 (aff_combination_add_cst): Likewise.
35034 (aff_combination_convert): Likewise.
35035 (add_elt_to_tree): Likewise. Remove unused argument.
35036 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
35038 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
35040 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
35042 * config/arm/arm.c (arm_default_short_enums): Use
35043 ARM_DEFAULT_SHORT_ENUMS.
35044 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
35046 2017-04-06 Jakub Jelinek <jakub@redhat.com>
35049 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
35050 members with redundant out-of-class redeclaration.
35052 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
35055 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
35056 * config/i386/i386.md (*zero_extendsidi2):
35057 Add (?*x,*x) and (?*v,*v) alternatives.
35059 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
35062 * config/i386/i386.c (ix86_expand_builtin)
35063 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
35064 mode from insn data. Convert operands to insn operand mode.
35065 Copy operands that don't satisfy insn predicate to a register.
35067 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
35069 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
35072 2017-04-06 Richard Biener <rguenther@suse.de>
35074 PR tree-optimization/80334
35075 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
35076 preserve alignment of accesses.
35078 2017-04-06 Richard Biener <rguenther@suse.de>
35080 PR tree-optimization/80262
35081 * tree-sra.c (build_ref_for_offset): Preserve address-space
35083 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
35084 Drop useless address-space information on MEM_REF offsets.
35086 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
35088 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
35090 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35092 PR rtl-optimization/70703
35093 * ira-color.c (update_conflict_hard_regno_costs): Use
35094 HOST_WIDE_INT instead of long.
35096 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
35099 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
35100 not defined for x86_64 target. Add -mmmx target option when __SSE2__
35102 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
35103 for x86_64 target. Handle -m3dnowa option.
35105 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35107 PR rtl-optimization/70703
35108 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
35109 (update_conflict_hard_regno_costs): Use long instead of unsigned
35110 arithmetic for cost calculation.
35112 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35113 Bernd Edlinger <bernd.edlinger@hotmail.de>
35116 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
35119 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
35122 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
35123 ptr_mode with Pmode throughout.
35124 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
35125 into probe_stack_range and use DImode.
35127 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
35130 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
35131 call_eh_return is true.
35133 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35135 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
35136 Initialize last_match_fntype_index.
35138 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35141 * tree-nvr.c: Include internal-fn.h.
35142 (pass_return_slot::execute): Ignore internal calls without
35145 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35146 Richard Biener <rguenther@suse.de>
35149 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
35150 captures used multiple times, except for the last use.
35151 * generic-match-head.c: Include gimplify.h.
35153 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35155 PR tree-optimization/79390
35156 * target.h (struct noce_if_info): Declare.
35157 * targhooks.h (default_noce_conversion_profitable_p): Declare.
35158 * target.def (noce_conversion_profitable_p): New target hook.
35159 * ifcvt.h (struct noce_if_info): New type, moved from ...
35160 * ifcvt.c (struct noce_if_info): ... here.
35161 (noce_conversion_profitable_p): Renamed to ...
35162 (default_noce_conversion_profitable_p): ... this. No longer
35164 (noce_try_store_flag_constants, noce_try_addcc,
35165 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
35166 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
35167 instead of noce_conversion_profitable_p.
35168 * config/i386/i386.c: Include ifcvt.h.
35169 (ix86_option_override_internal): Don't override
35170 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
35171 (ix86_noce_conversion_profitable_p): New function.
35172 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
35173 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
35174 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
35175 * doc/tm.texi: Regenerated.
35177 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35179 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
35182 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
35185 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
35186 instructions for small multiply cores.
35188 2017-04-04 Jeff Law <law@redhat.com>
35190 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
35192 (mips_expand_vec_perm_const): Initialize elements in orig_perm
35193 that are not set by the loop over the elements.
35195 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35198 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
35199 int mode, convert_modes it to mode as unsigned, otherwise use
35200 lowpart_subreg to mode rather than SImode.
35201 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
35202 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
35203 Use DImode instead of SImode for the shift count operand.
35204 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
35207 2017-04-04 Richard Biener <rguenther@suse.de>
35209 PR middle-end/80281
35210 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
35211 arithmetic done for the negate or the plus. Simplify.
35212 (A - (-B) -> A + B): Likewise.
35213 * fold-const.c (split_tree): Make sure to not negate pointers.
35215 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
35217 PR rtl-optimization/60818
35218 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
35219 a compare of comparisons with the thing compared if this results
35220 in a different machine mode.
35222 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35224 * alias.c (base_alias_check): Fix typo in comment.
35225 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35226 * cgraphunit.c (symbol_table::compile): Likewise.
35227 * collect2.c (maybe_run_lto_and_relink): Likewise.
35228 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35229 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35230 * config/avr/avr.c (avr_map_op_t): Likewise.
35231 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35232 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35233 * config/epiphany/epiphany.md (movcc): Likewise.
35234 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35235 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35237 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35238 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35239 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35240 * config/sh/sh.c (sh_rtx_costs): Likewise.
35241 * fold-const.c (fold_truth_andor): Likewise.
35242 * genautomata.c (collapse_flag): Likewise.
35243 * gengtype.h (struct type::u::s): Likewise.
35244 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35245 * input.c (FORMAT_AMOUNT): Likewise.
35246 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35247 (known_aggs_to_agg_replacement_list): Likewise.
35248 * ipa-inline-analysis.c: Likewise.
35249 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35250 * ipa-polymorphic-call.c
35251 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35252 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35253 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35255 * modulo-sched.c (apply_reg_moves): Likewise.
35256 * omp-expand.c (build_omp_regions_1): Likewise.
35257 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35258 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35259 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35260 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35261 * value-prof.c: Likewise.
35262 * var-tracking.c (val_reset): Likewise.
35264 2017-04-03 Richard Biener <rguenther@suse.de>
35266 PR tree-optimization/80275
35267 * fold-const.c (split_address_to_core_and_offset): Handle
35270 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35272 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35273 descriptors is at least equal to that of functions.
35275 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35277 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35279 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35282 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35283 (mov<IMOD4:mode>): New expander.
35284 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35286 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35288 PR rtl-optimization/79405
35289 * fwprop.c (propagations_left): New variable.
35290 (forward_propagate_into): Decrement it.
35291 (fwprop_init): Initialize it.
35292 (fw_prop): If the variable has reached zero, stop propagating.
35293 (fwprop_addr): Ditto.
35295 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35298 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35299 a FUNCTION_DECL, pass it as decl instead of origin to
35302 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35304 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35307 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35310 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35311 TARGET_VSX_SMALL_INTEGER.
35313 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35315 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35316 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35318 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35320 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35321 extraction from odd-numbered MSA register.
35323 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35325 PR middle-end/80173
35326 * expmed.c (store_bit_field_1): Don't attempt to create
35327 a word subreg out of hard registers wider than word if they
35328 have HARD_REGNO_NREGS of 1 for their mode.
35330 PR middle-end/80163
35331 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35332 conversions to integer types wider than word and pointer.
35335 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35336 (rtx_equal_for_cselib_p): Pass 0 to it.
35337 * cselib.c (cselib_hasher::equal): Likewise.
35338 (rtx_equal_for_cselib_1): Add depth argument. If depth
35339 is 128, don't look up VALUE locs and punt. Increment
35340 depth in recursive calls when walking VALUE locs.
35342 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35344 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35345 (make_gcov_file_name): Use the canonical path name for generating
35347 (read_line): Fix handling of files with ascii null bytes.
35349 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35351 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35352 to initialise a vector register instead
35353 of using a const_int.
35355 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35357 PR translation/80189
35358 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35359 diagnostic messages.
35361 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35364 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35365 (dfp_diex_<mode>): Update mode of operand 1.
35366 * doc/extend.texi (dxex, dxexq): Document change to return type.
35367 (diex, diexq): Document change to argument type.
35369 2017-03-30 Martin Jambor <mjambor@suse.cz>
35372 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35373 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35374 it reflects the signature changes performed at the callee side.
35375 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35376 to cgraph_build_function_type_skip_args.
35377 (build_function_decl_skip_args): Adjust call to the above function.
35379 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35382 * config/i386/sse.md
35383 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35384 register as dest whenever it is a MEM not rtx_equal_p to the
35385 corresponding dup operand, and when forcing into reg move the
35386 reg into the memory afterwards.
35387 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35388 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35389 for the force_reg mode.
35390 (avx512vl_vextractf128<mode>): Use register as dest either
35391 always when a MEM, or when it is a MEM not rtx_equal_p to the
35392 corresponding dup operand, or even not when it is a CONST_VECTOR
35393 depending on the mode and lo vs. hi.
35394 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35396 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35397 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35398 Likewise. Require that operands[2] is even.
35399 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35400 Remove extraneous parens. Require that operands[2] is a multiple
35402 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35403 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35404 disallow memory then.
35406 2017-03-30 Richard Biener <rguenther@suse.de>
35408 PR tree-optimization/77498
35409 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35410 to non-constants over backedges.
35412 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35414 PR rtl-optimization/80233
35415 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35416 as last_combined_insn. Do not test for BARRIER_P separately.
35418 2017-03-29 Andreas Schwab <schwab@suse.de>
35421 * calls.c (prepare_call_address): Convert funexp to Pmode before
35422 copying to temp reg.
35424 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35426 PR tree-optimization/80158
35427 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35428 Handle possible future case of more than one alternate
35430 (replace_rhs_if_not_dup): Likewise.
35431 (replace_one_candidate): Likewise.
35433 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35435 PR rtl-optimization/80193
35436 * ira.c (ira): Do not check allocation for LRA.
35438 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35440 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35441 (nvptx_output_simt_exit): Declare.
35442 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35443 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35444 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35445 (nvptx_declare_function_name): ...here. Emit declaration of local
35446 memory space buffer for omp_simt_enter insn.
35447 (nvptx_output_unisimt_switch): New.
35448 (nvptx_output_softstack_switch): New.
35449 (nvptx_output_simt_enter): New.
35450 (nvptx_output_simt_exit): New.
35451 * config/nvptx/nvptx.h (struct machine_function): New fields
35452 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35453 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35454 (UNSPECV_SIMT_EXIT): Ditto.
35455 (omp_simt_enter_insn): New insn.
35456 (omp_simt_enter): New expansion.
35457 (omp_simt_exit): New insn.
35458 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35460 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35461 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35462 (expand_GOMP_SIMT_EXIT): New.
35463 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35464 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35465 (GOMP_SIMT_EXIT): Ditto.
35466 * target-insns.def (omp_simt_enter): New insn.
35467 (omp_simt_exit): Ditto.
35468 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35470 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35471 (lower_rec_input_clauses): Likewise.
35472 (lower_lastprivate_clauses): Handle SIMT privatization.
35474 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35475 (ompdevlow_adjust_simt_enter): New.
35476 (find_simtpriv_var_op): New.
35477 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35478 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35480 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35481 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35482 (copy_decl_for_dup_finish): Ditto.
35484 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35486 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35489 * config/i386/i386.c (ix86_option_override_internal): Always
35490 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35492 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35494 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35496 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35498 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35499 mark new edge's irreducible flag accordign to it.
35500 (vect_do_peeling): Check loop preheader edge's irreducible flag
35501 and pass it to function slpeel_add_loop_guard.
35503 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35505 PR tree-optimization/80218
35506 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35507 Update block frequencies and counts.
35509 2017-03-28 Richard Biener <rguenther@suse.de>
35511 PR tree-optimization/78644
35512 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35513 of a simplification result we may not use it at all.
35515 2017-03-28 Richard Biener <rguenther@suse.de>
35518 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35519 without arguments, generate default definition of a SSA name.
35521 2017-03-28 Richard Biener <rguenther@suse.de>
35523 PR middle-end/80222
35524 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35525 TYPE_REF_CAN_ALIAS_ALL references.
35526 * fold-const.c (fold_indirect_ref_1): Likewise.
35528 2017-03-28 Martin Liska <mliska@suse.cz>
35531 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35532 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35534 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35535 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35537 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35538 (EXTRA_SPECS): Define.
35539 (SUBTARGET_EXTRA_SPECS): Likewise.
35540 (SUBTARGET_CPP_SPEC): Likewise.
35541 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35542 SUBTARGET_EXTRA_SPECS.
35543 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35545 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35547 * config/arc/simdext.md (vst64_insn): Update pattern.
35548 (vld32wh_insn): Likewise.
35549 (vld32wl_insn): Likewise.
35550 (vld64_insn): Likewise.
35551 (vld32_insn): Likewise.
35553 2017-03-28 Marek Polacek <polacek@redhat.com>
35556 * fold-const.c (fold_comparison): Use protected_set_expr_location
35557 instead of SET_EXPR_LOCATION.
35559 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35561 * tree.c (add_expr): Avoid name lookup warning.
35563 2017-03-27 Jeff Law <law@redhat.com>
35565 PR tree-optimization/80216
35566 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35567 function name. Limit recursion depth.
35568 (record_temporary_equivalences): Corresponding changes.
35570 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35572 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35575 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35578 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35580 * cfgcleanup.c (reg_note_cfa_p): New array.
35581 (insns_have_identical_cfa_notes): New function.
35582 (old_insns_match_p): Don't cross-jump in between /f
35583 and non-/f instructions. If both i1 and i2 are frame related,
35584 verify all CFA notes, their order and content.
35586 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35589 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35590 HImode and SImode with zero extend to DImode to one insn.
35591 (bswap<mode>2_extenddi): Likewise.
35592 (bswapsi2_extenddi): Likewise.
35593 (bswaphi2_extendsi): Likewise.
35594 (bswaphi2): Combine bswap HImode and SImode into one insn.
35595 Separate memory insns from swapping register.
35596 (bswapsi2): Likewise.
35597 (bswap<mode>2): Likewise.
35598 (bswaphi2_internal): Delete, no longer used.
35599 (bswapsi2_internal): Likewise.
35600 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35601 store, and gpr<-gpr swap insns.
35602 (bswap<mode>2_store): Likewise.
35603 (bswaphi2_reg): Register only splitter, combine with the splitter.
35604 (bswaphi2 splitter): Likewise.
35605 (bswapsi2_reg): Likewise.
35606 (bswapsi2 splitter): Likewise.
35607 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35608 the insns into load, store, and register/register insns.
35609 (bswapdi2_ldbrx): Likewise.
35610 (bswapdi2_load): Likewise.
35611 (bswapdi2_store): Likewise.
35612 (bswapdi2_reg): Likewise.
35614 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35616 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35617 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35619 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35622 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35624 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35625 special handling for target option conflicts between dform
35626 options (-mpower9-dform, -mpower9-dform-vector,
35627 -mpower9-dform-scalar) and -mno-direct-move.
35629 2017-03-27 Richard Biener <rguenther@suse.de>
35631 PR tree-optimization/80181
35632 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35634 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35636 * config/arc/predicates.md (move_double_src_operand): Replace the
35637 call to move_double_src_operand with a call to address_operand.
35639 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35641 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35642 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35643 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35645 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35647 * config/arc/predicates.md (long_immediate_loadstore_operand):
35648 Consider scaled addresses cases.
35650 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35652 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35653 restored when in interrupt.
35654 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35655 doesn't have delay slot.
35657 2017-03-27 Richard Biener <rguenther@suse.de>
35660 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35661 inlined thunk clones.
35663 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35666 * asan.c (instrument_derefs): Copy over last operand from
35667 original COMPONENT_REF to the new COMPONENT_REF with
35668 DECL_BIT_FIELD_REPRESENTATIVE.
35669 * ubsan.c (instrument_object_size): Likewise.
35671 2017-03-27 Richard Biener <rguenther@suse.de>
35673 PR tree-optimization/80170
35674 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35675 sure DR/SCEV didnt fold in constants we do not see when looking
35676 at the reference base alignment.
35678 2017-03-27 Richard Biener <rguenther@suse.de>
35680 PR middle-end/80171
35681 * gimple-fold.c (fold_ctor_reference): Properly guard against
35682 NULL return value from canonicalize_constructor_val.
35684 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35687 * config/i386/i386.c (ix86_expand_builtin)
35688 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35689 flags reg setting and flags reg using instructions.
35690 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35691 clobbering instructions to zero extend op2.
35693 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35695 * doc/install.texi (Configuration) <--with-aix-soname>:
35696 Update link to AIX ld.
35698 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35700 PR rtl-optimization/80160
35701 PR rtl-optimization/80159
35702 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35703 reg_alternate_class into account.
35705 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35708 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35709 to consider in curr_insn_transform.
35711 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35713 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35714 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35715 and emit_mode_inner.
35717 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35719 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35720 argument to the overloaded builtin variants. Use the new flag to
35721 deprecate certain builtin variants.
35722 * config/s390/s390-builtin-types.def: Add new builtin types.
35723 * config/s390/s390-builtins.h: Support new flags field for
35724 overloaded builtins.
35725 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35726 (s390_macro_to_expand): Enable vector float data type.
35727 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35728 builtins by incrementing the __VEC__ version number.
35729 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35731 (s390_resolve_overloaded_builtin): Emit error messages depending
35732 on the builtin flags.
35733 * config/s390/s390.c (s390_expand_builtin): Support additional
35734 flags argument. Change error message to match the messages
35735 emitted in s390-c.c.
35736 * config/s390/s390.md: New UNSPEC_* constants.
35737 (op_type): Add new instruction types.
35738 * config/s390/vecintrin.h: Add new builtins and test data class
35740 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35741 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35742 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35743 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35744 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35745 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35747 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35748 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35749 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35750 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35752 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35753 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35754 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35755 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35757 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35758 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35759 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35760 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35761 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35762 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35763 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35765 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35766 ("vec_scatter_element<V_HW_4:mode>_DI")
35767 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35768 ("vec_fpint<mode>", "vflls")
35769 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35770 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35771 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35772 ("*vec_cmphe<mode>_cc"): ... these.
35774 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35775 mode constant instead of magic value.
35777 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35779 * config/s390/s390.c (s390_expand_vec_compare): Support other
35780 vector floating point modes than just V2DF.
35781 (s390_expand_vcond): Likewise.
35782 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35783 (s390_cannot_change_mode_class): Prevent mode changes between TF
35784 and V1TF in vector registers.
35785 * config/s390/s390.md (DF, SF): New mode attributes.
35786 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35787 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35788 SFmode support for VRs.
35789 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35791 (VFT, VF_HW): New mode iterators.
35792 (vw, sdx): New mode attributes.
35793 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35794 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35795 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35796 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35797 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35798 also the new vector floating point modes. Renaming to ...
35800 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35801 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35802 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35803 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35804 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35805 ("vec_unordered<mode>"): ... these.
35807 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35808 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35809 ("*vec_extendv2df"): New insn definitions.
35811 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35813 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35814 ("mulditi3_2", "*muldi3_sign"): New patterns.
35815 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35816 rename the pattern definition.
35818 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35820 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35822 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35824 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35826 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35827 instruction if possible.
35828 * config/s390/vector.md (vec_halfnumelts): New mode
35830 ("*vec_vllezlf<mode>"): New pattern.
35832 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35834 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35835 ("popcountv4si2", "popcountv2di2"): Rename to ...
35836 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35837 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35839 ("popcount<mode>2_vxe"): New pattern.
35841 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35843 * common/config/s390/s390-common.c (processor_flags_table): Add
35845 * config.gcc: Add arch12.
35846 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35847 Default to arch12 for unknown CPU model numbers.
35848 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35849 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35850 PROCESSOR_max sanity check.
35851 * config/s390/s390-opts.h (enum processor_type): Add
35853 * config/s390/s390.c (processor_table): Add arch12.
35854 (s390_expand_builtin): Add check for B_VXE flag.
35855 (s390_issue_rate): Add PROCESSOR_ARCH12.
35856 (s390_get_sched_attrmask): Likewise.
35857 (s390_get_unit_mask): Likewise.
35858 (s390_sched_score): Enable z13 scheduling for arch12.
35859 (s390_sched_reorder): Likewise.
35860 (s390_sched_variable_issue): Likewise.
35861 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35863 (s390_tune_attr): Use z13 scheduling also for arch12.
35864 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35865 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35866 (TARGET_VXE_P): New macros.
35867 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35868 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35869 * config/s390/s390.opt: Add arch12 as processor_type.
35871 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35873 * config/s390/s390.md
35874 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35875 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35876 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35878 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35879 Rename expanders to ...
35881 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35882 ("fixuns_truncdddi2_emu"): ... these.
35884 ("fixuns_trunc<mode>si2_emu"): New expander.
35886 ("*fixuns_truncdfdi2_z13"): Rename to ...
35887 ("*fixuns_truncdfdi2_vx"): ... this.
35889 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35891 * config/s390/2964.md: Remove the single element vector compare
35892 instructions which are no longer used.
35893 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35895 (s390_canonicalize_comparison): Remove handling of DFmode
35897 (s390_expand_vec_compare_scalar): Remove function.
35898 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35899 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35901 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35903 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35905 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35907 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35908 will anyway by matched by mov<mode>_64dfp.
35910 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35912 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35913 vlef/vstef. Add missing operand to vleif.
35915 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35917 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35918 pair for all vector types with 64 bit elements.
35919 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35920 * config/s390/vector.md (V_HW_64): ... here.
35921 (V_128_NOSINGLE): New mode iterator.
35922 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35923 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35924 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35925 ("*vec_load_pairv2di"): Change to ...
35926 ("*vec_load_pair<mode>"): ... this one.
35928 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35930 * config/s390/constraints.md: Add comments.
35931 (jKK): Reject element sizes > 8 bytes.
35932 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35934 * config/s390/s390.md: Add the s_operand checks formerly in
35935 s390_split_ok_p to various splitters where they are still
35937 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35938 for 128 bit vectors. Plus two splitters.
35940 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35942 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35945 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35948 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35949 error if the boundary argument is not constant.
35951 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35953 PR rtl-optimization/80112
35954 * loop-doloop.c (doloop_condition_get): Don't check condition
35955 if cmp isn't SET with IF_THEN_ELSE src.
35957 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35959 PR tree-optimization/80158
35960 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35961 replacing a candidate statement, also replace it for the
35962 candidate's alternate interpretation.
35963 (replace_rhs_if_not_dup): Likewise.
35964 (replace_one_candidate): Likewise.
35966 2017-03-24 Richard Biener <rguenther@suse.de>
35968 PR tree-optimization/80167
35969 * graphite-isl-ast-to-gimple.c
35970 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35972 (translate_isl_ast_to_gimple::get_rename): Likewise.
35974 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35976 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35977 handling of certain combinations of target options, including the
35978 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35979 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35981 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35984 * config/arm/arm.md (*load_multiple): Add reload_completed to
35985 matching condition.
35987 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35988 Richard Biener <rguenth@suse.de>
35990 PR tree-optimization/79908
35991 PR tree-optimization/80136
35992 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35993 been cast away, gimplify_and_add suffices.
35995 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35997 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35999 2017-03-23 Richard Biener <rguenther@suse.de>
36001 PR tree-optimization/80032
36002 * gimplify.c (gimple_push_cleanup): Forced unconditional
36003 cleanups still have to go to the conditional_cleanups
36006 2017-03-22 Jakub Jelinek <jakub@redhat.com>
36008 PR tree-optimization/80072
36009 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
36011 (next_operand_entry_id): Change type to unsigned int.
36012 (sort_by_operand_rank): Make sure to return the right return value
36013 even if unsigned fields are bigger than INT_MAX.
36014 (struct oecount): Change cnt and id type to unsigned int.
36015 (oecount_hasher::equal): Formatting fix.
36016 (oecount_cmp): Make sure to return the right return value
36017 even if unsigned fields are bigger than INT_MAX.
36018 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
36021 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
36022 TREE_READONLY on result if writing it more than once.
36025 * doc/invoke.texi (-fsanitize=thread): Document that with
36026 -fnon-call-exceptions atomics are not able to throw
36030 * tsan.c: Include tree-eh.h.
36031 (instrument_builtin_call): Call maybe_clean_eh_stmt or
36032 maybe_clean_or_replace_eh_stmt where needed.
36033 (instrument_memory_accesses): Add cfg_changed argument.
36034 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
36035 if it returned true.
36036 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
36038 PR rtl-optimization/63191
36039 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
36040 wrapper function, moved the whole old content into ...
36041 (ix86_delegitimize_address_1): ... this. New inline function.
36042 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
36043 true as last argument instead of ix86_delegitimize_address.
36045 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
36047 * config/aarch64/aarch64.c (generic_branch_cost): Copy
36048 cortexa57_branch_cost.
36050 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
36052 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
36054 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36057 * doc/md.texi (Constraints): Document wA constraint.
36058 * config/rs6000/constraints.md (wA): New.
36059 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
36060 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
36061 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
36062 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
36064 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
36067 * gimplify.c (is_oacc_declared): New function.
36068 (oacc_default_clause): Use it to set default flags for acc declared
36069 variables inside parallel regions.
36070 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
36071 declared variables.
36072 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
36073 declare attribute to any decl as necessary.
36075 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
36078 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
36079 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
36080 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
36081 (arm_arch_lpae): This.
36082 * config/arm/arm.c (arm_arch7ve): Rename into ...
36083 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
36084 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
36087 2017-03-22 Martin Liska <mliska@suse.cz>
36090 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
36091 error message instead of an ICE.
36093 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36095 * doc/extend.texi (6.11 Additional Floating Types): Revise.
36097 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
36099 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36101 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36104 2017-03-21 Martin Sebor <msebor@redhat.com>
36106 * doc/extend.texi: Use "cannot" instead of "can't."
36107 * doc/hostconfig.texi: Same.
36108 * doc/install.texi: Same.
36109 * doc/invoke.texi: Same.
36110 * doc/loop.texi: Same.
36111 * doc/md.texi: Same.
36112 * doc/objc.texi: Same.
36113 * doc/rtl.texi: Same.
36114 * doc/tm.texi: Same.
36115 * doc/tm.texi.in: Same.
36116 * doc/trouble.texi: Same.
36118 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
36121 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
36122 (collect_checksum_attributes): Set it.
36123 (die_checksum_ordered): Use it.
36125 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36127 PR tree-optimization/79908
36128 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
36129 change: For a VA_ARG whose LHS has been cast away, use
36130 force_gimple_operand to construct the side effects.
36132 2017-03-21 David Malcolm <dmalcolm@redhat.com>
36134 PR translation/80001
36135 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
36136 more amenable to translation.
36137 (oacc_loop_auto_partitions): Likewise.
36139 2017-03-21 Marek Polacek <polacek@redhat.com>
36140 Martin Sebor <msebor@redhat.com>
36142 PR tree-optimization/80109
36143 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
36144 on INTEGRAL_TYPE_P.
36146 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36147 Segher Boessenkool <segher@kernel.crashing.org>
36150 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
36151 check reg_used_between_p between insn and one of succ or succ2
36152 depending on if succ is artificial insn not inserted into insn
36155 2017-03-21 Martin Liska <mliska@suse.cz>
36157 PR gcov-profile/80081
36158 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
36159 * doc/gcc.texi: Include gcov-dump stuff.
36160 * doc/gcov-dump.texi: New file.
36162 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
36164 PR rtl-optimization/79150
36165 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
36166 conditional jump, if the jump is the last insn of the loop.
36168 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36169 Richard Biener <rguenth@suse.de>
36171 PR tree-optimization/79908
36172 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36173 been cast away, use force_gimple_operand to construct the side
36176 2017-03-21 Martin Liska <mliska@suse.cz>
36178 PR libfortran/79956
36179 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
36182 2017-03-21 Brad Spengler <spender@grsecurity.net>
36185 * plugin.c (htab_hash_plugin): New function.
36186 (add_new_plugin): Use it and adjust.
36187 (parse_plugin_arg_opt): Adjust.
36188 (init_one_plugin): Likewise.
36190 2017-03-21 Richard Biener <rguenther@suse.de>
36192 PR tree-optimization/80032
36193 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
36194 if set force the cleanup to happen unconditionally.
36195 (gimplify_target_expr): Push inserted clobbers with force_uncond
36196 to avoid them being removed by control-dependent DCE.
36198 2017-03-21 Richard Biener <rguenther@suse.de>
36200 PR tree-optimization/80122
36201 * tree-inline.c (copy_bb): Do not expans va-arg packs or
36202 va_arg_pack_len when the inlined call stmt requires pack
36204 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
36206 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36209 * tsan.c (instrument_builtin_call): If the memory model argument
36210 is not a constant, assume it is valid.
36213 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
36216 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36218 PR rtl-optimization/79910
36219 * combine.c (can_combine_p): Do not allow combining an I0 or I1
36220 if its dest is used by an insn before I2 (other than the combined
36221 insns themselves, which are properly handled already).
36223 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36226 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36228 * combine.c (record_used_regs): New static function.
36229 (try_combine): Handle situations where there is an additional
36230 instruction between I2 and I3 which needs to have a LOG_LINK
36234 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36236 * combine.c (try_combine): Delete redundant i1 test. Call
36237 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36239 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36242 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36243 alternatives 13/14.
36245 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36247 PR tree-optimization/80054
36248 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36249 the optimization if a PHI or any of its arguments is not dominated
36250 by the candidate's basis. Use gphi* rather than gimple* as
36252 (replace_profitable_candidates): Clean up a gimple* variable that
36253 should be a gphi* variable.
36255 2017-03-20 Martin Sebor <msebor@redhat.com>
36258 * doc/extend.texi (attribute constructor): Document present limitation.
36260 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36263 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36264 __POWER9_VECTOR__ #ifdef control, change template definition to
36265 use Power9-specific built-in function.
36266 (vec_any_eq): Likewise.
36267 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36268 to control outcomes from this test.
36269 (vector_ae_<mode>p): For VEC_F modes, likewise.
36271 2017-03-20 Ian Lance Taylor <iant@google.com>
36273 * config/i386/i386.c (ix86_function_regparm): Save an extra
36274 register for -fsplit-stack with DECL_STATIC_CHAIN.
36276 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36279 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36280 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36282 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36284 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36286 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36289 2017-03-20 Marek Polacek <polacek@redhat.com>
36292 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36294 2017-03-20 Richard Biener <rguenther@suse.de>
36296 PR tree-optimization/80113
36297 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36298 allocate extra SSA name for PHI def.
36299 (add_close_phis_to_outer_loops): Likewise.
36300 (add_close_phis_to_merge_points): Likewise.
36301 (copy_loop_close_phi_args): Likewise.
36302 (copy_cond_phi_nodes): Likewise.
36304 2017-03-20 Martin Liska <mliska@suse.cz>
36306 PR middle-end/79753
36307 * tree-chkp.c (chkp_build_returned_bound): Do not build
36308 returned bounds for a LHS that's not a BOUNDED_P type.
36310 2017-03-20 Martin Liska <mliska@suse.cz>
36314 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36315 COMPLEX_CST and VECTOR_CST.
36317 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36320 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36321 target operand. A new splitter adds the clobber statement in case
36322 the target operand is dead anyway.
36324 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36326 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36327 to age-old versions of binutils and glibc.
36329 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36331 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36333 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36335 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36337 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36339 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36340 requirement for binutils 2.13.
36342 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36344 * combine.c (try_combine): Delete redundant i1 test. Call
36345 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36347 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36349 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36350 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36352 <riscv64-*-elf>: Re-arrange section
36353 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36354 <riscv32-*-linux>: Likewise.
36355 <riscv64-*-elf>: Likewise
36356 <riscv64-*-linux>: Likewise.
36358 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36361 * aarch64.opt(verbose-cost-dump): Fix typo.
36363 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36366 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36367 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36369 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36371 * reload.c (find_reloads): When reloading a nonoffsettable address,
36372 use RELOAD_OTHER for it and its address reloads.
36374 PR rtl-optimization/79910
36375 * combine.c (record_used_regs): New static function.
36376 (try_combine): Handle situations where there is an additional
36377 instruction between I2 and I3 which needs to have a LOG_LINK
36380 2017-03-17 Jeff Law <law@redhat.com>
36382 PR tree-optimization/71437
36383 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36384 conditional in the hash table first.
36385 (vrp_dom_walker::before_dom_children): Extract condition from
36386 ASSERT_EXPR. Record condition, its inverion and any implied
36387 conditions as well.
36389 2017-03-17 Marek Polacek <polacek@redhat.com>
36390 Markus Trippelsdorf <markus@trippelsdorf.de>
36392 PR tree-optimization/80079
36393 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36396 2017-03-17 Richard Biener <rguenther@suse.de>
36398 PR middle-end/80075
36399 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36400 Properly verify the LHS before the RHS possibly claims to be
36402 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36405 2017-03-17 Martin Jambor <mjambor@suse.cz>
36407 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36408 (List of -O2 options): Likewise.
36409 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36412 2017-03-17 Tom de Vries <tom@codesourcery.com>
36414 * gcov-dump.c (print_usage): Print bug_report_url.
36416 2017-03-17 Richard Biener <rguenther@suse.de>
36418 PR middle-end/80050
36419 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36420 (parser::peek): Likewise.
36422 2017-03-17 Richard Biener <rguenther@suse.de>
36424 PR tree-optimization/80048
36425 * sese.c (free_sese_info): Properly release rename_map and
36426 copied_bb_map elements.
36428 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36430 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36431 Add linked-list forward and backlinks. Insert on
36432 construction, remove on destruction.
36433 (class pass_store_merging): Add m_stores_head field.
36434 (pass_store_merging::terminate_and_process_all_chains):
36435 Iterate over m_stores_head list.
36436 (pass_store_merging::terminate_all_aliasing_chains):
36438 (pass_store_merging::execute): Check for debug stmts first.
36439 Push new chains onto the m_stores_head stack.
36441 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36444 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36445 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36446 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36448 2017-03-16 Jeff Law <law@redhat.com>
36450 PR tree-optimization/71437
36451 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36452 member function. Implementation moved into after_dom_children
36453 member function and into the threader's thread_outgoing_edges
36455 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36456 some code into new thread_outgoing_edges.
36457 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36458 definition. Simplify marker handling (do it here). Assume we always
36459 have the available expression and the const/copies tables.
36460 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36462 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36463 * tree-vrp.c (equiv_stack): No longer file scoped.
36464 (vrp_dom_walker): New class.
36465 (vrp_dom_walker::before_dom_children): New member function.
36466 (vrp_dom_walker::after_dom_children): Likewise.
36467 (identify_jump_threads): Setup domwalker. Use it rather than
36468 walking edges in a random order by hand. Simplify setup/finalization.
36469 (finalize_jump_threads): Remove.
36470 (vrp_finalize): Do not call identify_jump_threads here.
36471 (execute_vrp): Do it here instead and call thread_through_all_blocks
36474 PR tree-optimization/71437
36475 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36477 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36479 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36480 (dom_opt_dom_walker::thread_across_edge): Remove
36481 handle_dominating_asserts argument. All callers changed.
36482 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36483 changes. Remove calls to lhs_of_dominating_assert. Other
36484 uses of handle_dominating_asserts turn into unconditional code
36485 (simplify_control_stmt_condition_1): Likewise.
36486 (simplify_control_stmt_condition): Likewise.
36487 (thread_through_normal_block, thread_across_edge): Likewise.
36488 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36489 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36490 object if it is not an SSA_NAME.
36491 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36492 before calling into the VRP specific simplifiers.
36493 (identify_jump_threads): Remove handle_dominating_asserts
36496 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36499 * tree-diagnostic.c (default_tree_printer): No longer static.
36500 * tree-diagnostic.h (default_tree_printer): New prototype.
36502 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36504 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36505 Change ins into fmov.
36507 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36509 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36510 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36511 Use h_con constraint for operand 1.
36512 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36513 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36515 2017-03-15 Jeff Law <law@redhat.com>
36517 PR tree-optimization/71437
36518 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36519 (record_temporary_equivalences): Use it.
36521 PR tree-optimization/71437
36522 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36523 tree-ssa-scopedtables.
36524 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36525 (record_conditions, record_cond, vuse_eq): Likewise.
36526 (record_edge_info): Adjust to API tweak of record_conditions.
36527 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36528 (record_temporary_equivalences, optimize_stmt): Likewise.
36529 (eliminate_redundant_computations): Likewise.
36530 (record_equivalences_from_stmt): Likewise.
36531 * tree-ssa-scopedtables.c: Include options.h and params.h.
36532 (vuse_eq): New function, moved from tree-ssa-dom.c
36533 (build_and_record_new_cond): Likewise.
36534 (record_conditions): Likewise. Accept vector of conditions rather
36535 than edge_equivalence structure for first argument.
36536 for the first argument.
36537 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36538 from tree-ssa-dom.c.
36539 (avail_exprs_stack::record_cond): Likewise.
36540 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36541 from tree-ssa-dom.c.
36542 (avail_exprs_stack): Add new member functions lookup_avail_expr
36544 (record_conditions): Declare.
36546 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36549 * lra-constraints.c (process_alt_operands): Increase reject for
36550 reloading an input/output operand.
36552 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36555 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36556 insns to convert from signed/unsigned char/short to IEEE 128-bit
36558 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36560 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36563 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36564 subreg of inner mode for values already in registers.
36566 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36568 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36569 iteration reg is used after the loop.
36571 2017-03-14 Martin Sebor <msebor@redhat.com>
36573 PR tree-optimization/79800
36574 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36575 precision in negative-positive range.
36576 (format_floating): Call non-const overload with adjusted precision.
36578 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36581 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36584 2017-03-14 Martin Sebor <msebor@redhat.com>
36586 PR middle-end/80020
36587 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36588 * builtins.def (aligned_alloc): Use it.
36591 * Makefile.in (GTFILES): Add calls.c.
36592 * calls.c: Include "gt-calls.h".
36594 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36596 PR rtl-optimization/79728
36597 * regs.h (struct target_regs): New field
36598 x_contains_allocatable_regs_of_mode.
36599 (contains_allocatable_regs_of_mode): New macro.
36600 * reginfo.c (init_reg_sets_1): Initialize it, and change
36601 contains_reg_of_mode so it includes global regs as well.
36602 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36603 rather than contains_regs_of_mode.
36605 2017-03-14 Martin Liska <mliska@suse.cz>
36607 * doc/invoke.texi: Document options that can't be combined with
36608 -fcheck-pointer-bounds.
36610 2017-03-14 Martin Liska <mliska@suse.cz>
36612 PR middle-end/79831
36613 * doc/invoke.texi (-Wchkp): Document the option.
36615 2017-03-14 Martin Liska <mliska@suse.cz>
36617 * Makefile.in: Install gcov-dump.
36619 2017-03-14 Martin Liska <mliska@suse.cz>
36621 * multiple_target.c (expand_target_clones): Bail out for
36622 an invalid attribute.
36624 2017-03-14 Richard Biener <rguenther@suse.de>
36626 * alias.c (struct alias_set_entry): Pack properly.
36627 * cfgloop.h (struct loop): Likewise.
36628 * cse.c (struct set): Likewise.
36629 * ipa-utils.c (struct searchc_env): Likewise.
36630 * loop-invariant.c (struct invariant): Likewise.
36631 * lra-remat.c (struct cand): Likewise.
36632 * recog.c (struct change_t): Likewise.
36633 * rtl.h (struct address_info): Likewise.
36634 * symbol-summary.h (function_summary): Likewise.
36635 * tree-loop-distribution.c (struct partition): Likewise.
36636 * tree-object-size.c (struct object_size_info): Likewise.
36637 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36638 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36639 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36640 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36641 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36642 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36643 (struct _stmt_vec_info): Likewise.
36645 2017-03-14 Martin Liska <mliska@suse.cz>
36648 * multiple_target.c (create_dispatcher_calls): Check that
36649 a target can create a function dispatcher.
36651 2017-03-14 Martin Liska <mliska@suse.cz>
36654 * multiple_target.c (expand_target_clones): Drop local.local
36655 flag for default implementation.
36657 2017-03-14 Richard Biener <rguenther@suse.de>
36659 PR tree-optimization/80030
36660 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36662 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36664 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36665 gcc_fallthrough() instead of __attribute__((fallthrough));
36667 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36669 * doc/gcc.texi: Remove "up" link to (DIR).
36670 * doc/gccint.texi: Ditto.
36672 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36674 * doc/install.texi (Specific) <avr>: Remove reference to
36677 2017-03-13 Jeff Law <law@redhat.com>
36679 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36680 attribute rather than comments.
36682 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36683 match_scratch operand is highest.
36685 2017-03-13 Martin Liska <mliska@suse.cz>
36687 PR middle-end/78339
36688 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36689 is a CHKP clone, use original declaration.
36691 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36693 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36694 (arc_conditional_register_usage): Use a different allocation order
36695 when optimizing for size.
36696 * common/config/arc/arc-common.c (arc_option_optimization_table):
36697 Section anchors default on when optimizing for size.
36699 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36701 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36703 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36705 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36706 * config/arc/arc.md (cpu_facility): Add cd variant.
36707 (*movqi_insn): Add code density variant.
36708 (*movhi_insn): Likewise.
36709 (*movqi_insn): Likewise.
36710 (*addsi3_mixed): Likewise.
36711 (subsi3_insn): Likewise.
36713 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36715 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36717 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36719 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36720 expressions with MINUS and UNARY ops.
36722 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36725 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36727 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36728 between vec_select and vector argument.
36729 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36730 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36731 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36732 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36733 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36734 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36736 2017-03-13 Richard Biener <rguenther@suse.de>
36739 * params.def (vect-max-peeling-for-alignment): Fix typo.
36741 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36743 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36744 issue that only occurred with binutils below 2.18.
36746 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36748 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36749 refer to binutils 2.11/2.12 minimum.
36751 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36753 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36754 ftp.kernel.org and simplify binutils requirement.
36756 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36758 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36760 (Optimize Options): Ditto. Also remove redundancy.
36762 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36764 PR translation/79848
36765 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36767 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36768 to G_ to avoid double translation.
36770 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36772 PR translation/79923
36773 * auto-profile.c (get_combined_location): Convert leading
36774 character of diagnostics to lower case and remove trailing period.
36775 (read_profile): Likewise for various diagnostics.
36776 * config/arm/arm.c (arm_option_override): Remove trailing period
36777 from various diagnostics.
36778 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36779 (msp430_expand_delay_cycles): Likewise.
36781 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36784 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36785 full command-line argument, rather than just "str".
36786 (aarch64_validate_march): Likewise.
36787 (aarch64_validate_mtune): Likewise.
36789 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36791 PR rtl-optimization/78911
36792 * lra-assigns.c (must_not_spill_p): New function.
36793 (spill_for): Use it.
36795 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36797 PR tree-optimization/79981
36798 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36799 ATOMIC_COMPARE_EXCHANGE ifn result.
36800 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36801 IFN_ATOMIC_COMPARE_EXCHANGE.
36803 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36806 * opts.c (parse_sanitizer_options): Add missing question mark to
36807 "did you mean" message.
36809 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36811 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36813 (VMULEUH_UNS): Likewise.
36814 (VMULOUB_UNS): Likewise.
36815 (VMULOUH_UNS): Likewise.
36816 * config/rs6000/rs6000.c (builtin_function_type): Remove
36817 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36819 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36822 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36823 x with result of extra_parsing_for_operand_code_0.
36824 (function_reader::extra_parsing_for_operand_code_0): Convert
36825 return type from void to rtx, returning x. When reading
36826 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36827 larger size containing struct block_symbol.
36829 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36831 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36832 -mfloat128-hardware without -m64.
36834 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36837 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36838 entries to the case statement that marks unsigned arguments to
36839 overloaded functions.
36841 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36843 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36844 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36846 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36849 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36850 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36852 2017-03-10 Martin Liska <mliska@suse.cz>
36856 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36857 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36860 2017-03-10 Olivier Hainque <hainque@adacore.com>
36862 * tree-switch-conversion (array_value_type): Start by resetting
36863 candidate type to it's main variant.
36865 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36867 PR rtl-optimization/79909
36868 * combine.c (try_combine): Use simplify_replace_rtx on individual
36869 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36870 of the whole CALL_INSN_FUNCTION_USAGE.
36872 PR tree-optimization/79972
36873 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36874 get_range_info on SSA_NAMEs. Formatting fixes.
36876 2017-03-10 Richard Biener <rguenther@suse.de>
36877 Jakub Jelinek <jakub@redhat.com>
36879 PR tree-optimization/77975
36880 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36881 edge to be constant.
36882 (get_val_for): For constant x return it. Formatting fix.
36883 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36884 would use the same bases as the current one.
36886 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36888 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36889 instead of vec_select for V1TImode.
36890 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36892 (VSX_LE_128): Add V1TI to this mode iterator.
36893 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36894 (*vsx_le_perm_store_<mode>): Likewise.
36895 (pre-reload splitter for VSX stores): Likewise.
36896 (post-reload splitter for VSX stores): Likewise.
36897 (*vsx_xxpermdi2_le_<mode>): Likewise.
36898 (*vsx_lxvd2x2_le_<mode>): Likewise.
36899 (*vsx_stxvd2x2_le_<mode>): Likewise.
36901 2017-03-09 Michael Eager <eager@eagercon.com>
36903 Correct failures with --enable-checking=yes,rtl.
36905 * config/microblaze/microblaze.c (microblaze_expand_shift):
36906 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36907 test for const0_rtx.
36908 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36909 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36911 2017-03-09 Richard Biener <rguenther@suse.de>
36913 PR tree-optimization/79977
36914 * graphite-scop-detection.c (scop_detection::merge_sese):
36915 Handle the case of extra exits to blocks dominating the entry.
36917 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36919 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36922 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36924 PR rtl-optimization/79949
36925 * lra-constraints.c (process_alt_operands): Check memory when
36926 trying to predict a cycle. Print about the overall increase.
36928 2017-03-09 Richard Biener <rguenther@suse.de>
36930 PR middle-end/79971
36931 * gimple-expr.c (useless_type_conversion_p): Preserve
36932 TYPE_SATURATING for fixed-point types.
36934 2017-03-09 Richard Biener <rguenther@suse.de>
36937 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36938 alignment of BLKmode params.
36940 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36943 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36944 (VALL_NO_V2Q): Likewise.
36946 * config/aarch64/aarch64-simd.md
36947 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36949 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36950 VALL_NO_V2Q mode iterator.
36951 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36953 2017-03-09 Martin Liska <mliska@suse.cz>
36955 PR tree-optimization/79631
36956 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36957 tree_int_cst_sign_bit just for INTEGER constants.
36959 2017-03-09 Martin Liska <mliska@suse.cz>
36963 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36966 2017-03-09 Marek Polacek <polacek@redhat.com>
36969 * tree.c (inchash::add_expr): Handle TREE_VEC.
36971 2017-03-09 Martin Liska <mliska@suse.cz>
36974 (chkp_narrow_size_and_offset): New function.
36975 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36976 (void chkp_parse_bit_field_ref): New function.
36977 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36978 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36980 2017-03-09 Martin Liska <mliska@suse.cz>
36983 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36984 (chkp_find_bounds_1): Remove gcc_unreachable.
36986 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36989 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36990 BUILT_IN_SYNC*, determine the access type from the size suffix and
36991 always build a MEM_REF with that type. Handle forgotten
36992 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36995 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36996 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36997 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36998 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36999 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
37000 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
37001 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
37002 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
37003 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
37004 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
37005 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
37006 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
37007 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
37008 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
37009 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
37010 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
37011 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
37012 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
37013 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
37014 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
37015 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
37016 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
37017 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
37018 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
37019 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
37020 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
37021 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
37022 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
37023 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
37024 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
37025 definitions outside of __OPTIMIZE__ guarded section.
37028 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
37029 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
37030 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
37031 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
37034 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37036 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
37037 ("vfenez<mode>"): Add missing constraints.
37039 2017-03-08 Martin Sebor <msebor@redhat.com>
37042 * config/nds32/nds32.c (nds32_option_override):
37043 Fix misspelled diagnostic.
37045 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37048 * gimplify.c (gimplify_omp_for): Replace index var in outer
37049 taskloop statement with an artificial variable and add
37050 OMP_CLAUSE_PRIVATE clause for it.
37052 2017-03-08 Richard Biener <rguenther@suse.de>
37054 PR tree-optimization/79955
37055 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
37056 for accesses that are completely outside of the variable.
37058 2017-03-08 Andrew Haley <aph@redhat.com>
37060 PR tree-optimization/79943
37061 * tree-ssa-loop-split.c (compute_new_first_bound): When
37062 calculating the new upper bound, (END-BEG) should be added, not
37065 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37067 * config/avr/avr.md (setmemhi): Make sure match_dup
37068 operand number comes before match_scratch.
37070 2017-03-08 Richard Biener <rguenther@suse.de>
37072 PR tree-optimization/79920
37073 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
37074 with ncopies == 1 to ...
37075 (vect_transform_slp_perm_load): ... here. Properly compute
37076 all element loads by iterating VF times over the group. Do
37077 not handle ncopies (computed in a broken way) in
37078 vect_create_mask_and_perm.
37080 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37083 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
37084 is a uniform vector, use uniform_vector_p return value instead of
37085 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
37087 2017-03-07 Marek Polacek <polacek@redhat.com>
37089 PR middle-end/79809
37090 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
37091 (alloca_call_type): Likewise.
37093 2017-03-07 Martin Liska <mliska@suse.cz>
37095 * gcov.c (process_args): Put comment to correct location.
37097 2017-03-07 Martin Liska <mliska@suse.cz>
37099 PR middle-end/68270
37100 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
37101 Use array_at_struct_end_p instead of DECL_CHAIN (field).
37102 (chkp_narrow_bounds_for_field): Likewise.
37103 (chkp_parse_array_and_component_ref): Pass one more argument to
37106 2017-03-07 Richard Biener <rguenther@suse.de>
37108 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
37111 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
37113 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
37114 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
37116 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37119 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
37120 to end of description.
37121 (PARAM_MAX_STORES_TO_MERGE): Likewise.
37123 2017-03-07 Jakub Jelinek <jakub@redhat.com>
37125 PR rtl-optimization/79901
37126 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
37128 (*avx512f_<code><mode>3<mask_name>): ... this.
37129 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
37130 iterator instead of VI8_AVX2_AVX512BW.
37132 PR rtl-optimization/79901
37133 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
37134 min/max expander, expand it using expand_vec_cond_expr.
37137 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
37140 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37143 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
37144 to void * for PCH reasons.
37145 * dwarf2out.c (output_loc_operands, output_die): Cast
37146 v.val_vec.array to unsigned char *.
37148 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
37151 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
37154 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
37156 PR rtl-optimization/79571
37157 * lra-constraints.c (process_alt_operands): Calculate static
37158 reject and subtract it from overall when only addresses will be
37161 2017-03-06 Julia Koval <julia.koval@intel.com>
37164 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
37165 incoming stack boundary to 128 for 64-bit targets.
37167 2017-03-06 Richard Biener <rguenther@suse.de>
37169 PR tree-optimization/79894
37170 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
37171 to NULL after folding it.
37173 2017-03-06 Richard Biener <rguenther@suse.de>
37175 PR tree-optimization/79824
37176 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
37177 check disabling peeling for gaps.
37179 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
37181 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
37182 attributes): Document gettimeofday.
37184 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37186 * config/s390/s390.c (s390_option_override_internal): Set
37187 PARAM_MIN_VECT_LOOP_BOUND
37189 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37191 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
37192 * config/s390/s390.md: Likewise.
37194 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37197 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
37198 (<avx2_avx512>_perm<mode>): Rename to ...
37199 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
37201 (<avx512>_perm<mode>_mask): Rename to ...
37202 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
37204 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
37205 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
37206 instead of VI8F_256_512.
37207 (avx512f_perm<mode>): New define_expand.
37208 (avx512f_perm<mode>_mask): Likewise.
37209 (avx512f_perm<mode>_1<mask_name>): New define_insn.
37210 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
37212 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37214 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
37215 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
37217 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
37219 2017-03-06 Martin Liska <mliska@suse.cz>
37222 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37223 when having a SSA NAME w/o VAR_DECL assigned to it.
37225 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37227 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37228 msa_dpsub_<su>_d): Fix MODE for vec_select.
37230 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37232 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37234 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37236 2017-03-06 Richard Biener <rguenther@suse.de>
37238 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37239 * plugin.c (register_plugin_info): Likewise.
37240 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37242 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37244 * config/i386/sse.md (sse_storehps, sse_storelps,
37245 avx_<castmode><avxsizesuffix>_<castmode>,
37246 avx512f_<castmode><avxsizesuffix>_<castmode>,
37247 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37248 in condition that at least one operand is not a MEM.
37250 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37252 PR middle-end/79805
37253 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37254 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37256 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37257 gimple_call_nothrow_p flag based on whether original builtin can throw.
37258 If it can, emit following stmts on the fallthrough edge.
37259 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37260 don't create new bb if inserting just debug stmts on the edge, try to
37261 insert them on the fallthru bb or just reset debug stmts.
37263 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37266 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37267 restore recog_data (including the operand rtxes inside it) around
37268 the call to get_insn_template.
37270 2017-03-03 Martin Sebor <msebor@redhat.com>
37272 PR tree-optimization/79699
37273 * context.c (context::~context): Free MPFR caches to avoid
37274 a memory leak on program exit.
37276 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37278 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37279 Use wide_int::ulow () instead of .elt (0).
37281 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37283 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37284 (*pushxf): Limit oF constraint to 32bit targets and add oC
37285 constraint for 64bit targets.
37286 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37287 (*pushdf): Change rmF constraint to rmC.
37289 2017-03-03 Martin Liska <mliska@suse.cz>
37291 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37292 Remove unused variable.
37294 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37297 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37298 is a memory operand, increase num_memory.
37299 (ix86_expand_args_builtin): Likewise.
37301 2017-03-03 Jan Hubicka <jh@suse.cz>
37304 * ipa-devirt.c (maybe_record_node): Properly handle
37305 __cxa_pure_virtual visibility.
37307 2017-03-03 Martin Liska <mliska@suse.cz>
37309 PR tree-optimization/79803
37310 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37312 (pass_loop_prefetch::execute): Disabled optimization if an
37313 assumption about L1 cache size is not met.
37315 2017-03-03 Martin Liska <mliska@suse.cz>
37317 PR rtl-optimization/79574
37318 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37319 (hash_scan_set): Likewise.
37320 (dump_hash_table): Likewise.
37321 (hoist_code): Likewise.
37323 2017-03-03 Richard Biener <rguenther@suse.de>
37325 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37326 in place of uhigh ().
37327 (fixed_convert_from_real): Likewise.
37329 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37332 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37334 2017-03-03 Richard Biener <rguenther@suse.de>
37336 PR middle-end/79818
37337 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37338 TYPE_OVERFLOW_UNDEFINED check.
37340 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37342 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37344 (vector_ae_<mode>_p): Likewise.
37345 (vector_nez_<mode>_p): Likewise.
37346 (vector_ne_v2di_p): Likewise.
37347 (vector_ae_v2di_p): Likewise.
37348 (vector_ne_<mode>_p): Likewise.
37349 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37351 (vsx_tsqrt<mode>2_fe): Likewise.
37353 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37356 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37358 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37360 PR rtl-optimization/79780
37361 * cprop.c (one_cprop_pass): When second and further conditional trap
37362 in a single basic block is turned into an unconditional trap, turn it
37363 into a deleted note to avoid RTL verification failures.
37365 2017-03-02 Richard Biener <rguenther@suse.de>
37367 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37369 2017-03-02 Richard Biener <rguenther@suse.de>
37371 PR tree-optimization/79345
37373 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37374 param and abort the walk, returning -1 if it is hit.
37375 (walk_aliased_vdefs): Take a limit param and pass it on.
37376 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37377 defaulting to 0 and return a signed int.
37378 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37379 (check_defs): New helper.
37380 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37381 about uninitialized memory.
37382 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37383 bogus uninitialized warning.
37384 (fixed_convert_from_real): Likewise.
37386 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37388 PR tree-optimization/66768
37389 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37390 iv_use if base object can't be determined.
37392 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37394 PR tree-optimization/79345
37395 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37396 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37397 (get_pattern_stats): Initialize it.
37398 * genemit.c (gen_expand): Verify match_scratch numbers come after
37399 match_operand/match_dup numbers.
37400 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37401 match_scratch numbers.
37402 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37404 * config/s390/s390.md (trunctdsd2): Likewise.
37406 2017-03-02 Richard Biener <rguenther@suse.de>
37408 * wide-int.h (wide_int_storage::operator=): Implement in terms
37411 2017-03-02 Richard Biener <rguenther@suse.de>
37413 PR tree-optimization/79777
37414 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37415 the to insert expression to sth existing.
37417 2017-03-01 Martin Sebor <msebor@redhat.com>
37419 PR middle-end/79692
37420 * gimple-ssa-sprintf.c
37421 (directive::known_width_and_precision): New function.
37422 (format_integer): Use it.
37423 (get_mpfr_format_length): Consider the full range of precision
37424 when computing %g output with the # flag. Set the likely byte
37425 count to 3 rather than 1 when precision is indeterminate.
37426 (format_floating): Correct the lower bound of precision.
37428 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37430 * doc/invoke.texi: Document default code model for 64-bit Linux.
37432 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37435 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37436 udiv rather than div since input pattern is unsigned.
37438 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37440 * config/i386/i386.c (print_reg): Warn for values of
37441 unsupported size in integer register.
37443 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37446 * config/rs6000/predicates.md (current_file_function_operand): Do
37447 not allow self calls to be local if the function is replaceable.
37449 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37452 * config/rs6000/altivec.h (vec_ctz and others): Change the
37453 preprocessor macro that controls conditional compilation from
37454 _ARCH_PWR9 to __POWER9_VECTOR__.
37455 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37456 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37457 control (instead of _ARCH_PWR9 control) so that template
37458 definition uses power9-specific function.
37459 (vec_any_eq): Likewise.
37460 (vec_all_ne): Change macro definition to use a power9-specific
37461 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37462 _ARCH_PWR9 control).
37463 (vec_any_eq) Likewise.
37464 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37465 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37466 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37467 support for xvcmpnedp instruction.
37468 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37469 macro expansion so that Power9 implementation of vec_all_ne does
37470 not use the AltiVec predicate framework.
37471 (VCMPNEH_P): Likewise.
37472 (VCMPNEW_P): Likewise.
37473 (VCMPNED_P): Likewise.
37474 (VCMPNEFP_P): Likewise.
37475 (VCMPNEDP_P): Likewise.
37476 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37477 implementation of vec_any_eq to not use AltiVec predicate
37479 (VCMPAEH_P): Likewise.
37480 (VCMPAEW_P): Likewise.
37481 (VCMPAED_P): Likewise.
37482 (VCMPAEFP_P): Likewise.
37483 (VCMPAEDP_P): Likewise.
37484 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37485 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37486 not use the AltiVec predicate framework.
37487 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37488 of vec_any_eq to not use AltiVec predicate framework.
37489 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37490 support for predefined __POWER9_VECTOR__ macro to indicate that
37491 Power9 instruction selection is enabled.
37492 (altivec_overloaded_builtins): Remove extraneous
37493 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37494 function argument types RS6000_BTI_bool_V16QI and
37495 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37496 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37497 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37498 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37499 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37500 Power9 for implementations of vec_cmpne. Change the signature for
37501 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37502 (representing vec_all_ne) to remove the previously described first
37503 argument of type RS6000_BTI_INTSI, as this was an artifact of
37504 reliance on the AltiVec predicate framework, which is no longer
37505 used in the implementation of these functions. Add
37506 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37507 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37508 since, unlike the AltiVec predicate framework implementation, we
37509 do not share function descriptors between vec_alle and vec_anyeq.
37510 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37511 set of modes that receive special treatment even when
37512 TARGET_P9_VECTOR is true. The special treatment emits code that
37513 does not depend on Power9 instructions.
37514 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37515 define_expand to not rely on AltiVec predicate framework.
37516 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37518 (vector_ne_v2di_p): Change this define_expand to not rely on
37519 AltiVec predicate framework.
37520 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37522 (vector_ne_<mode>_p): Change this define_expand to not rely on
37523 AltiVec predicate framework.
37524 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37526 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37527 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37528 define_insn pattern.
37529 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37530 define_insn pattern because the xvcmpne<VSs>. instruction is not
37532 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37533 instruction is not supported.
37535 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37537 * config/nvptx/nvptx.c: Include intl.h.
37539 2017-03-01 Martin Jambor <mjambor@suse.cz>
37542 * ipa-prop.h (ipa_bits): Removed field known.
37543 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37544 to pointers. Adjusted their comments to warn about their sharing.
37545 (ipcp_transformation_summary): Change bits to a vector of pointers.
37546 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37547 (ipa_get_ipa_bits_for_value): Declare.
37548 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37549 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37550 (ipa_bits_hash_table): Likewise.
37551 (ipa_vr_ggc_hash_traits): Likewise.
37552 (ipa_vr_hash_table): Likewise.
37553 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37554 being pointers and vr_known being removed.
37555 (ipa_set_jf_unknown): Likewise.
37556 (ipa_get_ipa_bits_for_value): New function.
37557 (ipa_set_jfunc_bits): Likewise.
37558 (ipa_get_value_range): New overloaded functions.
37559 (ipa_set_jfunc_vr): Likewise.
37560 (ipa_compute_jump_functions_for_edge): Use the above functions to
37561 construct bits and vr parts of jump functions.
37562 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37563 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37565 (ipcp_grow_transformations_if_necessary): Also allocate
37566 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37568 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37569 them. Fix too long lines.
37570 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37571 vr_known being removed.
37572 (ipa_read_jump_function): Use new setter functions to construct bits
37573 and vr parts of jump functions or set them to NULL.
37574 (write_ipcp_transformation_info): Adjust for bits being pointers.
37575 (read_ipcp_transformation_info): Likewise.
37576 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37578 Include gt-ipa-prop.h.
37579 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37581 (ipcp_store_bits_results): Likewise.
37582 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37583 Do not write to existing jump functions but use a temporary instead.
37585 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37588 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37589 attempt to use its first operand as BIT_FIELD_REF base.
37591 2017-03-01 Richard Biener <rguenther@suse.de>
37593 PR middle-end/79721
37594 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37595 interpolating formula in wrapping arithmetic.
37596 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37598 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37600 PR tree-optimization/79734
37601 * tree-vect-generic.c (expand_vector_condition): Optimize
37602 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37603 Handle VEC_COND_EXPR where comparison has different inner width from
37604 type's inner width.
37606 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37608 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37609 markup, and similar issues. Remove @opindex entries for things
37610 that aren't options. Add missing -mmpy-option entries.
37612 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37614 PR tree-optimization/79737
37615 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37616 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37617 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37618 instead of byte_size. Formatting fix.
37619 (shift_bytes_in_array_right): Formatting fix.
37621 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37624 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37625 condition on optimize for the leaf function test.
37627 2017-02-28 Martin Liska <mliska@suse.cz>
37630 * read-rtl-function.c (function_reader::handle_unknown_directive):
37631 Bail out when one uses -flto.
37633 2017-02-28 Martin Liska <mliska@suse.cz>
37635 * common.opt: Replace space with tabular for options of <number>
37637 * config/i386/i386.opt: Show <number> value for
37638 -mlarge-data-threshold.
37639 * opts.c (print_filtered_help): Do not display number in hexadecimal
37642 2017-02-28 Martin Liska <mliska@suse.cz>
37644 * common.opt: Fix --help=option -Q for options which are of
37647 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37649 * config/i386/i386.c (print_reg): Error out for values
37650 of 8-bit size in invalid integer register.
37652 2017-02-28 Martin Sebor <msebor@redhat.com>
37654 PR tree-optimization/79691
37655 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37657 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37660 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37661 gcc_unreachable with output_operand_lossage.
37663 2017-02-28 Richard Biener <rguenther@suse.de>
37665 PR tree-optimization/79740
37666 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37668 (visit_nary_op): Insert the nary into the hashtable if we
37669 pattern-matched sth.
37670 * tree-ssa-pre.c (eliminate_insert): Robustify.
37672 2017-02-28 Richard Biener <rguenther@suse.de>
37674 PR middle-end/79731
37675 * fold-const.c (decode_field_reference): Reject out-of-bound
37678 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37680 * config/i386/i386.c: Include intl.h.
37681 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37682 instead of just cond ? "..." : "...".
37683 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37684 * coverage.c (read_counts_file): Likewise.
37685 * omp-offload.c: Include intl.h.
37686 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37687 of just cond ? "..." : "...".
37688 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37689 of just cond ? "..." : "...".
37691 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37694 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37696 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37697 'tune for' CPU name.
37698 * config/arm/arm-cpu-data.h: Regenerated.
37700 2017-02-28 Richard Biener <rguenther@suse.de>
37702 PR tree-optimization/79732
37703 * tree-inline.c (expand_call_inline): Do not shadow var.
37705 2017-02-28 Richard Biener <rguenther@suse.de>
37707 PR tree-optimization/79723
37708 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37709 address-space properly.
37711 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37713 * doc/optinfo.texi (Optimization groups): Fix option used for
37715 * doc/invoke.texi (-fopt-info): Document "omp".
37716 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37717 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37718 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37719 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37720 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37722 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37724 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37725 OPTGROUP_OMP with "omp".
37727 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37730 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37731 for arithmetic shift of unsigned V2DI.
37733 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37735 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37736 arc/linux.h headers.
37737 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37738 (LINK_SPEC): Likewise.
37739 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37740 (EXTRA_SPECS): Likewise.
37741 (STARTFILE_SPEC): Likewise.
37742 (ENDFILE_SPEC): Likewise.
37743 (LIB_SPEC): Likewise.
37744 (TARGET_SDATA_DEFAULT): Likewise.
37745 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37746 (MULTILIB_DEFAULTS): Likewise.
37747 (DWARF2_UNWIND_INFO): Likewise.
37748 * config/arc/big.h: New file.
37749 * config/arc/elf.h: Likewise.
37750 * config/arc/linux.h: Likewise.
37751 * config/arc/t-uClibc: Remove.
37753 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37755 PR tree-optimization/77536
37756 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37757 (tree_transform_and_unroll_loop): Use above function to compute the
37758 estimated niter of unrolled loop and use it when scaling profile.
37759 Also use count info rather than frequency if it's non-zero.
37760 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37761 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37762 (vect_transform_loop): Call above function.
37764 2017-02-27 Richard Biener <rguenther@suse.de>
37766 PR tree-optimization/45397
37767 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37768 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37769 (visit_nary_op): Add pattern matching for CSEing sign-changed
37770 or truncated operations with wider ones.
37772 2017-02-27 Richard Biener <rguenther@suse.de>
37774 PR tree-optimization/79690
37775 * tree-vect-stmts.c (vectorizable_store): Use vector type
37776 built from the DR with address-space.
37778 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37780 * doc/invoke.texi (Optimize Options): Refine the description
37781 of asan-use-after-return.
37783 2017-02-25 Alan Modra <amodra@gmail.com>
37785 PR rtl-optimization/79584
37786 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37787 base, not ad->base_term, the reg within base. Remove assertion
37788 that ad->base == ad->base_term. Replace gen_int_mode using
37789 bogus mode with const0_rtx.
37791 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37793 PR middle-end/79396
37794 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37795 FMA_EXPR like tcc_binary or tcc_unary.
37797 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37800 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37802 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37803 (output_loc_operands): Handle DW_OP_call_ref and
37804 DW_OP_GNU_variable_value.
37805 (struct variable_value_struct): New type.
37806 (struct variable_value_hasher): Likewise.
37807 (variable_value_hash): New variable.
37808 (string_types): Remove.
37809 (copy_loc_descr): New function.
37810 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37811 (prepend_loc_descr_to_each): New function.
37812 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37813 instead of add_loc_descr_to_each if the first argument is single
37814 location list and the second has multiple.
37815 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37816 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37817 when looking for variable value which doesn't have other location info.
37818 (loc_list_from_tree): Formatting fix.
37819 (gen_array_type_die): Simplify DW_AT_string_length handling.
37820 (adjust_string_types): Remove.
37821 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37822 string_types. Call resolve_variable_values.
37823 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37824 (resolve_addr_in_expr): Likewise. Add A argument.
37825 (copy_deref_exprloc): Remove deref argument. Adjust for the
37826 original expression being DW_OP_GNU_variable_value with optionally
37827 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37828 optionally after it.
37829 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37830 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37831 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37832 (variable_value_hasher::hash, variable_value_hasher::equal): New
37834 (resolve_variable_value_in_expr, resolve_variable_value,
37835 resolve_variable_values, note_variable_value_in_expr,
37836 note_variable_value): New functions.
37837 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37840 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37843 * opts.h (handle_generated_option): Add GENERATED_P argument.
37844 * opts-common.c (handle_option): Adjust function comment.
37845 (handle_generated_option): Add GENERATED_P argument, pass it to
37847 (control_warning_option): Pass false to handle_generated_option
37849 * opts.c (maybe_default_option): Pass true to handle_generated_option
37851 * optc-gen.awk: Likewise.
37853 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37855 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37856 a REG, look at the REG it is a SUBREG of.
37857 (splitter for cmpeqsi_t): Ditto.
37859 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37861 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37862 the special USEs with the pattern of the insn, not the insn itself.
37864 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37867 * doc/invoke.texi: Document -mload-store-pairs.
37869 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37870 Sandra Loosemore <sandra@codesourcery.com>
37872 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37873 argument isn't a CONST_INT.
37874 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37875 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37876 (nios2_validate_compare): Bypass alternate compare logic if *op2
37877 is not a CONST_INT.
37878 (ldstwm_operation_p): Return false if first_base is not a REG or
37879 if first_offset is not a CONST_INT.
37881 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37883 * config/cris/cris.md: Use correct operand in a define_peephole2.
37885 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37887 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37889 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37891 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37892 this_insn if it is an INSN or JUMP_INSN.
37893 (force_offsettable): Look at base, not at addr.
37894 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37895 on things that aren't necessarily CONST_INTs.
37897 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37899 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37900 -mfpmath=sse is the default also for x86-32 targets with SSE2
37901 instruction set when @option{-ffast-math} is enabled
37903 2017-02-24 Jeff Law <law@redhat.com>
37905 PR rtl-optimizatoin/79286
37906 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37909 2017-02-24 Richard Biener <rguenther@suse.de>
37911 PR tree-optimization/79389
37912 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37915 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37917 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37918 function comment to reflect reality.
37919 (loop_exits_before_overflow): Fix typo in function description.
37921 2017-02-24 Richard Biener <rguenther@suse.de>
37923 PR tree-optimization/79389
37924 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37925 properly that a threading opportunity exists. Detect conditional
37926 copy/constant propagation opportunities.
37928 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37930 * config/visium/visium.md (type): Add trap.
37931 (b): New mode attribute.
37932 (*btst): Rename into...
37933 (*btst<mode>): ...this and adjust.
37934 (*cbranchsi4_btst_insn): Rename into...
37935 (*cbranch<mode>4_btst_insn): ...this and adjust.
37936 (trap): New define_insn.
37938 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37940 PR tree-optimization/79389
37941 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37942 (noce_reversed_cond_code): New function.
37943 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37944 reversed_comparison_code. Formatting fix.
37945 (noce_try_store_flag): Test rev_cond != NULL in addition to
37946 reversed_comparison_code.
37947 (noce_try_store_flag_constants): Likewise.
37948 (noce_try_store_flag_mask): Likewise.
37949 (noce_try_addcc): Use rev_cond if non-NULL instead of
37950 reversed_comparison_code.
37951 (noce_try_cmove_arith): Likewise. Formatting fixes.
37952 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37953 (noce_find_if_block): Initialize rev_cond.
37954 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37955 instead of false as last argument never attempt to reverse it
37958 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37960 PR tree-optimization/79663
37961 * tree-predcom.c (combine_chains): Process refs in reverse order
37962 only for ZERO length chains, and add explaining comment.
37964 2017-02-23 Jeff Law <law@redhat.com>
37966 PR tree-optimization/79578
37967 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37968 in call to operand_equal_p.
37970 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37973 * config/i386/cpuid.h: Fix another undefined behavior.
37975 2017-02-23 Richard Biener <rguenther@suse.de>
37977 PR tree-optimization/79683
37978 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37979 vector types for data-refs.
37981 2017-02-23 Martin Liska <mliska@suse.cz>
37983 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37985 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37987 PR middle-end/79665
37988 * internal-fn.c (get_range_pos_neg): Moved to ...
37989 * tree.c (get_range_pos_neg): ... here. No longer static.
37990 * tree.h (get_range_pos_neg): New prototype.
37991 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37992 are known to be in between 0 and signed maximum inclusive, try to
37993 expand both unsigned and signed divmod and use the cheaper one from
37996 2017-02-22 Jeff Law <law@redhat.com>
37998 PR tree-optimization/79578
37999 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
38000 to compare base operands.
38002 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
38005 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
38006 gpc_reg_operand instead of fpr_reg_operand.
38008 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
38010 * config/mips/mips.c (mips_return_in_memory): Force FP
38011 vector types to be returned in memory for o32 ABI.
38013 2017-02-22 Jakub Jelinek <jakub@redhat.com>
38015 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
38016 instead of DW_TAG_member for static data member declarations and don't
38017 set no_linkage_name for static inline data members.
38018 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
38021 2017-02-22 Martin Liska <mliska@suse.cz>
38023 * doc/invoke.texi: Replace inequality signs with square brackets
38026 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
38029 * lra-constraints.c (simplify_operand_subreg): Handle
38030 WORD_REGISTER_OPERATIONS targets.
38032 2017-02-22 Jakub Jelinek <jakub@redhat.com>
38035 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
38036 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
38037 elimination by swapping fld*.
38039 2017-02-22 Richard Biener <rguenther@suse.de>
38041 PR tree-optimization/79673
38042 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
38043 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
38044 irrelevant address-space qualifiers and avoiding a
38045 ADDR_SPACE_CONVERT_EXPR from fold_convert.
38047 2017-02-22 Richard Biener <rguenther@suse.de>
38049 PR tree-optimization/79666
38050 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
38051 to not symbolically negate if that may introduce undefined
38054 2017-02-22 Martin Liska <mliska@suse.cz>
38057 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
38058 * data-streamer-out.c (streamer_write_gcov_count_stream):
38060 * value-prof.c (stream_out_histogram_value): Make assert more
38061 precise based on type of counter.
38063 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
38066 * config/i386/i386.md (standard_x87sse_constant_load splitter):
38067 Use nonimmediate_operand instead of memory_operand for operand 1.
38068 (float-extend standard_x87sse_constant_load splitter): Ditto.
38070 2017-02-21 Jeff Law <law@redhat.com>
38072 PR tree-optimization/79621
38073 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
38074 blocks with edges to themselves.
38076 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38079 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
38080 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
38081 Use gimple_call_builtin_p.
38084 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
38085 on temporarily removed DEBUG_INSNs.
38087 PR tree-optimization/79649
38088 * tree-loop-distribution.c (classify_partition): Give up on
38089 non-generic address space loads/stores.
38091 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
38093 * doc/loop.texi (Loop manipulation): Remove nonexistent
38094 tree_ssa_loop_version from the documentation.
38095 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
38097 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38100 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
38101 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
38102 * config/rs6000/rs6000.c: Include except.h.
38103 (rs6000_expand_split_stack_prologue): Call
38104 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
38106 2017-02-21 Martin Jambor <mjambor@suse.cz>
38109 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
38110 have been analyzed.
38112 2017-02-21 Martin Jambor <mjambor@suse.cz>
38114 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
38115 for backward compatibility only.
38116 * doc/invoke.texi (Option Summary): Remove all references to
38117 -fipa-cp-alignment.
38119 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
38123 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38125 * lra-constraints.c (curr_insn_transform): Handle
38126 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38128 2017-02-21 Martin Liska <mliska@suse.cz>
38130 * config/i386/i386.opt: Replace -masm-dialect with -masm.
38132 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
38134 PR translation/79638
38135 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
38137 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
38140 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
38141 (arm_function_ok_for_sibcall): Return false for an indirect call by
38142 descriptor if all the argument registers are used.
38143 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
38144 alignment of the function.
38146 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38148 PR tree-optimization/61441
38149 * simplify-rtx.c (simplify_const_unary_operation): For
38150 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
38151 the sNaN unmodified.
38153 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
38155 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
38156 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
38157 instead of SYSTEM_HEADER_DIR.
38159 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
38160 Martin Liška <mliska@suse.cz>
38162 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
38163 Fix typos and grammar, use active voice, and clarify.
38165 2017-02-20 Marek Polacek <polacek@redhat.com>
38167 PR middle-end/79537
38168 * gimplify.c (gimplify_expr): Handle unused *&&L;.
38171 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
38173 2017-02-20 Jakub Jelinek <jakub@redhat.com>
38176 * config/i386/i386.c (ix86_expand_builtin): Handle
38177 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
38178 ix86_builtins_isa[fcode].isa as a requirement of those
38179 flags and any other flag in the bitmask.
38180 (ix86_init_mmx_sse_builtins): Use 0 instead of
38181 ~OPTION_MASK_ISA_64BIT as mask.
38182 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
38183 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
38184 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
38185 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
38187 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38190 * lra-constraints.c (split_reg): Check requested split mode
38191 is supported by the register.
38193 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38195 * lra-constraints.c (simplify_operand_subreg): Remove early
38198 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38201 * lra-constraints.c (curr_insn_transform): Tighten condition
38202 for converting SUBREG reloads from OP_OUT to OP_INOUT.
38204 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38207 * lra-constraints.c (curr_insn_transform): Handle
38208 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38210 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
38213 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
38215 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
38217 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
38220 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38223 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38225 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38228 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38230 * final.c (last_columnnum, override_columnnum): New variables.
38231 (final_start_function): Set last_columnnum, pass it to begin_prologue
38232 hook and pass 0 to dwarf2out_begin_prologue.
38233 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38234 to source_line debug hook.
38235 (notice_source_line): Compute last_columnnum and for debug_column_info
38236 return true on column changes.
38237 * debug.h (struct gcc_debug_hooks): Add column argument to
38238 source_line and begin_prologue hooks.
38239 (debug_nothing_int_charstar_int_bool): Remove prototype.
38240 (debug_nothing_int_int_charstar,
38241 debug_nothing_int_int_charstar_int_bool): New prototypes.
38242 (dwarf2out_begin_prologue): Add column argument.
38243 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38244 begin_prologue hooks.
38245 (debug_nothing_int_charstar_int_bool): Remove.
38246 (debug_nothing_int_int_charstar,
38247 debug_nothing_int_int_charstar_int_bool): New functions.
38248 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38249 through to dwarf2out_source_line.
38250 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38251 (dwarf2out_source_line): Add column argument, emit it if requested.
38252 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38254 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38255 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38256 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38257 through to dwarf2out_begin_prologue.
38258 (vmsdbgout_source_line): Add column argument, pass it through to
38259 dwarf2out_source_line.
38260 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38261 dbxout_source_line caller.
38262 (dbxout_source_line): Add column argument.
38264 * common.opt (gno-column-info, gcolumn-info): New options.
38265 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38266 (check_die): Also test for multiple DW_AT_decl_column attributes.
38267 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38268 DW_AT_decl_column if requested.
38269 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38271 (gen_variable_die): Likewise.
38272 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38273 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38276 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38277 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38278 (ix86_handle_option): Handle OPT_m3dnowa.
38279 * doc/invoke.texi (-m3dnowa): Document.
38280 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38281 -m3dnowa instead of -m3dnow -march=athlon.
38284 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38285 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38287 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38290 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38291 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38292 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38293 generator for vsx_xxpermdi_<mode>_be.
38294 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38295 force big-endian semantics.
38296 (vsx_xxpermdi_<mode>_be): New define_expand with same
38297 implementation as previous version of vsx_xxpermdi_<mode>.
38299 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38301 PR tree-optimization/79327
38302 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38303 variable, its initialization and use.
38305 2017-02-17 Julia Koval <julia.koval@intel.com>
38307 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38308 (OPTION_MASK_ISA_PKU_UNSET): New.
38309 (ix86_handle_option): Handle -mrdpid.
38310 * config/i386/cpuid.h (bit_RDPID): New.
38311 * config/i386/driver-i386.c (host_detect_local_cpu):
38312 Detect RDPID feature.
38313 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38314 * config/i386/i386-c.c (ix86_target_macros_internal):
38316 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38317 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38318 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38319 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38320 * config/i386/i386.md (define_insn "rdpid"): New.
38321 * config/i386/i386.opt Add -mrdpid.
38322 * config/i386/immintrin.h (_rdpid_u32): New.
38324 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38326 PR rtl-optimization/79541
38327 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38328 instead of transforming it into USE.
38330 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38332 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38333 If HONOR_SNANS (SFmode) force the input to a register.
38334 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38335 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38336 an frsp or similar insn.
38338 2017-02-17 Martin Liska <mliska@suse.cz>
38340 PR rtl-optimization/79577
38341 * params.def (selsched-max-sched-times): Increase minimum to 1.
38343 2017-02-17 Martin Liska <mliska@suse.cz>
38345 PR rtl-optimization/79574
38346 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38348 2017-02-17 Martin Liska <mliska@suse.cz>
38350 PR tree-optimization/79529
38351 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38352 ssa_defined_default_def_p to handle cases which are implicitly
38354 * tree-ssa.c (ssa_defined_default_def_p): New function.
38355 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38356 which are implicitly defined.
38357 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38359 2017-02-17 Richard Biener <rguenther@suse.de>
38361 PR middle-end/79576
38362 * params.def (max-ssa-name-query-depth): Limit to 10.
38364 2017-02-17 Richard Biener <rguenther@suse.de>
38366 PR tree-optimization/79552
38367 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38370 2017-02-17 Richard Biener <rguenther@suse.de>
38373 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38375 2017-02-17 Marek Polacek <polacek@redhat.com>
38377 PR middle-end/79536
38378 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38379 (fold_negate_expr): New wrapper.
38381 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38383 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38384 Correct terminology and de-emphasize pre-standard behavior.
38386 2017-02-16 Alan Modra <amodra@gmail.com>
38388 PR rtl-optimization/79286
38389 * ira.c (def_dominates_uses): New function.
38390 (update_equiv_regs): Don't create an equivalence for insns that
38391 may trap where the register def does not dominate the use.
38393 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38395 PR rtl-optimization/78127
38396 * lra.c (lra): Call lra_eliminate before finish the loop after
38399 2017-02-16 Richard Biener <rguenther@suse.de>
38401 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38403 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38404 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38405 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38406 (isl_val_int_from_wi): New function.
38407 (extract_affine_gmp): Rename to ...
38408 (extract_affine_wi): ... this, take a widest_int.
38409 (extract_affine_int): Just wrap extract_affine_wi.
38410 (add_param_constraints): Use isl_val_int_from_wi.
38411 (add_loop_constraints): Likewise, and extract_affine_wi.
38413 2017-02-15 Jeff Law <law@redhat.com>
38415 PR middle-end/79521
38416 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38417 ira_init_register_move_cost_if_necessary.
38419 2017-02-15 Martin Sebor <msebor@redhat.com>
38421 PR middle-end/32003
38422 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38423 removed in a prior commit.
38425 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38427 PR tree-optimization/79347
38428 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38429 counters during peeling.
38431 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38433 * Makefile.in (site.exp): Remove "set ISLVER".
38435 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38438 * real.c (real_from_integer): Call real_convert even for decimal.
38440 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38443 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38445 2017-02-14 Andrew Pinski <apinski@cavium.com>
38447 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38448 cores and change the partno/implementer to be correct.
38449 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38450 the 'B" as the implementer.
38451 * config/aarch64/aarch64-tune.md: Regenerate.
38453 2017-02-14 Carl Love <cel@us.ibm.com>
38455 * config/rs6000/rs6000.c: Add case statement entry to make the
38456 xvcvuxdsp built-in argument unsigned.
38457 * config/rs6000/vsx.md: Fix the source and return operand types so they
38458 match the instruction definitions from the ISA document. Fix typo
38459 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38462 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38465 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38466 member early_clobber_alts.
38467 * lra-lives.c (reg_early_clobber_p): New.
38468 (process_bb_lives): Use it.
38469 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38470 (debug_operand_data): Initialize early_clobber_alts.
38471 (setup_operand_alternative): Set up early_clobber_alts.
38472 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38473 alternatives to new_insn_reg.
38474 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38476 (lra_update_insn_regno_info): Pass the new arg.
38478 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38480 PR middle-end/79505
38481 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38482 (new_oacc_loop_raw): Don't clear already cleared fields.
38485 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38486 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38487 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38489 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38492 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38494 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38497 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38498 the extra instruction to the right place to store 128-bit constant
38501 2017-02-14 Martin Sebor <msebor@redhat.com>
38503 PR middle-end/79448
38504 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38505 warning for strings of unknown length.
38507 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38509 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38511 2017-02-14 Jeff Law <law@redhat.com>
38514 * ira-costs.c (scan_one_insn): Initialize register move costs
38515 for pseudos seen in USE/CLOBBER insns.
38517 PR tree-optimization/79095
38518 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38519 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38520 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38521 if the operands are known to be not equal, then the resulting range
38523 (intersect_ranges): If the new range is ~[0,0] and the old range is
38524 wide, then prefer ~[0,0].
38525 * tree-vrp.c (overflow_comparison_p_1): New function.
38526 (overflow_comparison_p): New function.
38527 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38528 if NAME is used in an overflow test.
38529 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38530 overflow check that can be expressed as an equality test, then adjust
38531 ops to be that equality test.
38533 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38535 * config/s390/s390-builtin-types.def: Remove flags argument.
38536 * config/s390/s390.c (s390_init_builtins): Likewise.
38538 2017-02-14 Martin Liska <mliska@suse.cz>
38540 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38541 vector. Fix trailing white spaces.
38543 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38545 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38548 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38550 PR rtl-optimization/68664
38551 * config/arm/arm.c (arm_sched_can_speculate_insn):
38552 New function. Declare prototype.
38553 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38555 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38557 PR rtl-optimization/68664
38558 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38560 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38562 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38564 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38565 max skip bytes for function, loop and jump.
38567 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38569 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38570 ABS_EXPR for gimple dump.
38572 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38575 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38577 PR tree-optimization/79408
38578 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38579 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38580 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38581 also if rhs1 is INTEGER_CST.
38583 2017-02-14 Richard Biener <rguenther@suse.de>
38585 PR middle-end/79432
38586 * tree-into-ssa.c (insert_phi_nodes): When the function can
38587 have abnormal edges rewrite SSA names with broken use-def
38588 dominance out of SSA and register them for PHI insertion.
38590 2017-02-13 Martin Sebor <msebor@redhat.com>
38592 PR middle-end/79496
38593 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38594 clearing info.nowrite flag when snprintf size argument is a range.
38596 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38598 * cprop.c (cprop_jump): Add missing space in string literal.
38599 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38600 (get_constraint_for_component_ref): Likewise.
38601 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38602 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38603 * lra-constraints.c (process_alt_operands): Likewise.
38604 * ipa-inline.c (inline_small_functions): Likewise.
38605 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38606 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38607 * trans-mem.c (diagnose_tm_1_op): Likewise.
38608 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38609 (grid_parallel_clauses_gridifiable): Likewise.
38611 * config/nvptx/mkoffload.c (process): Add space in between
38614 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38615 "MOD4_SSE_REGS" and "ALL_REGS".
38617 * spellcheck.c (test_data): Add , in between "foo" and "food".
38619 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38622 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38623 boundary crossing check and subsequent code generation agree.
38625 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38627 * config/aarch64/aarch64.c (has_memory_op): Delete.
38628 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38631 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38633 PR rtl-optimization/79388
38634 PR rtl-optimization/79450
38635 * combine.c (distribute_notes): When removing TEM_INSN for which
38636 corresponding dest has last value recorded, invalidate that last
38639 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38641 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38642 of explicit '@'. Add missing assembly comment marker on branch costs
38645 2017-02-13 Nathan Sidwell <nathan@acm.org>
38647 * gengtype-lex.l (<in_struct>): Add '/'.
38649 2017-02-13 Martin Liska <mliska@suse.cz>
38652 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38654 2017-02-13 Richard Biener <rguenther@suse.de>
38656 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38658 * configure: Re-generate.
38659 * config.in: Likewise.
38660 * graphite-dependences.c: Simplify as if
38661 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38662 * graphite-isl-ast-to-gimple.c: Likewise.
38663 * graphite-optimize-isl.c: Likewise.
38664 * graphite-poly.c: Likewise.
38665 * graphite-sese-to-poly.c: Likewise.
38666 * graphite.h: Likewise.
38667 * toplev.c: Include isl/version.h and use isl_version () for
38668 printing the ISL version.
38669 * doc/install.texi: Update ISL requirement.
38671 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38673 * doc/standards.texi (Standards): Update reference to
38676 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38678 * doc/extend.texi (Named Address Spaces): sourceware.org now
38680 * doc/install.texi (Binaries): Ditto.
38683 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38685 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38686 standard terminology "stringize"/"stringizing" throughout.
38687 * doc/cppinternals.texi: Likewise.
38689 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38691 * doc/extend.texi: Fix some spelling mistakes and typos.
38692 * doc/invoke.texi: Likewise.
38694 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38697 * params.def (inline-min-speedup) Change from 10 to 8.
38699 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38701 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38704 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38707 * ipa-inline-analysis.c (get_minimal_bb): New function.
38708 (record_modified): Use it.
38709 (remap_edge_change_prob): Handle also ancestor functions.
38711 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38713 * doc/contrib.texi (Contributors): Remove broken link into
38714 the Mauve CVS repository.
38716 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38718 PR middle-end/79454
38719 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38720 result computation whenever lhs doesn't have vector mode, not
38721 just when it has BLKmode.
38723 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38725 * doc/makefile.texi (profiledbootstrap): Refer to the
38726 installation instructions only in textual form.
38728 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38731 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38733 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38735 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38736 (Specific): Update mingw-w64 reference.
38738 (Specific): Remove broken link to Renesas RX processor.
38740 2017-02-10 Richard Biener <rguenther@suse.de>
38742 * toplev.c (process_options): Do not mention obsolete graphite
38743 options when printing sorry message about missing graphite support.
38744 Mention -floop-nest-optimize.
38746 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38748 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38749 (vtst_p16): Likewise.
38750 (vtstq_p8): Likewise.
38751 (vtstq_p16): Likewise.
38753 (vtstq_p64): Likewise.
38754 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38755 (vset_lane_p64): New.
38756 (vsetq_lane_p64): New.
38758 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38760 PR tree-optimization/79411
38761 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38762 stmt operands are SSA_NAMEs used in abnormal phis.
38763 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38766 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38769 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38772 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38774 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38776 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38778 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38782 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38783 not arbitrary TREE_CONSTANT.
38786 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38787 "omp declare target link" attribute unless is_global_var.
38788 * omp-offload.c (find_link_var_op): Likewise.
38790 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38791 Chung-Lin Tang <cltang@codesourcery.com>
38793 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38795 (gimplify_adjust_omp_clauses): Don't delete TILE.
38796 (gimplify_omp_for): Deal with TILE.
38797 * internal-fn.c (expand_GOACC_TILE): New function.
38798 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38800 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38801 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38803 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38804 avoid DIV for outermost collapse var.
38805 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38806 Remove out of date comments, fix whitespace.
38807 * omp-general.c (omp_extract_for_data): Deal with tiling.
38808 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38809 adjust OLF_DIM_BASE value.
38810 (struct omp_for_data): Add tiling field.
38811 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38812 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38813 for auto loops. Remove default auto determining, moved to
38814 oacc_loop_fixed_partitions.
38815 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38816 stmts, add e_mask field.
38817 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38818 (oacc_thread_numbers): Use oacc_dim_call.
38819 (oacc_xform_tile): New.
38820 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38821 (finish_oacc_loop): Adjust for ifns vector.
38822 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38823 add case for GOACC_TILE fns.
38824 (oacc_loop_xform_loop): Delete.
38825 (oacc_loop_process): Iterate over call list directly, and add
38826 handling for GOACC_TILE fns.
38827 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38829 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38830 vector partitioning to outer loops. Assign 2 partitions to loops
38831 when available. Add TILE handling.
38832 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38833 (execite_oacc_device_lower): Process GOACC_TILE fns,
38834 ignore unknown specs.
38835 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38836 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38837 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38839 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38841 * configure.ac (ACX_BUGURL): Update.
38842 * configure: Regenerate.
38844 2017-02-09 Richard Biener <rguenther@suse.de>
38846 PR tree-optimization/69823
38847 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38848 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38850 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38852 * config/arc/arc-c.def: Add __NPS400__ definition.
38853 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38854 (TARGET_NPS400): Define.
38856 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38858 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38860 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38861 pointer, arch_info.
38862 (arc_cpu_types): Fill the arch_info field with a pointer into the
38863 arc_arch_types table.
38864 (arc_selected_cpu): Declare.
38865 * config/arc/arc.c (arc_selected_cpu): Make global.
38866 (arc_selected_arch): Delete.
38867 (arc_base_cpu): Delete.
38868 (arc_override_options): Remove references to deleted variables,
38869 update access to arch information.
38870 (ARC_OPT): Update access to arch information.
38871 (ARC_OPTX): Likewise.
38872 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38873 (TARGET_ARC600): Update access to arch information.
38874 (TARGET_ARC601): Likewise.
38875 (TARGET_ARC700): Likewise.
38876 (TARGET_EM): Likewise.
38877 (TARGET_HS): Likewise.
38878 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38881 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38884 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38885 condition/operands for integer GE/LE/GEU/LEU operations.
38887 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38889 PR translation/79397
38890 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38893 2017-02-08 Martin Jambor <mjambor@suse.cz>
38896 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38897 whether allocation happened.
38898 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38899 nothing was allocated.
38901 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38903 PR tree-optimization/79408
38904 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38905 constant, but SSA_NAME with a known integer range, use the minimum
38906 of that range instead of op1 to determine if modulo can be replaced
38907 with its first operand.
38909 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38911 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38913 2017-02-08 Richard Biener <rguenther@suse.de>
38915 PR tree-optimization/71824
38916 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38917 Check all loops contained in the merged region.
38919 2017-02-07 Andrew Pinski <apinski@cavium.com>
38921 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38923 2017-02-07 Andrew Pinski <apinski@cavium.com>
38925 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38926 (thunderxt88): Likewise.
38927 (thunderxt81): Disable LSE and change v8.1 to v8.
38928 (thunderxt83): Likewise.
38930 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38931 Richard Biener <rguenther@suse.de>
38933 PR middle-end/79399
38934 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38935 type from int to size_t.
38936 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38938 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38940 PR rtl-optimization/79386
38941 * cprop.c (bypass_conditional_jumps): Initialize
38942 bypass_last_basic_block already before splitting bbs after
38943 unconditional traps...
38944 (bypass_conditional_jumps): ... rather than here.
38947 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38948 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38949 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38950 fix -masm=intel patterns.
38952 2017-02-07 Richard Biener <rguenther@suse.de>
38954 PR tree-optimization/79256
38955 PR middle-end/79278
38956 * builtins.c (get_object_alignment_2): Use min_align_of_type
38957 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38958 and ADJUST_FIELD_ALIGN.
38960 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38962 * doc/tm.texi: Regenerate.
38963 * stor-layout.c (layout_decl): Adjust.
38964 (update_alignment_for_field): Likewise.
38965 (place_field): Likewise.
38966 (min_align_of_type): Likewise.
38967 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38968 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38969 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38970 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38971 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38972 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38973 * config/i386/i386.c (x86_field_alignment): Likewise.
38974 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38975 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38976 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38977 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38978 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38979 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38983 2017-01-30 Richard Biener <rguenther@suse.de>
38985 PR tree-optimization/79256
38986 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38987 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38990 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38992 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38993 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38994 builtins to SImode and emit a zero-extend, if necessary.
38996 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38998 * docs/invoke.texi (RISC-V Options): Alphabetize.
39000 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
39002 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
39005 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
39007 * config/riscv/riscv.c: New file.
39008 * common/config/riscv/riscv-common.c: Likewise.
39009 * config.gcc: Likewise.
39010 * config/riscv/constraints.md: Likewise.
39011 * config/riscv/elf.h: Likewise.
39012 * config/riscv/generic.md: Likewise.
39013 * config/riscv/linux.h: Likewise.
39014 * config/riscv/multilib-generator: Likewise.
39015 * config/riscv/peephole.md: Likewise.
39016 * config/riscv/pic.md: Likewise.
39017 * config/riscv/predicates.md: Likewise.
39018 * config/riscv/riscv-builtins.c: Likewise.
39019 * config/riscv/riscv-c.c: Likewise.
39020 * config/riscv/riscv-ftypes.def: Likewise.
39021 * config/riscv/riscv-modes.def: Likewise.
39022 * config/riscv/riscv-opts.h: Likewise.
39023 * config/riscv/riscv-protos.h: Likewise.
39024 * config/riscv/riscv.h: Likewise.
39025 * config/riscv/riscv.md: Likewise.
39026 * config/riscv/riscv.opt: Likewise.
39027 * config/riscv/sync.md: Likewise.
39028 * config/riscv/t-elf-multilib: Likewise.
39029 * config/riscv/t-linux: Likewise.
39030 * config/riscv/t-linux-multilib: Likewise.
39031 * config/riscv/t-riscv: Likewise.
39032 * configure.ac: Likewise.
39033 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
39034 Waterman as RISC-V maintainers.
39035 * doc/install.texi: Add RISC-V entries.
39036 * doc/invoke.texi: Add RISC-V options section.
39037 * doc/md.texi: Add RISC-V constraints section.
39038 * configure: Regenerated.
39040 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
39043 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
39044 false values to be constant vectors with all 0 or all 1 bits set.
39045 (vcondu<mode><mode>): Likewise.
39046 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
39048 (fpmask_comparison_operator): Update comment.
39049 (vecint_comparison_operator): New predicate.
39050 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
39051 vector conditionals when the true and false values are constant
39052 vectors with all 0 bits or all 1 bits set.
39054 2017-02-06 Martin Sebor <msebor@redhat.com>
39056 PR tree-optimization/79376
39057 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
39059 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
39061 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
39062 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
39063 to simplify split condition.
39065 2017-02-06 Jakub Jelinek <jakub@redhat.com>
39067 * omp-expand.c (oxpand_omp_atomic_fetch_op,
39068 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
39071 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
39073 PR rtl-optimization/68664
39074 * target.def (can_speculate_insn): New hook.
39075 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
39076 * doc/tm.texi: Regenerate.
39077 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
39078 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
39079 (rs6000_sched_can_speculate_insn): New function.
39081 2017-02-06 Jakub Jelinek <jakub@redhat.com>
39083 PR tree-optimization/79284
39084 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
39085 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
39086 vectorizable_mask_load_store, vectorizable_operation,
39087 vect_is_simple_cond, get_same_sized_vectype): Use it instead
39088 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
39089 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
39090 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
39091 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
39092 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
39093 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
39094 is_gimple_assign (stmt). Replace another such test with
39095 is_gimple_assign (stmt).
39097 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39100 * config/avr/avr.c (rtl-iter.h): Include it.
39101 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
39102 (avr_legitimate_combined_insn): ...and implementation.
39104 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
39106 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
39107 * config/s390/s390.c (s390_const_operand_ok)
39108 (s390_canonicalize_comparison, s390_extract_part)
39109 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
39110 (s390_contiguous_bitmask_p, s390_rtx_costs)
39111 (legitimize_pic_address): Likewise.
39112 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
39113 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
39114 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
39115 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
39116 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
39118 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39120 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
39121 REGNO($0) == REGNO($1).
39123 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39125 * config/s390/linux.h(SIZE_TYPE): Add comment.
39127 2017-02-06 Julian Brown <julian@codesourcery.com>
39128 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39129 Virendra Pathak <virendra.pathak@broadcom.com>
39131 * config/aarch64/aarch64-cores.def: Change the scheduler
39133 * config/aarch64/aarch64.md: Include thunderx2t99.md.
39134 * config/aarch64/thunderx2t99.md: New file.
39136 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39138 * doc/standards.texi (Go Language): Update link to language
39141 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39143 * tree-eh.c (lower_resx): Sanitize profile.
39144 (cleanup_empty_eh_move_lp): Likewise.
39146 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39149 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
39151 * cfgloopmanip.h (loop_version): Update prototype.
39152 * modulo-sched.c (sms_schedule): Update call of loop_version.
39153 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
39154 * tree-parloops.c (gen_parallel_loop): Likewise.
39155 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
39156 * tree-ssa-loop-split.c (split_loop): Likewise.
39157 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
39158 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
39160 2017-02-05 Martin Liska <mliska@suse.cz>
39163 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
39165 (print_operand_address): Initialize a struct to zero.
39167 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39169 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
39170 garbage collector only in textual form.
39172 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39174 * doc/extend.texi (x86 specific memory model extensions for
39175 transactional memory): Simplify a phrase.
39177 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
39180 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
39181 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
39182 (atomic_storedi_1): Likewise.
39184 2017-02-04 Jakub Jelinek <jakub@redhat.com>
39186 PR tree-optimization/79338
39187 * tree-parloops.c (gather_scalar_reductions): Don't call
39188 vect_analyze_loop_form for loop->inner before destroying loop's
39191 2017-02-03 Martin Sebor <msebor@redhat.com>
39193 PR tree-optimization/79327
39194 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
39195 when precision has resulted in leading zeros.
39196 (format_integer): Adjust the likely counter to assume an unknown
39197 argument that may be zero is non-zero.
39199 2017-02-03 Jason Merrill <jason@redhat.com>
39202 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
39203 avoid copying non-taken branch.
39205 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39207 PR tree-optimization/79340
39208 * tree-vect-loop.c (vectorizable_reduction): Release
39209 vec_defs elements after safe_splicing them into other vectors.
39212 PR tree-optimization/79327
39213 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
39214 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
39216 (format_integer): Use wide_int_to_tree instead of build_int_cst
39217 + to_?hwi. If argmin is NULL, just set argmin and argmax to
39218 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
39219 of shortest and longest sequence.
39221 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39223 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39224 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39226 2017-02-03 Walter Lee <walt@tilera.com>
39229 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39230 after initial stackframe link reg save.
39231 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39233 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39236 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39237 wu for stxssp alternative.
39239 2017-02-03 Martin Sebor <msebor@redhat.com>
39241 PR tree-optimization/79352
39242 * gimple-fold.c (get_range_strlen): Add argument.
39243 (get_range_strlen): Change return type to bool.
39244 (get_maxval_strlen): Pass in a dummy argument.
39245 * gimple-fold.h (get_range_strlen): Change return type to bool.
39246 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39247 * tree.h (array_at_struct_end_p): Add argument.
39248 * tree.c (array_at_struct_end_p): Handle it.
39250 2017-02-03 Martin Liska <mliska@suse.cz>
39253 * multiple_target.c (create_dispatcher_calls): Redirect edge
39254 from a caller of a dispatcher.
39255 (expand_target_clones): Make the clones local.
39256 (ipa_target_clone): Do both target clones and resolvers.
39257 (ipa_dispatcher_calls): Remove the pass.
39258 (pass_dispatcher_calls::gate): Likewise.
39259 (make_pass_dispatcher_calls): Likewise.
39260 * passes.def (pass_target_clone): Put as very first IPA early
39263 2017-02-03 Martin Liska <mliska@suse.cz>
39265 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39266 in case of a function with ifunc attribute.
39268 2017-02-03 Martin Liska <mliska@suse.cz>
39270 * cgraph.c (cgraph_node::dump): Dump function version info.
39271 * symtab.c (symtab_node::dump_base): Add missing new line.
39273 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39275 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39276 (ifcombine_ifandif): Use it.
39278 2017-02-03 Martin Liska <mliska@suse.cz>
39280 * doc/invoke.texi: Document default value for
39281 use-after-scope-direct-emission-threshold.
39283 2017-02-03 Martin Liska <mliska@suse.cz>
39285 PR tree-optimization/79339
39286 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39287 (format_floating): Likewise.
39289 2017-02-03 Martin Liska <mliska@suse.cz>
39292 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39294 (ipa_node_params_t::remove): Likewise.
39295 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39296 initialization from removed ipa_node_params_t::insert.
39297 (ipa_node_params::~ipa_node_params): Move from removed
39298 ipa_node_params_t::release.
39299 * symbol-summary.h (symbol_summary::m_released): New member.
39300 Do not release a summary twice. Do not allow to call finalizer
39301 for types of a summary that live in GGC memory.
39303 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39305 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39308 2017-02-02 Martin Sebor <msebor@redhat.com>
39310 PR middle-end/79275
39311 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39312 (format_string): Tighten up the range of output for non-constant
39313 strings and correct the expected range for wide non-constant strings.
39315 2017-02-02 Martin Sebor <msebor@redhat.com>
39317 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39319 PR middle-end/32003
39320 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39322 (-fdump-tree-@var): Add to index and document how to come up
39323 with pass-specific option and dump file names.
39324 (-fdump-passes): Clarify where to look for output.
39326 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39328 PR middle-end/77445
39329 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39330 statistics of the analyzed path; allow threading for speed when
39331 any of BBs along the path are optimized for speed.
39333 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39335 PR middle-end/78468
39336 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39337 settings of the virtual registers.
39340 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39342 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39343 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39346 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39348 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39349 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39351 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39353 * config/s390/s390.md: Add missing comments with the expanded
39355 * config/s390/vector.md: Likewise.
39356 * config/s390/vx-builtins.md: Likewise.
39358 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39361 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39362 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39363 conditions on a single line.
39365 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39367 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39368 __S390_VX__ to __VX__.
39370 2017-02-01 Andrew Pinski <apinski@cavium.com>
39372 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39373 stmt_info to record_stmt_cost.
39374 (vect_get_known_peeling_cost): Pass stmt_info if known to
39376 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39377 cpu_vector_cost field into
39378 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39379 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39380 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39381 splitting of scalar_stmt_cost and vec_stmt_cost.
39382 (thunderx_vector_cost): Likewise.
39383 (cortexa57_vector_cost): LIkewise.
39384 (exynosm1_vector_cost): Likewise.
39385 (xgene1_vector_cost): Likewise.
39386 (thunderx2t99_vector_cost): Improve after the splitting of the two
39388 (aarch64_builtin_vectorization_cost): Update for the splitting of
39389 scalar_stmt_cost and vec_stmt_cost.
39391 2017-02-01 Torvald Riegel <triegel@redhat.com>
39392 Richard Henderson <rth@redhat.com>
39394 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39395 conditional on existance of a fast atomic load.
39396 * optabs-query.c (can_atomic_load_p): New function.
39397 * optabs-query.h (can_atomic_load_p): Declare it.
39398 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39399 no fast atomic load is available for the particular size of access.
39400 (expand_atomic_compare_and_swap): Likewise.
39401 (expand_atomic_load): Likewise.
39402 (expand_atomic_store): Likewise.
39403 (expand_atomic_fetch_op): Likewise.
39404 * testsuite/lib/target-supports.exp
39405 (check_effective_target_sync_int_128): Remove x86 because it provides
39406 no fast atomic load.
39407 (check_effective_target_sync_int_128_runtime): Likewise.
39409 2017-02-01 Richard Biener <rguenther@suse.de>
39411 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39412 (graphite_transform_loops): Provide opt-info for optimized nests.
39413 * tree-parloop.c (parallelize_loops): Provide opt-info for
39414 parallelized loops.
39416 2017-02-01 Richard Biener <rguenther@suse.de>
39418 PR middle-end/79315
39419 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39420 was not set before.
39422 2017-02-01 Richard Biener <rguenther@suse.de>
39424 PR tree-optimization/71824
39425 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39426 Verify the loops are valid in the merged SESE region.
39427 (scop_detection::can_represent_loop_1): Check analyzing the
39428 evolution of the number of iterations in the region succeeds.
39430 2017-01-31 Ian Lance Taylor <iant@golang.org>
39432 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39433 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39435 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39437 PR preprocessor/79210
39438 * input.c (get_substring_ranges_for_loc): Replace line_width
39439 assertion with error-handling.
39441 2017-01-31 Richard Biener <rguenther@suse.de>
39443 PR tree-optimization/77318
39444 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39445 (create_pw_aff_from_tree): Take loop parameter.
39446 (add_condition_to_pbb): Pass loop of the condition to
39447 create_pw_aff_from_tree.
39449 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39451 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39452 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39454 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39458 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39460 (convert_int_to_float128): Likewise.
39461 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39462 (convert_int_to_float128): Likewise.
39463 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39464 (UNSPEC_IEEE128_CONVERT): Likewise.
39465 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39466 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39467 Use local variables for IBM extended format.
39468 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39469 (fix_trunc<mode>si2_fprs): Likewise.
39470 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39471 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39472 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39473 to know that we can now have integers of all sizes in vector
39475 (fix<uns>_<mode>di2_hw): Likewise.
39476 (float<uns>_<mode>si2_hw): Likewise.
39477 (fix_<mode>si2_hw): Likewise.
39478 (fixuns_<mode>si2_hw): Likewise.
39479 (float<uns>_<mode>di2_hw): Likewise.
39480 (float_<mode>di2_hw): Likewise.
39481 (float_<mode>si2_hw): Likewise.
39482 (floatuns_<mode>di2_hw): Likewise.
39483 (floatuns_<mode>si2_hw): Likewise.
39484 (xscvqp<su>wz_<mode>): Delete, no longer used.
39485 (xscvqp<su>dz_<mode>): Likewise.
39486 (xscv<su>dqp_<mode>): Likewise.
39487 (ieee128_mfvsrd_64bit): Likewise.
39488 (ieee128_mfvsrd_32bit): Likewise.
39489 (ieee128_mfvsrwz): Likewise.
39490 (ieee128_mtvsrw): Likewise.
39491 (ieee128_mtvsrd_64bit): Likewise.
39492 (ieee128_mtvsrd_32bit): Likewise.
39494 2017-01-31 Martin Liska <mliska@suse.cz>
39497 * ipa-prop.c (ipa_free_all_node_params): Call release method
39498 instead of ~sumbol_summary to not to trigger double times
39501 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39503 PR tree-optimization/71691
39504 * bitmap.h (class auto_bitmap): New.
39505 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39506 is_maybe_undefined instead of ssa_undefined_value_p.
39508 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39510 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39511 __S390_ARCH_LEVEL__ to __ARCH__.
39513 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39515 PR tree-optimization/79267
39516 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39517 if should_remove_lhs_p is true.
39519 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39522 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39523 (add_alignment_attribute): New.
39524 (base_type_die): Add alignment attribute.
39525 (subrange_type_die): Likewise.
39526 (modified_type_die): Likewise.
39527 (gen_array_type_die): Likewise.
39528 (gen_descr_array_type_die: Likewise.
39529 (gen_enumeration_type_die): Likewise.
39530 (gen_subprogram_die): Likewise.
39531 (gen_variable_die): Likewise.
39532 (gen_field_die): Likewise.
39533 (gen_ptr_to_mbr_type_die): Likewise.
39534 (gen_struct_or_union_type_die): Likewise.
39535 (gen_subroutine_type_die): Likewise.
39536 (gen_typedef_die): Likewise.
39537 (base_type_cmp): Compare alignment attribute.
39539 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39542 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39543 (setb_unsigned) New pattern for setb with CCUNS.
39544 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39545 subfc./subfe sequence to avoid overflow problems. Generate a
39546 shorter sequence with cmpld/setb for power9.
39547 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39548 for generating subfc. instruction.
39549 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39550 now uses this instruction.
39552 2017-01-30 Ian Lance Taylor <iant@google.com>
39555 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39556 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39558 2017-01-30 Martin Sebor <msebor@redhat.com>
39560 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39561 Move constant to the right of a relational operator.
39562 (get_mpfr_format_length, format_character, format_string): Ditto.
39563 (should_warn_p, maybe_warn): Same.
39565 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39567 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39570 * asan.c (get_translation_unit_decl): Remove function.
39571 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39573 2017-01-30 Martin Liska <mliska@suse.cz>
39575 PR gcov-profile/79259
39576 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39577 -fprofile-generate.
39579 2017-01-30 Martin Liska <mliska@suse.cz>
39582 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39583 Initialize variables with NULL value.
39585 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39588 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39590 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39592 2017-01-30 Richard Biener <rguenther@suse.de>
39594 PR tree-optimization/79276
39595 * tree-vrp.c (process_assert_insertions): Properly adjust common
39596 when removing a duplicate.
39598 2017-01-30 Richard Biener <rguenther@suse.de>
39600 PR tree-optimization/79256
39601 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39602 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39604 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39606 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39609 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39610 ("*r<noxa>sbg_<mode>_sll_bitmask")
39611 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39612 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39613 Use contiguous_bitmask_nowrap_operand.
39615 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39618 * config/rs6000/altivec.h (vec_xl): Revise #define.
39619 (vec_xst): Likewise.
39621 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39623 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39625 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39627 PR rtl-optimization/79194
39628 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39629 traps before call to bypass_conditional_jumps.
39631 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39633 PR tree-optimization/71374
39634 * lra-constraints.c (check_conflict_input_operands): New.
39635 (match_reload): Use it.
39637 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39640 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39641 account to calculate conflict_set.
39643 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39645 PR rtl-optimization/78559
39646 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39647 other_insn in combine.
39649 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39651 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39652 uint16_type_node for BT_UINT16.
39654 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39656 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39657 "RTL Tests" to menu.
39658 (GIMPLE Tests): New node.
39659 (RTL Tests): New node.
39661 2017-01-27 Richard Biener <rguenther@suse.de>
39663 PR tree-optimization/79245
39664 * tree-loop-distribution.c (distribute_loop): Apply cost
39665 modeling also to detected patterns.
39667 2017-01-27 Richard Biener <rguenther@suse.de>
39669 PR tree-optimization/71433
39670 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39671 (compare_assert_loc): New function.
39672 (process_assert_insertions): Sort and optimize assert locations
39673 to remove duplicates and push down identical assertions on
39674 edges to their destination block.
39676 2017-01-27 Richard Biener <rguenther@suse.de>
39678 PR tree-optimization/79244
39679 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39680 out SSA names even if abnormal.
39682 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39684 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39685 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39686 instead of MPFR_RNDN.
39688 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39691 * arm.c (arm_option_override): Don't call build_target_option_node
39692 until after doing all option overrides.
39693 (arm_valid_target_attribute_tree): Likewise.
39695 2017-01-27 Martin Liska <mliska@suse.cz>
39697 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39698 for {cd}tors and C++ {cd}tors.
39700 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39702 * config/s390/s390.md ("*setmem_long_and")
39703 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39705 2017-01-26 Martin Sebor <msebor@redhat.com>
39707 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39710 2017-01-26 Martin Sebor <msebor@redhat.com>
39712 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39713 HAVE_DFmode before using XFmode or DFmode.
39714 (parse_directive): Avoid using the z length modifier to avoid
39715 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39717 PR middle-end/78703
39718 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39719 to accept adjustment as an array.
39720 (get_int_range): New function.
39721 (struct directive): Make width and prec arrays.
39722 (directive::set_width, directive::set_precision): Call get_int_range.
39723 (format_integer, format_floating): Handle width and precision ranges.
39724 (format_string, parse_directive): Same.
39726 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39729 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39730 comdat_type_p set, just clone them, but keep the children in the
39734 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39735 which have direct callers with -fvar-tracking-assignments enabled
39737 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39738 inside of type units.
39740 2017-01-26 Martin Sebor <msebor@redhat.com>
39742 PR middle-end/78703
39743 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39745 (struct format_result): Replace number_chars, number_chars_min,
39746 and number_chars_max with a single member of struct result_range.
39748 (format_result::operator+=): Adjust.
39749 (struct fmtresult): Remove bounded. Handle likely and unlikely
39751 (fmtresult::adjust_for_width_or_precision): New function.
39752 (fmtresult:type_max_digits): New function.
39753 (bytes_remaining): Handle likely and unlikely counters.
39754 (min_bytes_remaining): Remove.
39755 (format_percent): Simplify.
39756 (format_integer, format_floating): Set likely and unlikely counters.
39757 (get_string_length, format_character, format_string): Same.
39758 (format_plain, should_warn_p): New function.
39759 (maybe_warn): Call should_warn_p. Update diagnostic messages
39760 and handle those for all directives, including plain strings.
39761 (format_directive): Handle likely and unlikely counters.
39762 Remove unnecessary quoting from diagnostics. Add an informational
39764 (add_bytes): Remove.
39765 (pass_sprintf_length::compute_format_length): Simplify.
39766 (try_substitute_return_value): Handle likely and unlikely counters.
39768 2017-01-26 Carl Love <cel@us.ibm.com>
39770 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39771 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39773 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39776 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39777 endianess for subregs into account.
39778 * lra-constraints.c (lra_constraints): Do risky transformations
39779 always on the first iteration.
39780 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39781 last_call_used_reg_set.
39782 (process_bb_lives): Define and use last_call_used_reg_set.
39783 * lra.c (lra): Always continue after lra_constraints on the first
39786 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39788 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39790 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39792 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39794 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39795 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39796 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39797 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39798 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39799 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39800 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39801 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39802 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39804 2017-01-26 Marek Polacek <polacek@redhat.com>
39807 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39808 for the third operand.
39810 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39812 PR middle-end/79236
39813 * omp-low.c (struct omp_context): Add simt_stmt field.
39814 (scan_omp_for): Return omp_context *.
39815 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39816 context to the _simt_ SIMD stmt.
39817 (lower_omp_for): For combined SIMD with sibling _simt_
39818 SIMD, make sure to use the same decls in _looptemp_
39819 clauses as in the sibling.
39821 2017-01-26 David Sherwood <david.sherwood@arm.com>
39823 PR middle-end/79212
39824 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39827 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39830 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39831 emit fld b; fld a; if possible.
39833 * brig-builtins.def: Update copyright years.
39834 * config/arm/arm_acle_builtins.def: Update copyright years.
39836 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39839 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39840 constraint instead of o for the stxsd instruction.
39842 2017-01-25 Carl Love <cel@us.ibm.com>
39844 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39845 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39847 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39849 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39851 2017-01-25 Richard Biener <rguenther@suse.de>
39853 PR tree-optimization/69264
39854 * target.def (vector_alignment_reachable): Improve documentation.
39855 * doc/tm.texi: Regenerate.
39856 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39858 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39859 earlier changes with respect to TYPE_USER_ALIGN.
39860 (vector_alignment_reachable_p): Likewise. Improve dumping.
39862 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39865 * config/arm/arm.md (xordi3): Force constant operand into a register
39868 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39870 * doc/invoke.texi (-fstore-merging): Correct default optimization
39871 levels at which it is enabled.
39872 (-O): Move -fstore-merging from list to...
39875 2017-01-25 Richard Biener <rguenther@suse.de>
39878 * omp-expand.c: Include debug.h.
39879 (expand_omp_taskreg): Make sure to generate early debug before
39880 outlining anything from a function.
39881 (expand_omp_target): Likewise.
39882 (grid_expand_target_grid_body): Likewise.
39884 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39887 * asan.c (get_translation_unit_decl): New function.
39888 (asan_add_global): Extract modules file name from globals
39889 TRANSLATION_UNIT_DECL name.
39891 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39894 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39895 for long calls with APCS frame and VFP.
39897 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39899 * cfg.c (original_copy_tables_initialized_p): New function.
39900 * cfg.h (original_copy_tables_initialized_p): New decl.
39901 * cfgrtl.c (relink_block_chain): Guard the call to
39902 free_original_copy_tables with a call to
39903 original_copy_tables_initialized_p.
39904 * cgraph.h (symtab_node::native_rtl_p): New decl.
39905 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39906 (symtab_node::needed_p): Don't assert for early assembly output
39907 for __RTL functions.
39908 (cgraph_node::finalize_function): Set "force_output" for __RTL
39910 (cgraph_node::analyze): Bail out early for __RTL functions.
39911 (analyze_functions): Update assertion to support __RTL functions.
39912 (cgraph_node::expand): Bail out early for __RTL functions.
39913 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39915 * function.h (struct function): Update comment for field
39917 * gimple-expr.c: Include "tree-pass.h".
39918 (gimple_has_body_p): Return false for __RTL functions.
39919 * Makefile.in (OBJS): Add run-rtl-passes.o.
39920 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39922 (gcc::pass_manager::get_clean_slate): New accessor.
39923 * passes.c: Include "insn-addr.h".
39924 (should_skip_pass_p): Add logging. Update logic for running
39925 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39926 property-provider override so it is only done for gimple passes.
39928 (skip_pass): New function.
39929 (execute_one_pass): Call skip_pass when skipping passes.
39930 * read-md.c (md_reader::read_char): Support filtering
39931 the input to a subset of line numbers.
39932 (md_reader::md_reader): Initialize fields
39933 m_first_line and m_last_line.
39934 (md_reader::read_file_fragment): New function.
39935 * read-md.h (md_reader::read_file_fragment): New decl.
39936 (md_reader::m_first_line): New field.
39937 (md_reader::m_last_line): New field.
39938 * read-rtl-function.c (function_reader::create_function): Only
39939 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39940 curr_properties. Set DECL_INITIAL to a dummy block.
39941 (read_rtl_function_body_from_file_range): New function.
39942 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39944 * run-rtl-passes.c: New file.
39945 * run-rtl-passes.h: New file.
39947 2017-01-24 Jeff Law <law@redhat.com>
39949 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39952 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39954 PR tree-optimization/79159
39955 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39956 (record_nonwrapping_iv): Improve boundary using above function if no
39957 value range information.
39959 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39960 Martin Jambor <mjambor@suse.cz>
39962 * brig-builtins.def: New file.
39963 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39964 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39965 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39966 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39967 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39968 * builtin-types.def (BT_INT8): New.
39969 (BT_INT16): Likewise.
39970 (BT_UINT8): Likewise.
39971 (BT_UINT16): Likewise.
39972 (BT_FN_ULONG): Likewise.
39973 (BT_FN_UINT_INT): Likewise.
39974 (BT_FN_UINT_ULONG): Likewise.
39975 (BT_FN_UINT_LONG): Likewise.
39976 (BT_FN_UINT_PTR): Likewise.
39977 (BT_FN_ULONG_PTR): Likewise.
39978 (BT_FN_INT8_FLOAT): Likewise.
39979 (BT_FN_INT16_FLOAT): Likewise.
39980 (BT_FN_UINT32_FLOAT): Likewise.
39981 (BT_FN_UINT16_FLOAT): Likewise.
39982 (BT_FN_UINT8_FLOAT): Likewise.
39983 (BT_FN_UINT64_FLOAT): Likewise.
39984 (BT_FN_UINT16_UINT32): Likewise.
39985 (BT_FN_UINT32_UINT16): Likewise.
39986 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39987 (BT_FN_INT_PTR_INT): Likewise.
39988 (BT_FN_UINT_PTR_UINT): Likewise.
39989 (BT_FN_LONG_PTR_LONG): Likewise.
39990 (BT_FN_ULONG_PTR_ULONG): Likewise.
39991 (BT_FN_VOID_UINT64_UINT64): Likewise.
39992 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39993 (BT_FN_INT8_INT8_INT8): Likewise.
39994 (BT_FN_INT16_INT16_INT16): Likewise.
39995 (BT_FN_INT_INT_INT): Likewise.
39996 (BT_FN_UINT_FLOAT_UINT): Likewise.
39997 (BT_FN_FLOAT_UINT_UINT): Likewise.
39998 (BT_FN_ULONG_UINT_UINT): Likewise.
39999 (BT_FN_ULONG_UINT_PTR): Likewise.
40000 (BT_FN_ULONG_ULONG_ULONG): Likewise.
40001 (BT_FN_UINT_UINT_UINT): Likewise.
40002 (BT_FN_VOID_UINT_PTR): Likewise.
40003 (BT_FN_UINT_UINT_PTR: Likewise.
40004 (BT_FN_UINT32_UINT64_PTR): Likewise.
40005 (BT_FN_INT_INT_UINT_UINT): Likewise.
40006 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
40007 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
40008 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
40009 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
40010 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
40011 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
40012 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
40013 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
40014 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
40015 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
40016 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
40017 * doc/frontends.texi: List BRIG FE.
40018 * doc/install.texi (Testing): Add BRIG tesring requirements.
40019 * doc/invoke.texi (Overall Options): Mention BRIG.
40020 * doc/standards.texi (Standards): Doucment BRIG HSA version.
40022 2017-01-24 Richard Biener <rguenther@suse.de>
40024 PR translation/79208
40025 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
40027 2017-01-24 Martin Jambor <mjambor@suse.cz>
40030 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
40031 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
40032 and known_contexts.
40034 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
40036 PR middle-end/79123
40037 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
40038 casts from signed to unsigned really don't have a range.
40040 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
40042 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
40043 GMP_RNDx for compatiblity.
40045 2017-01-24 Martin Liska <mliska@suse.cz>
40048 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
40049 that would prevent us to call alloca with -1 as argument.
40051 2017-01-24 Jakub Jelinek <jakub@redhat.com>
40053 * dwarf2out.c (output_compilation_unit_header, output_file_names):
40054 Avoid -Wformat-security warning.
40056 2017-01-23 Andrew Pinski <apinski@cavium.com>
40058 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
40061 2017-01-23 Martin Sebor <msebor@redhat.com>
40063 PR middle-end/78703
40064 * gimple-ssa-sprintf.c (warn_level): New global.
40065 (format_integer): Use it here and throughout the rest of the file.
40066 Use the same switch to compute sign as base.
40067 (maybe_warn): New function.
40068 (format_directive): Factor out warnings into maybe_warn.
40069 Add debugging output. Use warn_level.
40070 (add_bytes): Use warn_level.
40071 (pass_sprintf_length::compute_format_length): Add debugging output.
40072 (try_substitute_return_value): Same.
40073 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
40075 PR middle-end/78703
40076 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
40077 (struct fmtresult, format_integer, format_floating): Adjust.
40078 (fmtresult::fmtresult): Set max correctly in two argument ctor.
40079 (get_string_length, format_string,format_directive): Same.
40080 (pass_sprintf_length::compute_format_length): Same.
40081 (try_substitute_return_value): Simplify slightly.
40083 PR middle-end/78703
40084 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
40085 (fmtresult::operator+=): Outlined.
40086 (struct fmtresult): Add ctors.
40087 (struct conversion_spec): Rename...
40088 (struct directive): ...to this. Add and remove data members.
40089 (directive::set_width, directive::set_precision): New functions.
40090 (format_percent): Use fmtresult ctor.
40091 (get_width_and_precision): Remove.
40092 (format_integer): Make naming changes. Avoid computing width and
40094 (format_floating): Same. Adjust indentation.
40095 (format_character, format_none): New functions.
40096 (format_string): Moved character handling to format_character.
40097 (format_directive): Remove arguments, change return type.
40098 (parse_directive): New function.
40099 (pass_sprintf_length::compute_format_length): Move directive
40100 parsing to parse_directive.
40102 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40104 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
40105 (assign_assembler_name_if_needed): ... this.
40106 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
40107 (assign_assembler_name_if_needed): ... this.
40108 (free_lang_data_in_cgraph): Adjust callers.
40109 * cgraphunit.c (cgraph_node::analyze): Likewise.
40110 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
40113 2017-01-23 Richard Biener <rguenther@suse.de>
40115 PR tree-optimization/79088
40116 PR tree-optimization/79188
40117 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
40118 resetting loop bounds after last path deletion. Reset loop
40119 bounds of the target loop, make code match the comments.
40120 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
40121 Make sure loops need no fixups.
40123 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
40125 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
40126 exponent support with double type for first argument.
40127 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
40128 type returned by __builtin_vec_extract_sig,
40129 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
40130 functions from "vector int" to "vector unsigned int" or from
40131 "vector long long int" to "vector unsigned long long int".
40132 Changed type returned by __builtin_vec_extract_exp,
40133 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
40134 functions from "vector int" to "vector unsigned int" or from
40135 "vector long long int" to "vector unsigned long long int".
40136 Changed return type of __builtin_vec_test_data_class,
40137 __builtin_vec_test_data_class_sp, and
40138 __builtin_vec_test_data_class_dp from "vector int" to
40139 "vector bool int" or from "vector long long int" to "vector bool
40140 long long int" and changed second argument type from "unsigned
40141 int" to "int". Added new overloaded function forms "vector float
40142 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
40143 "vector float __builtin_vec_insert_exp_sp (vector float, vector
40144 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
40145 double, vector unsigned long long int)" and "vector double
40146 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
40147 long int)". Changed return type of
40148 __builtin_scalar_test_data_class and
40149 __builtin_scalar_test_data_class_sp and
40150 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
40151 int" and changed second argument from "unsigned int" to "int".
40152 Changed type returned by __builtin_scalar_test_neg,
40153 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
40154 from "int" to "bool int". Added new overloaded function form
40155 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
40156 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
40157 exponent double-precision with floating point first argument.
40158 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
40159 documentation of scalar_test_data_class, scalar_test_neg,
40160 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
40161 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
40162 vec_test_data_class built-in functions to reflect refinements in
40163 their type signatures.
40165 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
40167 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
40169 (aarch64_elf_asm_destructor): Likewise.
40171 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
40173 PR rtl-optimization/78634
40174 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
40175 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
40176 * ifcvt.c (noce_try_cmove): Add missing cost check.
40178 PR rtl-optimization/71724
40179 * combine.c (if_then_else_cond): Look for situations where it is
40180 beneficial to undo the work of one of the recursive calls.
40182 2017-01-23 Bin Cheng <bin.cheng@arm.com>
40184 PR tree-optimization/70754
40185 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
40186 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
40187 combined stmt before it if not NULL.
40188 (combine_chains): Process refs reversely and compute dominance point
40191 2017-01-23 Martin Liska <mliska@suse.cz>
40193 PR tree-optimization/79196
40194 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
40195 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
40197 (strlen_optimize_stmt): Call the renamed function.
40199 2017-01-23 Michael Matz <matz@suse.de>
40201 PR tree-optimization/78384
40202 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
40204 2017-01-23 Richard Biener <rguenther@suse.de>
40206 PR tree-optimization/79186
40207 * tree-vrp.c (register_new_assert_for): Make sure we've seen
40208 both incoming edges before moving an assert.
40210 2017-01-23 Martin Jambor <mjambor@suse.cz>
40212 * ipa-prop.c (load_from_param_1): Removed.
40213 (load_from_unmodified_param): Bits from load_from_param_1 put back
40215 (load_from_param): Removed.
40216 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
40217 with stmt. Reverted back to use of load_from_unmodified_param.
40219 2017-01-23 Martin Jambor <mjambor@suse.cz>
40222 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40223 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40224 field a pointer to garbage collected vector, mark lattices and
40225 ipcp_orig_node with GTY((skip)).
40226 (ipa_get_param_count): Adjust to descriptors being a pointer.
40227 (ipa_get_param): Likewise.
40228 (ipa_get_type): Likewise.
40229 (ipa_get_param_move_cost): Likewise.
40230 (ipa_set_param_used): Likewise.
40231 (ipa_get_controlled_uses): Likewise.
40232 (ipa_set_controlled_uses): Likewise.
40233 (ipa_is_param_used): Likewise.
40234 (ipa_node_params_t): Move into garbage collector. New methods insert
40236 (ipa_node_params_sum): Annotate wth GTY(()).
40237 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40239 (ipa_load_from_parm_agg): Adjust declaration.
40240 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40241 * ipa-profile.c (ipa_profile): Likewise.
40242 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40243 (ipa_populate_param_decls): Make descriptors parameter garbage
40245 (ipa_dump_param): Adjust to descriptors being a pointer.
40246 (ipa_alloc_node_params): Likewise.
40247 (ipa_initialize_node_params): Likewise.
40248 (load_from_param_1): Make descriptors parameter garbage collected.
40249 (load_from_unmodified_param): Likewise.
40250 (load_from_param): Likewise.
40251 (ipa_load_from_parm_agg): Likewise.
40252 (ipa_node_params::~ipa_node_params): Removed.
40253 (ipa_free_all_node_params): Remove call to delete operator.
40254 (ipa_node_params_t::insert): New.
40255 (ipa_node_params_t::remove): Likewise.
40256 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40257 copy known_csts and known_contexts vectors.
40258 (ipa_read_node_info): Adjust to descriptors being a pointer.
40259 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40261 (ipcp_transform_function): Make descriptors variable garbage
40264 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40266 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40267 * config/i386/avx512dqintrin.h: Ditto.
40268 * config/i386/avx512fintrin.h: Ditto.
40269 * config/i386/i386.c: Handle new builtins.
40270 * config/i386/i386-builtin.def: Add new builtins.
40271 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40272 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40274 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40275 Martin Liska <mliska@suse.cz>
40277 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40278 * asan.c (asan_expand_poison_ifn): Support stores and use
40279 appropriate ASAN report function.
40280 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40281 * internal-fn.def (ASAN_POISON_USE): Declare.
40282 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40283 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40284 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40285 ASAN_POISON calls w/o LHS.
40286 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40287 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40288 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40289 * gimplify.c (asan_poison_variables): Add attribute
40290 use_after_scope_memory to variables that really needs to live
40292 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40293 having the attribute.
40295 2017-01-23 Martin Liska <mliska@suse.cz>
40297 * asan.c (create_asan_shadow_var): New function.
40298 (asan_expand_poison_ifn): Likewise.
40299 * asan.h (asan_expand_poison_ifn): New declaration.
40300 * internal-fn.c (expand_ASAN_POISON): Likewise.
40301 * internal-fn.def (ASAN_POISON): New builtin.
40302 * sanopt.c (pass_sanopt::execute): Expand
40303 asan_expand_poison_ifn.
40304 * tree-inline.c (copy_decl_for_dup_finish): Make function
40306 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40307 * tree-ssa.c (is_asan_mark_p): New function.
40308 (execute_update_addresses_taken): Rewrite local variables
40309 (identified just by use-after-scope as addressable) into SSA.
40311 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40313 * doc/install.texi (Specific): opensource.apple.com uses https
40314 now. Remove trailing slash.
40316 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40318 * README.Portability: Remove note on an Irix compatibility issue.
40320 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40322 * gcov.c (INCLUDE_ALGORITHM): Define.
40323 (INCLUDE_VECTOR): Define.
40324 No longer include <vector> and <algorithm> directly.
40326 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40328 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40330 * doc/invoke.texi (Code Gen Options): Ditto.
40332 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40335 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40337 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40339 rtl-optimization/79125
40340 * cprop.c (local_cprop_pass): Handle cases where we make an
40341 unconditional trap.
40343 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40347 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40348 read from, for big endian.
40350 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40352 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40353 register pauth builtins for LP64 only.
40355 2017-01-20 Marek Polacek <polacek@redhat.com>
40358 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40361 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40363 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40365 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40366 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40367 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40369 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40372 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40373 in big-endian mode when they are not a single duplicated value.
40375 2017-01-20 Richard Biener <rguenther@suse.de>
40377 * BASE-VER: Bump to 7.0.1.
40379 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40381 * omp-low.c (omplow_simd_context): New struct. Use it...
40382 (lower_rec_simd_input_clauses): ...here and...
40383 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40384 references to idx, lane, max_vf, is_simt.
40386 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40388 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40391 2017-01-20 Martin Jambor <mjambor@suse.cz>
40393 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40394 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40396 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40397 (GTFILES): Rename hsa.c to hsa-common.c.
40398 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40399 * hsa-dump.c: Likewise.
40400 * hsa-gen.c: Likewise.
40401 * hsa-regalloc.c: Likewise.
40402 * ipa-hsa.c: Likewise.
40403 * omp-expand.c: Likewise.
40404 * omp-low.c: Likewise.
40405 * toplev.c: Likewise.
40407 2017-01-20 Marek Polacek <polacek@redhat.com>
40410 * doc/invoke.texi: Document -Wduplicated-branches.
40411 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40412 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40413 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40414 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40415 return 0 only when not OEP_LEXICOGRAPHIC.
40416 (fold_build_cleanup_point_expr): Use the expression
40417 location when building CLEANUP_POINT_EXPR.
40418 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40419 * tree.c (add_expr): Handle error_mark_node.
40421 2017-01-20 Martin Liska <mliska@suse.cz>
40424 * tree-profile.c (init_ic_make_global_vars): Do not call
40426 (gimple_init_gcov_profiler): Likewise.
40428 2017-01-20 Martin Liska <mliska@suse.cz>
40431 * cgraph.h (maybe_create_reference): Remove argument and
40433 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40435 * ipa-cp.c (create_specialized_node): Likewise.
40436 * symtab.c (symtab_node::maybe_create_reference): Handle
40437 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40439 2017-01-20 Martin Liska <mliska@suse.cz>
40441 * read-rtl-function.c (function_reader::create_function): Use
40442 build_decl instread of build_decl_stat.
40444 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40446 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40447 * config/i386/avx512dqintrin.h: Ditto.
40448 * config/i386/avx512fintrin.h: Ditto.
40449 * config/i386/i386-builtin-types.def: Add new types.
40450 * config/i386/i386.c: Handle new types.
40451 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40452 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40453 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40454 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40455 (__builtin_ia32_kshiftridi): New.
40456 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40458 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40462 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40463 define to rs6000_init_stack_protect_guard.
40464 (rs6000_init_stack_protect_guard): New function.
40466 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40467 Yunqiang Su <yunqiang.su@imgtec.com>
40469 * config.gcc (supported_defaults): Add madd4.
40470 (with_madd4): Add validation.
40471 (all_defaults): Add madd4.
40472 * config/mips/mips.opt (mmadd4): New option.
40473 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40475 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40477 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40478 (ISA_HAS_FUSED_MADD4): Likewise.
40479 * doc/invoke.texi (-mmadd4): Document the new option.
40480 * doc/install.texi (--with-madd4): Document the new option.
40482 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40484 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40485 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40486 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40487 (aarch64_init_pauth_hint_builtins): New.
40488 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40489 (aarch64_expand_builtin): Expand new builtins.
40491 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40493 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40494 * combine-stack-adj.c (no_unhandled_cfa): Handle
40495 REG_CFA_TOGGLE_RA_MANGLE.
40496 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40497 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40498 info for return address signing.
40499 (aarch64_expand_epilogue): Likewise.
40501 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40503 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40504 * config/aarch64/aarch64-protos.h
40505 (aarch64_return_address_signing_enabled): New declaration.
40506 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40508 (aarch64_expand_prologue): Sign return address before it's pushed onto
40510 (aarch64_expand_epilogue): Authenticate return address fetched from
40512 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40513 (aarch64_attributes): New function attributes for "sign-return-address".
40514 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40515 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40516 ("*do_return"): Generate combined instructions according to key index.
40517 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40518 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40520 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40521 * config/aarch64/aarch64.opt (msign-return-address=): New.
40522 * doc/extend.texi (AArch64 Function Attributes): Documents
40523 "sign-return-address=".
40524 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40526 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40528 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40529 overall option summary.
40531 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40533 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40534 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40535 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40536 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40538 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40540 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40541 -mpower9-minmax by default for -mcpu=power9.
40542 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40543 128-bit floating point.
40545 2017-01-20 Alan Modra <amodra@gmail.com>
40547 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40548 optimizing for size.
40550 2017-01-20 Alan Modra <amodra@gmail.com>
40553 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40554 for strcmp and strncmp from corresponding builtin decl.
40556 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40558 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40559 instead of i386/rtems-64.h.
40560 * config/i386/rtems-64.h: Remove.
40562 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40566 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40568 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40570 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40572 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40573 Change int to HOST_WIDE_INT.
40574 * config/aarch64/aarch64-protos.h
40575 (aarch64_simd_gen_const_vector_dup): Likewise.
40576 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40578 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40580 * langhooks-def.h (lhd_type_for_size): New decl.
40581 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40582 * langhooks.c (lhd_type_for_size): New function, taken from
40585 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40587 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40588 define_bypass for CR latency.
40589 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40590 (power9-alu2): Add define_bypass for CR latency.
40592 (power9-mul): Update insn latency.
40593 (power9-mul-compare): Update insn latency, bypass latency and remove
40596 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40598 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40600 * config/aarch64/aarch64.md
40601 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40602 aarch64_nopcrelative_literal_loads.
40603 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40605 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40607 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40608 TARGET_LOONGSON_3A.
40609 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40611 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40614 * config.gcc (supported_defaults): Add lxc1-sxc1.
40615 (with_lxc1_sxc1): Add validation.
40616 (all_defaults): Add lxc1-sxc1.
40617 * config/mips/mips.opt (mlxc1-sxc1): New option.
40618 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40620 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40621 __mips_no_lxc1_sxc1.
40622 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40623 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40624 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40626 2017-01-19 Richard Biener <rguenther@suse.de>
40628 PR tree-optimization/72488
40629 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40630 sure to restore SSA info.
40631 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40633 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40635 PR rtl-optimization/79121
40636 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40637 of the inner type when shifting an extended value.
40639 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40642 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40643 interposable aliases.
40645 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40648 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40649 Use the evmergelohi instruction.
40650 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40651 (mov_sitf_e500_subreg8_2_be): Likewise.
40652 (mov_sitf_e500_subreg12_2_le): Likewise.
40653 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40654 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40655 (mov_sitf_e500_subreg8_2_le): Likewise.
40656 (mov_sitf_e500_subreg12_2_be): Likewise.
40658 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40660 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40661 attribute from vecsimple to vecperm.
40662 (altivec_vbpermq2): Likewise.
40664 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40667 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40669 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40670 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40671 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40672 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40673 case where N arg is SIZE_MAX.
40674 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40675 (cmpstrsi): Add pattern.
40677 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40679 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40680 __builtin_vec_revb builtins.
40681 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40682 built-in functions to support generation of the ISA 3.0 XXBR<x>
40683 vector byte reverse instructions.
40684 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40685 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40686 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40687 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40688 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40689 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40690 (P9V_BUILTIN_VEC_REVB): Likewise.
40691 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40692 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40693 (p9_xxbrq_v16qi): Likewise.
40694 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40695 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40696 (p9_xxbrh_v8hi): Likewise.
40697 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40698 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40699 vec_revb built-in functions.
40701 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40703 PR rtl-optimization/78952
40704 * config/i386/i386.md (any_extract): New code iterator.
40705 (*insvqi_2): Use any_extract for source operand.
40706 (*insvqi_3): Use any_shiftrt for source operand.
40708 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40710 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40712 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40714 2017-01-18 Matthias Klose <doko@ubuntu.com>
40716 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40718 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40720 * config/rs6000/altivec.h (vec_bperm): Change #define.
40721 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40722 (altivec_vbpermq2): New define_insn.
40723 (altivec_vbpermd): Likewise.
40724 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40725 function interface.
40726 (VBPERMD): Likewise.
40727 (VBPERM): New polymorphic function interface.
40728 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40729 Add entries for P9V_BUILTIN_VEC_VBPERM.
40730 * doc/extend.texi: Add interfaces for vec_bperm.
40732 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40734 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40735 first letter of error messages.
40736 (s390_resolve_overloaded_builtin): Likewise.
40737 * config/s390/s390.c (s390_expand_builtin): Likewise.
40738 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40739 (s390_valid_target_attribute_inner_p): Likewise.
40740 * config/s390/s390.md ("tabort"): Likewise.
40742 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40744 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40745 (ISA_AVOID_DIV_HILO): New macro.
40746 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40747 (ISA_HAS_DDIV): Likewise.
40749 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40751 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40753 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40755 * doc/invoke.texi (fabi-version): Spelling fix.
40757 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40760 * doc/invoke.texi (fabi-version): Mention mangling fix for
40763 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40766 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40768 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40771 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40772 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40774 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40776 (stack_protect_test): Ditto.
40777 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40778 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40780 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40781 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40782 -mstack-protector-guard-offset=.
40783 (RS/6000 and PowerPC Options): Ditto.
40785 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40787 * config/i386/i386.h (MASK_CLASS_P): New define.
40788 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40789 there are no registers from different register sets also when
40790 mask registers are used. Update function comment.
40791 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40792 to (*k/*r) and (*k/*km) alternatives.
40794 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40796 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40797 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40798 (EH_RETURN_HANDLER_RTX): New define.
40799 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40800 Force frame pointer in EH return functions.
40801 (aarch64_expand_epilogue): Add barrier for eh_return.
40802 (aarch64_final_eh_return_addr): Remove.
40803 (aarch64_eh_return_handler_rtx): New function.
40804 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40806 (aarch64_eh_return_handler_rtx): New prototype.
40808 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40810 * config/rs6000/altivec.h (vec_rlmi): New #define.
40811 (vec_vrlnm): Likewise.
40812 (vec_rlnm): Likewise.
40813 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40814 (UNSPEC_VRLNM): Likewise.
40815 (VIlong): New mode iterator.
40816 (altivec_vrl<VI_char>mi): New define_insn.
40817 (altivec_vrl<VI_char>nm): Likewise.
40818 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40820 (VRLDNM): Likewise.
40821 (RLNM): New polymorphic function entry.
40822 (VRLWMI): New monomorphic function entry.
40823 (VRLDMI): Likewise.
40824 (RLMI): New polymorphic function entry.
40825 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40826 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40827 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40830 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40833 * dwarf2out.c (field_byte_offset): Restore the
40834 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40835 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40836 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40839 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40842 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40844 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40847 * dwarf2out.c (add_data_member_location_attribute): For constant
40848 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40849 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40850 DW_AT_byte_size attributes.
40852 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40854 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40855 after forcing to constant memory when the code model is medium.
40857 2017-01-17 Julia Koval <julia.koval@intel.com>
40860 * config/i386/avx512fintrin.h
40861 (_mm512_i32gather_ps): Change __addr type to void const*.
40862 (_mm512_mask_i32gather_ps): Ditto.
40863 (_mm512_i32gather_pd): Ditto.
40864 (_mm512_mask_i32gather_pd): Ditto.
40865 (_mm512_i64gather_ps): Ditto.
40866 (_mm512_mask_i64gather_ps): Ditto.
40867 (_mm512_i64gather_pd): Ditto.
40868 (_mm512_mask_i64gather_pd): Ditto.
40869 (_mm512_i32gather_epi32): Ditto.
40870 (_mm512_mask_i32gather_epi32): Ditto.
40871 (_mm512_i32gather_epi64): Ditto.
40872 (_mm512_mask_i32gather_epi64): Ditto.
40873 (_mm512_i64gather_epi32): Ditto.
40874 (_mm512_mask_i64gather_epi32): Ditto.
40875 (_mm512_i64gather_epi64): Ditto.
40876 (_mm512_mask_i64gather_epi64): Ditto.
40877 (_mm512_i32scatter_ps): Change __addr type to void*.
40878 (_mm512_mask_i32scatter_ps): Ditto.
40879 (_mm512_i32scatter_pd): Ditto.
40880 (_mm512_mask_i32scatter_pd): Ditto.
40881 (_mm512_i64scatter_ps): Ditto.
40882 (_mm512_mask_i64scatter_ps): Ditto.
40883 (_mm512_i64scatter_pd): Ditto.
40884 (_mm512_mask_i64scatter_pd): Ditto.
40885 (_mm512_i32scatter_epi32): Ditto.
40886 (_mm512_mask_i32scatter_epi32): Ditto.
40887 (_mm512_i32scatter_epi64): Ditto.
40888 (_mm512_mask_i32scatter_epi64): Ditto.
40889 (_mm512_i64scatter_epi32): Ditto.
40890 (_mm512_mask_i64scatter_epi32): Ditto.
40891 (_mm512_i64scatter_epi64): Ditto.
40892 (_mm512_mask_i64scatter_epi64): Ditto.
40893 * config/i386/avx512pfintrin.h
40894 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40895 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40896 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40897 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40898 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40899 (_mm512_prefetch_i32scatter_ps): Ditto.
40900 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40901 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40902 (_mm512_prefetch_i64scatter_pd): Ditto.
40903 (_mm512_prefetch_i64scatter_ps): Ditto.
40904 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40905 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40906 * config/i386/avx512vlintrin.h
40907 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40908 (_mm_mmask_i32gather_ps): Ditto.
40909 (_mm256_mmask_i32gather_pd): Ditto.
40910 (_mm_mmask_i32gather_pd): Ditto.
40911 (_mm256_mmask_i64gather_ps): Ditto.
40912 (_mm_mmask_i64gather_ps): Ditto.
40913 (_mm256_mmask_i64gather_pd): Ditto.
40914 (_mm_mmask_i64gather_pd): Ditto.
40915 (_mm256_mmask_i32gather_epi32): Ditto.
40916 (_mm_mmask_i32gather_epi32): Ditto.
40917 (_mm256_mmask_i32gather_epi64): Ditto.
40918 (_mm_mmask_i32gather_epi64): Ditto.
40919 (_mm256_mmask_i64gather_epi32): Ditto.
40920 (_mm_mmask_i64gather_epi32): Ditto.
40921 (_mm256_mmask_i64gather_epi64): Ditto.
40922 (_mm_mmask_i64gather_epi64): Ditto.
40923 (_mm256_i32scatter_ps): Change __addr type to void*.
40924 (_mm256_mask_i32scatter_ps): Ditto.
40925 (_mm_i32scatter_ps): Ditto.
40926 (_mm_mask_i32scatter_ps): Ditto.
40927 (_mm256_i32scatter_pd): Ditto.
40928 (_mm256_mask_i32scatter_pd): Ditto.
40929 (_mm_i32scatter_pd): Ditto.
40930 (_mm_mask_i32scatter_pd): Ditto.
40931 (_mm256_i64scatter_ps): Ditto.
40932 (_mm256_mask_i64scatter_ps): Ditto.
40933 (_mm_i64scatter_ps): Ditto.
40934 (_mm_mask_i64scatter_ps): Ditto.
40935 (_mm256_i64scatter_pd): Ditto.
40936 (_mm256_mask_i64scatter_pd): Ditto.
40937 (_mm_i64scatter_pd): Ditto.
40938 (_mm_mask_i64scatter_pd): Ditto.
40939 (_mm256_i32scatter_epi32): Ditto.
40940 (_mm256_mask_i32scatter_epi32): Ditto.
40941 (_mm_i32scatter_epi32): Ditto.
40942 (_mm_mask_i32scatter_epi32): Ditto.
40943 (_mm256_i32scatter_epi64): Ditto.
40944 (_mm256_mask_i32scatter_epi64): Ditto.
40945 (_mm_i32scatter_epi64): Ditto.
40946 (_mm_mask_i32scatter_epi64): Ditto.
40947 (_mm256_i64scatter_epi32): Ditto.
40948 (_mm256_mask_i64scatter_epi32): Ditto.
40949 (_mm_i64scatter_epi32): Ditto.
40950 (_mm_mask_i64scatter_epi32): Ditto.
40951 (_mm256_i64scatter_epi64): Ditto.
40952 (_mm256_mask_i64scatter_epi64): Ditto.
40953 (_mm_i64scatter_epi64): Ditto.
40954 (_mm_mask_i64scatter_epi64): Ditto.
40955 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40956 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40957 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40958 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40959 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40960 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40961 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40962 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40963 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40964 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40965 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40966 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40967 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40968 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40969 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40970 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40971 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40972 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40973 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40974 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40975 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40976 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40977 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40978 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40979 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40980 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40981 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40982 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40983 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40984 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40985 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40986 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40987 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40988 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40989 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40990 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40991 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40992 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40993 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40994 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40995 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40996 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40997 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40998 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40999 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
41000 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
41001 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
41002 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
41003 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
41004 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
41005 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
41006 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
41007 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
41008 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
41009 definitions accordingly.
41011 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
41012 Kuan-Lin Chen <kuanlinchentw@gmail.com>
41015 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
41018 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
41021 * ira-conflicts.c (ira_build_conflicts): Update total conflict
41022 hard regs for inner regno.
41024 2017-01-17 Martin Liska <mliska@suse.cz>
41027 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
41028 assumption and add comment.
41030 2017-01-17 Nathan Sidwell <nathan@acm.org>
41032 * ipa-visibility.c (localize_node): New function, broken out of ...
41033 (function_and_variable_visibility): ... here. Call it.
41035 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
41037 PR middle-end/77445
41038 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
41039 correctly set frequency of oudgoing edge.
41040 (duplicate_thread_path): Fix profile updating.
41042 2017-01-17 Jakub Jelinek <jakub@redhat.com>
41045 * configure.ac: Add GCC_BASE_VER.
41046 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
41047 version from BASE-VER file.
41048 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
41049 (gcc.o): Depend on $(BASEVER).
41050 * common.opt (dumpfullversion): New option.
41051 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
41052 * doc/invoke.texi: Document -dumpfullversion.
41053 * doc/install.texi: Document --with-gcc-major-version-only.
41054 * configure: Regenerated.
41056 2017-01-17 Richard Biener <rguenther@suse.de>
41058 PR tree-optimization/71433
41059 * tree-vrp.c (register_new_assert_for): Merge same asserts
41060 on all incoming edges.
41061 (process_assert_insertions_for): Handle insertions at the
41064 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
41066 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
41067 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
41069 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
41072 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
41075 2017-01-17 Alan Modra <amodra@gmail.com>
41078 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
41079 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
41080 symbolic stack limit when pic.
41082 2017-01-16 Martin Sebor <msebor@redhat.com>
41084 PR tree-optimization/78608
41085 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
41087 2017-01-16 Jeff Law <law@redhat.com>
41090 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
41091 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
41092 for several include directories that may be relative to sysroot.
41093 * config/i386/x-mingw32 (gplus_includedir): Define.
41094 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
41095 (native_system_includedir): Likewise.
41096 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
41097 override if TARGET_SYSTEM_ROOT is defined.
41098 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
41100 PR tree-optimization/79090
41101 PR tree-optimization/33562
41102 PR tree-optimization/61912
41103 PR tree-optimization/77485
41104 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
41105 and computed trims into the dump file.
41107 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
41109 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
41111 2017-01-16 Jakub Jelinek <jakub@redhat.com>
41114 * gimplify.c (gimplify_init_constructor): If want_value and
41115 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
41119 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
41120 sequence. Formatting fixes.
41121 (doloop_optimize): Formatting fixes.
41124 * gcc.c (debug_level_greater_than_spec_func): New function.
41125 (static_spec_functions): Add debug-level-gt spec function.
41126 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
41128 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41129 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41130 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
41131 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
41132 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
41133 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
41135 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
41137 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
41138 QImode fixups to general and mask registers only.
41140 2017-01-16 Carl Love <cel@us.ibm.com>
41142 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
41143 for built-in functions
41144 vector signed char vec_nabs (vector signed char)
41145 vector signed short vec_nabs (vector signed short)
41146 vector signed int vec_nabs (vector signed int)
41147 vector signed long long vec_nabs (vector signed long long)
41148 vector float vec_nabs (vector float)
41149 vector double vec_nabs (vector double)
41150 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
41152 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
41153 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
41154 * doc/extend.texi: Update the documentation file for the new built-in
41157 2017-01-16 Martin Sebor <msebor@redhat.com>
41159 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
41162 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41164 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
41165 UNSPEC_VSX__XXSPLTD to require special splat handling.
41167 2017-01-16 David Malcolm <dmalcolm@redhat.com>
41170 * system.h: Poison strndup.
41172 2017-01-16 Alan Modra <amodra@gmail.com>
41175 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
41178 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
41180 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
41182 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
41184 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
41185 call recog here. Assert that INSN_CODE (insn) is non-negative.
41187 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41190 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
41192 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
41193 in the currently scheduled RTL fragment.
41195 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41197 PR rtl-optimization/78751
41198 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
41201 2017-01-14 Jeff Law <law@redhat.com>
41203 PR tree-optimization/79090
41204 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
41205 variable length stores.
41206 (compute_trims): Delete dead assignment to *trim_tail.
41207 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
41210 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
41212 PR rtl-optimization/78626
41213 PR rtl-optimization/78727
41214 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
41215 of a block, and split such blocks after everything else is finished.
41217 2017-01-14 Alan Modra <amodra@gmail.com>
41220 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
41221 target legitimate_combined_insn.
41222 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41223 (rs6000_legitimate_combined_insn): New function.
41224 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41226 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41227 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41228 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41230 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41232 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41234 2017-01-13 Jeff Law <law@redhat.com>
41236 PR tree-optimization/33562
41237 PR tree-optimization/61912
41238 PR tree-optimization/77485
41239 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41241 (delete_dead_assignment): Likewise.
41242 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41243 statement to delete_dead_call and delete_dead_assignment.
41245 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41248 * substring-locations.c (format_warning_va): Strengthen case 1 so
41249 that both endpoints of the substring must be within the format
41250 range for just the substring to be printed.
41252 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41254 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41255 * config/i386/i386.c (ix86_target_string): Add missing options
41256 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41257 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41258 flags_other and ix86_target_other to flags2_other. Display unknown
41260 (ix86_valid_target_attribute_inner_p): Add missing options and
41261 reorder options by implied ISAs, as in ix86_target_string.
41263 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41265 * hash-table.h (hash_table::too_empty_p): New function.
41266 (hash_table::expand): Use it.
41267 (hash_table::traverse): Likewise.
41268 (hash_table::empty_slot): Use sizeof (value_type) instead of
41269 sizeof (PTR) to convert bytes to elements. Shrink the table
41270 if the current size is excessive for the current number of
41273 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41275 * ira-costs.c (record_reg_classes): Break from the inner loop
41276 early once alt_fail is known to be true. Update outer loop
41277 handling accordingly.
41279 2017-01-13 Jeff Law <law@redhat.com>
41281 * tree-ssa-dse.c (decrement_count): New function.
41282 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41283 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41284 when we know the partially dead statement is a mem* function.
41286 PR tree-optimization/61912
41287 PR tree-optimization/77485
41288 * tree-ssa-dse.c: Include expr.h.
41289 (maybe_trim_constructor_store): New function.
41290 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41292 PR tree-optimization/33562
41293 PR tree-optimization/61912
41294 PR tree-optimization/77485
41295 * doc/invoke.texi: Document new dse-max-object-size param.
41296 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41297 * tree-ssa-dse.c: Include params.h.
41298 (dse_store_status): New enum.
41299 (initialize_ao_ref_for_dse): New, partially extracted from
41301 (valid_ao_ref_for_dse, normalize_ref): New.
41302 (setup_live_bytes_from_ref, compute_trims): Likewise.
41303 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41304 (maybe_trim_partially_dead_store): Likewise.
41305 (maybe_trim_complex_store): Likewise.
41306 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41307 Track what bytes live from the original store. Return tri-state
41308 for dead, partially dead or live.
41309 (dse_dom_walker): Add constructor, destructor and new private members.
41310 (delete_dead_call, delete_dead_assignment): New extracted from
41312 (dse_optimize_stmt): Make a member of dse_dom_walker.
41313 Use initialize_ao_ref_for_dse.
41315 PR tree-optimization/33562
41316 PR tree-optimization/61912
41317 PR tree-optimization/77485
41318 * sbitmap.h (bitmap_count_bits): Prototype.
41319 (bitmap_clear_range, bitmap_set_range): Likewise.
41320 * sbitmap.c (bitmap_clear_range): New function.
41321 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41323 2017-01-13 Martin Liska <mliska@suse.cz>
41326 * function.c (set_cfun): Add new argument force.
41327 * function.h (set_cfun): Likewise.
41328 * ipa-inline-transform.c (inline_call): Use the function when
41329 strict alising from is dropped for function we inline to.
41331 2017-01-13 Richard Biener <rguenther@suse.de>
41333 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41334 for dumping GIMPLE INTEGER_CSTs.
41336 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41338 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41339 to 201112L since C++17.
41341 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41344 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41345 if -fsanitize=kernel-address is present.
41347 2017-01-13 Richard Biener <rguenther@suse.de>
41349 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41350 as _Literal ( type ) number in case usual suffixes do not
41351 preserve all information.
41353 2017-01-13 Richard Biener <rguenther@suse.de>
41355 PR tree-optimization/77283
41356 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41357 and ssa-iterators.h.
41358 (is_feasible_trace): Implement a cost model based on joiner
41361 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41364 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41365 char or short to __float128/_Float128 directly.
41367 2017-01-12 Martin Sebor <msebor@redhat.com>
41369 to -Wformat-overflow.
41370 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41371 (min_bytes_remaining): Same.
41372 (get_string_length): Same.
41373 (format_string): Same.
41374 (format_directive): Same.
41376 (pass_sprintf_length::handle_gimple_call): Same.
41378 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41380 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41381 info.nowrite calls with no lhs that can't throw. Return bool
41382 whether gsi_remove has been called or not.
41383 (pass_sprintf_length::handle_gimple_call): Return bool whether
41384 try_substitute_return_value called gsi_remove. Formatting fix.
41385 (pass_sprintf_length::execute): Don't use gsi_remove if
41386 handle_gimple_call returned true.
41389 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41390 be removed due to side-effects, don't remove following barrier nor
41391 turn the successor edge into fallthru edge.
41393 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41396 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41397 element-reversing loads and stores as not swappable.
41399 2017-01-12 Nathan Sidwell <nathan@acm.org>
41400 Nicolai Stange <nicstange@gmail.com>
41402 * combine.c (try_combine): Don't ignore result of overlap checking
41403 loop. Combine overlap & asm check into single loop.
41405 2017-01-12 Richard Biener <rguenther@suse.de>
41407 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41408 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41410 2017-01-12 Richard Biener <rguenther@suse.de>
41412 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41413 and TS_TARGET_OPTION directly derive from TS_BASE.
41414 * tree-core.h (tree_optimization_option): Derive from tree_base.
41415 (tree_target_option): Likewise.
41417 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41419 * config/i386/i386.c (memory_address_length): Increase len
41420 only when rip_relative_addr_p returns false.
41422 2017-01-11 Julia Koval <julia.koval@intel.com>
41424 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41425 (OPTION_MASK_ISA_SGX_SET): New.
41426 (ix86_handle_option): Handle OPT_msgx.
41427 * config.gcc: Added sgxintrin.h.
41428 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41429 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41430 * config/i386/i386.c (ix86_target_string): Add -msgx.
41432 (ix86_option_override_internal): Handle new options.
41433 (ix86_valid_target_attribute_inner_p): Add sgx.
41434 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41435 * config/i386/i386.opt: Add msgx.
41436 * config/i386/sgxintrin.h: New file.
41437 * config/i386/x86intrin.h: Add sgxintrin.h.
41439 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41442 * fold-const.c (maybe_nonzero_address): Return 1 for function
41444 (tree_single_nonzero_warnv_p): Don't handle function local objects
41448 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41451 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41454 * opts.c: Include "spellcheck.h"
41455 (struct string_fragment): New struct.
41456 (struct edit_distance_traits<const string_fragment &>): New
41458 (get_closest_sanitizer_option): New function.
41459 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41461 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41463 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41465 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41466 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41467 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41468 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41469 for initial die_offset if dwarf_split_debug_info.
41470 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41471 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41473 (output_skeleton_debug_sections): Formatting fix. Use
41474 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41475 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41477 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41479 * config/arm/cortex-a53.md: Add bypasses for
41480 cortex_a53_r2f_cvt.
41481 (cortex_a53_r2f): Only use for transfers.
41482 (cortex_a53_f2r): Likewise.
41483 (cortex_a53_r2f_cvt): Add reservation for conversions.
41484 (cortex_a53_f2r_cvt): Likewise.
41486 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41488 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41489 to all inlined functions, change static to extern.
41491 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41494 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41496 (arm_assemble_integer): Likewise.
41498 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41500 * config.gcc: Use new awk script to check CPU, FPU and architecture
41501 parameters for --with-... options.
41502 * config/arm/parsecpu.awk: New file
41503 * config/arm/arm-cpus.in: New file.
41504 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41506 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41508 * config/arm/t-arm: Update dependency rules.
41509 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41510 of processing .def files.
41511 * config/arm/genopt.sh: Deleted.
41512 * config/arm/gentune.sh: Deleted.
41513 * config/arm/arm-cores.def: Deleted.
41514 * config/arm/arm-arches.def: Deleted.
41515 * config/arm/arm-fpus.def: Deleted.
41516 * config/arm/arm-tune.md: Regenerated.
41517 * config/arm/arm-tables.opt: Regenerated.
41518 * config/arm/arm-cpu.h: New generated file.
41519 * config/arm/arm-cpu-data.h: New generated file.
41520 * config/arm/arm-cpu-cdata.h: New generated file.
41522 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41525 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41527 (input_varpool_node): Unpack dynamically_initialized bit.
41529 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41531 PR rtl-optimization/79032
41532 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41533 the alignment of the adjusted memory reference against that of MODE,
41534 instead of the alignment of the original memory reference.
41536 2017-01-11 Martin Jambor <mjambor@suse.cz>
41538 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41540 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41541 decorated functions.
41543 2017-01-11 Richard Biener <rguenther@suse.de>
41545 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41546 set range/nonnull info for PHI results. Do not set it on
41547 stmts marked for removal.
41549 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41551 * expr.c (store_field): In the bitfield case, fetch the return value
41552 from the registers before applying a single big-endian adjustment.
41553 Always do a final load for a BLKmode value not larger than a word.
41555 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41558 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41559 that we correctly handle column numbers greater than
41560 LINE_MAP_MAX_COLUMN_NUMBER.
41562 2017-01-10 Martin Sebor <msebor@redhat.com>
41564 PR middle-end/78245
41565 * gimple-ssa-sprintf.c (get_destination_size): Call
41566 {init,fini}object_sizes.
41567 * tree-object-size.c (addr_object_size): Adjust.
41568 (pass_through_call): Adjust.
41569 (pass_object_sizes::execute): Adjust.
41570 * tree-object-size.h (fini_object_sizes): Declare.
41572 2017-01-10 Martin Sebor <msebor@redhat.com>
41574 PR tree-optimization/78775
41575 * builtins.c (get_size_range): Move...
41576 * calls.c: ...to here.
41577 (alloc_max_size): Accept zero argument.
41578 (operand_signed_p): Remove.
41579 (maybe_warn_alloc_args_overflow): Call get_size_range.
41580 * calls.h (get_size_range): Declare.
41582 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41584 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41585 from TI's devices.csv file as of September 2016.
41586 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41588 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41590 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41591 * doc/invoke.texi: Likewise.
41592 * doc/md.texi: Likewise.
41593 * doc/objc.texi: Likewise.
41595 2017-01-10 Joshua Conner <joshconner@google.com>
41597 * config/arm/fuchsia-elf.h: New file.
41598 * config/fuchsia.h: New file.
41599 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41600 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41602 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41604 2016-01-10 Richard Biener <rguenther@suse.de>
41606 PR tree-optimization/79034
41607 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41608 Propagate out degenerate PHIs in the joiner.
41610 2017-01-10 Martin Liska <mliska@suse.cz>
41612 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41613 (sort_congruence_classes_by_decl_uid): Likewise.
41614 (sort_congruence_class_groups_by_decl_uid): Likewise.
41615 (sem_item_optimizer::merge_classes): Sort class, groups in these
41616 classes and members in the groups by DECL_UID of declarations.
41617 This would make merge operations stable.
41619 2017-01-10 Martin Liska <mliska@suse.cz>
41621 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41622 usage of m_classes_vec.
41623 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41624 (sem_item_optimizer::get_group_by_hash): Likewise.
41625 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41626 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41627 (sem_item_optimizer::verify_classes): Likewise.
41628 (sem_item_optimizer::process_cong_reduction): Likewise.
41629 (sem_item_optimizer::dump_cong_classes): Likewise.
41630 (sem_item_optimizer::merge_classes): Likewise.
41631 * ipa-icf.h (congruence_class_hash): Rename from
41632 congruence_class_group_hash. Remove declaration of m_classes_vec.
41634 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41636 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41637 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41638 * config.gcc: Add avx512vpopcntdqintrin.h.
41639 * config/i386/avx512vpopcntdqintrin.h: New.
41640 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41641 * config/i386/i386-builtin-types.def: Add new types.
41642 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41643 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41644 __builtin_ia32_vpopcountq_v8di_mask): New.
41645 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41646 __AVX512VPOPCNTDQ__.
41647 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41648 (PTA_AVX512VPOPCNTDQ): Define.
41649 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41650 TARGET_AVX512VPOPCNTDQ_P): Define.
41651 * config/i386/i386.opt: Add mavx512vpopcntdq.
41652 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41653 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41655 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41657 PR middle-end/77484
41658 * predict.def (PRED_CALL): Set to 67.
41660 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41662 * expr.c (store_field): In the bitfield case, if the value comes from
41663 a function call and is of an aggregate type returned in registers, do
41664 not modify the field mode; extract the value in all cases if the mode
41665 is BLKmode and the size is not larger than a word.
41667 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41670 * config/i386/cpuid.h: Fix undefined behavior.
41672 2017-01-04 Jeff Law <law@redhat.com>
41674 PR tree-optimization/79007
41675 PR tree-optimization/67955
41676 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41677 conservative for pt.null when flag_non_call_exceptions is on.
41679 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41681 PR translation/79019
41682 PR translation/79020
41683 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41684 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41685 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41687 * config/avr/avr.opt (maccumulate-args): Likewise.
41688 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41689 * common.opt (freport-bug): Likewise.
41690 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41691 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41692 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41693 translatable string.
41694 * config/i386/i386.c (function_value_32): Likewise.
41695 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41696 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41698 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41699 * common/config/msp430/msp430-common.c (msp430_handle_option):
41701 * symtab.c (symtab_node::verify_base): Likewise.
41702 * opts.c (set_debug_level): Likewise.
41703 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41704 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41705 missing whitespace to translatable strings.
41706 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41707 * config/sh/superh.h: Likewise.
41708 * config/i386/xopintrin.h: Likewise.
41709 * config/i386/znver1.md: Likewise.
41710 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41711 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41712 * double-int.h (struct double_int): Likewise.
41713 * double-int.c (div_and_round_double): Likewise.
41714 * wide-int.cc: Likewise.
41715 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41716 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41717 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41718 (crossjumps_occurred): ... this.
41719 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41722 PR tree-optimization/78899
41723 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41724 returning bool return struct loop *, NULL for failure and the new
41726 (versionable_outer_loop_p): Don't version outer loop if it has
41727 dont_vectorized bit set.
41728 (tree_if_conversion): When versioning outer loop, ensure
41729 tree_if_conversion is performed also on the inner loop of the
41730 non-vectorizable outer loop copy.
41731 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41732 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41733 prevent vectorization of it.
41734 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41735 the outer loop vectorization of the non-scalar version is attempted
41736 before vectorization of the inner loop in scalar version. If
41737 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41738 vectorization of its inner loop.
41739 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41740 has 2 inner loops, rename also on edges from bb whose single pred
41741 is outer_loop->header. Fix typo in function comment.
41743 2017-01-09 Martin Sebor <msebor@redhat.com>
41746 * asan.c (asan_emit_stack_protection): Increase local buffer size
41747 to avoid snprintf truncation warning.
41749 2017-01-09 Andrew Pinski <apinski@cavium.com>
41751 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41752 to reference thunderx2t99 for the tuning structure
41753 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41755 (thunderx2t99_extra_costs): This.
41756 * config/aarch64/aarch64-tune.md: Regenerate.
41757 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41758 (vulcan_addrcost_table): This.
41759 (vulcan_regmove_cost): Rename to ...
41760 (thunderx2t99_regmove_cost): This.
41761 (vulcan_vector_cost): Rename to ...
41762 (thunderx2t99_vector_cost): this.
41763 (vulcan_branch_cost): Rename to ...
41764 (thunderx2t99_branch_cost): This.
41765 (vulcan_tunings): Rename to ...
41766 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41767 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41769 2017-01-09 Martin Jambor <mjambor@suse.cz>
41773 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41774 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41775 (propagate_vr_accross_jump_function): Use the above function for all
41776 value range computations for pass-through jump functions and type
41777 converasion from explicit value range values.
41778 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41779 parameters from TYPE_ARG_TYPES.
41780 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41781 (ipa_write_node_info): Stream type of the actual argument.
41782 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41784 2017-01-09 Martin Liska <mliska@suse.cz>
41787 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41788 (lookup_compiler): Do not show error message with have_E.
41790 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41792 PR tree-optimization/78938
41793 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41794 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41795 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41796 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41799 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41801 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41804 2017-01-09 Richard Biener <rguenther@suse.de>
41806 PR tree-optimization/78997
41807 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41808 name condition properly.
41810 2017-01-09 Richard Biener <rguenther@suse.de>
41813 * dwarf2out.c (is_cxx): New overload with context.
41814 (is_naming_typedef_decl): Use it.
41816 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41818 * invoke.texi (Option Summary): Correct spacing in option lists
41819 and add line breaks to fix over-long lines.
41821 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41823 PR middle-end/17660
41825 * extend.texi (Common Variable Attributes): Add xref to GCC
41826 Internals manual to explain mode attribute keywords.
41828 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41831 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41832 and Preprocessor Options.
41833 (Options for Linking): Document -pthread here....
41834 (RS/6000 and PowerPC Options): ...not here.
41835 (Solaris 2 Options): ...or here.
41836 * doc/cppopts.texi: Document -pthread.
41838 2017-01-08 Martin Sebor <msebor@redhat.com>
41840 PR middle-end/77708
41841 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41842 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41843 New member functions.
41844 (format_directive): Used them.
41846 (pass_sprintf_length::handle_gimple_call): Same.
41847 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41848 to avoid truncation for any argument.
41849 (extract_affine_mul): Same.
41850 * tree.c (get_file_function_name): Same.
41852 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41854 PR middle-end/77484
41855 * predict.def (PRED_INDIR_CALL): Set to 86.
41857 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41859 PR preprocessor/54124
41860 * doc/cppopts.texi: Reformat -d subtable to list the full name
41861 of the options. Add cross-reference to the docs for the general
41862 compiler -d options.
41863 * doc/invoke.texi (Developer Options): Add cross-reference to the
41864 preprocessor-specific -d option documentation.
41866 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41868 PR preprocessor/13498
41869 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41870 redudant material, and reflect new command-line options.
41871 (System Headers): Likewise.
41873 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41875 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41876 -isystem, and -idirafter. Copy-edit.
41877 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41878 default for -ftrack-macro-expansion. Delete obsolete and
41879 badly-formatted implementation details about -fdebug-cpp output.
41880 * doc/cppwarnopts.texi: Copy-edit.
41882 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41885 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41886 that the transition from a max line width >= 1<<10 to narrower
41887 lines works correctly.
41889 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41891 * doc/options.texi (PerFunction): New.
41892 * opt-functions.awk (switch_flags): Map both Optimization and
41893 PerFunction to CL_OPTIMIZATION.
41894 * opth-gen.awk: Test for PerFunction flag along with
41896 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41897 it only when the latter is present. Skip those that don't in
41898 the hash function generator.
41899 * common.opt (fvar-tracking): Mark as PerFunction instead of
41901 (fvar-tracking-assignments): Likewise.
41902 (fvar-tracking-assignments-toggle): Likewise.
41903 (fvar-tracking-uninit): Likewise.
41905 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41907 PR translation/79018
41908 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41911 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41914 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41915 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41916 TARGET_LONG_JUMP_TABLE_OFFSETS.
41917 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41918 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41919 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41920 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41921 * config/m68k/m68k.md (tablejump expander): Likewise.
41922 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41923 TARGET_LONG_JUMP_TABLE_OFFSETS.
41924 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41925 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41927 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41928 David Holsgrove <david.holsgrove@xilinx.com>
41930 * common/config/microblaze/microblaze-common.c
41931 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41932 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41934 * config/microblaze/microblaze.c (microblaze_must_save_register)
41935 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41937 (microblaze_eh_return): New function.
41938 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41939 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41940 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41941 * config/microblaze/microblaze.md (eh_return): New pattern.
41943 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41945 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41946 GCC_DIAGNOSTIC_STRINGIFY): Define.
41948 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41950 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41952 * config/arm/arm.md (<mcrr>): New.
41954 * config/arm/arm.c (arm_arch5te): New.
41955 (arm_option_override): Set arm_arch5te.
41956 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41958 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41959 (arm_mcrr_qualifiers): ... this. New.
41960 (MRRC_QUALIFIERS): Define to...
41961 (arm_mrrc_qualifiers): ... this. New.
41962 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41964 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41965 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41966 (MRRCI, mrrc, MRRC): New.
41967 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41968 VUNSPEC_MRRC2): New.
41970 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41972 * config/arm/arm.md (<mcr>): New.
41974 * config/arm/arm.c (arm_coproc_builtin_available): Add
41975 support for mcr, mrc, mcr2 and mrc2.
41976 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41977 (arm_mcr_qualifiers): ... this. New.
41978 (MRC_QUALIFIERS): Define to ...
41979 (arm_mrc_qualifiers): ... this. New.
41980 (MCR_QUALIFIERS): Define to ...
41981 (arm_mcr_qualifiers): ... this. New.
41982 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41984 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41985 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41986 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41987 VUNSPEC_MRC2): New.
41989 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41991 * config/arm/arm.md (*ldc): New.
41995 * config/arm/arm.c (arm_coproc_builtin_available): Add
41996 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41997 (arm_coproc_ldc_stc_legitimate_address): New.
41998 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41999 'qualifier_const_pointer'.
42000 (LDC_QUALIFIERS): Define to...
42001 (arm_ldc_qualifiers): ... this. New.
42002 (STC_QUALIFIERS): Define to...
42003 (arm_stc_qualifiers): ... this. New.
42004 * config/arm/arm-protos.h
42005 (arm_coproc_ldc_stc_legitimate_address): New.
42006 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
42007 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
42008 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
42009 stc2, stcl, stc2l): New.
42010 * config/arm/constraints.md (Uz): New.
42011 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
42012 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
42013 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
42014 VUNSPEC_STC2L): New.
42016 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42018 * config/arm/arm.md (<cdp>): New.
42019 * config/arm/arm.c (neon_const_bounds): Rename this ...
42020 (arm_const_bounds): ... this.
42021 (arm_coproc_builtin_available): New.
42022 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
42023 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
42024 (CDP_QUALIFIERS): Define to...
42025 (arm_cdp_qualifiers): ... this. New.
42027 (arm_expand_builtin_args): Add case for 6 arguments.
42028 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
42029 (arm_const_bounds): ... this.
42030 (arm_coproc_builtin_available): New.
42031 * config/arm/arm_acle.h (__arm_cdp): New.
42033 * config/arm/arm_acle_builtins.def (cdp): New.
42035 * config/arm/iterators.md (CDPI,CDP,cdp): New.
42036 * config/arm/neon.md: Rename all 'neon_const_bounds' to
42037 'arm_const_bounds'.
42038 * config/arm/types.md (coproc): New.
42039 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
42040 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
42041 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
42042 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
42044 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42046 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
42047 (UBINOP_QUALIFIERS): New.
42049 (acle_builtin_data): New. Change comment.
42050 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
42051 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
42052 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
42053 arm_acle_builtins.def.
42054 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
42055 (arm_init_acle_builtins): New.
42056 (CRC32_BUILTIN): Remove.
42057 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
42058 crc32cb, crc32ch and crc32cw.
42059 (arm_init_crc32_builtins): Remove.
42060 (arm_init_builtins): Use arm_init_acle_builtins rather
42061 than arm_init_crc32_builtins.
42062 (arm_expand_acle_builtin): New.
42063 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
42064 * config/arm/arm_acle_builtins.def: New.
42066 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42068 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
42069 (arm_builtin_datum): ... this.
42070 (arm_init_neon_builtin): Rename to ...
42071 (arm_init_builtin): ... this. Add a new parameters PREFIX
42072 and USE_SIG_IN_NAME.
42073 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
42074 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
42075 'arm_builtin_datum'.
42076 (arm_init_vfp_builtins): Likewise.
42077 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
42078 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
42079 (arm_expand_neon_args): Rename to ...
42080 (arm_expand_builtin_args): ... this. Rename builtin_arg
42081 enum values and differentiate between ARG_BUILTIN_MEMORY
42082 and ARG_BUILTIN_NEON_MEMORY.
42083 (arm_expand_neon_builtin_1): Rename to ...
42084 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
42085 values, arm_expand_builtin_args and add bool parameter NEON.
42086 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
42087 (arm_expand_vfp_builtin): Likewise.
42088 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
42090 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42092 PR middle-end/77484
42093 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
42094 * predict.c (tree_estimate_probability_bb): Reverse direction of
42095 polymorphic call predictor.
42097 2017-01-06 David Malcolm <dmalcolm@redhat.com>
42099 * passes.c (execute_one_pass): Split out pass-skipping logic into...
42100 (determine_pass_name_match): ...this new function and...
42101 (should_skip_pass_p): ...this new function.
42103 2017-01-06 Nathan Sidwell <nathan@acm.org>
42105 * ipa-visibility.c (function_and_variable_visibility): Reformat
42106 comments and long lines. Remove extrneous if.
42107 * symtab.c (symtab_node::make_decl_local): Fix code format.
42108 (symtab_node::set_section_for_node): Fix comment typo.
42110 2017-01-06 Martin Liska <mliska@suse.cz>
42113 * lra-constraints.c: Rename invariant to lra_invariant.
42114 * predict.c (set_even_probabilities): Initialize e to NULL.
42116 2017-01-05 Martin Sebor <msebor@redhat.com>
42118 PR tree-optimization/78910
42119 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
42120 (format_integer): Correct off-by-one error in the handling
42121 of precision with negative numbers in signed conversions..
42123 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
42125 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
42127 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42129 PR tree-optimization/71016
42130 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
42131 factor_out_conditional_conversion. Formatting fix.
42132 (factor_out_conditional_conversion): Add cond_stmt argument.
42133 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
42134 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
42137 2017-01-05 David Malcolm <dmalcolm@redhat.com>
42139 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
42140 read-rtl-function.o, and selftest-rtl.o.
42141 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
42142 (selftest::aarch64_test_loading_full_dump): New function.
42143 (selftest::aarch64_run_selftests): New function.
42144 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
42145 selftest::aarch64_run_selftests.
42146 * config/i386/i386.c
42147 (selftest::ix86_test_loading_dump_fragment_1): New function.
42148 (selftest::ix86_test_loading_call_insn): New function.
42149 (selftest::ix86_test_loading_full_dump): New function.
42150 (selftest::ix86_test_loading_unspec): New function.
42151 (selftest::ix86_run_selftests): Call the new functions.
42152 * emit-rtl.c (maybe_set_max_label_num): New function.
42153 * emit-rtl.h (maybe_set_max_label_num): New decl.
42154 * function.c (instantiate_decls): Guard call to
42155 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
42156 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
42158 * gensupport.c (gen_reader::gen_reader): Pass "false"
42159 for new "compact" param of rtx_reader.
42160 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
42161 rather than an empty string for NULL strings.
42162 * read-md.c: Potentially include config.h rather than bconfig.h.
42163 Wrap include of errors.h with #ifdef GENERATOR_FILE.
42164 (have_error): New global, copied from errors.c.
42165 (md_reader::read_name): Rename to...
42166 (md_reader::read_name_1): ...this, adding "out_loc" param,
42167 and converting "missing name or number" to returning false, rather
42169 (md_reader::read_name): Reimplement in terms of read_name_1.
42170 (md_reader::read_name_or_nil): New function.
42171 (md_reader::read_string): Handle "(nil)" by returning NULL.
42172 (md_reader::md_reader): Add new param "compact".
42173 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
42174 (md_reader::read_file): New method.
42175 * read-md.h (md_reader::md_reader): Add new param "compact".
42176 (md_reader::read_file): New method.
42177 (md_reader::is_compact): New accessor.
42178 (md_reader::read_name): Convert return type from void to file_location.
42179 (md_reader::read_name_or_nil): New decl.
42180 (md_reader::read_name_1): New decl.
42181 (md_reader::m_compact): New field.
42182 (noop_reader::noop_reader): Pass "false" for new "compact" param
42184 (rtx_reader::rtx_reader): Add new "compact" param.
42185 (rtx_reader::read_rtx_operand): Make virtual and convert return
42186 type from void to rtx.
42187 (rtx_reader::read_until): New decl.
42188 (rtx_reader::handle_any_trailing_information): New virtual function.
42189 (rtx_reader::postprocess): New virtual function.
42190 (rtx_reader::finalize_string): New virtual function.
42191 (rtx_reader::m_in_call_function_usage): New field.
42192 (rtx_reader::m_reuse_rtx_by_id): New field.
42193 * read-rtl-function.c: New file.
42194 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
42195 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
42196 (selftest::verify_three_block_rtl_cfg): New decl.
42197 * read-rtl-function.h: New file.
42198 * read-rtl.c: Potentially include config.h rather than bconfig.h.
42199 For host, include function.h, memmodel.h, and emit-rtl.h.
42200 (one_time_initialization): New function.
42201 (struct compact_insn_name): New struct.
42202 (compact_insn_names): New array.
42203 (find_code): Handle insn codes in compact dumps.
42204 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
42205 (bind_subst_iter_and_attr): Likewise.
42206 (add_condition_to_string): Likewise.
42207 (add_condition_to_rtx): Likewise.
42208 (apply_attribute_uses): Likewise.
42209 (add_current_iterators): Likewise.
42210 (apply_iterators): Likewise.
42211 (initialize_iterators): Guard usage of apply_subst_iterator with
42212 #ifdef GENERATOR_FILE.
42213 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
42214 (md_reader::read_mapping): Likewise.
42215 (add_define_attr_for_define_subst): Likewise.
42216 (add_define_subst_attr): Likewise.
42217 (read_subst_mapping): Likewise.
42218 (check_code_iterator): Likewise.
42219 (rtx_reader::read_rtx): Likewise. Move one-time initialization
42221 (one_time_initialization): New function.
42222 (rtx_reader::read_until): New method.
42223 (read_flags): New function.
42224 (parse_reg_note_name): New function.
42225 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42226 Handle reuse_rtx ids.
42227 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42228 Add parsing support for RTL dumps, mirroring the special-cases in
42229 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42230 values, and calling handle_any_trailing_information.
42231 (rtx_reader::read_rtx_operand): Convert return type from void
42232 to rtx, returning return_rtx. Handle case 'e'. Call
42233 finalize_string on XSTR and XTMPL fields.
42234 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42235 "(nil)" values were omitted. Call the postprocess vfunc on the
42237 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42238 class ctor. Initialize m_in_call_function_usage. Call
42239 one_time_initialization.
42240 * rtl-tests.c (selftest::test_uncond_jump): Call
42241 set_new_first_and_last_insn.
42242 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42243 * selftest-rtl.c: New file.
42244 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42245 (selftest::get_insn_by_uid): New decl.
42246 * selftest-run-tests.c (selftest::run_tests): Call
42247 read_rtl_function_c_tests.
42248 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42249 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42252 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42254 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42255 operands in a special way. Assert that pos+len <= mode precision.
42257 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42259 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42260 3 argument Alias with unlimited for the negative form.
42261 (fno-vect-cost-model): Removed.
42263 2017-01-05 Martin Liska <mliska@suse.cz>
42265 * hsa-gen.c (gen_hsa_divmod): New function.
42266 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42268 2017-01-05 Martin Liska <mliska@suse.cz>
42271 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42274 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42276 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42277 small constant length operands.
42279 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42281 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42282 between loop iterations.
42284 2017-01-05 Martin Liska <mliska@suse.cz>
42287 * gimplify.c (gimplify_decl_expr): Compare to
42288 asan_poisoned_variables instread of checking flags.
42289 (gimplify_target_expr): Likewise.
42290 (gimplify_expr): Likewise.
42291 (gimplify_function_tree): Conditionally initialize
42292 asan_poisoned_variables.
42294 2017-01-04 Jeff Law <law@redhat.com>
42296 PR tree-optimizatin/78812
42297 * rtl.h (contains_mem_rtx_p): Prototype.
42298 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42299 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42300 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42301 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42302 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42304 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42306 * input.c (assert_char_at_range): Default-initialize actual_range.
42308 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42310 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42311 to match the caller.
42313 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42315 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42316 insns after final jump in test to emit dummy move.
42318 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42320 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42321 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42323 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42325 * multiple_target.c (create_dispatcher_calls): Init e_next.
42326 * tree-ssa-loop-split.c (split_loop): Init border.
42327 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42330 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42335 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42336 (altivec_register_operand): Do not return true if the operand
42337 contains a SUBREG mixing SImode and SFmode.
42338 (vsx_register_operand): Likewise.
42339 (vsx_reg_sfsubreg_ok): New predicate.
42340 (vfloat_operand): Do not return true if the operand contains a
42341 SUBREG mixing SImode and SFmode.
42342 (vint_operand): Likewise.
42343 (vlogical_operand): Likewise.
42344 (gpc_reg_operand): Likewise.
42345 (int_reg_operand): Likewise.
42346 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42347 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42348 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42350 (rs6000_emit_move_si_sf_subreg): New helper function.
42351 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42352 fixup SUBREGs involving SImode and SFmode.
42353 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42354 numbers for the new peephole2 optimization.
42355 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42356 the GLIBC math library that do AND/IOR/XOR operations on single
42357 precision floating point.
42358 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42359 target macros to say whether we need to avoid SUBREGs mixing
42361 (TARGET_ALLOW_SF_SUBREG): Likewise.
42362 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42363 (UNSPEC_SI_FROM_SF): Likewise.
42364 (iorxor): Change spacing.
42365 (and_ior_xor): New iterator for AND, IOR, and XOR.
42366 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42367 (movdi_from_sf_zero_ext): Likewise.
42368 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42369 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42370 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42371 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42372 (fms<mode>4): Likewise.
42373 (fnma<mode>4): Likewise.
42374 (fnms<mode>4): Likewise.
42375 (nfma<mode>4): Likewise.
42376 (nfms<mode>4): Likewise.
42378 2017-01-04 Marek Polacek <polacek@redhat.com>
42381 * doc/invoke.texi: Document -Wpointer-compare.
42383 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42385 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42388 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42389 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42392 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42395 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42396 documentation of the powerpc_popcntb_ok attribute.
42397 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42398 code to issue warning messages if a requested CPU configuration is
42399 not supported by the binary (assembler and loader) toolchain.
42400 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42401 made to define a built-in function that has been disabled.
42402 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42403 made to define a built-in function that has been disabled.
42404 (altivec_init_builtins): Add comment explaining why definition
42405 of the DST built-in functions is not preceded by an assertion
42406 check. Add assertions to prevent ICE if attempts are made to
42407 define an altivec predicate or an abs* built-in function that has
42409 (htm_init_builtins): Add comment explaining why definition of the
42410 htm built-in functions is not preceded by an assertion check.
42412 2017-01-04 Jeff Law <law@redhat.com>
42414 PR tree-optimizatin/67955
42415 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42416 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42417 the points-to solution does not include pt_null. Use DECL_PT_UID
42420 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42422 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42423 Use gen_int_mode instead of gen_lopwart for const_int operands.
42425 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42427 PR tree-optimization/71563
42428 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42429 out of range value - has low bits known to be zero.
42431 2017-01-04 Alan Modra <amodra@gmail.com>
42433 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42434 * configure: Regenerate.
42435 * config.in: Regenerate.
42437 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42440 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42441 a substring of the message, but strcmp with the whole message. Ifdef
42442 ENABLE_NLS, translate the message first using dgettext.
42444 2017-01-03 Jeff Law <law@redhat.com>
42446 PR tree-optimizatin/78856
42447 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42448 (mark_threaded_blocks): Remove code to truncate thread paths that
42449 cross multiple loop headers. Instead invalidate the cached loop
42450 iteration information and handle case of a thread path walking
42451 into an irreducible region.
42453 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42456 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42457 assertions. Add support for doing the signbit if the IEEE 128-bit
42458 floating point value is in a GPR.
42459 * config/rs6000/rs6000.md (Fsignbit): Delete.
42460 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42461 Update the length attribute if the value is in a GPR.
42462 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42463 the sign or zero extension instruction, since the value is always 0/1.
42464 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42467 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42468 extracting SImode to a GPR register so that we can generate a
42469 store, limit the vector to be in a traditional Altivec register
42470 for the vextuwrx instruction.
42472 2017-01-03 Ian Lance Taylor <iant@google.com>
42474 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42476 2017-01-03 Martin Sebor <msebor@redhat.com>
42478 PR tree-optimization/78696
42479 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42480 precision. Use MPFR for %f for greater fidelity. Correct handling
42482 (pass_sprintf_length::compute_format_length): Set width and precision
42483 specified by asrerisk to void_node for vararg functions.
42484 (try_substitute_return_value): Adjust dump output.
42486 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42488 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42490 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42492 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42493 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42494 -mlra/-mno-lra was passed to the compiler.
42496 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42498 PR rtl-optimization/65618
42499 * emit-rtl.c (try_split): Move initialization of "before" and
42500 "after" to just before the call to emit_insn_after_setloc.
42502 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42504 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42506 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42508 * dwarf2out.c (gen_enumeration_type_die): When
42509 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42511 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42513 PR tree-optimization/78965
42514 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42515 Change first argument from const call_info & to call_info &. For %n
42516 set info.nowrite to false.
42518 PR middle-end/78901
42519 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42520 possibly throwing calls.
42522 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42523 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42524 and fns handling, rather than in a separate case SSA_NAME.
42526 2017-01-02 Jeff Law <law@redhat.com>
42528 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42529 fixes for first_period and second_period variables.
42531 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42534 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42535 (*insvqi_1): New insn pattern.
42536 (*insvqi_1_mem_rex64): Ditto.
42537 (*insvqi_2): Ditto.
42538 (*insvqi_3): Rename from *insvqi.
42540 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42542 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42544 * doc/cfg.texi (Edges): Remove reference to Java.
42545 (Maintaining the CFG): Ditto.
42547 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42549 PR middle-end/77674
42550 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42551 transparent aliases.
42553 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42555 PR middle-end/77484
42556 * predict.def (PRED_CALL): Update hitrate.
42557 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42558 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42559 into direct/indirect/polymorphic variants.
42561 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42563 Update copyright years.
42565 * gcc.c (process_command): Update copyright notice dates.
42566 * gcov-dump.c (print_version): Ditto.
42567 * gcov.c (print_version): Ditto.
42568 * gcov-tool.c (print_version): Ditto.
42569 * gengtype.c (create_file): Ditto.
42570 * doc/cpp.texi: Bump @copying's copyright year.
42571 * doc/cppinternals.texi: Ditto.
42572 * doc/gcc.texi: Ditto.
42573 * doc/gccint.texi: Ditto.
42574 * doc/gcov.texi: Ditto.
42575 * doc/install.texi: Ditto.
42576 * doc/invoke.texi: Ditto.
42578 Copyright (C) 2017 Free Software Foundation, Inc.
42580 Copying and distribution of this file, with or without modification,
42581 are permitted in any medium without royalty provided the copyright
42582 notice and this notice are preserved.