1 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
6 (vect_compile_time_alias): ...this new function. Do the calculation
7 on poly_ints rather than trees.
8 (vect_prune_runtime_alias_test_list): Update call accordingly.
10 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
11 Alan Hayward <alan.hayward@arm.com>
12 David Sherwood <david.sherwood@arm.com>
14 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
16 (vect_schedule_slp_instance): Likewise.
18 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19 Alan Hayward <alan.hayward@arm.com>
20 David Sherwood <david.sherwood@arm.com>
22 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
23 constant and extern definitions for variable-length vectors.
24 (vect_get_constant_vectors): Note that the number of units
25 is known to be constant.
27 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28 Alan Hayward <alan.hayward@arm.com>
29 David Sherwood <david.sherwood@arm.com>
31 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
32 of units as polynomial. Choose between WIDE and NARROW based
35 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
36 Alan Hayward <alan.hayward@arm.com>
37 David Sherwood <david.sherwood@arm.com>
39 * tree-vect-stmts.c (simd_clone_subparts): New function.
40 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
42 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
43 Alan Hayward <alan.hayward@arm.com>
44 David Sherwood <david.sherwood@arm.com>
46 * tree-vect-stmts.c (vectorizable_call): Treat the number of
47 vectors as polynomial. Use build_index_vector for
50 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
51 Alan Hayward <alan.hayward@arm.com>
52 David Sherwood <david.sherwood@arm.com>
54 * tree-vect-stmts.c (get_load_store_type): Treat the number of
55 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
56 for variable-length vectors.
57 (vectorizable_mask_load_store): Treat the number of units as
58 polynomial, asserting that it is constant if the condition has
59 already been enforced.
60 (vectorizable_store, vectorizable_load): Likewise.
62 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
63 Alan Hayward <alan.hayward@arm.com>
64 David Sherwood <david.sherwood@arm.com>
66 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
67 of units as polynomial. Punt if we can't tell at compile time
68 which vector contains the final result.
70 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
71 Alan Hayward <alan.hayward@arm.com>
72 David Sherwood <david.sherwood@arm.com>
74 * tree-vect-loop.c (vectorizable_induction): Treat the number
75 of units as polynomial. Punt on SLP inductions. Use an integer
76 VEC_SERIES_EXPR for variable-length integer reductions. Use a
77 cast of such a series for variable-length floating-point
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.h (build_index_vector): Declare.
85 * tree.c (build_index_vector): New function.
86 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
87 of units as polynomial, forcibly converting it to a constant if
88 vectorizable_reduction has already enforced the condition.
89 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
90 to create a {1,2,3,...} vector.
91 (vectorizable_reduction): Treat the number of units as polynomial.
92 Choose vectype_in based on the largest scalar element size rather
93 than the smallest number of units. Enforce the restrictions
96 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
97 Alan Hayward <alan.hayward@arm.com>
98 David Sherwood <david.sherwood@arm.com>
100 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
101 number of units as polynomial.
103 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
104 Alan Hayward <alan.hayward@arm.com>
105 David Sherwood <david.sherwood@arm.com>
107 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
108 * target.def (autovectorize_vector_sizes): Return the vector sizes
109 by pointer, using vector_sizes rather than a bitmask.
110 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
111 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
112 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
114 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
115 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
116 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
117 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
118 * omp-general.c (omp_max_vf): Likewise.
119 * omp-low.c (omp_clause_aligned_alignment): Likewise.
120 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
121 * tree-vect-loop.c (vect_analyze_loop): Likewise.
122 * tree-vect-slp.c (vect_slp_bb): Likewise.
123 * doc/tm.texi: Regenerate.
124 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
126 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
127 the vector size as a poly_uint64 rather than an unsigned int.
128 (current_vector_size): Change from an unsigned int to a poly_uint64.
129 (get_vectype_for_scalar_type): Update accordingly.
130 * tree.h (build_truth_vector_type): Take the size and number of
131 units as a poly_uint64 rather than an unsigned int.
132 (build_vector_type): Add a temporary overload that takes
133 the number of units as a poly_uint64 rather than an unsigned int.
134 * tree.c (make_vector_type): Likewise.
135 (build_truth_vector_type): Take the number of units as a poly_uint64
136 rather than an unsigned int.
138 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
139 Alan Hayward <alan.hayward@arm.com>
140 David Sherwood <david.sherwood@arm.com>
142 * target.def (get_mask_mode): Take the number of units and length
143 as poly_uint64s rather than unsigned ints.
144 * targhooks.h (default_get_mask_mode): Update accordingly.
145 * targhooks.c (default_get_mask_mode): Likewise.
146 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
147 * doc/tm.texi: Regenerate.
149 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
150 Alan Hayward <alan.hayward@arm.com>
151 David Sherwood <david.sherwood@arm.com>
153 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
154 * omp-general.c (omp_max_vf): Likewise.
155 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
156 (expand_omp_simd): Handle polynomial safelen.
157 * omp-low.c (omplow_simd_context): Add a default constructor.
158 (omplow_simd_context::max_vf): Change from int to poly_uint64.
159 (lower_rec_simd_input_clauses): Update accordingly.
160 (lower_rec_input_clauses): Likewise.
162 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
163 Alan Hayward <alan.hayward@arm.com>
164 David Sherwood <david.sherwood@arm.com>
166 * tree-vectorizer.h (vect_nunits_for_cost): New function.
167 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
168 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
169 (vect_analyze_slp_cost): Likewise.
170 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
171 (vect_model_load_cost): Likewise.
173 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
174 Alan Hayward <alan.hayward@arm.com>
175 David Sherwood <david.sherwood@arm.com>
177 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
178 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
179 from an unsigned int * to a poly_uint64_pod *.
180 (calculate_unrolling_factor): New function.
181 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
183 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
184 Alan Hayward <alan.hayward@arm.com>
185 David Sherwood <david.sherwood@arm.com>
187 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
188 from an unsigned int to a poly_uint64.
189 (_loop_vec_info::slp_unrolling_factor): Likewise.
190 (_loop_vec_info::vectorization_factor): Change from an int
192 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
193 (vect_get_num_vectors): New function.
194 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
195 (vect_get_num_copies): Use vect_get_num_vectors.
196 (vect_analyze_data_ref_dependences): Change max_vf from an int *
197 to an unsigned int *.
198 (vect_analyze_data_refs): Change min_vf from an int * to a
200 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
201 than an unsigned HOST_WIDE_INT.
202 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
203 (vect_analyze_data_ref_dependence): Change max_vf from an int *
204 to an unsigned int *.
205 (vect_analyze_data_ref_dependences): Likewise.
206 (vect_compute_data_ref_alignment): Handle polynomial vf.
207 (vect_enhance_data_refs_alignment): Likewise.
208 (vect_prune_runtime_alias_test_list): Likewise.
209 (vect_shift_permute_load_chain): Likewise.
210 (vect_supportable_dr_alignment): Likewise.
211 (dependence_distance_ge_vf): Take the vectorization factor as a
212 poly_uint64 rather than an unsigned HOST_WIDE_INT.
213 (vect_analyze_data_refs): Change min_vf from an int * to a
215 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
216 vfm1 as a poly_uint64 rather than an int. Make the same change
217 for the returned bound_scalar.
218 (vect_gen_vector_loop_niters): Handle polynomial vf.
219 (vect_do_peeling): Likewise. Update call to
220 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
221 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
223 * tree-vect-loop.c (vect_determine_vectorization_factor)
224 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
225 (vect_get_known_peeling_cost): Likewise.
226 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
227 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
228 (vect_transform_loop): Likewise. Use the lowest possible VF when
229 updating the upper bounds of the loop.
230 (vect_min_worthwhile_factor): Make static. Return an unsigned int
232 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
233 polynomial unroll factors.
234 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
235 (vect_make_slp_decision): Likewise.
236 (vect_supported_load_permutation_p): Likewise, and polynomial
238 (vect_analyze_slp_cost): Handle polynomial vf.
239 (vect_slp_analyze_node_operations): Likewise.
240 (vect_slp_analyze_bb_1): Likewise.
241 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
242 than an unsigned HOST_WIDE_INT.
243 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
244 (vectorizable_load): Handle polynomial vf.
245 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
247 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
249 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
250 Alan Hayward <alan.hayward@arm.com>
251 David Sherwood <david.sherwood@arm.com>
253 * match.pd: Handle bit operations involving three constants
254 and try to fold one pair.
256 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
258 * tree-vect-loop-manip.c: Include gimple-fold.h.
259 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
260 niters_maybe_zero parameters. Handle other cases besides a step of 1.
261 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
262 Add a path that uses a step of VF instead of 1, but disable it
264 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
265 and niters_no_overflow parameters. Update calls to
266 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
267 Create a new SSA name if the latter choses to use a ste other
268 than zero, and return it via niters_vector_mult_vf_var.
269 * tree-vect-loop.c (vect_transform_loop): Update calls to
270 vect_do_peeling, vect_gen_vector_loop_niters and
271 slpeel_make_loop_iterate_ntimes.
272 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
273 (vect_gen_vector_loop_niters): Update declarations after above changes.
275 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
277 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
278 128-bit round to integer instructions.
279 (ceil<mode>2): Likewise.
280 (btrunc<mode>2): Likewise.
281 (round<mode>2): Likewise.
283 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
285 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
286 unaligned VSX load/store on P8/P9.
287 (expand_block_clear): Allow the use of unaligned VSX
290 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
292 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
294 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
295 swap associated with both a load and a store.
297 2018-01-02 Andrew Waterman <andrew@sifive.com>
299 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
300 * config/riscv/riscv.md (clear_cache): Use it.
302 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
304 * web.c: Remove out-of-date comment.
306 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
308 * expr.c (fixup_args_size_notes): Check that any existing
309 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
310 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
311 (emit_single_push_insn): ...here.
313 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
315 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
316 (const_vector_encoded_nelts): New function.
317 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
318 (const_vector_int_elt, const_vector_elt): Declare.
319 * emit-rtl.c (const_vector_int_elt_1): New function.
320 (const_vector_elt): Likewise.
321 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
324 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
326 * expr.c: Include rtx-vector-builder.h.
327 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
328 directly on the tree encoding.
329 (const_vector_from_tree): Likewise.
330 * optabs.c: Include rtx-vector-builder.h.
331 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
332 sequence of "u" values.
333 * vec-perm-indices.c: Include rtx-vector-builder.h.
334 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
335 directly on the vec_perm_indices encoding.
337 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
339 * doc/rtl.texi (const_vector): Describe new encoding scheme.
340 * Makefile.in (OBJS): Add rtx-vector-builder.o.
341 * rtx-vector-builder.h: New file.
342 * rtx-vector-builder.c: Likewise.
343 * rtl.h (rtx_def::u2): Add a const_vector field.
344 (CONST_VECTOR_NPATTERNS): New macro.
345 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
346 (CONST_VECTOR_DUPLICATE_P): Likewise.
347 (CONST_VECTOR_STEPPED_P): Likewise.
348 (CONST_VECTOR_ENCODED_ELT): Likewise.
349 (const_vec_duplicate_p): Check for a duplicated vector encoding.
350 (unwrap_const_vec_duplicate): Likewise.
351 (const_vec_series_p): Check for a non-duplicated vector encoding.
352 Say that the function only returns true for integer vectors.
353 * emit-rtl.c: Include rtx-vector-builder.h.
354 (gen_const_vec_duplicate_1): Delete.
355 (gen_const_vector): Call gen_const_vec_duplicate instead of
356 gen_const_vec_duplicate_1.
357 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
358 (gen_const_vec_duplicate): Use rtx_vector_builder.
359 (gen_const_vec_series): Likewise.
360 (gen_rtx_CONST_VECTOR): Likewise.
361 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
362 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
363 Build a new vector rather than modifying a CONST_VECTOR in-place.
364 (handle_special_swappables): Update call accordingly.
365 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
366 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
367 Build a new vector rather than modifying a CONST_VECTOR in-place.
368 (handle_special_swappables): Update call accordingly.
370 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
372 * simplify-rtx.c (simplify_const_binary_operation): Use
373 CONST_VECTOR_ELT instead of XVECEXP.
375 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
377 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
378 the selector elements to be different from the data elements
379 if the selector is a VECTOR_CST.
380 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
381 ssizetype for the selector.
383 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
385 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
386 before testing each element individually.
387 * tree-vect-generic.c (lower_vec_perm): Likewise.
389 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
391 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
392 * selftest-run-tests.c (selftest::run_tests): Call it.
393 * vector-builder.h (vector_builder::operator ==): New function.
394 (vector_builder::operator !=): Likewise.
395 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
396 (vec_perm_indices::all_from_input_p): New function.
397 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
398 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
399 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
400 instead of reading the VECTOR_CST directly. Detect whether both
401 vector inputs are the same before constructing the vec_perm_indices,
402 and update the number of inputs argument accordingly. Use the
403 utility functions added above. Only construct sel2 if we need to.
405 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
407 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
408 the broadcast of the low byte.
409 (expand_mult_highpart): Use an explicit encoding for the permutes.
410 * optabs-query.c (can_mult_highpart_p): Likewise.
411 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
412 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
413 (vectorizable_bswap): Likewise.
414 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
415 explicit encoding for the power-of-2 permutes.
416 (vect_permute_store_chain): Likewise.
417 (vect_grouped_load_supported): Likewise.
418 (vect_permute_load_chain): Likewise.
420 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
422 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
423 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
424 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
425 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
426 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
427 (vect_gen_perm_mask_any): Likewise.
429 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
431 * int-vector-builder.h: New file.
432 * vec-perm-indices.h: Include int-vector-builder.h.
433 (vec_perm_indices): Redefine as an int_vector_builder.
434 (auto_vec_perm_indices): Delete.
435 (vec_perm_builder): Redefine as a stand-alone class.
436 (vec_perm_indices::vec_perm_indices): New function.
437 (vec_perm_indices::clamp): Likewise.
438 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
439 (vec_perm_indices::new_vector): New function.
440 (vec_perm_indices::new_expanded_vector): Update for new
441 vec_perm_indices class.
442 (vec_perm_indices::rotate_inputs): New function.
443 (vec_perm_indices::all_in_range_p): Operate directly on the
444 encoded form, without computing elided elements.
445 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
446 encoding. Update for new vec_perm_indices class.
447 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
448 the given vec_perm_builder.
449 (expand_vec_perm_var): Update vec_perm_builder constructor.
450 (expand_mult_highpart): Use vec_perm_builder instead of
451 auto_vec_perm_indices.
452 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
453 vec_perm_indices instead of auto_vec_perm_indices. Use a single
454 or double series encoding as appropriate.
455 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
456 vec_perm_indices instead of auto_vec_perm_indices.
457 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
458 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
459 (vect_permute_store_chain): Likewise.
460 (vect_grouped_load_supported): Likewise.
461 (vect_permute_load_chain): Likewise.
462 (vect_shift_permute_load_chain): Likewise.
463 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
464 (vect_transform_slp_perm_load): Likewise.
465 (vect_schedule_slp_instance): Likewise.
466 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
467 (vectorizable_mask_load_store): Likewise.
468 (vectorizable_bswap): Likewise.
469 (vectorizable_store): Likewise.
470 (vectorizable_load): Likewise.
471 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
472 vec_perm_indices instead of auto_vec_perm_indices. Use
473 tree_to_vec_perm_builder to read the vector from a tree.
474 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
475 vec_perm_builder instead of a vec_perm_indices.
476 (have_whole_vector_shift): Use vec_perm_builder and
477 vec_perm_indices instead of auto_vec_perm_indices. Leave the
478 truncation to calc_vec_perm_mask_for_shift.
479 (vect_create_epilog_for_reduction): Likewise.
480 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
481 from auto_vec_perm_indices to vec_perm_indices.
482 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
483 instead of changing individual elements.
484 (aarch64_vectorize_vec_perm_const): Use new_vector to install
485 the vector in d.perm.
486 * config/arm/arm.c (expand_vec_perm_d::perm): Change
487 from auto_vec_perm_indices to vec_perm_indices.
488 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
489 instead of changing individual elements.
490 (arm_vectorize_vec_perm_const): Use new_vector to install
491 the vector in d.perm.
492 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
493 Update vec_perm_builder constructor.
494 (rs6000_expand_interleave): Likewise.
495 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
496 (rs6000_expand_interleave): Likewise.
498 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
500 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
501 to qimode could truncate the indices.
502 * optabs.c (expand_vec_perm_var): Likewise.
504 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
506 * Makefile.in (OBJS): Add vec-perm-indices.o.
507 * vec-perm-indices.h: New file.
508 * vec-perm-indices.c: Likewise.
509 * target.h (vec_perm_indices): Replace with a forward class
511 (auto_vec_perm_indices): Move to vec-perm-indices.h.
512 * optabs.h: Include vec-perm-indices.h.
513 (expand_vec_perm): Delete.
514 (selector_fits_mode_p, expand_vec_perm_var): Declare.
515 (expand_vec_perm_const): Declare.
516 * target.def (vec_perm_const_ok): Replace with...
517 (vec_perm_const): ...this new hook.
518 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
519 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
520 * doc/tm.texi: Regenerate.
521 * optabs.def (vec_perm_const): Delete.
522 * doc/md.texi (vec_perm_const): Likewise.
523 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
524 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
525 expand_vec_perm for constant permutation vectors. Assert that
526 the mode of variable permutation vectors is the integer equivalent
527 of the mode that is being permuted.
528 * optabs-query.h (selector_fits_mode_p): Declare.
529 * optabs-query.c: Include vec-perm-indices.h.
530 (selector_fits_mode_p): New function.
531 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
532 is defined, instead of checking whether the vec_perm_const_optab
533 exists. Use targetm.vectorize.vec_perm_const instead of
534 targetm.vectorize.vec_perm_const_ok. Check whether the indices
535 fit in the vector mode before using a variable permute.
536 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
537 vec_perm_indices instead of an rtx.
538 (expand_vec_perm): Replace with...
539 (expand_vec_perm_const): ...this new function. Take the selector
540 as a vec_perm_indices rather than an rtx. Also take the mode of
541 the selector. Update call to shift_amt_for_vec_perm_mask.
542 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
543 Use vec_perm_indices::new_expanded_vector to expand the original
544 selector into bytes. Check whether the indices fit in the vector
545 mode before using a variable permute.
546 (expand_vec_perm_var): Make global.
547 (expand_mult_highpart): Use expand_vec_perm_const.
548 * fold-const.c: Includes vec-perm-indices.h.
549 * tree-ssa-forwprop.c: Likewise.
550 * tree-vect-data-refs.c: Likewise.
551 * tree-vect-generic.c: Likewise.
552 * tree-vect-loop.c: Likewise.
553 * tree-vect-slp.c: Likewise.
554 * tree-vect-stmts.c: Likewise.
555 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
557 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
558 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
559 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
560 (aarch64_vectorize_vec_perm_const): ...this new function.
561 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
562 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
563 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
564 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
565 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
566 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
567 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
569 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
570 check for NEON modes.
571 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
572 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
573 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
574 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
576 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
577 the old VEC_PERM_CONST conditions.
578 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
579 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
580 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
581 (ia64_vectorize_vec_perm_const_ok): Merge into...
582 (ia64_vectorize_vec_perm_const): ...this new function.
583 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
584 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
585 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
586 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
587 * config/mips/mips.c (mips_expand_vec_perm_const)
588 (mips_vectorize_vec_perm_const_ok): Merge into...
589 (mips_vectorize_vec_perm_const): ...this new function.
590 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
591 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
592 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
593 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
594 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
595 (rs6000_expand_vec_perm_const): Delete.
596 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
598 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
599 (altivec_expand_vec_perm_const_le): Take each operand individually.
600 Operate on constant selectors rather than rtxes.
601 (altivec_expand_vec_perm_const): Likewise. Update call to
602 altivec_expand_vec_perm_const_le.
603 (rs6000_expand_vec_perm_const): Delete.
604 (rs6000_vectorize_vec_perm_const_ok): Delete.
605 (rs6000_vectorize_vec_perm_const): New function.
606 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
607 an element count and rtx array.
608 (rs6000_expand_extract_even): Update call accordingly.
609 (rs6000_expand_interleave): Likewise.
610 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
611 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
612 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
613 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
614 (rs6000_expand_vec_perm_const): Delete.
615 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
616 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
617 (altivec_expand_vec_perm_const_le): Take each operand individually.
618 Operate on constant selectors rather than rtxes.
619 (altivec_expand_vec_perm_const): Likewise. Update call to
620 altivec_expand_vec_perm_const_le.
621 (rs6000_expand_vec_perm_const): Delete.
622 (rs6000_vectorize_vec_perm_const_ok): Delete.
623 (rs6000_vectorize_vec_perm_const): New function. Remove stray
624 reference to the SPE evmerge intructions.
625 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
626 an element count and rtx array.
627 (rs6000_expand_extract_even): Update call accordingly.
628 (rs6000_expand_interleave): Likewise.
629 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
630 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
632 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
634 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
636 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
637 vector mode and that that mode matches the mode of the data
639 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
640 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
641 directly using expand_vec_perm_1 when forcing selectors into
643 (expand_vec_perm_var): New function, split out from expand_vec_perm.
645 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
647 * optabs-query.h (can_vec_perm_p): Delete.
648 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
649 * optabs-query.c (can_vec_perm_p): Split into...
650 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
651 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
652 particular selector is valid.
653 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
654 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
655 (vect_grouped_load_supported): Likewise.
656 (vect_shift_permute_load_chain): Likewise.
657 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
658 (vect_transform_slp_perm_load): Likewise.
659 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
660 (vectorizable_bswap): Likewise.
661 (vect_gen_perm_mask_checked): Likewise.
662 * fold-const.c (fold_ternary_loc): Likewise. Don't take
663 implementations of variable permutation vectors into account
664 when deciding which selector to use.
665 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
666 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
667 with a false third argument.
668 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
669 to test whether the constant selector is valid and can_vec_perm_var_p
670 to test whether a variable selector is valid.
672 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
674 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
675 * optabs-query.c (can_vec_perm_p): Likewise.
676 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
677 instead of vec_perm_indices.
678 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
679 (vect_gen_perm_mask_checked): Likewise,
680 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
681 (vect_gen_perm_mask_checked): Likewise,
683 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
685 * optabs-query.h (qimode_for_vec_perm): Declare.
686 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
687 (qimode_for_vec_perm): ...this new function.
688 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
690 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
692 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
693 does not have a conditional at the top.
695 2018-01-02 Richard Biener <rguenther@suse.de>
697 * ipa-inline.c (big_speedup_p): Fix expression.
699 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
702 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
705 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
709 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
710 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
711 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
712 cond_taken_branch_cost 3->4.
714 2018-01-01 Jakub Jelinek <jakub@redhat.com>
716 PR tree-optimization/83581
717 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
718 TODO_cleanup_cfg if any changes have been made.
721 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
722 convert_modes if target mode has the right side, but different mode
726 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
727 last argument when extracting from CONCAT. If either from_real or
728 from_imag is NULL, use expansion through memory. If result is not
729 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
730 the parts directly to inner mode, if even that fails, use expansion
734 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
735 check for bswap in mode rather than HImode and use that in expand_unop
738 2017-12-31 Jakub Jelinek <jakub@redhat.com>
741 * config/i386/i386.c (ix86_attribute_table): Remove excess
742 initializer for "shared" attribute.
744 2017-12-30 Tom de Vries <tom@codesourcery.com>
747 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
749 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
750 offload_funcs entries that no longer have a corresponding cgraph_node.
751 Mark the remaining ones as DECL_PRESERVE_P.
752 (output_lto): Call prune_offload_funcs.
754 2017-12-30 Jakub Jelinek <jakub@redhat.com>
756 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
757 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
758 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
759 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
760 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
761 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
762 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
763 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
764 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
765 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
766 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
767 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
768 vaesenc_<mode>, vpclmulqdq_<mode>,
769 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
771 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
773 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
774 _Float<N> and _Float<N>X variants for rounding built-in
776 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
777 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
778 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
779 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
780 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
781 * builtins.c (mathfn_built_in_2): Likewise.
782 * internal-fn.def (CEIL): Likewise.
784 (NEARBYINT): Likewise.
788 * convert.c (convert_to_integer_1): Likewise.
789 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
790 (integer_valued_real_call_p): Likewise.
791 * fold-const-call.c (fold_const_call_ss): Likewise.
792 * gencfn-macros.c (print_case_cfn): Change CFN and operator
793 printers to take a const char * suffix instead of a bool.
794 (print_define_operator_list): Likewise.
795 (fltall_suffixes): New list of suffixes, that include the
796 traditional suffixes as well as all of the _Float<N> and
798 (main): For _Float<N> and _Float<N>X functions, emit both
799 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
800 has the _Float<N> and _Float<N>X case names or operators. The
801 <name>_ALL variant has both the traditional and the
802 _Float<N>/_Float<N>X case names or operators.
803 * match.pd (COPYSIGN optimizations): Provide optimizations for
804 _Float<N> and _Float<N>X types where possible.
805 (MIN/MAX optimizations): Likewise.
806 (sqrt optimizations): Likewise.
807 (rounding optimizations): Likewise.
809 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
811 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
812 gen_int_mode rather than GEN_INT.
813 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
815 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
817 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
818 instead of CONSTANT_P.
819 (gen_vec_series): Likewise.
820 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
822 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
824 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
827 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
829 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
830 (aarch64_ld1x2<VDC:mode>): Likewise.
831 (aarch64_simd_ld1<mode>_x2): Likewise.
832 (aarch64_simd_ld1<mode>_x2): Likewise.
833 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
834 (vld1_s8_x2): Likewise.
835 (vld1_u16_x2): Likewise.
836 (vld1_s16_x2): Likewise.
837 (vld1_u32_x2): Likewise.
838 (vld1_s32_x2): Likewise.
839 (vld1_u64_x2): Likewise.
840 (vld1_s64_x2): Likewise.
841 (vld1_f16_x2): Likewise.
842 (vld1_f32_x2): Likewise.
843 (vld1_f64_x2): Likewise.
844 (vld1_p8_x2): Likewise.
845 (vld1_p16_x2): Likewise.
846 (vld1_p64_x2): Likewise.
847 (vld1q_u8_x2): Likewise.
848 (vld1q_s8_x2): Likewise.
849 (vld1q_u16_x2): Likewise.
850 (vld1q_s16_x2): Likewise.
851 (vld1q_u32_x2): Likewise.
852 (vld1q_s32_x2): Likewise.
853 (vld1q_u64_x2): Likewise.
854 (vld1q_s64_x2): Likewise.
855 (vld1q_f16_x2): Likewise.
856 (vld1q_f32_x2): Likewise.
857 (vld1q_f64_x2): Likewise.
858 (vld1q_p8_x2): Likewise.
859 (vld1q_p16_x2): Likewise.
860 (vld1q_p64_x2): Likewise.
862 2017-12-27 Martin Liska <mliska@suse.cz>
864 PR tree-optimization/83552
865 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
866 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
868 2017-12-27 Tom de Vries <tom@codesourcery.com>
870 * config/nvptx/nvptx.c (nvptx_option_override): Disable
871 -gstatement-frontiers.
873 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
875 PR rtl-optimization/83513
876 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
877 before priority comparison.
879 2017-12-25 Jakub Jelinek <jakub@redhat.com>
882 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
883 ix86_isa_flags2 to ix86_isa_flags.
884 * config/i386/i386-c.c (ix86_target_macros_internal): Test
885 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
886 isa_flags rather than isa_flags2.
887 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
888 and -mavx512bitalg from isa2_opts to isa_opts.
889 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
890 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
891 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
893 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
894 OPTION_MASK_ISA_AVX512F ored with another option similarly to
895 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
896 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
897 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
898 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
899 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
900 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
901 builtins from bdesc_args2 to bdesc_args section.
902 (__builtin_ia32_compressstoreuqi512_mask,
903 __builtin_ia32_compressstoreuhi512_mask,
904 __builtin_ia32_compressstoreuqi256_mask,
905 __builtin_ia32_expandloadqi512_mask,
906 __builtin_ia32_expandloadqi512_maskz,
907 __builtin_ia32_expandloadhi512_mask,
908 __builtin_ia32_expandloadhi512_maskz,
909 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
910 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
911 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
912 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
913 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
914 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
915 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
916 __builtin_ia32_vpshldv_v32hi_maskz,
917 __builtin_ia32_vpopcountb_v64qi_mask,
918 __builtin_ia32_vpopcountw_v32hi_mask,
919 __builtin_ia32_vpshufbitqmb512_mask,
920 __builtin_ia32_vpshufbitqmb256_mask): Add
921 " | OPTION_MASK_ISA_AVX512BW".
922 (__builtin_ia32_expandloadqi256_mask,
923 __builtin_ia32_expandloadqi256_maskz,
924 __builtin_ia32_vpopcountb_v32qi_mask): Add
925 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
926 (__builtin_ia32_expandloadhi256_mask,
927 __builtin_ia32_expandloadhi256_maskz,
928 __builtin_ia32_expandloadqi128_mask,
929 __builtin_ia32_expandloadqi128_maskz,
930 __builtin_ia32_expandloadhi128_mask,
931 __builtin_ia32_expandloadhi128_maskz,
932 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
933 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
934 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
935 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
936 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
937 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
938 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
939 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
940 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
941 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
942 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
943 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
944 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
945 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
946 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
947 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
948 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
949 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
950 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
951 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
952 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
953 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
954 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
955 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
956 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
957 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
958 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
959 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
960 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
961 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
962 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
963 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
964 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
965 __builtin_ia32_vpopcountw_v8hi_mask): Add
966 " | OPTION_MASK_ISA_AVX512VL".
967 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
968 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
969 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
970 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
971 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
972 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
973 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
974 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
975 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
976 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
977 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
978 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
979 _mm_mask_bitshuffle_epi64_mask): Likewise.
980 * common/config/i386/i386-common.c
981 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
982 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
983 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
984 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
985 OPTION_MASK_ISA_AVX512BITALG_UNSET.
986 (OPTION_MASK_ISA2_AVX512F_UNSET,
987 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
988 (ix86_handle_option): For -mno-general-regs-only, clear from
989 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
990 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
991 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
992 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
993 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
994 for moving from ix86_isa_flags2 to ix86_isa_flags.
996 2017-12-23 Jakub Jelinek <jakub@redhat.com>
999 * fold-const.c (struct contains_label_data): New type.
1000 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
1001 inside of a SWITCH_BODY seen during the walk.
1002 (contains_label_p): Use walk_tree instead of
1003 walk_tree_without_duplicates, prepare data for contains_label_1 and
1006 2017-12-22 Jakub Jelinek <jakub@redhat.com>
1009 * tree-iterator.c (alloc_stmt_list): Start with cleared
1010 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
1014 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
1015 enable avx512f explicitly in #pragma GCC target.
1016 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
1017 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
1018 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
1019 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
1020 __builtin_ia32_vpdpbusds_v8si_mask,
1021 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
1022 __builtin_ia32_vpdpbusds_v4si_mask,
1023 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
1024 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
1025 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
1026 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
1027 __builtin_ia32_vpdpwssds_v8si_mask,
1028 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
1029 __builtin_ia32_vpdpwssds_v4si_mask,
1030 __builtin_ia32_vpdpwssds_v4si_maskz): Use
1031 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
1032 just OPTION_MASK_ISA_AVX512VNNI.
1034 2017-12-22 Martin Jambor <mjambor@suse.cz>
1037 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
1040 2017-12-22 Julia Koval <julia.koval@intel.com>
1041 Sebastian Peryt <sebastian.peryt@intel.com>
1043 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
1044 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
1045 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
1046 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
1047 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
1048 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
1049 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
1050 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
1051 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
1052 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
1053 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
1054 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
1055 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
1056 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
1057 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
1058 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
1059 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
1060 _mm256_maskz_popcnt_epi64): New intrinsics.
1061 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
1062 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
1063 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
1064 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
1065 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
1066 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
1067 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
1068 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
1069 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
1070 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
1071 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
1072 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
1073 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
1074 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
1075 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1076 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
1077 __builtin_ia32_vpshufbitqmb256_mask,
1078 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
1079 * config/i386/i386-c.c (__AVX512BITALG__): New.
1080 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
1081 (ix86_valid_target_attribute_inner_p): Ditto.
1082 (ix86_expand_args_builtin): Handle new types.
1083 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
1084 * config/i386/i386.opt: Add -mavx512bitalg.
1085 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
1086 avx512bitalgintrin.h.
1087 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
1088 (vpopcount<mode><mask_name>): Add more types.
1089 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
1090 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
1092 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1094 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
1095 Or in OPTION_MASK_ISA_AVX512F_SET.
1096 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1097 OPTION_MASK_ISA_AVX512VNNI_UNSET.
1098 (ix86_handle_option): Adjust for
1099 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
1100 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
1102 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
1103 isa_flag instead of isa_flag2.
1104 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
1105 isa_opts2 to isa_opts.
1106 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
1109 2017-12-22 Mike Stump <mikestump@comcast.net>
1110 Eric Botcazou <ebotcazou@adacore.com>
1112 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
1114 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1117 * config/i386/i386.c (ix86_function_arg_boundary): Return
1118 PARM_BOUNDARY for TYPE_EMPTY_P types.
1121 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
1122 if navail is >= dir.len.
1124 2017-12-21 Steve Ellcey <sellcey@cavium.com>
1126 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
1129 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1131 PR rtl-optimization/80747
1132 PR rtl-optimization/83512
1133 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
1134 succ edge from ENTRY, copy partition from e->dest to the newly
1136 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
1137 ENTRY, use BB_PARTITION of its successor block as current_partition.
1138 Don't copy partition when splitting succ edge from ENTRY.
1140 PR tree-optimization/83523
1141 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
1142 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
1143 (convert_mult_to_fma): Likewise.
1145 PR tree-optimization/83521
1146 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
1147 gimple_build_assign without code on result of
1148 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
1149 a VIEW_CONVERT_EXPR.
1151 2017-12-21 Andrew Pinski <apinski@cavium.com>
1152 Steve Ellcey <sellcey@cavium.com>
1154 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
1155 multi-arch for ilp32.
1157 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
1160 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1161 constraints to operand 2.
1162 (*ashl<mode>3_mask_1): Ditto.
1163 (*<shift_insn><mode>3_mask): Ditto.
1164 (*<shift_insn><mode>3_mask_1): Ditto.
1165 (*<rotate_insn><mode>3_mask): Ditto.
1166 (*<rotate_insn><mode>3_mask_1): Ditto.
1168 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1170 * reorg.c (make_return_insns): Reemit each insn with its own location.
1172 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1174 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1175 generation for cases where splatting a value is not useful.
1176 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1177 across a vec_duplicate and a paradoxical subreg forming a vector
1178 mode to a vec_concat.
1180 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1182 * combine.c (simplify_set): Do not transform subregs to zero_extends
1183 if the destination is not a scalar int mode.
1185 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1188 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1189 shared zero if the input has overflowed.
1191 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1193 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1194 support for Cortex-A55 and Cortex-A75.
1196 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1198 * common/config/arm/arm-common.c (compare_opt_names): Add function
1199 comment. Use strcmp instead of manual loop.
1201 2017-12-21 Martin Liska <mliska@suse.cz>
1203 PR gcov-profile/83509
1204 * gcov-dump.c (dump_gcov_file): Do not read info about
1205 support_unexecuted_blocks for gcda files.
1207 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1209 PR rtl-optimization/82973
1210 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1211 (valid_for_const_vector_p): ... this.
1212 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1213 (valid_for_const_vector_p): ... this. Adjust function comment.
1214 (gen_vec_duplicate): Adjust caller.
1215 * optabs.c (expand_vector_broadcast): Likewise.
1216 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1217 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1219 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1220 instead of GET_CODE == CONST_FIXED.
1221 (simplify_subreg): Use CONST_FIXED_P macro instead of
1222 GET_CODE == CONST_FIXED.
1225 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1226 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1227 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1228 (ix86_option_override_internal): Adjust for
1229 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1230 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1231 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1232 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1233 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1234 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1235 def_builtin2 for CET builtins.
1236 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1237 up formatting in IX86_BUILTIN_RDPID code.
1238 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1239 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1240 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1241 ix86_isa_flags2 to ix86_isa_flags.
1242 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1244 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1245 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1246 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1248 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1249 Or in OPTION_MASK_ISA_AVX512F_SET.
1250 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1251 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1252 (ix86_handle_option): Adjust for
1253 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1254 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
1256 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1257 Alan Hayward <alan.hayward@arm.com>
1258 David Sherwood <david.sherwood@arm.com>
1260 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
1261 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1262 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1263 Track polynomial offsets.
1265 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1266 Alan Hayward <alan.hayward@arm.com>
1267 David Sherwood <david.sherwood@arm.com>
1269 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1270 Treat drb->init as a poly_int. Fail if its misalignment wrt
1271 vector_alignment isn't known.
1273 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1274 Alan Hayward <alan.hayward@arm.com>
1275 David Sherwood <david.sherwood@arm.com>
1277 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1279 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1280 (vect_loop_versioning): Take the loop versioning threshold as a
1282 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1283 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1284 versioning_threshold.
1285 (vect_analyze_loop_2): Compute the loop versioning threshold
1286 whenever loop versioning is needed, and store it in the new
1287 field rather than combining it with the cost model threshold.
1288 (vect_transform_loop): Update call to vect_loop_versioning.
1289 Try to combine the loop versioning and cost thresholds here.
1291 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1292 Alan Hayward <alan.hayward@arm.com>
1293 David Sherwood <david.sherwood@arm.com>
1295 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1296 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1297 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1298 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1299 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1300 (group_compare_offset): Update accordingly.
1301 (split_small_address_groups_p): Likewise.
1302 (record_use): Take addr_offset as a poly_uint64 rather than
1303 an unsigned HOST_WIDE_INT.
1304 (strip_offset): Return the offset as a poly_uint64 rather than
1305 an unsigned HOST_WIDE_INT.
1306 (record_group_use, split_address_groups): Track polynomial offsets.
1307 (add_iv_candidate_for_use): Likewise.
1308 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1309 than a HOST_WIDE_INT.
1310 (strip_offset_1): Return the offset as a poly_int64 rather than
1313 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1314 Alan Hayward <alan.hayward@arm.com>
1315 David Sherwood <david.sherwood@arm.com>
1317 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1318 rather than a HOST_WIDE_INT.
1319 * tree.c (get_binfo_at_offset): Likewise.
1321 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1322 Alan Hayward <alan.hayward@arm.com>
1323 David Sherwood <david.sherwood@arm.com>
1325 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1326 rather than a HOST_WIDE_INT.
1327 * tree-sra.c (build_ref_for_offset): Likewise.
1329 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1330 Alan Hayward <alan.hayward@arm.com>
1331 David Sherwood <david.sherwood@arm.com>
1333 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1335 * tree.c (mem_ref_offset): Likewise.
1336 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1337 * builtins.c (get_object_alignment_2): Likewise.
1338 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1339 * gimple-fold.c (get_base_constructor): Likewise.
1340 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1341 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1343 * ipa-polymorphic-call.c
1344 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1345 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1346 (get_ancestor_addr_info): Likewise.
1347 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1348 * match.pd: Likewise.
1349 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1350 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1351 * tree-eh.c (tree_could_trap_p): Likewise.
1352 * tree-object-size.c (addr_object_size): Likewise.
1353 * tree-ssa-address.c (copy_ref_info): Likewise.
1354 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1355 (indirect_refs_may_alias_p): Likewise.
1356 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1357 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1358 (non_rewritable_mem_ref_base): Likewise.
1359 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1360 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1361 * varasm.c (decode_addr_const): Likewise.
1363 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1364 Alan Hayward <alan.hayward@arm.com>
1365 David Sherwood <david.sherwood@arm.com>
1367 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1368 offsets for MEM_REFs.
1370 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1371 Alan Hayward <alan.hayward@arm.com>
1372 David Sherwood <david.sherwood@arm.com>
1374 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1375 as a poly_uint64 rather than an unsigned int.
1376 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1378 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1379 Alan Hayward <alan.hayward@arm.com>
1380 David Sherwood <david.sherwood@arm.com>
1382 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1384 (decode_addr_const): Update accordingly.
1386 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1387 Alan Hayward <alan.hayward@arm.com>
1388 David Sherwood <david.sherwood@arm.com>
1390 * tree.h (bit_field_size, bit_field_offset): New functions.
1391 * hsa-gen.c (gen_hsa_addr): Use them.
1392 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1393 (simplify_vector_constructor): Likewise.
1394 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1395 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1396 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1397 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1400 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1401 Alan Hayward <alan.hayward@arm.com>
1402 David Sherwood <david.sherwood@arm.com>
1404 * expr.h (emit_group_load, emit_group_load_into_temps)
1405 (emit_group_store): Take the size as a poly_int64 rather than an int.
1406 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1407 (emit_group_load_into_temp, emit_group_store): Likewise.
1409 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1410 Alan Hayward <alan.hayward@arm.com>
1411 David Sherwood <david.sherwood@arm.com>
1413 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1414 an unsigned int to a poly_uint64.
1415 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1416 sizes as poly_uint64s rather than unsigned ints.
1417 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1420 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1421 Alan Hayward <alan.hayward@arm.com>
1422 David Sherwood <david.sherwood@arm.com>
1424 * lra-constraints.c (emit_inc): Change inc_amount from an int
1427 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1428 Alan Hayward <alan.hayward@arm.com>
1429 David Sherwood <david.sherwood@arm.com>
1431 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1433 (add_stack_var, stack_var_cmp, partition_stack_vars)
1434 (dump_stack_var_partition): Update accordingly.
1435 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1436 than a HOST_WIDE_INT.
1437 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1438 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1439 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1440 rather than a HOST_WIDE_INT.
1442 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1443 Alan Hayward <alan.hayward@arm.com>
1444 David Sherwood <david.sherwood@arm.com>
1446 * target.def (return_pops_args): Treat both the input and output
1447 sizes as poly_int64s rather than HOST_WIDE_INTS.
1448 * targhooks.h (default_return_pops_args): Update accordingly.
1449 * targhooks.c (default_return_pops_args): Likewise.
1450 * doc/tm.texi: Regenerate.
1451 * emit-rtl.h (incoming_args): Change pops_args, size and
1452 outgoing_args_size from int to poly_int64_pod.
1453 * function.h (expr_status): Change x_pending_stack_adjust and
1454 x_stack_pointer_delta from int to poly_int64.
1455 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1456 (ARGS_SIZE_RTX): Update accordingly.
1457 * calls.c (highest_outgoing_arg_in_use): Change from int to
1459 (stack_usage_watermark, stored_args_watermark): New variables.
1460 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1461 (emit_call_1): Change the stack_size and rounded_stack_size
1462 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1464 (save_fixed_argument_area): Check stack_usage_watermark.
1465 (initialize_argument_information): Change old_pending_adj from
1466 a HOST_WIDE_INT * to a poly_int64_pod *.
1467 (compute_argument_block_size): Return the size as a poly_int64
1469 (finalize_must_preallocate): Track polynomial argument sizes.
1470 (compute_argument_addresses): Likewise.
1471 (internal_arg_pointer_based_exp): Track polynomial offsets.
1472 (mem_overlaps_already_clobbered_arg_p): Rename to...
1473 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1474 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1475 Check stored_args_used_watermark.
1476 (load_register_parameters): Update accordingly.
1477 (check_sibcall_argument_overlap_1): Likewise.
1478 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1479 args size as a poly_int64 rather than an int. Return a bool
1480 indicating whether the optimization was possible and return
1481 the new adjustment by reference.
1482 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1483 Update stored_args_watermark.
1484 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1485 (expand_call): Likewise. Maintain stack_usage_watermark and
1486 stored_args_watermark. Update calls to
1487 combine_pending_stack_adjustment_and_call.
1488 (emit_library_call_value_1): Handle polynomial argument sizes.
1489 Call stack_region_maybe_used_p and mark_stack_region_used.
1490 Maintain stack_usage_watermark.
1491 (store_one_arg): Likewise. Update call to
1492 mem_overlaps_already_clobbered_arg_p.
1493 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1495 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1496 * config/microblaze/microblaze.c (microblaze_function_prologue):
1498 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1499 TARGET_RETURN_POPS_ARGS interface.
1500 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1502 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1503 * config/i386/i386.c (ix86_return_pops_args): Update for new
1504 TARGET_RETURN_POPS_ARGS interface.
1505 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1506 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1507 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1508 TARGET_RETURN_POPS_ARGS interface.
1509 * config/vax/vax.c (vax_return_pops_args): Likewise.
1510 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1511 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1512 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1513 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1514 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1515 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1516 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1517 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1518 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1519 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1520 and x_stack_pointer_delta from int to poly_int64.
1521 * dojump.c (do_pending_stack_adjust): Update accordingly.
1522 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1523 stack_pointer_deltas.
1524 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1525 (pad_to_arg_alignment): Track polynomial offsets.
1526 (assign_parm_find_stack_rtl): Likewise.
1527 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1528 * toplev.c (output_stack_usage): Update reference to
1529 current_function_pushed_stack_size.
1531 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1532 Alan Hayward <alan.hayward@arm.com>
1533 David Sherwood <david.sherwood@arm.com>
1535 * function.c (in_arg_offset, var_offset, dynamic_offset)
1536 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1537 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1538 rather than a HOST_WIDE_INT.
1539 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1540 (instantiate_virtual_regs_in_insn): Likewise.
1542 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1543 Alan Hayward <alan.hayward@arm.com>
1544 David Sherwood <david.sherwood@arm.com>
1546 * rtl.h (get_args_size, add_args_size_note): New functions.
1547 (find_args_size_adjust): Return a poly_int64 rather than a
1549 (fixup_args_size_notes): Likewise. Make the same change to the
1550 end_args_size parameter.
1551 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1552 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1553 * calls.c (emit_call_1): Likewise.
1554 * explow.c (adjust_stack_1): Likewise.
1555 * cfgcleanup.c (old_insns_match_p): Update use of
1556 find_args_size_adjust.
1557 * combine.c (distribute_notes): Track polynomial arg sizes.
1558 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1559 end_true_args_size, beg_delay_args_size and end_delay_args_size
1560 from HOST_WIDE_INT to poly_int64.
1561 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1562 than a HOST_WIDE_INT.
1563 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1564 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1565 polynomial arg sizes.
1566 * emit-rtl.c (try_split): Use get_args_size.
1567 * recog.c (peep2_attempt): Likewise.
1568 * reload1.c (reload_as_needed): Likewise.
1569 * expr.c (find_args_size_adjust): Return the adjustment as a
1570 poly_int64 rather than a HOST_WIDE_INT.
1571 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1572 to a poly_int64 and change the return type in the same way.
1573 (emit_single_push_insn): Track polynomial arg sizes.
1575 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1576 Alan Hayward <alan.hayward@arm.com>
1577 David Sherwood <david.sherwood@arm.com>
1579 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1580 from HOST_WIDE_INT to poly_int64.
1581 * expr.c (push_block, emit_push_insn): Likewise.
1583 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1584 Alan Hayward <alan.hayward@arm.com>
1585 David Sherwood <david.sherwood@arm.com>
1587 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1589 (get_frame_size): Return the size as a poly_int64 rather than a
1591 (frame_offset_overflow): Take the offset as a poly_int64 rather
1592 than a HOST_WIDE_INT.
1593 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1594 (assign_stack_temp): Likewise for the size.
1595 * function.c (get_frame_size): Return a poly_int64 rather than
1597 (frame_offset_overflow): Take the offset as a poly_int64 rather
1598 than a HOST_WIDE_INT.
1599 (try_fit_stack_local): Take the start, length and size as poly_int64s
1600 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1601 rather than a HOST_WIDE_INT.
1602 (add_frame_space): Take the start and end as poly_int64s rather than
1604 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1605 (assign_stack_temp): Likewise for the size.
1606 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1608 (find_temp_slot_from_address): Handle polynomial offsets.
1609 (combine_temp_slots): Likewise.
1610 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1612 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1613 poly_int64 rather than a HOST_WIDE_INT.
1614 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1615 than a HOST_WIDE_INT.
1616 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1617 polynomial frame offsets.
1618 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1619 as a poly_int64 rather than an int.
1620 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1621 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1622 * config/v850/v850.c (compute_frame_size): Likewise.
1623 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1624 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1625 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1626 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1627 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1628 rather than a HOST_WIDE_INT.
1629 * explow.c (get_dynamic_stack_base): Likewise.
1630 * final.c (final_start_function): Use the constant lower bound
1631 of the frame size for -Wframe-larger-than.
1632 * ira.c (do_reload): Adjust for new get_frame_size return type.
1633 * lra.c (lra): Likewise.
1634 * reload1.c (reload): Likewise.
1635 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1636 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1637 * rtlanal.c (get_initial_register_offset): Return the offset as
1638 a poly_int64 rather than a HOST_WIDE_INT.
1640 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1641 Alan Hayward <alan.hayward@arm.com>
1642 David Sherwood <david.sherwood@arm.com>
1644 * reload1.c (elim_table): Change initial_offset, offset and
1645 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1646 (offsets_at): Change the target array's element type from
1647 HOST_WIDE_INT to poly_int64_pod.
1648 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1649 (elimination_costs_in_insn, update_eliminable_offsets)
1650 (verify_initial_elim_offsets, set_offsets_for_label)
1651 (init_eliminable_invariants): Update after above changes.
1653 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1654 Alan Hayward <alan.hayward@arm.com>
1655 David Sherwood <david.sherwood@arm.com>
1657 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1658 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1659 (decomposition): Change start and end from HOST_WIDE_INT
1661 (decompose, immune_p): Update accordingly.
1662 (find_inc_amount): Return a poly_int64 rather than an int.
1663 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1666 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1667 Alan Hayward <alan.hayward@arm.com>
1668 David Sherwood <david.sherwood@arm.com>
1670 * tree.h (get_inner_reference): Return the bitsize and bitpos
1671 as poly_int64_pods rather than HOST_WIDE_INT.
1672 * fold-const.h (ptr_difference_const): Return the pointer difference
1673 as a poly_int64_pod rather than a HOST_WIDE_INT.
1674 * expr.c (get_inner_reference): Return the bitsize and bitpos
1675 as poly_int64_pods rather than HOST_WIDE_INT.
1676 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1678 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1679 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1680 (optimize_bit_field_compare): Update call to get_inner_reference.
1681 (decode_field_reference): Likewise.
1682 (fold_unary_loc): Track polynomial offsets and sizes.
1683 (split_address_to_core_and_offset): Return the bitpos as a
1684 poly_int64_pod rather than a HOST_WIDE_INT.
1685 (ptr_difference_const): Likewise for the pointer difference.
1686 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1687 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1688 * dbxout.c (dbxout_expand_expr): Likewise.
1689 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1690 (loc_list_from_tree_1, fortran_common): Likewise.
1691 * gimple-laddress.c (pass_laddress::execute): Likewise.
1692 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1693 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1694 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1695 * tree-affine.c (tree_to_aff_combination): Likewise.
1696 (get_inner_reference_aff): Likewise.
1697 * tree-data-ref.c (split_constant_offset_1): Likewise.
1698 (dr_analyze_innermost): Likewise.
1699 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1700 * tree-sra.c (ipa_sra_check_caller): Likewise.
1701 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1702 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1703 (instrument_bool_enum_load, instrument_object_size): Likewise.
1704 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1705 to get_inner_reference.
1706 * hsa-gen.c (gen_hsa_addr): Likewise.
1707 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1708 * tsan.c (instrument_expr): Likewise.
1709 * match.pd: Update call to ptr_difference_const.
1711 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1712 Alan Hayward <alan.hayward@arm.com>
1713 David Sherwood <david.sherwood@arm.com>
1715 * fold-const.c (fold_comparison): Track sizes and offsets as
1716 poly_int64s rather than HOST_WIDE_INTs when folding address
1719 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1720 Alan Hayward <alan.hayward@arm.com>
1721 David Sherwood <david.sherwood@arm.com>
1723 * expr.h (get_bit_range): Return the bitstart and bitend as
1724 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1725 as a poly_int64 rather than a HOST_WIDE_INT.
1726 * expr.c (get_bit_range): Likewise.
1727 (expand_assignment): Update call accordingly.
1728 * fold-const.c (optimize_bit_field_compare): Likewise.
1730 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1731 Alan Hayward <alan.hayward@arm.com>
1732 David Sherwood <david.sherwood@arm.com>
1734 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1735 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1736 as an optional final parameter.
1737 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1738 (aarch64_legitimate_address_p): Likewise.
1739 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1740 instead of an rtx code.
1741 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1742 (aarch64_legitimate_address_hook_p): Likewise.
1743 (aarch64_print_ldpstp_address): Likewise.
1744 (aarch64_print_operand_address): Likewise.
1745 (aarch64_address_cost): Likewise.
1746 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1747 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1748 (aarch64_mem_pair_lanes_operand): Likewise.
1750 2017-12-20 Richard Biener <rguenther@suse.de>
1752 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1753 update_stmt_if_modified.
1755 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1757 PR tree-optimization/83491
1758 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1759 before walking uses. Improve coding style and comments.
1761 2017-12-20 Tom de Vries <tom@codesourcery.com>
1763 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1765 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1768 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1770 (pass_ipa_free_fn_summary): Add small_p private data member,
1771 initialize to false in the ctor.
1772 (pass_ipa_free_fn_summary::clone,
1773 pass_ipa_free_fn_summary::set_pass_param,
1774 pass_ipa_free_fn_summary::gate): New methods.
1775 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1776 | TODO_dump_symtab if small_p.
1777 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1778 entry and add another instance of the pass with false parm after
1780 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1781 ipa_free_fn_summary here.
1783 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1785 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1787 2017-12-20 Martin Sebor <msebor@redhat.com>
1790 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1791 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1792 (expand_builtin_strncmp): Same.
1794 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1797 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1798 allowed debug stmts before labels.
1799 (expand_gimple_basic_block): Likewise.
1800 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1801 * gimple-iterator.h (gsi_after_labels): Likewise.
1802 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1803 rename reused variable, and simplify using gsi_move_before.
1804 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1805 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1806 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1807 (gimple_verify_flow_info, gimple_block_label): Likewise.
1808 (make_blocks): Move debug markers after adjacent labels.
1809 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1810 allowed debug insns outside blocks.
1811 * df-scan.c (df_insn_delete): Likewise.
1812 * lra-constraints.c (update_ebb_live_info): Likewise.
1813 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1814 (vt_initialize, delete_vta_debug_insns): Likewise.
1815 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1817 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1818 Alan Hayward <alan.hayward@arm.com>
1819 David Sherwood <david.sherwood@arm.com>
1821 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1822 (round_up_to_byte_boundary): Likewise.
1823 * expr.h (get_bit_range): Add temporary shim.
1824 * gimple-ssa-store-merging.c (store_operand_info): Change the
1825 bitsize, bitpos, bitregion_start and bitregion_end fields from
1826 unsigned HOST_WIDE_INT to poly_uint64.
1827 (merged_store_group): Likewise load_align_base.
1828 (compatible_load_p, compatible_load_p): Update accordingly.
1829 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1830 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1831 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1832 bitregion_start and bitregion_end as poly_uint64s rather than
1833 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1834 (handled_load): Take the bitsize, bitpos,
1835 bitregion_start and bitregion_end as poly_uint64s rather than
1836 unsigned HOST_WIDE_INTs.
1837 (pass_store_merging::process_store): Update call to
1838 mem_valid_for_store_merging.
1840 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1841 Alan Hayward <alan.hayward@arm.com>
1842 David Sherwood <david.sherwood@arm.com>
1844 * builtins.c (get_object_alignment_2): Track polynomial offsets
1845 and sizes. Update the alignment handling.
1847 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1848 Alan Hayward <alan.hayward@arm.com>
1849 David Sherwood <david.sherwood@arm.com>
1851 * tree.h (get_inner_reference): Add a version that returns the
1852 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1853 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1854 the case in which bitpos is not associated with the first byte.
1856 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1857 Alan Hayward <alan.hayward@arm.com>
1858 David Sherwood <david.sherwood@arm.com>
1860 * tree-affine.h (get_inner_reference_aff): Return the size as a
1862 * tree-affine.c (get_inner_reference_aff): Likewise.
1863 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1864 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1866 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1867 Alan Hayward <alan.hayward@arm.com>
1868 David Sherwood <david.sherwood@arm.com>
1870 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1871 HOST_WIDE_INT rather than a poly_int64.
1873 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1874 Alan Hayward <alan.hayward@arm.com>
1875 David Sherwood <david.sherwood@arm.com>
1877 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1878 HOST_WIDE_INT to poly_int64_pod.
1879 (perform_symbolic_merge): Update accordingly.
1880 (bswap_replace): Likewise.
1882 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1883 Alan Hayward <alan.hayward@arm.com>
1884 David Sherwood <david.sherwood@arm.com>
1886 * tree-affine.h (aff_tree::offset): Change from widest_int
1888 (wide_int_ext_for_comb): Delete.
1889 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1890 constants as poly_widest_int rather than widest_int.
1891 (aff_combination_constant_multiple_p): Return the multiplier
1892 as a poly_widest_int.
1893 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1895 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1896 version static and add an overload for poly_widest_int.
1897 (aff_combination_const, aff_combination_add_cst)
1898 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1899 the constants as poly_widest_int rather than widest_int.
1900 (tree_to_aff_combination): Generalize INTEGER_CST case to
1902 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1903 (aff_combination_add_product, aff_combination_mult): Handle
1905 (aff_combination_constant_multiple_p): Return the multiplier
1906 as a poly_widest_int.
1907 * tree-predcom.c (determine_offset): Return the offset as a
1909 (split_data_refs_to_components, suitable_component_p): Update
1911 (valid_initializer_p): Update call to
1912 aff_combination_constant_multiple_p.
1913 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1914 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1915 as a poly_int64 rather than a HOST_WIDE_INT.
1916 (get_address_cost): Handle polynomial offsets.
1917 (iv_elimination_compare_lt): Likewise.
1918 (rewrite_use_nonlinear_expr): Likewise.
1920 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1921 Alan Hayward <alan.hayward@arm.com>
1922 David Sherwood <david.sherwood@arm.com>
1924 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1925 as a poly_int64_pod rather than a HOST_WIDE_INT.
1926 (get_addr_base_and_unit_offset): Likewise.
1927 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1928 (get_addr_base_and_unit_offset): Likewise.
1929 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1930 to poly_int64 in example.
1931 * fold-const.c (fold_binary_loc): Update call to
1932 get_addr_base_and_unit_offset.
1933 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1934 (maybe_canonicalize_mem_ref_addr): Likewise.
1935 (gimple_fold_stmt_to_constant_1): Likewise.
1936 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1938 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1939 * match.pd: Likewise.
1940 * omp-low.c (lower_omp_target): Likewise.
1941 * tree-sra.c (build_ref_for_offset): Likewise.
1942 (build_debug_ref_for_model): Likewise.
1943 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1944 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1945 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1946 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1947 (constant_pointer_difference): Likewise.
1948 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1949 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1950 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1951 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1952 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1953 (set_ssa_val_to): Likewise.
1954 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1955 (maybe_diag_stxncpy_trunc): Likewise.
1956 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1957 * tree.c (build_simple_mem_ref_loc): Likewise.
1958 (array_at_struct_end_p): Likewise.
1960 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1961 Alan Hayward <alan.hayward@arm.com>
1962 David Sherwood <david.sherwood@arm.com>
1964 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1965 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1966 (get_ref_base_and_extent_hwi): Declare.
1967 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1968 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1969 (get_ref_base_and_extent_hwi): New function.
1970 * cfgexpand.c (expand_debug_expr): Update call to
1971 get_ref_base_and_extent.
1972 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1973 * gimple-fold.c (get_base_constructor): Return the offset as a
1974 poly_int64_pod rather than a HOST_WIDE_INT.
1975 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1976 * ipa-polymorphic-call.c
1977 (ipa_polymorphic_call_context::set_by_invariant)
1978 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1979 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1980 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1981 rather than get_ref_base_and_extent.
1982 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1983 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1984 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1986 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1987 call to get_ref_base_and_extent.
1988 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1989 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1990 (stmt_kills_ref_p): Likewise.
1991 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1992 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1994 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1995 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1996 when calling native_encode_expr.
1997 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1998 call to get_ref_base_and_extent.
1999 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
2000 get_ref_base_and_extent.
2001 * var-tracking.c (track_expr_p): Likewise.
2003 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2004 Alan Hayward <alan.hayward@arm.com>
2005 David Sherwood <david.sherwood@arm.com>
2007 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
2008 HOST_WIDE_INT to poly_int64_pod.
2009 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
2010 polynomail parameter offsets.
2012 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2013 Alan Hayward <alan.hayward@arm.com>
2014 David Sherwood <david.sherwood@arm.com>
2016 * gengtype.c (main): Handle poly_int64_pod.
2017 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
2018 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
2019 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
2020 from HOST_WIDE_INT to poly_int64_pod.
2021 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
2022 (copy_cfa): New function.
2023 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
2024 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
2025 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
2026 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
2027 (initial_return_save): Treat offsets as poly_ints.
2028 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
2029 offset is nonconstant.
2030 (reg_save): Take the offset as a poly_int64. Fall back to
2031 DW_CFA_expression for nonconstant offsets.
2032 (queue_reg_save): Take the offset as a poly_int64.
2033 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
2035 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2036 Alan Hayward <alan.hayward@arm.com>
2037 David Sherwood <david.sherwood@arm.com>
2039 * rtl.h (operand_subword, operand_subword_force): Take the offset
2040 as a poly_uint64 an unsigned int.
2041 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
2043 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2044 Alan Hayward <alan.hayward@arm.com>
2045 David Sherwood <david.sherwood@arm.com>
2047 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
2048 'p' format code. Use INT_LIST rather than SUBREG as the example of
2049 a code with an XINT and an XEXP. Remove the implication that
2050 accessing an rtx field using XINT is expected to work.
2051 * rtl.def (SUBREG): Change format from "ei" to "ep".
2052 * rtl.h (rtunion::rt_subreg): New field.
2053 (XCSUBREG): New macro.
2054 (SUBREG_BYTE): Use it.
2055 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
2056 Update constructor accordingly.
2057 (subreg_shape::operator ==): Update accordingly.
2058 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
2059 than an unsigned int.
2060 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
2061 a poly_uint64 rather than an unsigned int.
2062 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
2063 than an unsigned int.
2064 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
2065 (subreg_size_highpart_offset): Return a poly_uint64 rather than
2066 an unsigned int. Take the sizes as poly_uint64s.
2067 (subreg_offset_from_lsb): Return a poly_uint64 rather than
2068 an unsigned int. Take the shift as a poly_uint64 rather than
2070 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
2071 as a poly_uint64 rather than an unsigned int.
2072 (simplify_subreg_regno): Likewise.
2073 (byte_lowpart_offset): Return the memory offset as a poly_int64
2075 (subreg_memory_offset): Likewise. Take the subreg offset as a
2076 poly_uint64 rather than an unsigned int.
2077 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
2078 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
2079 poly_uint64 rather than an unsigned int.
2080 * rtl.c (rtx_format): Describe 'p' in comment.
2081 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
2082 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
2083 offset as a poly_uint64 rather than an unsigned int.
2084 (byte_lowpart_offset): Return the memory offset as a poly_int64
2086 (subreg_memory_offset): Likewise. Take the subreg offset as a
2087 poly_uint64 rather than an unsigned int.
2088 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
2089 mode sizes as poly_uint64s rather than unsigned ints. Return a
2090 poly_uint64 rather than an unsigned int.
2091 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
2092 (copy_insn_1): Handle 'p'.
2093 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
2094 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
2095 an unsigned int. Return the shift in the same way.
2096 (subreg_lsb): Return the shift as a poly_uint64 rather than an
2098 (subreg_size_offset_from_lsb): Take the sizes and shift as
2099 poly_uint64s rather than unsigned ints. Return the offset as
2101 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
2102 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
2104 * rtlhash.c (add_rtx): Handle 'p'.
2105 * genemit.c (gen_exp): Likewise.
2106 * gengenrtl.c (type_from_format, gendef): Likewise.
2107 * gensupport.c (subst_pattern_match, get_alternatives_number)
2108 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
2109 (subst_dup): Likewise.
2110 * gengtype.c (adjust_field_rtx_def): Likewise.
2111 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
2112 (match_pattern_2): Likewise.
2113 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
2114 (rtx_test::subreg_field): New function.
2115 (operator ==, safe_to_hoist_p, transition_parameter_type)
2116 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
2117 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
2118 * genpeep.c (match_rtx): Likewise.
2119 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
2120 (rtx_writer::print_rtx_operand): Handle 'p'.
2121 (print_value): Handle SUBREG.
2122 * read-rtl.c (apply_int_iterator): Likewise.
2123 (rtx_reader::read_rtx_operand): Handle 'p'.
2124 * alias.c (rtx_equal_for_memref_p): Likewise.
2125 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
2126 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
2128 * calls.c (expand_call): Likewise.
2129 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
2130 (make_extraction, gen_lowpart_for_combine): Likewise.
2131 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
2133 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
2134 rather than an unsigned int. Treat subreg offsets as poly_ints.
2135 (exp_equiv_p): Handle 'p'.
2136 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
2137 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
2138 * dse.c (find_shift_sequence): Likewise.
2139 * dwarf2out.c (rtl_for_decl_location): Likewise.
2140 * expmed.c (extract_low_bits): Likewise.
2141 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
2142 (expand_expr_real_2): Likewise.
2143 * final.c (alter_subreg): Likewise.
2144 (leaf_renumber_regs_insn): Handle 'p'.
2145 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
2146 Treat subreg offsets as poly_ints.
2147 * fwprop.c (forward_propagate_and_simplify): Likewise.
2148 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
2149 * ira.c (get_subreg_tracking_sizes): Likewise.
2150 * ira-conflicts.c (go_through_subreg): Likewise.
2151 * ira-lives.c (process_single_reg_class_operands): Likewise.
2152 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
2153 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
2154 as a poly_uint64 rather than an unsigned int.
2155 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
2156 subreg offsets as poly_ints.
2157 * lra-constraints.c (operands_match_p): Handle 'p'.
2158 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
2159 * lra-spills.c (assign_mem_slot): Likewise.
2160 * postreload.c (move2add_valid_value_p): Likewise.
2161 * recog.c (general_operand, indirect_operand): Likewise.
2162 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2163 (copyprop_hardreg_forward_1): Likewise.
2164 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2165 (record_subregs_of_mode): Likewise.
2166 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2167 * reload.c (operands_match_p): Handle 'p'.
2168 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2169 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2170 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2171 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2172 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2173 (simplify_const_poly_int_tests<N>::run): Likewise.
2174 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2175 a poly_uint64 rather than an unsigned int.
2176 * valtrack.c (debug_lowpart_subreg): Likewise.
2177 * var-tracking.c (var_lowpart): Likewise.
2178 (loc_cmp): Handle 'p'.
2180 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2181 Alan Hayward <alan.hayward@arm.com>
2182 David Sherwood <david.sherwood@arm.com>
2184 * ira.c (get_subreg_tracking_sizes): New function.
2185 (init_live_subregs): Take an integer size rather than a register.
2186 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2187 to init_live_subregs.
2189 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2190 Alan Hayward <alan.hayward@arm.com>
2191 David Sherwood <david.sherwood@arm.com>
2193 * expr.c (store_constructor_field): Change bitsize from a
2194 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2195 HOST_WIDE_INT to a poly_int64.
2196 (store_constructor): Change size from a HOST_WIDE_INT to
2198 (store_field): Likewise bitsize and bitpos.
2200 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2201 Alan Hayward <alan.hayward@arm.com>
2202 David Sherwood <david.sherwood@arm.com>
2204 * expmed.h (store_bit_field): Change bitregion_start and
2205 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2206 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2207 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2208 (store_fixed_bit_field, store_split_bit_field): Likewise.
2209 * expr.c (store_constructor_field, store_field): Likewise.
2210 (optimize_bitfield_assignment_op): Likewise. Make the same change
2211 to bitsize and bitpos.
2212 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2213 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2214 same change in the constructor arguments.
2215 (get_best_mode): Change bitregion_start and bitregion_end from
2216 unsigned HOST_WIDE_INT to poly_uint64.
2217 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2218 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2220 (bit_field_mode_iterator::next_mode): Update for new types
2221 of m_bitregion_start and m_bitregion_end.
2222 (get_best_mode): Change bitregion_start and bitregion_end from
2223 unsigned HOST_WIDE_INT to poly_uint64.
2225 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2226 Alan Hayward <alan.hayward@arm.com>
2227 David Sherwood <david.sherwood@arm.com>
2229 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2230 accepts poly_uint64 offsets.
2231 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2232 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2233 * expmed.c (lowpart_bit_field_p): Likewise.
2234 (extract_bit_field_as_subreg): New function, split out from...
2235 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2236 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2237 extractions, check that BITSIZE matches the size of the extracted
2238 value and that BITNUM is an exact multiple of that size.
2239 If all else fails, try forcing the value into memory if
2240 BITNUM is variable, and adjusting the address so that the
2241 offset is constant. Split the part that can only handle constant
2242 bitsize and bitnum out into...
2243 (extract_integral_bit_field): ...this new function.
2244 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2245 rather than unsigned HOST_WIDE_INTs.
2247 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2248 Alan Hayward <alan.hayward@arm.com>
2249 David Sherwood <david.sherwood@arm.com>
2251 * expmed.h (store_bit_field): Take bitsize and bitnum as
2252 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2253 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2254 that returns the byte size.
2255 (store_bit_field_1): Take bitsize and bitnum as
2256 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2257 to simple_mem_bitfield_p. Split the part that can only handle
2258 constant bitsize and bitnum out into...
2259 (store_integral_bit_field): ...this new function.
2260 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2261 than unsigned HOST_WIDE_INTs.
2262 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2264 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2265 Alan Hayward <alan.hayward@arm.com>
2266 David Sherwood <david.sherwood@arm.com>
2268 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2269 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2271 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2272 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2273 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2274 offset as a poly_int64 rather than an int.
2275 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2276 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2277 * lra-constraints.c (equiv_address_substitution): Track offsets
2279 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2280 (curr_insn_transform): Handle the new form of sp_offset.
2281 * lra-eliminations.c (lra_elim_table): Change previous_offset
2282 and offset from HOST_WIDE_INT to poly_int64.
2283 (print_elim_table, update_reg_eliminate): Update accordingly.
2284 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2285 (get_elimination): Update accordingly.
2286 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2287 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2288 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2289 poly_int64 offsets generally.
2290 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2291 (mark_not_eliminable, init_elimination): Update accordingly.
2292 (remove_reg_equal_offset_note): Return a bool and pass the new
2293 offset back by pointer as a poly_int64.
2294 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2295 rather than a HOST_WIDE_INT.
2296 (do_remat): Track offsets poly_int64s.
2297 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2299 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2300 Alan Hayward <alan.hayward@arm.com>
2301 David Sherwood <david.sherwood@arm.com>
2303 * rtl.h (mem_attrs): Add a default constructor. Change size and
2304 offset from HOST_WIDE_INT to poly_int64.
2305 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2306 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2307 (widen_memory_access): Take the sizes and offsets as poly_int64s
2308 rather than HOST_WIDE_INTs.
2309 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2310 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2312 (adjust_offset_for_component_ref): Change the offset from a
2313 HOST_WIDE_INT to a poly_int64.
2314 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2315 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2316 * dce.c (find_call_stack_args): Likewise.
2317 * dse.c (record_store): Likewise.
2318 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2319 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2320 * read-rtl-function.c (test_loading_mem): Likewise.
2321 * rtlanal.c (may_trap_p_1): Likewise.
2322 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2323 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2324 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2325 (mem_attrs::mem_attrs): New function.
2326 (set_mem_attributes_minus_bitpos): Change bitpos from a
2327 HOST_WIDE_INT to poly_int64.
2328 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2329 (clear_mem_offset, clear_mem_size, change_address)
2330 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2331 initialize mem_attrs.
2332 (set_mem_offset, set_mem_size, adjust_address_1)
2333 (adjust_automodify_address_1, offset_address, widen_memory_access):
2334 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2336 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2337 Alan Hayward <alan.hayward@arm.com>
2338 David Sherwood <david.sherwood@arm.com>
2340 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2341 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2342 rather than 0 to represent an unknown size. Assert that the size
2343 is known when the mode isn't BLKmode.
2344 (may_trap_p_1): Use -1 for unknown sizes.
2345 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2347 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2348 Alan Hayward <alan.hayward@arm.com>
2349 David Sherwood <david.sherwood@arm.com>
2351 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2352 to poly_int64. Update commentary for positions_needed.large.
2353 (read_info_type): Change offset and width from HOST_WIDE_INT
2355 (set_usage_bits): Likewise.
2356 (canon_address): Return the offset as a poly_int64 rather than
2357 a HOST_WIDE_INT. Use strip_offset_and_add.
2358 (set_all_positions_unneeded, any_positions_needed_p): Use
2359 positions_needed.large to track stores with non-constant widths.
2360 (all_positions_needed_p): Likewise. Take the offset and width
2361 as poly_int64s rather than ints. Assert that rhs is nonnull.
2362 (record_store): Cope with non-constant offsets and widths.
2363 Nullify the rhs of an earlier store if we can't tell which bytes
2365 (find_shift_sequence): Take the access_size and shift as poly_int64s
2367 (get_stored_val): Take the read_offset and read_width as poly_int64s
2368 rather than HOST_WIDE_INTs.
2369 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2370 non-constant offsets and widths.
2372 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2373 Alan Hayward <alan.hayward@arm.com>
2374 David Sherwood <david.sherwood@arm.com>
2376 * inchash.h (inchash::hash::add_poly_int): New function.
2377 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2378 Use poly_int64 rather than HOST_WIDE_INT.
2379 (ao_ref::max_size_known_p): New function.
2380 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2381 rather than HOST_WIDE_INT.
2382 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2383 to temporaries until its interface is adjusted to match.
2384 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2385 (aliasing_component_refs_p, decl_refs_may_alias_p)
2386 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2387 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2388 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2390 * alias.c (ao_ref_from_mem): Likewise.
2391 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2392 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2393 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2394 (maybe_trim_complex_store, maybe_trim_constructor_store)
2395 (live_bytes_read, dse_classify_store): Likewise.
2396 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2397 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2398 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2399 (vn_reference_lookup_3): Likewise.
2400 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2402 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2403 Alan Hayward <alan.hayward@arm.com>
2404 David Sherwood <david.sherwood@arm.com>
2406 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2407 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2408 instead of ranges_overlap_p.
2410 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2411 Alan Hayward <alan.hayward@arm.com>
2412 David Sherwood <david.sherwood@arm.com>
2414 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2415 sizes as poly_int64s rather than HOST_WIDE_INTs.
2417 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2418 Alan Hayward <alan.hayward@arm.com>
2419 David Sherwood <david.sherwood@arm.com>
2421 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2422 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2423 * gimple-fold.c (fold_ctor_reference): Likewise.
2425 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2426 Alan Hayward <alan.hayward@arm.com>
2427 David Sherwood <david.sherwood@arm.com>
2429 * target.def (dwarf_poly_indeterminate_value): New hook.
2430 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2431 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2432 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2433 * doc/tm.texi: Regenerate.
2434 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2435 offset as a poly_int64.
2436 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2437 offset as a poly_int64.
2438 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2439 Take the offset as a poly_int64.
2440 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2441 (frame_pointer_fb_offset): Change to a poly_int64.
2442 (int_loc_descriptor): Take the offset as a poly_int64. Use
2443 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2444 (based_loc_descr): Take the offset as a poly_int64.
2445 Use strip_offset_and_add to handle (plus X (const)).
2446 Use new_reg_loc_descr instead of an open-coded version of the
2447 previous implementation.
2448 (mem_loc_descriptor): Handle CONST_POLY_INT.
2449 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2450 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2452 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2453 Alan Hayward <alan.hayward@arm.com>
2454 David Sherwood <david.sherwood@arm.com>
2456 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2457 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2458 * inchash.h (inchash::hash::add_poly_hwi): New function.
2459 * gengtype.c (main): Register poly_int64.
2460 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2461 offset as a poly_int.
2462 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2463 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2464 offset as a poly_int64.
2465 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2466 * print-rtl.c (print_poly_int): New function.
2467 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2469 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2471 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2472 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2473 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2474 Enforce track_offset_p for parts of a PARALLEL.
2475 (vt_add_function_parameter): Use const_offset for the final
2476 offset to track. Use get_tracked_reg_offset for the parts
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 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2484 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2485 * doc/tm.texi: Regenerate.
2486 * hooks.h (hook_bool_uint_uint_true): Delete.
2487 (hook_bool_puint64_puint64_true): Declare.
2488 * hooks.c (hook_bool_uint_uint_true): Delete.
2489 (hook_bool_puint64_puint64_true): New function.
2490 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2491 instead of unsigned ints.
2492 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2493 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2495 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2496 Alan Hayward <alan.hayward@arm.com>
2497 David Sherwood <david.sherwood@arm.com>
2499 * optabs.h (expand_operand): Add an int_value field.
2500 (create_expand_operand): Add an int_value parameter and use it
2501 to initialize the new expand_operand field.
2502 (create_integer_operand): Replace with a declaration of a function
2503 that accepts poly_int64s. Move the implementation to...
2504 * optabs.c (create_integer_operand): ...here.
2505 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2506 the mode preserves the value of int_value, instead of calling
2507 const_int_operand on the rtx. Use gen_int_mode to generate
2510 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2511 Alan Hayward <alan.hayward@arm.com>
2512 David Sherwood <david.sherwood@arm.com>
2514 * dumpfile.h (dump_dec): Declare.
2515 * dumpfile.c (dump_dec): New function.
2516 * pretty-print.h (pp_wide_integer): Turn into a function and
2517 declare a poly_int version.
2518 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2520 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2521 Alan Hayward <alan.hayward@arm.com>
2522 David Sherwood <david.sherwood@arm.com>
2524 * doc/generic.texi (POLY_INT_CST): Document.
2525 * tree.def (POLY_INT_CST): New tree code.
2526 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2527 * tree-core.h (tree_poly_int_cst): New struct.
2528 (tree_node): Add a poly_int_cst field.
2529 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2530 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2531 instead of a wide_int_ref.
2532 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2534 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2535 instead of an unsigned HOST_WIDE_INT.
2536 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2537 (ptrdiff_tree_p): Declare.
2538 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2539 extern inline implementations if the target doesn't use POLY_INT_CST.
2540 (poly_int_tree_p): New function.
2541 (wi::unextended_tree): New class.
2542 (wi::int_traits <unextended_tree>): New override.
2543 (wi::extended_tree): Add a default constructor.
2544 (wi::extended_tree::get_tree): New function.
2545 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2546 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2547 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2548 (wi::tree_to_poly_wide_ref): New typedefs.
2549 (wi::ints_for): Provide overloads for extended_tree and
2551 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2552 (wi::to_wide): New functions.
2553 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2554 * tree.c (poly_int_cst_hasher): New struct.
2555 (poly_int_cst_hash_table): New variable.
2556 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2557 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2559 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2560 (init_ttree): Initialize poly_int_cst_hash_table.
2561 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2562 a poly_int64 instead of a HOST_WIDE_INT.
2563 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2564 instead of an unsigned HOST_WIDE_INT.
2565 (wide_int_to_tree): Rename to...
2566 (wide_int_to_tree_1): ...this.
2567 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2568 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2569 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2570 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2571 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2572 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2574 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2575 (streamer_read_tree_body): Likewise.
2576 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2577 (streamer_write_tree_body): Likewise.
2578 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2579 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2580 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2581 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2582 * gimple-expr.h (is_gimple_constant): Likewise.
2583 * gimplify.c (maybe_with_size_expr): Likewise.
2584 * print-tree.c (print_node): Likewise.
2585 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2586 * tree-pretty-print.c (dump_generic_node): Likewise.
2587 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2588 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2589 * tree-vrp.c (compare_values_warnv): Likewise.
2590 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2591 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2592 (force_expr_to_var_cost): Likewise.
2593 * tree-ssa-loop.c (for_each_index): Likewise.
2594 * fold-const.h (build_invariant_address, size_int_kind): Take a
2595 poly_int64 instead of a HOST_WIDE_INT.
2596 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2597 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2599 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2600 (int_const_binop_2): New function, split out from...
2601 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2602 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2603 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2604 * gimple-ssa-strength-reduction.c (slsr_process_add)
2605 (slsr_process_mul): Check for INTEGER_CSTs before using them
2607 * stor-layout.c (bits_from_bytes): New function.
2608 (bit_from_pos): Use it.
2609 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2610 by BITS_PER_UNIT to get the TYPE_SIZE.
2611 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2612 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2614 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2615 Alan Hayward <alan.hayward@arm.com>
2616 David Sherwood <david.sherwood@arm.com>
2618 * doc/rtl.texi (const_poly_int): Document. Also document the
2619 rtl sharing behavior.
2620 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2621 * rtl.h (const_poly_int_def): New struct.
2622 (rtx_def::u): Add a cpi field.
2623 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2624 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2625 (wi::rtx_to_poly_wide_ref): New typedef
2626 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2627 (poly_int_rtx_p): New functions.
2628 (trunc_int_for_mode): Declare a poly_int64 version.
2629 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2630 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2632 (strip_offset): Declare.
2633 (strip_offset_and_add): New function.
2634 * rtl.def (CONST_POLY_INT): New rtx code.
2635 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2636 (shared_const_p): Use poly_int_rtx_p.
2637 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2639 (gen_int_shift_amount): Likewise.
2640 * emit-rtl.c (const_poly_int_hasher): New class.
2641 (const_poly_int_htab): New variable.
2642 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2643 (const_poly_int_hasher::hash): New function.
2644 (const_poly_int_hasher::equal): Likewise.
2645 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2646 (immed_wide_int_const): Rename to...
2647 (immed_wide_int_const_1): ...this and make static.
2648 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2649 instead of a wide_int_ref.
2650 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2651 (gen_lowpart_common): Handle CONST_POLY_INT.
2652 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2653 * cselib.c (cselib_hash_rtx): Likewise.
2654 * dwarf2out.c (const_ok_for_output_1): Likewise.
2655 * expr.c (convert_modes): Likewise.
2656 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2657 * rtlhash.c (add_rtx): Likewise.
2658 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2659 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2660 Handle existing CONST_POLY_INT rtxes.
2661 * expmed.h (expand_shift): Take a poly_int64 instead of a
2663 * expmed.c (expand_shift): Likewise.
2664 * rtlanal.c (strip_offset): New function.
2665 (commutative_operand_precedence): Give CONST_POLY_INT the same
2666 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2668 * rtl-tests.c (const_poly_int_tests): New struct.
2669 (rtl_tests_c_tests): Use it.
2670 * simplify-rtx.c (simplify_const_unary_operation): Handle
2672 (simplify_const_binary_operation): Likewise.
2673 (simplify_binary_operation_1): Fold additions of symbolic constants
2674 and CONST_POLY_INTs.
2675 (simplify_subreg): Handle extensions and truncations of
2677 (simplify_const_poly_int_tests): New struct.
2678 (simplify_rtx_c_tests): Use it.
2679 * wide-int.h (storage_ref): Add default constructor.
2680 (wide_int_ref_storage): Likewise.
2681 (trailing_wide_ints): Use GTY((user)).
2682 (trailing_wide_ints::operator[]): Add a const version.
2683 (trailing_wide_ints::get_precision): New function.
2684 (trailing_wide_ints::extra_size): Likewise.
2686 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2687 Alan Hayward <alan.hayward@arm.com>
2688 David Sherwood <david.sherwood@arm.com>
2690 * emit-rtl.h (gen_int_shift_amount): Declare.
2691 * emit-rtl.c (gen_int_shift_amount): New function.
2692 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2694 * calls.c (shift_return_value): Likewise.
2695 * cse.c (fold_rtx): Likewise.
2696 * dse.c (find_shift_sequence): Likewise.
2697 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2698 (expand_shift, expand_smod_pow2): Likewise.
2699 * lower-subreg.c (shift_cost): Likewise.
2700 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2701 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2702 (expand_vec_perm_var): Likewise.
2703 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2704 (simplify_binary_operation_1): Likewise.
2705 * combine.c (try_combine, find_split_point, force_int_to_mode)
2706 (simplify_shift_const_1, simplify_shift_const): Likewise.
2707 (change_zero_ext): Likewise. Use simplify_gen_binary.
2709 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2711 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2713 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2715 * doc/invoke.texi (ARM Options): Document accepted extension options
2716 for -march=armv8.3-a.
2718 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2721 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2722 or --with-float=softfp, set the default CPU to arm10e.
2724 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2726 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2727 * config/visium/predicates.md (const_shift_operand): Likewise.
2728 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2729 (visium_legitimize_reload_address): Likewise.
2731 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2734 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2736 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2739 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2741 2017-12-20 Julia Koval <julia.koval@intel.com>
2743 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2744 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2745 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2746 * config.gcc: Include vpclmulqdqintrin.h.
2747 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2748 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2750 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2751 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2752 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2753 * config/i386/i386.c (isa2_opts): Add -mcx16.
2754 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2755 (ix86_option_override_internal): Move mcx16 to flags2.
2756 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2757 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2758 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2759 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2760 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2761 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2762 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2763 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2764 * doc/invoke.texi: Add -mvpclmulqdq.
2766 2017-12-20 Tom de Vries <tom@codesourcery.com>
2769 * config/i386/i386.c (ix86_static_chain): Move
2770 DECL_STATIC_CHAIN test ...
2771 * calls.c (rtx_for_static_chain): ... here. New function.
2772 * calls.h (rtx_for_static_chain): Declare.
2773 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2774 instead of targetm.calls.static_chain.
2775 * df-scan.c (df_get_entry_block_def_set): Same.
2777 2017-12-19 Tom de Vries <tom@codesourcery.com>
2779 PR tree-optimization/83493
2780 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2783 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2785 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2786 inform with hardcoded english plural handling.
2788 2017-12-18 Jeff Law <law@redhat.com>
2790 PR tree-optimization/83477
2791 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2792 a non-virtual PHI, always push a new range.
2794 2017-12-19 Martin Sebor <msebor@redhat.com>
2797 * builtins.c (compute_objsize): Handle non-constant offsets.
2799 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2801 PR tree-optimization/83444
2802 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2803 character load case, if get_stridx on MEM_REF's operand doesn't
2804 look usable, retry with get_addr_stridx.
2806 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2809 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2810 (variable_tracking_main_1): Keep markers even when VTA fails.
2813 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2814 even if there are markers before it.
2815 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2817 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2819 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2820 typical order conditions.
2821 * sel-sched.c (extract_new_fences_from): Likewise.
2822 * config/visium/constraints.md (J, K, L): Likewise.
2823 * config/visium/predicates.md (const_shift_operand): Likewise.
2824 * config/visium/visium.c (visium_legitimize_address,
2825 visium_legitimize_reload_address): Likewise.
2826 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2827 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2828 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2829 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2830 SET_DUMP_DETAIL): Likewise.
2831 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2832 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2833 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2834 avr_set_core_architecture, avr_set_current_function,
2835 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2836 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2837 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2838 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2839 avr_map_decompose, avr_fold_builtin): Likewise.
2840 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2841 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2842 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2843 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2844 * config/m32c/m32c.c (m32c_conditional_register_usage,
2845 m32c_address_cost): Likewise.
2846 * config/m32c/predicates.md (shiftcount_operand,
2847 longshiftcount_operand): Likewise.
2848 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2849 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2850 can_use_cdx_ldstw): Likewise.
2851 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2852 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2854 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2855 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2856 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2857 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2858 * config/vax/vax.c (adjacent_operands_p): Likewise.
2859 * config/ft32/constraints.md (L, b, KA): Likewise.
2860 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2862 * cfgexpand.c (expand_stack_alignment): Likewise.
2863 * gcse.c (insert_expr_in_table): Likewise.
2864 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2865 * cgraphunit.c (cgraph_node::expand): Likewise.
2866 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2867 * emit-rtl.c (add_insn): Likewise.
2868 * input.c (dump_location_info): Likewise.
2869 * passes.c (NEXT_PASS): Likewise.
2870 * read-rtl-function.c (parse_note_insn_name,
2871 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2873 * sched-rgn.c (sched_rgn_init): Likewise.
2874 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2875 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2876 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2878 * explow.c (eliminate_constant_term): Likewise.
2879 * final.c (leaf_renumber_regs_insn): Likewise.
2880 * cfgrtl.c (print_rtl_with_bb): Likewise.
2881 * genhooks.c (emit_init_macros): Likewise.
2882 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2883 * tree-data-ref.c (conflict_fn): Likewise.
2884 * selftest.c (assert_streq): Likewise.
2885 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2886 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2887 fold_binary_loc, multiple_of_p): Likewise.
2888 * reload.c (push_reload, find_equiv_reg): Likewise.
2889 * et-forest.c (et_nca, et_below): Likewise.
2890 * dbxout.c (dbxout_symbol_location): Likewise.
2891 * reorg.c (relax_delay_slots): Likewise.
2892 * dojump.c (do_compare_rtx_and_jump): Likewise.
2893 * gengtype-parse.c (type): Likewise.
2894 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2895 simplify_const_relational_operation): Likewise.
2896 * reload1.c (do_output_reload): Likewise.
2897 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2898 * gengtype.c (type_for_name): Likewise.
2899 * gimple-ssa-sprintf.c (format_directive): Likewise.
2901 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2904 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2905 accessing it. Adjust comment.
2907 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2910 * predict.c (zero_one_minusone): New function.
2911 (apply_return_prediction): Avoid return prediction for functions
2912 returning only -1, 0 and 1 values, unless they only return -1 and 0
2915 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2917 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2918 fall-through warning.
2919 (arc_compute_frame_size): Remove unused variables.
2920 (arc_print_operand): Fix fprintif format.
2921 (arc_can_follow_jump): Clean fall-through warning.
2923 2017-12-19 Marek Polacek <polacek@redhat.com>
2926 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2929 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2931 * config/arc/arc.c (overriderregs): New variable.
2932 (arc_override_options): Track fixed/call saved/call options.
2933 (arc_conditional_register_usage): Check against overrideregs
2934 variable whenever we change register properties.
2936 2017-12-19 Nathan Sidwell <nathan@acm.org>
2938 * opts.c (finish_options): Don't prefix dump_base_name if it
2939 already contains directories.
2941 2017-12-19 Martin Liska <mliska@suse.cz>
2943 PR rtl-optimization/82675
2944 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2945 more element in sbitmap.
2947 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2949 * gimplify.c (gimplify_expr): Use error_operand_p.
2951 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2954 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2957 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2959 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2960 the jump is to a label.
2962 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2964 PR tree-optimization/83444
2965 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2970 * ipa-inline.c (flatten_remove_node_hook): New function.
2971 (ipa_inline): Keep only nodes with flatten attribute at the end of
2972 the array in the order from ipa_reverse_postorder, only walk that
2973 portion of array for flattening, if there is more than one such
2974 node, temporarily register a removal hook and ignore removed nodes.
2976 PR tree-optimization/80631
2977 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2978 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2979 IFN_REDUC_MAX or IFN_REDUC_MIN.
2981 2017-12-18 Jeff Law <law@redhat.com>
2983 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2984 record symbolic equivalences from backedges in the CFG.
2987 2017-11-19 Jeff Law <law@redhat.com>
2989 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2990 of degenerates resulting from ignoring an edge.
2992 2017-12-18 Martin Sebor <msebor@redhat.com>
2995 PR tree-optimization/78450
2996 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2997 (handle_builtin_strlen): Call it.
2999 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
3001 PR rtl-optimization/83424
3002 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
3004 2017-12-18 Marek Polacek <polacek@redhat.com>
3007 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
3008 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
3011 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
3013 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
3014 (umaddsidi4, umaddsidi_split): Likewise.
3016 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
3018 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
3021 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3024 * config/s390/s390.c (s390_option_override): Avoid strncpy.
3026 2017-12-18 Richard Biener <rguenther@suse.de>
3028 PR tree-optimization/81877
3029 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
3030 (outermost_indep_loop): Adjust.
3031 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
3032 (can_sm_ref_p): Adjust.
3034 2017-12-18 Richard Biener <rguenther@suse.de>
3037 * tree.c (array_at_struct_end_p): Return true if the underlying
3038 object has space for at least one element in excess of what
3039 the array domain specifies.
3041 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3043 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
3044 example to avoid overfull hbox.
3045 * doc/invoke.texi (Option Summary): Add missing @gol.
3046 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
3049 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3050 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3052 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
3053 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
3054 (AArch64 Options): Document -mverbose-cost-dump.
3055 (ARM Options): Likewise, plus -mflip-thumb.
3057 2017-12-17 Martin Sebor <msebor@redhat.com>
3060 * gimple-ssa-warn-restrict.c
3061 (builtin_memref::offset_out_of_bounds): Correct the handling of
3064 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3066 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
3068 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
3070 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
3072 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
3074 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
3077 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
3079 * doc/invoke.texi: Fix some typos.
3081 2017-12-16 Martin Sebor <msebor@redhat.com>
3083 PR tree-optimization/78918
3084 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
3085 * builtins.c (check_sizes): Rename...
3086 (check_access): ...to this. Rename function arguments for clarity.
3087 (check_memop_sizes): Adjust names.
3088 (expand_builtin_memchr, expand_builtin_memcpy): Same.
3089 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
3090 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
3091 (check_strncat_sizes, expand_builtin_strncat): Same.
3092 (expand_builtin_strncpy, expand_builtin_memset): Same.
3093 (expand_builtin_bzero, expand_builtin_memcmp): Same.
3094 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
3095 (maybe_emit_sprintf_chk_warning): Same.
3096 (expand_builtin_strcpy): Adjust.
3097 (expand_builtin_stpcpy): Same.
3098 (expand_builtin_with_bounds): Detect out-of-bounds accesses
3099 in pointer-checking forms of memcpy, memmove, and mempcpy.
3100 (gcall_to_tree_minimal, max_object_size): Define new functions.
3101 * builtins.h (max_object_size): Declare.
3102 * calls.c (alloc_max_size): Call max_object_size instead of
3103 hardcoding ssizetype limit.
3104 (get_size_range): Handle new argument.
3105 * calls.h (get_size_range): Add a new argument.
3106 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
3107 * doc/invoke.texi (-Wrestrict): Adjust, add example.
3108 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
3110 (gimple_fold_builtin_memory_chk): Same.
3111 (gimple_fold_builtin_stxcpy_chk): New function.
3112 * gimple-ssa-warn-restrict.c: New source.
3113 * gimple-ssa-warn-restrict.h: New header.
3114 * gimple.c (gimple_build_call_from_tree): Propagate location.
3115 * passes.def (pass_warn_restrict): Add new pass.
3116 * tree-pass.h (make_pass_warn_restrict): Declare.
3117 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
3119 (handle_builtin_strcat): Same.
3120 (strlen_optimize_stmt): Rename...
3121 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
3122 stpncpy, strncpy, and their checking forms.
3124 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
3126 PR rtl-optimization/82849
3127 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
3128 and get_max_loop_iterations_int.
3130 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3131 Alan Hayward <alan.hayward@arm.com>
3132 David Sherwood <david.sherwood@arm.com>
3134 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
3135 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
3136 size as a poly_uint64.
3137 (mode_for_vector, mode_for_int_vector): Take the number of vector
3138 elements as a poly_uint64.
3139 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
3140 size as a poly_uint64.
3141 (mode_for_vector, mode_for_int_vector): Take the number of vector
3142 elements as a poly_uint64.
3144 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3145 Alan Hayward <alan.hayward@arm.com>
3146 David Sherwood <david.sherwood@arm.com>
3148 * machmode.h (MACRO_MODE): New macro.
3149 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
3150 * targhooks.c (default_libcall_value, default_secondary_reload)
3151 (default_memory_move_cost, default_register_move_cost)
3152 (default_class_max_nregs): Likewise.
3154 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3155 Alan Hayward <alan.hayward@arm.com>
3156 David Sherwood <david.sherwood@arm.com>
3158 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
3159 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3161 * genautomata.c (main): Likewise.
3162 * genconditions.c (write_header): Likewise.
3163 * genemit.c (main): Likewise.
3164 * genextract.c (print_header): Likewise.
3165 * genopinit.c (main): Likewise.
3166 * genoutput.c (output_prologue): Likewise.
3167 * genpeep.c (main): Likewise.
3168 * genpreds.c (write_insn_preds_c): Likewise.
3169 * genrecog.c (writer_header): Likewise.
3170 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3171 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3172 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3173 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3174 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3175 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3176 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3177 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3178 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3179 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3180 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3181 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3182 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3183 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3184 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3185 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3186 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3187 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3188 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3189 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3190 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3191 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3192 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3193 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3194 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3195 * config/darwin.c (IN_TARGET_CODE): Likewise.
3196 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3197 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3198 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3199 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3200 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3201 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3202 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3203 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3204 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3205 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3206 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3207 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3208 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3209 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3210 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3211 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3212 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3213 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3214 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3215 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3216 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3217 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3218 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3219 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3220 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3221 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3222 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3223 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3224 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3225 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3226 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3227 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3228 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3229 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3230 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3231 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3232 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3233 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3234 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3235 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3236 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3237 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3238 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3239 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3240 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3241 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3242 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3243 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3244 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3245 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3246 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3247 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3248 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3249 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3250 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3251 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3252 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3253 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3254 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3255 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3256 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3257 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3258 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3259 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3260 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3261 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3262 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3263 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3264 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3265 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3266 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3267 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3268 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3269 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3270 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3271 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3272 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3273 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3274 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3275 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3276 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3277 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3278 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3279 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3280 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3281 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3282 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3283 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3284 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3285 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3286 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3287 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3288 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3289 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3290 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3291 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3292 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3293 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3294 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3295 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3296 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3297 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3298 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3299 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3300 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3301 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3303 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3304 Alan Hayward <alan.hayward@arm.com>
3305 David Sherwood <david.sherwood@arm.com>
3307 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3310 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3311 Alan Hayward <alan.hayward@arm.com>
3312 David Sherwood <david.sherwood@arm.com>
3314 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3315 * doc/md.texi (vec_series@var{m}): Document.
3316 * tree.def (VEC_SERIES_EXPR): New tree code.
3317 * tree.h (build_vec_series): Declare.
3318 * tree.c (build_vec_series): New function.
3319 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3320 * tree-pretty-print.c (dump_generic_node): Likewise.
3321 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3322 * tree-inline.c (estimate_operator_cost): Likewise.
3323 * expr.c (expand_expr_real_2): Likewise.
3324 * optabs-tree.c (optab_for_tree_code): Likewise.
3325 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3326 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3327 * expmed.c (make_tree): Handle VEC_SERIES.
3328 * optabs.def (vec_series_optab): New optab.
3329 * optabs.h (expand_vec_series_expr): Declare.
3330 * optabs.c (expand_vec_series_expr): New function.
3331 * tree-vect-generic.c (expand_vector_operations_1): Check that
3332 the operands also have vector type.
3334 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3335 Alan Hayward <alan.hawyard@arm.com>
3336 David Sherwood <david.sherwood@arm.com>
3338 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3339 (VEC_COND_EXPR): Add missing @tindex.
3340 * doc/md.texi (vec_duplicate@var{m}): Document.
3341 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3342 * tree.c (build_vector_from_val): Add stubbed-out handling of
3343 variable-length vectors, using VEC_DUPLICATE_EXPR.
3344 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3345 * cfgexpand.c (expand_debug_expr): Likewise.
3346 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3347 * tree-inline.c (estimate_operator_cost): Likewise.
3348 * tree-pretty-print.c (dump_generic_node): Likewise.
3349 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3350 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3351 (test_vec_duplicate_folding): New function.
3352 (fold_const_c_tests): Call it.
3353 * optabs.def (vec_duplicate_optab): New optab.
3354 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3355 * optabs.h (expand_vector_broadcast): Declare.
3356 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3357 vec_duplicate_optab.
3358 * expr.c (store_constructor): Try using vec_duplicate_optab for
3360 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3362 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3365 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3366 div/mod latencies a bit.
3368 2017-12-15 Jeff Law <law@redhat.com>
3370 PR tree-optimization/36550
3371 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3372 (mark_threaded_blocks): Rewrite code to avoid block copying when
3373 optimizing for size. Don't pessimize blocks which will be
3374 copied, but all the statements will be dead.
3376 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3378 PR tree-optimization/81165
3379 * tree-ssa-threadupdate.c (uses_in_bb): New.
3380 (estimate_threading_killed_stmts): New.
3381 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3382 * tree-ssa-threadedge.c
3383 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3386 2017-12-15 Jeff Law <law@redhat.com>
3388 PR tree-optimization/83410
3389 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3390 threads when parallelizing loops.
3392 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3394 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3396 * config/alpha/alpha.c (vms_attribute_table): Swap
3397 affects_type_identity and handler fields, adjust comments.
3398 * config/mips/mips.c (mips_attribute_table): Likewise.
3399 * config/visium/visium.c (visium_attribute_table): Likewise.
3400 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3401 * config/microblaze/microblaze.c (microblaze_attribute_table):
3403 * config/spu/spu.c (spu_attribute_table): Likewise.
3404 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3405 * config/arc/arc.c (arc_attribute_table): Likewise.
3406 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3407 * config/v850/v850.c (v850_handle_interrupt_attribute,
3408 v850_handle_data_area_attribute): Formatting fixes.
3409 (v850_attribute_table): Swap affects_type_identity and handler
3410 fields, adjust comments.
3411 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3412 * config/arm/arm.c (arm_attribute_table): Likewise.
3413 * config/avr/avr.c (avr_attribute_table): Likewise.
3414 * config/s390/s390.c (s390_attribute_table): Likewise.
3415 * config/sh/sh.c (sh_attribute_table): Likewise.
3416 * config/i386/i386.c (ix86_handle_cconv_attribute,
3417 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3418 (ix86_attribute_table): Swap affects_type_identity and handler
3419 fields, adjust comments.
3420 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3421 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3422 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3423 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3424 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3425 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3426 * config/rx/rx.c (rx_attribute_table): Likewise.
3427 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3428 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3429 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3430 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3431 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3432 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3433 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3434 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3435 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3436 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3437 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3439 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3441 (register_scoped_attributes, decl_attributes): Formatting fixes.
3443 PR tree-optimization/83269
3444 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3445 subtraction in arg0's type if type is signed and arg0 is unsigned.
3449 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3450 convert? on @0 instead of convert. Check type of @1, not @0.
3451 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3452 convert? on @0 instead of convert. Check type of @1, not @0.
3453 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3454 only optimize if either both @1 and @2 types are narrower
3455 precision, or both are wider or equal precision, and in the former
3456 case only if both have undefined overflow.
3458 2017-12-15 Richard Biener <rguenther@suse.de>
3461 * internal-fn.def (IFN_NOP): Add.
3462 * internal-fn.c (expand_NOP): Do nothing.
3463 * lto-streamer-in.c (input_function): Instead of removing
3464 sanitizer calls replace them with IFN_NOP calls.
3466 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3467 Alan Hayward <alan.hayward@arm.com>
3468 David Sherwood <david.sherwood@arm.com>
3470 * dse.c (store_info, read_info_type): Replace begin and end with
3472 (print_range): New function.
3473 (set_all_positions_unneeded, any_positions_needed_p)
3474 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3476 (record_store): Likewise. Optimize the case in which all positions
3478 (get_stored_val): Replace read_begin and read_end with read_offset
3480 (replace_read): Update call accordingly.
3482 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3484 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3485 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3486 (loop_cand::loop_cand): Initialize above members.
3487 (loop_cand::supported_operations): Delete.
3488 (loop_cand::can_interchange_p): Inline above function.
3489 (loop_cand::classify_simple_reduction): Record number of constant
3490 initialized simple reductions.
3491 (should_interchange_loops): New parameters. Check stmt cost of loops
3493 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3494 Update call to should_interchange_loops.
3495 (should_interchange_loop_nest): Update call to
3496 should_interchange_loops.
3498 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3501 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3502 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3503 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3505 2017-12-15 Julia Koval <julia.koval@intel.com>
3507 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3508 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3510 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3511 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3512 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3514 2017-12-15 Julia Koval <julia.koval@intel.com>
3516 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3517 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3518 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3519 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3520 _mm_aesenc_epi128): New intrinsics.
3522 2017-12-15 Julia Koval <julia.koval@intel.com>
3524 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3525 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3527 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3528 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3529 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3531 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3533 * gimple-ssa-strength-reduction.c (analyze_increments):
3534 Distinguish replacement costs for constant strides from those for
3537 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3539 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3541 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3543 * doc/invoke.texi: Document -Wcast-function-type.
3544 * recog.h (stored_funcptr): Change signature.
3545 * tree-dump.c (dump_node): Avoid warning.
3546 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3548 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3551 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3554 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3556 PR tree-optimization/83312
3557 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3558 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3559 "val" param, and to cope with arbitrary basic blocks.
3560 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3561 handle NULL_TREE for "val", dropping "bb" param.
3562 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3563 drop "bb" param. Handle NULL_TREE for "val".
3564 (find_case_label_for_value): Make "switch_stmt" param const.
3565 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3567 (vrp_prop::check_all_array_refs): Reimplement as...
3568 (check_array_bounds_dom_walker::before_dom_children): ...this new
3569 vfunc. Replace linear search through BB block list, excluding
3570 those with non-executable in-edges via dominator walk.
3572 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3574 * config/arm/arm.opt (mverbose-cost-dump): New option.
3575 * config/arm/arm.c (arm_rtx_costs): Use it.
3577 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3580 * reload1.c (emit_input_reload_insns): Skip debug markers.
3582 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3584 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3588 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3589 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3590 * config/sh/sh.c: Skip debug insns besides notes.
3591 * config/sh/sh.md: Likewise.
3592 * config/sh/sh_treg_combine.cc: Likewise.
3593 * config/sh/sync.md: Likewise.
3595 2017-12-14 Tom de Vries <tom@codesourcery.com>
3597 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3598 Add item for weak_undefined.
3600 2017-12-14 Richard Biener <rguenther@suse.de>
3602 PR tree-optimization/67842
3603 * sese.h (bb_in_region): Remove #if 0'ed code.
3605 2017-12-14 Richard Biener <rguenther@suse.de>
3607 PR tree-optimization/83326
3608 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3609 may_be_zero parameter and handle it by not marking the first
3610 peeled copy as not exiting the loop.
3611 (try_peel_loop): Likewise.
3612 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3613 to handle the case of constant or zero iterations and perform
3614 loop header copying on-the-fly.
3616 2017-12-14 Richard Biener <rguenther@suse.de>
3618 PR tree-optimization/83418
3619 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3620 Instead of asserting we don't get unfolded comparisons deal with them.
3622 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3625 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3626 basic blocks. Assert debug bind insns don't appear outside of bbs,
3627 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3630 PR tree-optimization/83198
3631 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3632 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3633 value if arg is a REAL_CST with incompatible type.
3635 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3636 Bin Cheng <bin.cheng@arm.com>
3639 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3641 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3644 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3646 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3647 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3649 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3650 Alan Hayward <alan.hayward@arm.com>
3651 David Sherwood <david.sherwood@arm.com>
3653 * poly-int.h: New file.
3654 * poly-int-types.h: Likewise.
3655 * coretypes.h: Include them.
3656 (POLY_INT_CONVERSION): Define.
3657 * target.def (estimated_poly_value): New hook.
3658 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3659 * doc/tm.texi: Regenerate.
3660 * doc/poly-int.texi: New file.
3661 * doc/gccint.texi: Include it.
3662 * doc/rtl.texi: Describe restrictions on subreg modes.
3663 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3664 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3665 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3666 * targhooks.h (default_estimated_poly_value): Declare.
3667 * targhooks.c (default_estimated_poly_value): New function.
3668 * target.h (estimated_poly_value): Likewise.
3669 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3670 (wi::unary_traits): Delete.
3671 (wi::binary_traits::signed_shift_result_type): Define for
3672 offset_int << HOST_WIDE_INT, etc.
3673 (generic_wide_int::operator <<=): Define for all types and use
3674 wi::lshift instead of <<.
3675 (wi::hwi_with_prec): Add a default constructor.
3676 (wi::ints_for): New class.
3677 (operator <<): Define for all wide-int types.
3678 (operator /): New function.
3679 (operator %): Likewise.
3680 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3681 (ASSERT_MAYBE_NE_AT): New macros.
3683 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3684 Dominik Vogt <vogt@linux.vnet.ibm.com>
3687 * emit-rtl.c (init_emit): Remove ??? comment.
3688 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3689 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3690 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3691 alignment of 3 virtual registers to BITS_PER_WORD.
3693 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3695 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3697 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3698 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3700 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3703 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3705 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3707 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3708 with the exception of debug begin stmt markers appear before
3712 * final.c (rest_of_handle_final): Call variable_tracking_main only
3713 if !flag_var_tracking.
3715 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3716 Jakub Jelinek <jakub@redhat.com>
3720 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3721 labels debug stmts that can only appear after labels.
3723 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3725 PR rtl-optimization/82398
3726 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3727 EXPR_USEFULNESS in priority comparison.
3729 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3731 PR rtl-optimization/83393
3732 * combine.c (move_deaths): If reg_stat points to a too new insn in
3733 last_death, do not use it: find the proper insn instead.
3735 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3738 * config/rs6000/rs6000-string.c (expand_block_compare,
3739 expand_strn_compare): Fix set_mem_size() calls.
3741 2017-12-12 Jeff Law <law@redhat.com>
3743 PR tree-optimization/83298
3744 PR tree-optimization/83362
3745 PR tree-optimization/83383
3746 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3747 push_value_range a public interface. Add new argument to
3748 record_ranges_from_stmt.
3749 * gimple-ssa-evrp-analyze.c
3750 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3751 Update comments. Handle recording temporary equivalences.
3752 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3753 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3754 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3755 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3756 gimple-ssa-evrp-analyze.h.
3757 (record_temporary_equivalences_from_phis): Add new argument. When
3758 the PHI arg is an SSA_NAME, set the result's range to the range
3760 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3761 from statements too.
3762 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3763 Pass it down to children as needed.
3764 (thread_outgoing_edges): Likewise.
3765 (thread_across_edge): Likewise. Push/pop range state as needed.
3766 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3768 2017-12-12 Julia Koval <julia.koval@intel.com>
3770 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3771 (PTA_CANNONLAKE): Remove PTA_CLWB.
3773 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3776 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3777 (vcondv2div2df): Likewise.
3778 (vconduv2dfv2di): Likewise.
3780 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3782 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3783 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3784 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3785 ATTR_MATHFN_FPROUNDING_ERRNO.
3787 2017-12-12 Richard Biener <rguenther@suse.de>
3789 PR tree-optimization/83385
3790 * graphite-scop-detection.c (get_order, order): Remove.
3791 (bb_to_rpo): New global.
3793 (build_scops): Sort pbbs in RPO order.
3795 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3797 * combine.c (simplify_set): Do not transform subregs to zero_extends
3798 if the destination mode is a vector mode.
3800 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3802 PR tree-optimization/83359
3803 * tree-cfg.h (fold_loop_internal_call): Declare.
3804 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3805 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3806 (find_loop_dist_alias): New function.
3807 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3808 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3811 PR tree-optimization/80631
3812 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3813 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3814 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3815 hardcoding zero as the value if COND_EXPR is never true. For
3816 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3817 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3818 hardcoding MAX_EXPR as the reduction operation.
3819 (is_nonwrapping_integer_induction): Allow negative step.
3820 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3821 vect_create_epilog_for_reduction, if no value is suitable, don't
3822 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3824 2017-12-12 Richard Biener <rguenther@suse.de>
3826 PR tree-optimization/81889
3827 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3828 range info from the non-wrapping IV instead of just the range
3831 2017-12-12 Julia Koval <julia.koval@intel.com>
3833 * config.gcc: Add vaesintrin.h.
3834 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3836 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3837 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3839 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3840 * config/i386/immintrin.h: Include vaesintrin.h.
3841 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3842 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3843 _mm_aesdec_epi128): New intrinsics.
3845 2017-12-12 Julia Koval <julia.koval@intel.com>
3847 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3848 OPTION_MASK_ISA_VAES_UNSET): New.
3849 (ix86_handle_option): Handle -mvaes.
3850 * config/i386/cpuid.h: Define bit_VAES.
3851 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3852 * config/i386/i386-c.c (__VAES__): New.
3853 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3854 (ix86_valid_target_attribute_inner_p): Ditto.
3855 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3856 * config/i386/i386.opt: Add -mvaes.
3857 * doc/invoke.texi: Ditto.
3859 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3861 * debug.h (gcc_debug_hooks): Add inline_entry.
3862 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3863 * debug.c (do_nothing_debug_hooks): Likewise.
3864 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3865 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3866 (dwarf2_lineno_debug_hooks): Likewise.
3868 * common.opt (gstatement-frontiers): New, setting
3869 debug_nonbind_markers_p.
3870 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3871 * toplev.c (process_options): Autodetect value for debug statement
3873 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3874 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3876 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3877 markers. Integrate source bind into debug stmt expand loop.
3878 (pass_expand::execute): Check debug marker limit. Avoid deep
3879 TER and expand debug locations for debug bind insns only.
3880 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3882 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3883 * final.c (reemit_insn_block_notes): Take current block from
3884 nonbind markers. Declare note where it's first set.
3885 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3886 begin stmt markers if enabled.
3887 (notice_source_line): Handle nonbind markers. Fail if their
3888 location is unknown or that of builtins.
3889 (rest_of_handle_final): Convert begin stmt markers to notes if
3890 var-tracking didn't run.
3891 (rest_of_clean_state): Skip begin stmt markers.
3892 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3894 * function.c (allocate_struct_function): Set begin_stmt_markers.
3895 * function.h (struct function): Add debug_marker_count counter
3896 and debug_nonbind_markers flag.
3897 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3898 * gimple-low.c (lower_function_body): Adjust
3899 debug_nonbind_markers.
3900 (lower_stmt): Drop or skip gimple debug stmts.
3901 (lower_try_catch): Skip debug stmts.
3902 * gimple.c (gimple_build_debug_begin_stmt): New.
3903 (gimple_copy): Increment debug_marker_count if copying one.
3904 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3905 * gimplify.c (rexpr_location): New.
3906 (rexpr_has_location): New.
3907 (warn_switch_unreachable_r): Handle gimple debug stmts.
3908 (shortcut_cond_r): Call expr_location.
3910 (find_goto_label): New.
3911 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3913 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3915 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3916 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3917 (LANG_HOOKS_INITIALIZER): ... this.
3918 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3919 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3921 (update_ebb_live_info): Skip debug insn markers.
3922 * lra.c (debug_insn_static_data): Rename to...
3923 (debug_bind_static_data): ... this.
3924 (debug_marker_static_data): New.
3925 (lra_set_insn_recog_data): Select one of the above depending
3927 (lra_update_isn_regno_info): Don't assume debug insns have
3929 (push_insns): Skip debug insns.
3930 * lto-streamer-in.c (input_function): Drop debug stmts
3931 depending on active options. Adjust debug_nonbind_markers.
3932 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3933 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3934 begin stmt marker notes.
3935 (print_insn): Likewise.
3936 * recog.c (extract_insn): Recognize rtl for debug markers.
3937 * rtl.def (DEBUG_MARKER): New.
3938 * tree-inline.c: Include params.h.
3939 (remap_gimple_stmt): Handle nonbind markers.
3940 (maybe_move_debug_stmts_to_successors): Likewise.
3941 (copy_debug_stmt): Likewise.
3942 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3943 markers regardless of no side effects.
3944 (tsi_link_before): Don't update container's side effects when adding
3945 a begin stmt marker.
3946 (tsi_link_after): Likewise.
3947 (expr_first): Skip begin stmt markers.
3948 (expr_last): Likewise.
3949 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3950 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3951 Disregard nonbind markers.
3952 * tree.c (make_node_stat): Don't set side effects for begin stmt
3954 (build1_stat): Likewise.
3955 * tree.def (DEBUG_BEGIN_STMT): New.
3956 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3957 * var-tracking.c (delete_debug_insns): Renamed to...
3958 (delete_vta_debug_insns): ... this.
3959 (reemit_marker_as_note): New.
3960 (vt_initialize): Reemit markers.
3961 (delete_vta_debug_insns): Likewise.
3962 (vt_debug_insns_local): Reemit or delete markers.
3963 (variable_tracking_main_1): Likewise.
3964 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3965 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3966 (gimple_debug_nonbind_marker_p): New.
3967 (gimple_build_debug_bind): Adjust.
3968 (gimple_build_debug_begin_stmt): New.
3969 * doc/invoke.texi (max-debug-marker-count): New param.
3970 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3971 (debug_parameter_ref, debug_marker): New.
3972 (NOTE_INSN_BEGIN_STMT): New.
3973 (DEBUG_INSN): Describe begin stmt markers.
3975 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3976 without debug insns, we wouldn't, but clean up debug insns
3977 after a control flow insn nevertheless.
3979 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3980 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3981 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3982 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3983 (rtl_tidy_fallthru_edge): Likewise.
3984 (rtl_verify_fallthru): Likewise.
3985 (rtl_verify_bb_layout): Likewise.
3986 (skip_insns_after_block): Likewise.
3987 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3988 * dwarf2out.c: Include print-rtl.h.
3989 (dwarf2out_next_real_insn): New.
3990 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3991 Dump debug binds in asm comments.
3992 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3993 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3994 callers to use gsi_start_nondebug_bb instead.
3995 (gsi_after_labels): Skip gimple debug stmts.
3996 (gsi_start_nondebug): New.
3997 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3998 (proper_loop_form_for_interchange): Adjust.
3999 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
4000 * gimple.h (gimple_seq_last_nondebug_stmt): New.
4001 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
4002 (collect_fallthrough_labels): Likewise.
4003 (should_warn_for_implicit_fallthrough): Likewise.
4004 (warn_implicit_fallthrough_r): Likewise.
4005 (expand_FALLTHROUGH_r): Likewise.
4006 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
4007 (graphite_copy_stmts_from_block): Skip nonbind markers.
4008 * haifa-sched.c (sched_extend_bb): Skip debug insns.
4009 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
4010 * jump.c (clean_barriers): Skip debug insns.
4011 * omp-expand.c (expand_parallel_call): Skip debug insns.
4012 (expand_task_call): Likewise.
4013 (remove_exit_barrier): Likewise.
4014 (expand_omp_taskreg): Likewise.
4015 (expand_omp_for_init_counts): Likewise.
4016 (expand_omp_for_generic): Likewise.
4017 (expand_omp_for_static_nochunk): Likewise.
4018 (expand_omp_for_static_chunk): Likewise.
4019 (expand_omp_simd): Likewise.
4020 (expand_omp_taskloop_for_outer): Likewise.
4021 (expand_omp_taskloop_for_inner): Likewise.
4022 (expand_oacc_for): Likewise.
4023 (expand_omp_sections): Likewise.
4024 (expand_omp_single): Likewise.
4025 (expand_omp_synch): Likewise.
4026 (expand_omp_atomic_load): Likewise.
4027 (expand_omp_atomic_store): Likewise.
4028 (expand_omp_atomic_fetch_op): Likewise.
4029 (expand_omp_atomic_pipeline): Likewise.
4030 (expand_omp_atomic_mutex): Likewise.
4031 (expand_omp_target): Likewise.
4032 (grid_expand_omp_for_loop): Likewise.
4033 (grid_expand_target_grid_body): Likewise.
4034 (build_omp_regions_1): Likewise.
4035 * omp-low.c (check_combined_parallel): Skip debug stmts.
4036 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
4037 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
4038 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
4040 * tree-cfg.c (make_blobs_1): Skip debug stmts.
4041 (make_edges): Likewise.
4042 (cleanup_dead_labels): Likewise.
4043 (gimple_can_merge_blocks_p): Likewise.
4044 (stmt_starts_bb_p): Likewise.
4045 (gimple_block_label): Likewise.
4046 (gimple_redirect_edge_and_branch): Likewise.
4047 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
4049 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
4051 * tree-pretty-print (print_declaration): Omit initializer in slim
4053 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
4055 (eliminate_unnecessary_stmts): Stabilize block removal order.
4056 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
4057 * var-tracking.c (get_first_insn): New.
4058 (vt_emit_notes): Call it.
4059 (vt_initialize): Walk any insns before the first BB.
4060 (delete_debug_insns): Likewise.
4062 * gimple.h (enum gimple_debug_subcode): Add
4063 GIMPLE_DEBUG_BEGIN_STMT.
4064 (gimple_debug_begin_stmt_p): New.
4065 (gimple_debug_nonbind_marker_p): New.
4066 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
4067 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
4068 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
4069 * insn-notes.def (BEGIN_STMT): New.
4070 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
4071 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
4072 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
4073 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
4074 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
4075 (INSN_DEBUG_MARKER_KIND): New.
4076 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
4077 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
4078 (INSN_VAR_LOCATION_PTR): New.
4079 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
4081 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
4082 deps only if debug bind insns are enabled.
4083 (pass_expand::execute): Avoid deep TER and expand
4084 debug locations for debug bind insns only.
4085 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
4086 debug stmts special handling down to debug bind stmts.
4087 * combine.c (try_combine): Narrow debug insns special handling
4088 down to debug bind insns.
4089 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
4090 Narrow debug insns preexisting special handling down to debug
4092 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
4093 handling down to debug bind insns.
4094 * function.c (instantiate_virtual_regs): Skip debug markers,
4095 adjust handling of debug binds.
4096 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
4097 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
4098 * haifa-sched.c (schedule_insn): Narrow special handling of debug
4099 insns to debug bind insns.
4100 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
4101 special handling of debug stmts to debug bind stmts.
4102 * ipa-split.c (split_function): Likewise.
4103 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
4104 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
4106 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
4107 * regrename.c (build_def_use): Likewise.
4108 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
4109 (pass_cprop_hardreg): Narrow special casing of debug insns to
4111 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
4112 * reload1.c (reload): Likewise.
4113 * sese.c (sese_insert_phis_for_liveouts): Narrow special
4114 casing of debug stmts to debug bind stmts.
4115 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
4116 * ssa-iterators.h (num_imm_uses): Likewise.
4117 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
4118 debug stmts to debug bind stmts.
4119 * tree-inline.c (tree_function_versioning): Narrow special casing
4120 of debug stmts to debug bind stmts.
4121 * tree-loop-distribution.c (generate_loops_for_partition):
4122 Narrow special casing of debug stmts to debug bind stmts.
4123 * tree-sra.c (analyze_access_subtree): Narrow special casing
4124 of debug stmts to debug bind stmts.
4125 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
4126 stmts to debug bind stmts.
4127 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
4128 casing of debug stmts to debug bind stmts.
4129 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
4130 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
4131 casing of debug stmts to debug bind stmts.
4132 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4134 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
4135 debug stmts to debug bind stmts.
4136 (gimple_replace_ssa_lhs): Likewise.
4137 (insert_debug_temp_for_var_def): Likewise.
4138 (insert_debug_temps_for_defs): Likewise.
4139 (reset_debug_uses): Likewise.
4140 * tree-ssanames.c (release_ssa_name_fn): Likewise.
4141 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
4142 (adjust_debug_stmts): Likewise.
4143 (adjust_phi_and_debug_stmts): Likewise.
4144 (vect_do_peeling): Likewise.
4145 * tree-vect-loop.c (vect_transform_loop): Likewise.
4146 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
4147 * var-tracking.c (adjust_mems): Narrow special casing of debug
4148 insns to debug bind insns.
4149 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
4150 (compute_bb_dataflow, vt_find_locations): Likewise.
4151 (vt_expand_loc, emit_notes_for_changes): Likewise.
4152 (vt_init_cfa_base): Likewise.
4153 (vt_emit_notes): Likewise.
4154 (vt_initialize): Likewise.
4155 (vt_finalize): Likewise.
4157 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
4158 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
4159 (next_nonnote_nondebug_insn_bb): New.
4160 (prev_nonnote_nondebug_insn_bb): New.
4161 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4162 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4163 (prev_nonnote_nondebug_insn_bb): Declare.
4164 (next_nonnote_nondebug_insn_bb): Declare.
4165 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4166 * cfgrtl.c (get_last_bb_insn): Likewise.
4167 * lra.c (push_insns): Likewise.
4169 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4172 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4174 * selftest.c (selftest::test_runner::test_runner): New ctor.
4175 (selftest::test_runner::~test_runner): New dtor.
4176 * selftest.h (class selftest::test_runner): New class.
4178 2017-12-11 Carl Love <cel@us.ibm.com>
4180 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4181 vec_extract_fp32_from_shortl]): Add #defines.
4182 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4183 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4184 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4185 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4186 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4187 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4188 * doc/extend.texi: Add documentation for the added builtins.
4190 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4192 PR rtl-optimization/80693
4193 PR rtl-optimization/81019
4194 PR rtl-optimization/81020
4195 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4196 are not mentioned in i3. Place the REG_UNUSED note on i2,
4197 possibly modified to REG_DEAD, if it did not originate in i3.
4199 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4201 * recog.c (store_data_bypass_p_1): New function.
4202 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4203 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4205 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4207 PR rtl-optimization/83361
4208 * ifcvt.c (if_convert): Call fixup_partitions.
4210 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4212 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4213 early folding of splat_u{8,16,32}.
4215 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4217 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4218 output_operand_lossage first argument with capital letter.
4219 (aarch64_override_options): Don't start error and sorry first argument
4220 with capital letter.
4222 2017-12-11 Andi Kleen <ak@linux.intel.com>
4224 PR gcov-profile/83355
4225 * auto-profile.c (string_table::get_index_by_decl): Don't
4226 recurse when abstract origin points to itself.
4228 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4230 PR tree-optimization/83320
4231 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4232 (prune_datarefs_not_in_loop): Ditto.
4234 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4236 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4238 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4240 PR tree-optimization/83337
4241 * gimple-loop-interchange.cc (compute_access_stride): Handle
4242 bitfield DRs properly.
4244 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4246 PR tree-optimization/83338
4247 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4248 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4249 vector of unsigned integers to vector of signed integers.
4251 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4253 PR rtl-optimization/83317
4254 * lra-constraints.c (process_address_1): Add insn code check.
4256 2017-12-08 Michael Matz <matz@suse.de>
4258 Fix PR tree-optimization/83323
4259 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4260 head-controlled loops and loop BBs.
4261 * common.opt (funroll-and-jam): Remove, instead ...
4262 (floop-unroll-and-jam): ... reuse this option.
4263 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4264 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4265 (-floop-unroll-and-jam): ... this option.
4267 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4269 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4270 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4271 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4273 2017-12-08 Richard Biener <rguenther@suse.de>
4276 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4277 handle accesses outside of zero-sized vars.
4279 2017-12-08 Martin Jambor <mjambor@suse.cz>
4281 PR tree-optimization/83141
4282 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4283 test for MEM_REFs implicitely changing types with padding. Remove
4285 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4287 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4289 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4291 * config/avr/avr.c (avr_attribute_table): Likewise.
4292 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4293 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4294 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4295 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4296 of struct attribute_spec.
4297 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4299 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4302 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4304 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4307 * profile-count.c (profile_count::from_gcov_type): Move from
4308 profile-count.h; handle overflow.
4309 * profile-count.h (profile_count::from_gcov_type): Move offline.
4311 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4313 PR rtl-optimization/83304
4314 * combine.c (move_deaths): If we do not know where a register died,
4317 2017-12-08 Richard Biener <rguenther@suse.de>
4319 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4320 Provide -fopt-info-loop feedback when we interchange in a nest.
4322 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4324 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4325 for armv6 ARM CPU IDs.
4327 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4329 * common/config/arm/arm-common.c: Include <algorithm>.
4330 (INCLUDE_VECTOR): Define.
4331 (compare_opt_names): New function.
4332 (arm_rewrite_selected_arch): Only strip out extensions that can be
4333 expressed through -mfpu. Sort the remaining extensions
4336 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4338 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4339 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4340 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4341 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4343 2017-06-08 Tristan Gingold <gindold@adacore.com>
4346 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4347 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4349 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4351 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4352 of struct attribute_spec.
4354 2017-12-08 Julia Koval <julia.koval@intel.com>
4356 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4357 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4358 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4359 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4360 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4361 _mm_maskz_dpwssds_epi32): Ditto.
4363 2017-12-08 Richard Biener <rguenther@suse.de>
4365 PR tree-optimization/81303
4366 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4367 conditions try to create a comparison vector type matching
4368 the data vector type.
4369 (vectorizable_condition): Adjust.
4370 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4371 Leave invariant conditions alone in case we can vectorize those.
4373 2017-12-08 Julia Koval <julia.koval@intel.com>
4375 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4376 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4377 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4378 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4379 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4381 2017-12-08 Julia Koval <julia.koval@intel.com>
4383 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4384 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4385 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4386 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4387 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4388 _mm_maskz_dpbusds_epi32): New intrinsics.
4390 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4392 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4394 (ld<bh>io_signed): Likewise.
4395 (st<bhw>io): Likewise.
4396 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4397 SMALL_INT12 constant integer operand.
4399 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4401 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4403 (nios2_large_constant_allowed): ...this. Adjust uses.
4404 (nios2_plus_symbolic_constant_p): Rename to...
4405 (nios2_plus_large_constant_p): ...this. Adjust uses.
4406 (nios2_legitimate_address_p): Correct CONST_INT handling.
4407 (nios2_symbolic_memory_operand_p): Rename to...
4408 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4409 (nios2_large_constant_p): Check for large constant integers too.
4410 (nios2_split_large_constant): Handle constant integers.
4411 (nios2_split_symbolic_memory_operand): Rename to...
4412 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4413 (nios2_legitimize_constant_address): Handle constant integers.
4414 (r0rel_constant_p): Handle small constant integers.
4415 (nios2_print_operand_address): Handle r0-relative integer addresses.
4416 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4417 * config/nios2/nios2.md: Adjust for renamed functions.
4419 2017-12-07 Andrew Waterman <andrew@sifive.com>
4421 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4422 (TARGET_HAVE_SRODATA_SECTION): New define.
4423 (riscv_select_section): New function.
4425 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4426 Alexander Monakov <amonakov@ispras.ru>
4427 Jakub Jelinek <jakub@redhat.com>
4430 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4432 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4434 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4435 a bool success value. Don't call output_operand_lossage here.
4436 (aarch64_print_ldpstp_address): Return a bool success value.
4437 (aarch64_print_operand_address): Call output_addr_const if
4438 aarch64_print_address_internal fails.
4439 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4440 'y'; call output_operand_lossage instead. Call output_operand_lossage
4441 if aarch64_print_ldpstp_address fails.
4443 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4445 * tree-vector-builder.h
4446 (tree_vector_builder::binary_encoded_nelts): Declare.
4447 * tree-vector-builder.c
4448 (tree_vector_builder::binary_encoded_nelts): New function.
4449 * fold-const.c (negate_expr_p): Likewise.
4450 (operand_equal_p, fold_checksum_tree): Likewise.
4451 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4452 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4453 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4454 (uniform_vector_p): Likewise.
4455 * varasm.c (const_hash_1, compare_constant): Likewise.
4456 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4457 (valid_lattice_transition): Operate directly on the VECTOR_CST
4459 * ipa-icf.c: Include tree-vector-builder.h.
4460 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4461 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4463 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4465 * tree.c (build_vector): Delete.
4466 * tree.h (build_vector): Make static and move into the self-testing
4469 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4471 * vector-builder.h (vector_builder::derived): New const overload.
4472 (vector_builder::elt): New function.
4473 * tree-vector-builder.h (tree_vector_builder::type): New function.
4474 (tree_vector_builder::apply_step): Declare.
4475 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4477 * gimple-fold.h (tree_vector_builder): Declare.
4478 (gimple_build_vector): Take a tree_vector_builder instead of a
4479 type and vector of elements.
4480 * gimple-fold.c (gimple_build_vector): Likewise.
4481 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4483 (get_initial_defs_for_reduction): Likewise.
4484 (vectorizable_induction): Likewise.
4486 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4488 * tree-vector-builder.h
4489 (tree_vector_builder::new_binary_operation): Declare.
4490 * tree-vector-builder.c
4491 (tree_vector_builder::new_binary_operation): New function.
4492 * fold-const.c (fold_relational_const): Use it.
4493 (const_binop): Likewise. Check that both input vectors have
4494 the same number of elements, thus excluding things like WIDEN_SUM.
4495 Check whether it is possible to operate directly on the encodings
4498 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4500 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4501 new_unary_operation, operating only on the encoded elements.
4502 (const_unop): Likewise.
4503 (exact_inverse): Likewise.
4504 (distributes_over_addition_p): New function.
4505 (const_binop): Use tree_vector_builder and new_unary_operation
4506 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4507 on the encoded elements unless the encoding is strided and the
4508 operation does not distribute over addition.
4509 (fold_convert_const): Use tree_vector_builder and
4510 new_unary_operation. Operate only on the encoded elements
4511 for truncating integer conversions, or for non-stepped encodings.
4513 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4515 * config/sparc/sparc.c: Include tree-vector-builder.h.
4516 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4517 * expmed.c: Include tree-vector-builder.h.
4518 (make_tree): Use tree_vector_builder instead of build_vector.
4519 * fold-const.c: Include tree-vector-builder.h.
4520 (const_binop): Use tree_vector_builder instead of build_vector.
4521 (const_unop): Likewise.
4522 (native_interpret_vector): Likewise.
4523 (fold_vec_perm): Likewise.
4524 (fold_ternary_loc): Likewise.
4525 * gimple-fold.c: Include tree-vector-builder.h.
4526 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4528 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4529 (simplify_vector_constructor): Use tree_vector_builder instead
4531 * tree-vect-generic.c: Include tree-vector-builder.h.
4532 (add_rshift): Use tree_vector_builder instead of build_vector.
4533 (expand_vector_divmod): Likewise.
4534 (optimize_vector_constructor): Likewise.
4535 * tree-vect-loop.c: Include tree-vector-builder.h.
4536 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4537 of build_vector. Explicitly use a stepped encoding for
4539 * tree-vect-slp.c: Include tree-vector-builder.h.
4540 (vect_get_constant_vectors): Use tree_vector_builder instead
4542 (vect_transform_slp_perm_load): Likewise.
4543 (vect_schedule_slp_instance): Likewise.
4544 * tree-vect-stmts.c: Include tree-vector-builder.h.
4545 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4546 (vect_gen_perm_mask_any): Likewise.
4547 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4548 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4550 (build_vector_from_val): Likewise. Explicitly use a duplicate
4553 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4555 * doc/generic.texi (VECTOR_CST): Describe new representation of
4557 * vector-builder.h: New file.
4558 * tree-vector-builder.h: Likewise.
4559 * tree-vector-builder.c: Likewise.
4560 * Makefile.in (OBJS): Add tree-vector-builder.o.
4561 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4562 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4563 (tree_vector): Change the number of elements to
4564 vector_cst_encoded_nelts.
4565 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4566 (VECTOR_CST_ELTS): Delete.
4567 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4568 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4569 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4570 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4571 (VECTOR_CST_ENCODED_ELT): Likewise.
4572 (vector_cst_encoded_nelts): New function.
4573 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4574 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4575 (vector_cst_int_elt, vector_cst_elt): Declare.
4576 * tree.c: Include tree-vector-builder.h.
4577 (tree_code_size): Abort if passed VECTOR_CST.
4578 (tree_size): Update for new VECTOR_CST layout.
4579 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4580 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4581 (build_vector): Use tree_vector_builder.
4582 (vector_cst_int_elt, vector_cst_elt): New functions.
4583 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4584 encoded elements and then create the vector in the canonical form.
4585 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4586 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4587 (tree_c_tests): Call test_vector_cst_patterns.
4588 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4590 (hash_tree): Likewise.
4591 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4592 (streamer_write_tree_header): Likewise.
4593 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4594 (streamer_alloc_tree): Likewise. Update call to make_vector.
4595 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4597 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4599 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4600 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4601 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4603 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4604 Richard Biener <rguenther@suse.de>
4606 PR tree-optimization/81303
4607 * Makefile.in (gimple-loop-interchange.o): New object file.
4608 * common.opt (floop-interchange): Reuse the option from graphite.
4609 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4610 -floop-interchange and mention it for -O3.
4611 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4612 * gimple-loop-interchange.cc: New file.
4613 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4614 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4615 * passes.def (pass_linterchange): New pass.
4616 * timevar.def (TV_LINTERCHANGE): New time var.
4617 * tree-pass.h (make_pass_linterchange): New declaration.
4618 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4619 interchange. Record IV before/after increment in new parameters.
4620 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4621 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4623 (check_reduction_path): ...New function here.
4624 * tree-vectorizer.h (check_reduction_path): New declaration.
4626 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4629 PR rtl-optimization/80818
4630 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4631 always early clobbered.
4632 * lra-lives.c (process_bb_lives): Check input hard regs for early
4633 clobbered non-operand hard reg.
4635 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4638 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4639 types_compatible_p, just that TYPE_MODE is the same.
4641 2017-12-07 Martin Sebor <msebor@redhat.com>
4644 * attribs.c (empty_attribute_table): Initialize new member of
4645 struct attribute_spec.
4646 (decl_attributes): Add argument. Handle mutually exclusive
4647 combinations of attributes.
4648 (selftests::test_attribute_exclusions): New function.
4649 (selftests::attribute_c_tests): Ditto.
4650 * attribs.h (decl_attributes): Add default argument.
4651 * selftest.h (attribute_c_tests): Declare.
4652 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4653 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4655 * doc/extend.texi (Common Function Attributes): Update const and pure.
4656 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4657 of struct attribute_spec.
4658 * config/arc/arc.c (arc_attribute_table): Same.
4659 * config/arm/arm.c (arm_attribute_table): Same.
4660 * config/avr/avr.c ( avr_attribute_table): Same.
4661 * config/bfin/bfin.c (bfin_attribute_table): Same.
4662 * config/cr16/cr16.c (cr16_attribute_table): Same.
4663 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4664 * config/h8300/h8300.c (h8300_attribute_table): Same.
4665 * config/i386/i386.c (ix86_attribute_table): Same.
4666 * config/ia64/ia64.c (ia64_attribute_table): Same.
4667 * config/m32c/m32c.c (m32c_attribute_table): Same.
4668 * config/m32r/m32r.c (m32r_attribute_table): Same.
4669 * config/m68k/m68k.c (m68k_attribute_table): Same.
4670 * config/mcore/mcore.c (mcore_attribute_table): Same.
4671 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4672 * config/mips/mips.c (mips_attribute_table): Same.
4673 * config/msp430/msp430.c (msp430_attribute_table): Same.
4674 * config/nds32/nds32.c (nds32_attribute_table): Same.
4675 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4676 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4677 * config/rl78/rl78.c (rl78__attribute_table): Same.
4678 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4679 * onfig/rx/rx.c (rx_attribute_table): Same.
4680 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4681 * config/sh/sh.c (sh_attribute_table): Same.
4682 * config/sparc/sparc.c (sparc_attribute_table): Same.
4683 * config/spu/spu.c (spu_attribute_table): Same.
4684 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4685 * config/v850/v850.c (v850_attribute_table): Same.
4686 * config/visium/visium.c (visium_attribute_table): Same.
4688 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4691 * config/arm/arm.c (INCLUDE_STRING): Define.
4692 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4693 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4696 2017-12-07 Michael Matz <matz@suse.de>
4698 Add unroll and jam pass
4700 * gimple-loop-jam.c: New file.
4701 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4702 * common.opt (funroll-and-jam): New option.
4703 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4704 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4705 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4706 * passes.def: Add pass_loop_jam.
4707 * timevar.def (TV_LOOP_JAM): Add.
4708 * tree-pass.h (make_pass_loop_jam): Declare.
4709 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4710 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4711 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4712 to flow_loop_tree_node_add.
4713 (duplicate_subloops, copy_loops_to): Append to sibling list.
4714 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4715 * doc/invoke.texi (-funroll-and-jam): Document new option.
4716 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4718 2017-12-07 Richard Biener <rguenther@suse.de>
4720 PR tree-optimization/83296
4721 PR tree-optimization/67769
4722 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4723 flow sensitive info in an unrelated BB.
4724 (value_replacement): Use reset_flow_sensitive_info.
4725 (minmax_replacement): Reset flow sensitive info on the def
4726 we move. Do not reset flow sensitive info in the whole BB
4727 we move the stmt to.
4728 (abs_replacement): Likewise.
4730 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4733 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4734 rs6000_cpu to the given -mcpu=, or to the default processor.
4736 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4738 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4739 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4740 instead of rs6000_cpu_attr.
4741 (rs6000_adjust_cost): Ditto.
4742 (is_microcoded_insn): Ditto.
4743 (rs6000_adjust_priority): Ditto.
4744 (rs6000_issue_rate): Ditto.
4745 (rs6000_use_sched_lookahead): Ditto.
4746 (rs6000_use_sched_lookahead_guard): Ditto.
4747 (rs6000_sched_reorder): Ditto.
4748 (force_new_group): Ditto.
4749 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4750 (group_ending_nop): Ditto.
4752 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4754 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4755 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4756 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4757 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4759 (rs6000_reassociation_width): Ditto.
4760 (rs6000_emit_epilogue): Ditto.
4761 (rs6000_adjust_cost): Ditto.
4762 (is_microcoded_insn): Ditto.
4763 (is_cracked_insn): Ditto.
4764 (rs6000_adjust_priority): Ditto.
4765 (rs6000_sched_reorder): Ditto.
4766 (rs6000_sched_reorder2): Ditto.
4767 (insn_must_be_first_in_group): Ditto.
4768 (insn_must_be_last_in_group): Ditto.
4769 (rs6000_register_move_cost): Ditto.
4770 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4773 2017-12-07 Julia Koval <julia.koval@intel.com>
4775 * config.gcc: Add vaesintrin.h.
4776 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4778 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4779 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4781 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4782 * config/i386/immintrin.h: Include vaesintrin.h.
4783 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4784 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4785 _mm_aesdec_epi128): New intrinsics.
4787 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4789 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4790 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4791 c-family/c-spellcheck.cc.
4792 (best_macro_match::best_macro_match): Likewise.
4794 (struct edit_distance_traits<cpp_hashnode *>): Move to
4795 c-family/c-spellcheck.h.
4796 (class best_macro_match): Likewise.
4798 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4800 PR tree-optimization/83293
4801 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4802 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4803 might insert into empty bb.
4806 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4807 simplify for plus with :c added, and pointer_plus without that.
4808 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4809 with undefined overflow and the conversion is not widening,
4810 perform negation in utype and only convert to type afterwards.
4811 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4812 simplify for plus with :c added, and pointer_plus without that.
4813 If type is integral with undefined overflow and the conversion is
4814 not widening, perform minus in utype and only convert to type
4815 afterwards. Move the last pointer_diff_expr simplify into the
4818 2017-12-06 Martin Sebor <msebor@redhat.com>
4820 PR tree-optimization/82646
4821 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4822 strncpy, not maxlen.
4824 2017-12-06 Martin Sebor <msebor@redhat.com>
4826 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4829 PR tree-optimization/83075
4830 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4831 strncat/strncpy don't change length of source string.
4833 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4836 2017-11-29 Martin Aberg <maberg@gaisler.com>
4838 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4839 to prevent b2bst errata sequence.
4840 (sqrtdf2_fix): Likewise.
4842 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4844 PR tree-optimization/81945
4845 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4846 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4847 to dest_cfun has orig_loop_num set, either remap it to the new
4848 loop number if the loop got moved too, or clear it.
4850 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4852 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4854 (thunderx2t99_nothing): New insn reservation.
4855 (thunderx2t99_mrs): New insn reservation.
4856 (thunderx2t99_multiple): New insn reservation.
4857 (thunderx2t99_alu_basi): Add bfx to reservation.
4858 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4860 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4863 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4865 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4867 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4868 tree-ssa-pre.c::remove_dead_inserted_code.
4869 * tree-ssa-dce.h: New file.
4870 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4871 (remove_dead_inserted_code): Move and rename to function
4872 tree-ssa-dce.c::simple_dce_from_worklist.
4873 (pass_pre::execute): Update use.
4875 2017-12-05 Richard Biener <rguenther@suse.de>
4877 PR tree-optimization/83277
4878 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4879 to code-gen liveout vars.
4881 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4883 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4884 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4885 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4888 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4890 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4892 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4893 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4896 2017-12-05 Richard Biener <rguenther@suse.de>
4898 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4899 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4900 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4901 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4903 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4905 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4907 2017-12-05 Martin Liska <mliska@suse.cz>
4908 Jakub Jelinek <jakub@redhat.com>
4910 * doc/invoke.texi: Document the options.
4911 * flag-types.h (enum sanitize_code): Add
4912 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4913 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4914 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4915 * opts.c: Define new sanitizer options.
4916 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4917 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4919 2017-12-05 Julia Koval <julia.koval@intel.com>
4921 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4922 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4923 (ix86_handle_option): Handle -mavx512vnni.
4924 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4925 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4926 * config/i386/i386-c (__AVX512VNNI__): New.
4927 * config/i386/i386.c (ix86_target_string): Handle new option.
4928 (ix86_valid_target_attribute_inner_p): Handle new option.
4929 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4930 * config/i386/i386.opt (mavx512vnni): New option.
4932 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4935 * athlon.md: Disable for generic.
4936 * haswell.md: Enable for generic.
4937 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4938 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4940 (ix86_adjust_cost): Move generic to haswell path.
4942 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4944 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4945 instead of MEM_P in a couple more places. Fix formatting issues.
4947 2017-12-04 Jim Wilson <jimw@sifive.com>
4949 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4950 instead of GP_REG_LAST-1.
4951 (riscv_adjust_libcall_cfi_prologue): Likewise.
4952 (riscv_adjust_libcall_cri_epilogue): Likewise.
4953 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4956 2017-12-04 Luis Machado <luis.machado@linaro.org>
4958 * ipa-pure-const.c (check_decl): Add missing newline.
4959 (state_from_flags): Likewise.
4961 2017-12-04 Jeff Law <law@redhat.com>
4963 PR tree-optimizatin/78496
4964 * gimple-ssa-evrp-analyze.h
4965 (evrp_range_analyzer::get_vr_values): Simplify.
4966 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4967 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4968 and gimple-ssa-evrp-analyze.h.
4969 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4970 (simplify_stmt_for_jump_threading): Copy a blob of code from
4971 tree-vrp.c to use ranges to simplify statements.
4972 (dom_opt_dom_walker::before_dom_children): Call
4973 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4974 (dom_opt_dom_walker::after_dom_children): Similarly for
4975 evrp_range_analyzer::leave.
4976 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4979 * gimple-ssa-evrp-analyze.c
4980 (evrp_range_analyzer::extract_range_from_stmt): Always use
4981 vr_values::update_value_range so preexisting range info is
4982 medged with new range info, even if the new range is VR_VARYING.
4984 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4986 * combine.c: Adjust comment.
4987 (use_crosses_set_p): Delete.
4988 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4989 (try_combine): Ditto.
4991 2017-12-04 Richard Biener <rguenther@suse.de>
4993 PR tree-optimization/83255
4994 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4995 Re-add zero-iteration check.
4997 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4999 PR rtl-optimization/83245
5000 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
5001 hard registers as earlyclobber, also if not in an asm.
5003 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5007 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5010 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5011 rs6000_cpu based on cpu_index, not tune_index.
5013 2017-12-04 Richard Biener <rguenther@suse.de>
5015 PR tree-optimization/83238
5016 * graphite-scop-detection.c (scop_detection::merge_sese): Make
5017 code match comment, rejecting invalid SESE regions.
5019 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
5021 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
5022 require base operand is a REG_POINTER prior to reload on targets
5023 with non-equivalent space registers.
5025 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5027 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
5028 (update_specialized_profile): Fix updating of counts.
5029 (perhaps_add_new_callers): Likewise.
5031 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5034 * x86-tune.def: Remove obsolette FIXMEs.
5035 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
5036 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
5037 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
5039 (X86_TUNE_PAD_RETURNS): Disable for generic.
5041 2017-12-02 Jakub Jelinek <jakub@redhat.com>
5043 PR tree-optimization/83170
5044 PR tree-optimization/83241
5045 * gimple-ssa-store-merging.c
5046 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
5047 gimple_vuse (ins_stmt) in case it has changed.
5048 (imm_store_chain_info::output_merged_store): Likewise.
5050 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
5054 * tree-cfg.c (pass_warn_function_return::execute): Handle
5055 __builtin_ubsan_handle_missing_return like __builtin_unreachable
5056 with BUILTINS_LOCATION.
5060 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
5061 is BLKmode for vector field with vector raw mode, use TYPE_MODE
5062 instead of DECL_MODE.
5064 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
5065 last argument to rtx pointer.
5066 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
5067 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
5068 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
5069 depending on the chosen ISAs.
5070 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
5071 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
5072 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
5074 * config/i386/sse.md (mov<mode>_internal): Likewise.
5075 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
5077 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5079 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
5081 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5083 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
5084 parameters from prototype.
5085 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
5086 parameters. Don't print a comment.
5087 (emit_fusion_gpr_load): Adjust.
5088 (emit_fusion_load_store): Adjust.
5089 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
5090 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
5091 comment on the second line.
5093 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5096 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5097 rs6000_cpu based on cpu_index, not tune_index.
5099 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5101 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
5102 which_alternative instead of which_alternative + 1.
5103 (output_asm_insn): Print an extra tab if the template is short.
5105 2017-12-01 Jim Wilson <jimw@sifive.com>
5107 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
5109 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
5110 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
5112 * doc/tm.texi: Regenerate.
5114 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
5117 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
5118 whether we can allocate pseudos before trying to fix an address.
5119 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
5120 memory address is indexed or indirect.
5121 (floatuns_<mode>si2_hw2): Likewise.
5123 2017-12-01 Jason Merrill <jason@redhat.com>
5125 * Makefile.in (TAGS): Add c-family/*.cc.
5127 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
5129 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
5130 (call_value_insn): Likewise.
5131 (sibcall_insn): Likewise.
5132 (sibcall_value_insn): Likewise.
5133 (movsi_aarch64): Likewise.
5134 (movdi_aarch64): Likewise.
5135 (add_losym_): Likewise.
5136 (ldr_got_small_): Likewise.
5137 (ldr_got_small_sidi): Likewise.
5138 (ldr_got_small_28k_): Likewise.
5139 (ldr_got_small_28k_sidi): Likewise.
5140 * config/aarch64/aarch64.c (aarch64_print_address_internal):
5141 Move output_addr_const to symbolic case. Add error check.
5143 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5145 * config/s390/predicates.md (plus16_Q_operand): New predicate.
5146 * config/s390/s390.md: Disable MVC merging peephole if it would
5147 disable operand forwarding.
5148 (new peephole2): Split MVCs if it would turn them into up to 2
5151 2017-12-01 Richard Biener <rguenther@suse.de>
5153 PR tree-optimization/83232
5154 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
5155 detection of same access. Instead of breaking the group here
5156 do not consider the duplicate. Add comment explaining real fix.
5158 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
5160 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5162 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5164 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5166 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5168 * function.h (struct function): Remove cilk_frame_decl,
5169 is_cilk_function and calls_cilk_spawn fields.
5170 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5172 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5173 cilk_elemental field.
5174 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5175 * target.def: Adjust comment.
5176 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5177 Don't test cilk_elemental.
5179 PR tree-optimization/83233
5180 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5181 bswap_stat name for the struct.
5184 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5185 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5186 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5187 SWITCH_BREAK_LABEL_P set on the label.
5188 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5189 added for default case if it was missing and not all cases covered.
5190 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5191 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5193 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5194 to avoid -Wimplicit-fallthrough warning.
5195 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5196 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5198 PR tree-optimization/83221
5199 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5201 (init_reassoc): Formatting fix.
5204 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5205 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5209 * tree-cfg.c: Include asan.h.
5210 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5212 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5214 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5216 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5218 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5219 vpcmpeqd instruction.
5221 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5223 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5224 registers type for 128bit mode.
5226 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5228 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5230 * gengtype-state.c (read_state_pair): Likewise.
5231 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5232 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5233 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5234 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5235 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5236 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5237 * builtins.c (fold_builtin_3): Likewise.
5238 * graphite-scop-detection.c
5239 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5240 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5243 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5246 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5247 parameter. Return true if ignore_decl_rtl_set_p is true and other
5248 conditions are satisfied.
5249 * asan.h (asan_protect_global): Add new parameter.
5250 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5251 to asan_protect_global calls.
5253 2017-11-30 Jim Wilson <jimw@sifive.com>
5255 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5256 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5257 -mstrict-align, add info on default value. Delete redundant lines for
5258 -mabi. Add missing -mexplicit-relocs docs.
5260 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5262 * config/arc/arc.md (trap): New pattern.
5264 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5266 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5267 instruction to end into a delay slot.
5268 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5269 can be placed into a delay slot against reg_note.
5271 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5273 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5274 labels number of usages.
5276 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5278 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5280 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5282 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5285 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5286 multiplication by power of 2 constant into two shifts + comparison.
5288 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5291 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5293 2017-11-30 Richard Biener <rguenther@suse.de>
5295 PR tree-optimization/83202
5296 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5297 allow_peel argument and guard peeling.
5298 (canonicalize_loop_induction_variables): Likewise.
5299 (canonicalize_induction_variables): Pass false.
5300 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5301 peeling from cunrolli.
5303 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5305 * combine.c (try_combine): Print a message to dump file whenever
5306 I0, I1, or I2 cannot be combined into I3.
5308 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5310 PR rtl-optimization/83156
5311 PR rtl-optimization/82621
5312 * combine.c (try_combine): Don't split an I2 if one of the dests is
5313 set again before I3. Allow unused dests.
5315 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5317 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5319 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5321 PR rtl-optimization/80818
5322 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5323 recursively. Use insn code for clobber.
5324 (lra_set_insn_recog_data): Pass the new arg to
5325 collect_non_operand_hard_regs.
5326 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5328 (lra_update_insn_regno_info): Pass insn to
5329 add_regs_to_insn_regno_info.
5331 2017-11-29 Jim Wilson <jimw@sifive.com>
5332 Andrew Waterman <andrew@sifive.com>
5334 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5335 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5336 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5337 use SINGLE_SHIFT_COST.
5338 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5339 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5341 2017-11-29 Julia Koval <julia.koval@intel.com>
5343 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5344 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5345 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5346 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5347 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5348 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5349 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5350 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5351 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5352 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5353 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5354 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5355 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5356 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5357 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5358 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5359 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5360 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5361 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5362 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5363 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5364 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5365 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5366 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5367 __builtin_ia32_vpshldv_v2di_mask,
5368 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5369 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5370 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5372 2017-11-29 Julia Koval <julia.koval@intel.com>
5374 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5375 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5376 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5377 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5378 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5379 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5380 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5381 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5382 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5383 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5384 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5385 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5386 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5387 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5388 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5389 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5390 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5391 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5392 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5393 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5394 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5396 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5398 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5399 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5402 2017-11-29 Martin Aberg <maberg@gaisler.com>
5404 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5405 to prevent b2bst errata sequence.
5406 (sqrtdf2_fix): Likewise.
5408 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5410 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5411 (div_sqrt_insn_p): New function.
5412 (sparc_do_work_around_errata): Insert NOP instructions to
5413 prevent sequences that could trigger the TN-0013 errata for
5414 certain LEON3 processors.
5415 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5416 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5417 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5418 (in_branch_delay): Prevent div and sqrt in delay slot if
5420 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5422 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5424 * config/sparc/sparc.c (atomic_insn_p): New function.
5425 (sparc_do_work_around_errata): Insert NOP instructions to
5426 prevent sequences that could trigger the TN-0010 errata for
5428 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5429 instruction referable in atomic_insns_p.
5431 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5433 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5434 (atomic_compare_and_swap_leon3_1): Likewise.
5437 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5439 * config/sparc/sparc.c (fpop_insn_p): New function.
5440 (sparc_do_work_around_errata): Insert NOP instructions to
5441 prevent sequences that could trigger the TN-0012 errata for
5443 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5444 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5445 (in_branch_annul_delay): Prevent floating-point instructions
5446 in delay slot of annulled integer branch.
5448 2017-11-29 Richard Biener <rguenther@suse.de>
5450 PR tree-optimization/83202
5451 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5453 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5454 nodes vectorized to the same stmts multiple times.
5455 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5456 (vect_analyze_slp_instance): Adjust.
5457 (scalar_stmts_to_slp_tree_map_t): New typedef.
5458 (vect_schedule_slp_instance): Add a map recording the SLP node
5459 representing the vectorized stmts for a set of scalar stmts.
5460 Avoid code-generating redundancies.
5461 (vect_schedule_slp): Allocate map and pass it down.
5463 2017-11-29 Nathan Sidwell <nathan@acm.org>
5466 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5469 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5471 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5472 cases for printing LDP/STP memory addresses.
5473 (aarch64_print_address_internal): Renamed from
5474 aarch64_print_operand_address, added parameter, add Pmode check.
5475 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5476 (aarch64_print_operand_address): Indirect to
5477 aarch64_print_address_internal.
5478 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5479 'y' operand output specifier.
5481 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5484 * tree.c (build_simple_mem_ref_loc): Handle
5485 get_addr_base_and_unit_offset returning a MEM_REF.
5488 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5491 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5494 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5496 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5498 (layout::move_to_column): Likewise.
5500 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5503 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5504 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5506 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5507 Martin Jambor <mjambor@suse.cz>
5510 * tree.h (expr_type_first_operand_type_p): Declare
5511 * tree.c (expr_type_first_operand_type_p): New function.
5512 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5513 (ipa_value_from_jfunc): Adjust declaration.
5514 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5515 Use it as result type for arithmetics, unless it is NULL in which case
5516 be more conservative.
5517 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5518 ipa_get_jf_pass_through_result.
5519 (propagate_vals_across_pass_through): Likewise.
5520 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5521 is to propagate_vals_across_pass_through.
5522 (propagate_constants_across_call): Pass PARM_TYPE to
5523 propagate_scalar_across_jump_function.
5524 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5525 ipa_value_from_jfunc.
5526 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5527 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5528 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5529 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5530 target_type, pass it to ipa_value_from_jfunc.
5531 (update_indirect_edges_after_inlining): Pass parameter type to
5532 try_make_edge_direct_simple_call.
5534 2017-11-28 Jeff Law <law@redhat.com>
5536 * gimple-ssa-evrp-analyze.c
5537 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5538 refine ranges if scev_initialized_p returns true.
5539 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5541 2017-11-28 Julia Koval <julia.koval@intel.com>
5543 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5544 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5545 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5546 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5547 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5548 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5549 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5550 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5551 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5552 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5553 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5554 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5555 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5556 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5557 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5558 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5559 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5560 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5561 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5562 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5563 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5564 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5566 2017-11-28 Julia Koval <julia.koval@intel.com>
5568 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5569 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5570 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5571 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5572 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5573 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5574 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5575 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5576 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5577 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5578 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5579 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5580 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5581 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5582 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5583 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5584 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5585 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5586 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5587 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5588 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5589 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5590 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5591 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5592 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5593 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5594 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5595 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5596 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5597 __builtin_ia32_vpshld_v2di_mask): New builtins.
5598 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5599 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5601 2017-11-28 Richard Biener <rguenther@suse.de>
5603 PR tree-optimization/80776
5604 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5606 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5608 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5609 If the incoming edge is an effective fallthru because the other
5610 edge only reaches a __builtin_unreachable () then record ranges
5611 derived from the controlling condition in SSA info.
5612 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5613 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5615 2017-11-28 Olivier Hainque <hainque@adacore.com>
5617 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5618 on mingw build hosts.
5620 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5623 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5625 2017-11-28 Richard Biener <rguenther@suse.de>
5628 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5629 copies generated from memcpy use a character array as reference
5632 2017-11-28 Julia Koval <julia.koval@intel.com>
5633 Sebastian Peryt <sebastian.peryt@intel.com>
5635 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5636 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5637 cilk-common.o, cilk.h, cilk-common.c): Remove.
5639 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5640 * builtins.c (is_builtin_name): Remove cilkplus condition.
5641 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5642 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5643 cilk-builtins.def, cilkplus.def): Remove.
5644 * cif-code.def (CILK_SPAWN): Remove.
5645 * cilk-builtins.def: Delete.
5646 * cilk-common.c: Ditto.
5648 * cilkplus.def: Ditto.
5649 * config/darwin.h (fcilkplus): Delete.
5650 * cppbuiltin.c: Ditto.
5651 * doc/extend.texi: Remove cilkplus doc.
5652 * doc/generic.texi: Ditto.
5653 * doc/invoke.texi: Ditto.
5654 * doc/passes.texi: Ditto.
5655 * gcc.c (fcilkplus): Remove.
5656 * gengtype.c (cilk.h): Remove.
5657 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5659 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5661 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5663 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5664 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5665 cilkplus conditions.
5666 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5667 inline_read_section): Ditto.
5668 * ipa-inline-analysis.c (cilk.h): Remove.
5669 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5670 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5671 append_linker_options): Remove condition for fcilkplus.
5672 * lto/lto-lang.c (cilk.h): Remove.
5673 (lto_init): Remove condition for fcilkplus.
5674 * omp-expand.c (expand_cilk_for_call): Delete.
5675 (expand_omp_taskreg, expand_omp_for_static_chunk,
5676 expand_omp_for): Remove cilkplus
5678 (expand_cilk_for): Delete.
5679 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5680 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5681 execute_lower_omp, diagnose_sb_0): Ditto.
5682 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5683 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5684 * tree-nested.c: Ditto.
5685 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5686 (dump_generic_node): Ditto.
5687 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5688 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5690 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5692 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5694 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5695 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5696 for vector mode and !TARGET_SIMD.
5698 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5700 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5702 * tree.h (SWITCH_LABELS): Remove.
5703 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5704 assert SWITCH_BODY is non-NULL.
5705 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5707 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5709 PR tree-optimization/80788
5710 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5711 has TREE_OVERFLOW set, call drop_tree_overflow.
5713 2017-11-28 Richard Biener <rguenther@suse.de>
5715 PR tree-optimization/83158
5716 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5718 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5721 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5722 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5724 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5726 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5727 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5729 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5732 * varasm.c (compare_constant): Take the mode of the constants into
5733 account when comparing floating point constants.
5735 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5737 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5738 from explicit instantiation of debug_helper.
5739 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5741 2017-11-27 Richard Biener <rguenther@suse.de>
5743 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5746 2017-11-27 Richard Biener <rguenther@suse.de>
5748 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5749 (type_hash_canon): Also clear int_cst_hash_table entry for
5751 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5753 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5755 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5756 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5757 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5759 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5761 * hash-map.h (gt_cleare_cache): Avoid UB.
5763 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5765 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5766 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5767 (decide_unroll_runtime_iterations): Likewise.
5768 (decide_unroll_stupid): Likewise.
5770 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5773 * config/i386/i386.md: Add a loop with incssp.
5775 2017-11-27 Martin Jambor <mjambor@suse.cz>
5777 PR tree-optimization/81248
5778 * tree-sra.c (splice_param_accesses): Remove size check.
5779 (decide_one_param_reduction): Fix size check.
5780 * gimple-pretty-print.c (dump_profile): Silence warning.
5781 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5783 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5786 * dbxout.c (lastlineno): New variable.
5787 (dbx_debug_hooks): Use dbxout_switch_text_section as
5788 switch_text_section debug hook.
5789 (dbxout_function_end): Switch to current_function_section
5790 rather than function_section. If crtl->has_bb_partition,
5791 output just one N_FUN, depending on in_cold_section_p.
5792 (dbxout_source_line): Remember last lineno in lastlineno.
5793 (dbxout_switch_text_section): New function.
5794 (dbxout_function_decl): Adjust dbxout_block caller.
5795 (dbx_block_with_cold_children): New function.
5796 (dbxout_block): Return true if any LBRAC/RBRAC have been
5797 emitted. Use dbx_block_with_cold_children at depth == 0
5798 in second partition. Add PARENT_BLOCKNUM argument, pass
5799 it optionally adjusted to children. Output LBRAC/RBRAC
5800 around recursive call only if the block is in the current
5801 partition, if not and anything was output, emit empty
5803 * final.c (final_scan_insn): Compute cold_function_name
5804 before calling switch_text_section debug hook. Call
5805 that hook even if dwarf2out_do_frame if not emitting
5809 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5810 TREE_READONLY decls.
5812 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5814 PR rtl-optimization/82488
5815 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5817 2017-11-26 Julia Koval <julia.koval@intel.com>
5819 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5821 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5824 2017-11-26 Julia Koval <julia.koval@intel.com>
5826 * config/i386/driver-i386.c (host_detect_local_cpu):
5827 Detect skylake-avx512.
5829 2017-11-26 Julia Koval <julia.koval@intel.com>
5831 * config.gcc: Add -march=cannonlake.
5832 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5833 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5834 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5835 (PTA_CANNONLAKE): New.
5836 (processor_target_table): Add cannonlake.
5837 (ix86_option_override_internal): Ditto.
5838 (fold_builtin_cpu): Ditto.
5839 (get_builtin_code_for_version): Handle cannonlake.
5840 (M_INTEL_COREI7_CANNONLAKE): New.
5841 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5842 * doc/invoke.texi: Add -march=cannonlake.
5844 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5846 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5847 (try_init_one_plugin): Alternative implementation for MinGW.
5848 * Makefile.in (plugin_implib): New.
5849 (gengtype-lex.c): Fix broken AIX workaround.
5850 * configure: Regenerate.
5851 * doc/plugins.texi: Document support for MinGW.
5853 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5855 PR rtl-optimization/81553
5856 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5857 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5858 is a shift where C1 has different mode than the whole shift, use C1's
5859 mode for MULT rather than the shift's mode.
5862 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5863 builtins not enabled in the currently selected ISA.
5865 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5867 PR tree-optimization/71026
5868 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5869 (insert_reciprocals): Change to insert reciprocals before a division
5870 by a square and to insert the square of a reciprocal.
5871 (execute_cse_reciprocals_1): Change to consider division by a square.
5872 (register_division_in): Add importance parameter.
5874 2017-11-24 Richard Biener <rguenther@suse.de>
5876 PR tree-optimization/82402
5877 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5878 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5880 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5882 * match.pd (0-ptr): New transformation.
5884 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5887 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5888 when skipping checking.
5890 2017-11-24 Richard Biener <rguenther@suse.de>
5892 PR tree-optimization/83128
5893 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5894 (vn_reference_lookup_3): Likewise.
5896 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5899 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5900 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5903 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5904 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5905 check for the latter with gimple_call_builtin_p. Do not handle
5906 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5908 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5910 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5913 2017-11-23 Julia Koval <julia.koval@intel.com>
5915 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5916 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5917 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5918 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5919 _mm512_maskz_expandloadu_epi16): New intrinsics.
5920 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5921 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5922 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5923 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5924 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5925 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5926 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5927 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5928 _mm256_maskz_expandloadu_epi8): New intrinsics.
5929 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5930 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5931 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5932 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5933 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5934 * config/i386/sse.md (VI248_VLBW): New iterator.
5935 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5937 2017-11-23 Julia Koval <julia.koval@intel.com>
5939 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5940 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5941 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5942 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5943 _mm512_mask_compressstoreu_epi16): New.
5944 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5945 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5946 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5947 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5948 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5949 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5950 _mm256_mask_compressstoreu_epi8): New.
5951 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5952 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5953 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5954 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5955 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5956 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5957 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5958 __builtin_ia32_compresshi128_mask,
5959 __builtin_ia32_compressstoreuqi512_mask,
5960 __builtin_ia32_compressstoreuhi512_mask,
5961 __builtin_ia32_compressstoreuqi256_mask,
5962 __builtin_ia32_compressstoreuqi128_mask,
5963 __builtin_ia32_compressstoreuhi256_mask,
5964 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5965 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5967 (ix86_expand_special_args_builtin): Handle new types.
5968 (s4fma_expand): Handle new builtin array.
5969 * config/i386/immintrin.h: Include new headers.
5970 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5971 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5973 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5975 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5976 fixes. Declare temp and g variables at the top in order to avoid
5977 {} in most of the cases.
5979 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5981 * match.pd (ptr-0): New transformation.
5983 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5985 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5986 (struct addr_mode_cost_table): New.
5987 (struct tune_params): Add field addr_mode_costs.
5988 * config/arm/arm.c (generic_addr_mode_costs): New.
5989 (arm_slowmul_tune): Initialise addr_mode_costs field.
5990 (arm_fastmul_tune): Likewise.
5991 (arm_strongarm_tune): Likewise.
5992 (arm_xscale_tune): Likewise.
5993 (arm_9e_tune): Likewise.
5994 (arm_marvell_pj4_tune): Likewise.
5995 (arm_v6t2_tune): Likewise.
5996 (arm_cortex_tune): Likewise.
5997 (arm_cortex_a8_tune): Likewise.
5998 (arm_cortex_a7_tune): Likewise.
5999 (arm_cortex_a15_tune): Likewise.
6000 (arm_cortex_a35_tune): Likewise.
6001 (arm_cortex_a53_tune): Likewise.
6002 (arm_cortex_a57_tune): Likewise.
6003 (arm_exynosm1_tune): Likewise.
6004 (arm_xgene1_tune): Likewise.
6005 (arm_cortex_a5_tune): Likewise.
6006 (arm_cortex_a9_tune): Likewise.
6007 (arm_cortex_a12_tune): Likewise.
6008 (arm_cortex_a73_tune): Likewise.
6009 (arm_v7m_tune): Likewise.
6010 (arm_cortex_m7_tune): Likewise.
6011 (arm_v6m_tune): Likewise.
6012 (arm_fa726te_tune): Likewise.
6013 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
6015 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
6017 * config/arm/arm.c (arm_mem_costs): New function.
6018 (arm_rtx_costs_internal): Use arm_mem_costs.
6020 2017-11-23 Mark Wielaard <mark@klomp.org>
6022 * dwarf2out.c (init_sections_and_labels): Use generation to create
6023 unique ranges_section_label and ranges_base_label. Return generation.
6024 (output_rnglists): Add generation argument. Use generation to create
6025 unique ranges labels.
6026 (dwarf2out_finish): Get generation from init_sections_and_labels
6027 and pass generation to output_rnglists.
6029 2017-11-23 Mike Stump <mikestump@comcast.net>
6030 Eric Botcazou <ebotcazou@adacore.com>
6032 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
6033 * cfgloop.h (struct loop): Add unroll field.
6034 * function.h (struct function): Add has_unroll bitfield.
6035 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
6036 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
6037 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
6038 (pass_rtl_unroll_loops::gate): Likewise.
6039 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
6040 for which loop->unroll==1.
6041 (decide_unroll_constant_iterations): Use note for consistency and
6042 take loop->unroll into account. Return early if loop->unroll is set.
6043 Fix thinko in existing test.
6044 (decide_unroll_runtime_iterations): Use note for consistency and
6045 take loop->unroll into account.
6046 (decide_unroll_stupid): Likewise.
6047 * lto-streamer-in.c (input_cfg): Read loop->unroll.
6048 * lto-streamer-out.c (output_cfg): Write loop->unroll.
6049 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
6051 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
6052 (print_loop): Print loop->unroll if set.
6053 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
6054 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
6055 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
6057 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
6058 loop->unroll is set and smaller than the trip count. Otherwise bypass
6059 entirely the heuristics if loop->unroll is set. Remove dead note.
6060 Fix off-by-one bug in other note.
6061 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
6062 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
6064 (tree_unroll_loops_completely): Make static.
6065 (pass_complete_unroll::execute): Use correct type for variable.
6066 (pass_complete_unrolli::execute): Fix formatting.
6067 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
6069 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
6071 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
6072 enable when TARGET_PREFER_AVX128 is set.
6074 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6076 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
6078 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
6079 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
6081 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6083 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
6084 frequencies. Fix estimation of aggregate parameters.
6086 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6088 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
6091 2017-11-23 Tom de Vries <tom@codesourcery.com>
6093 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
6095 2017-11-23 Tom de Vries <tom@codesourcery.com>
6097 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
6100 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
6103 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
6104 sibcall_value_pcrel_fdpic): Use local variable instead of
6106 (calli_tbr_rel): Add missing operand 2.
6107 (call_valuei_tbr_rel): Add missing operand 3.
6109 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6112 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
6113 bitpos/bitsize covering the whole destination, use store_expr only if
6114 the complex mode is the same. Otherwise, use expand_normal and if
6115 it returns CONCAT, subreg each part separately instead of trying to
6116 subreg the whole result.
6118 2017-11-23 Richard Biener <rguenther@suse.de>
6120 PR tree-optimization/23094
6121 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
6122 come here from walking over backedges in the first iteration.
6123 (vn_reference_lookup_3): Skip clobbers that store the same value.
6125 2017-11-23 Richard Biener <rguenther@suse.de>
6127 PR tree-optimization/81403
6128 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
6129 a block we need a leader relative to.
6130 (phi_translate_1): For nary processing require a leader from
6131 get_representative_for given we run expression simplification
6132 using match-and-simplify. Remove previous fix.
6134 2017-11-22 Jeff Law <law@redhat.com>
6136 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
6137 Use new method allocate_value_range rather than accessing the
6138 vrp_value_range_pool data member directly.
6139 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
6140 to use extract_range_from_stmt method to avoid need for
6141 extract_range_from_assignment method.
6142 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
6143 method rather than setting values_propgated data member directly.
6144 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
6145 and values propagated data members and extract_range_from_assignment
6146 method. Reorder private data members to conform to standards.
6147 Add new methods set_lattice_propagation_complete and
6148 allocate_value_range.
6150 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
6152 PR rtl-optimization/83030
6153 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
6154 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
6155 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
6156 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
6157 CROSSING_JUMP_P flag is already set before setting it.
6158 * cfgrtl.c (fixup_partition_crossing): Likewise.
6159 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6162 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6164 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6165 Handle the case where both arguments are using gen_const_vec_series.
6167 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6170 * pretty-print.c (pp_format): Move quoting implementation to
6171 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6172 to pass address of "quote" local.
6173 (pp_begin_quote): New function.
6174 (pp_end_quote): New function.
6175 * pretty-print.h (printer_fn): Convert penultimate param from bool
6177 (pp_begin_quote): New decl.
6178 (pp_end_quote): New decl.
6179 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6180 param from bool to bool *.
6181 * tree-diagnostic.h (default_tree_printer): Likewise.
6183 2017-11-22 Jeff Law <law@redhat.com>
6185 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6186 Initialize vr_values.
6187 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6188 vr_values via vr_values class instance rather than delegators.
6189 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6190 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6191 (evrp_range_analyzer::push_value_range): Likewise.
6192 (evrp_range_analyzer::pop_value_range): Likewise.
6193 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6194 most delegators. Those remaining are exposed as public interfaces.
6195 Make vr_values a pointer and private.
6196 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6198 (evrp_range_analyzer::get_vr_value): New method.
6199 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6200 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6201 (class evrp_dom_walker): Attach evrp_folder class, initialize
6202 it in the ctor. Remove temporary delegators.
6203 (evrp_dom_walker::before_dom_children): Call methods in attached
6204 evrp_range_analyzer class via class instance pointer. Use
6205 free value_range_constant_singleton to remove need for
6206 op_with_constant_singleton_value delegator method. Do not
6207 create a vrp_prop class instance for every call! Narrow
6208 scope of a couple variables.
6209 (evrp_dom_walker::cleanup): Call methods in attached
6210 evrp_range_analyzer class via class instance pointer.
6211 * vr-values.h (class vr_values): Privatize many methods and
6214 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6216 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6217 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6218 * expr.c (expand_expr_real_2): Likewise.
6219 * fold-const.c (const_unop): Likewise.
6220 * optabs-tree.c (optab_for_tree_code): Likewise.
6221 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6222 * tree-inline.c (estimate_operator_cost): Likewise.
6223 * tree-pretty-print.c (dump_generic_node): Likewise.
6224 (op_code_prio): Likewise.
6225 (op_symbol_code): Likewise.
6226 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6227 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6228 * internal-fn.c (direct_internal_fn_optab): New function.
6229 (direct_internal_fn_array, direct_internal_fn_supported_p
6230 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6231 * fold-const-call.c (fold_const_reduction): New function.
6232 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6234 * tree-vect-loop.c: Include internal-fn.h.
6235 (reduction_code_for_scalar_code): Rename to...
6236 (reduction_fn_for_scalar_code): ...this and return an internal
6238 (vect_model_reduction_cost): Take an internal_fn rather than
6240 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6242 (vectorizable_reduction): Use internal functions rather than tree
6243 codes for the reduction operation. Update calls to the functions
6245 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6246 Use calls to internal functions rather than REDUC tree codes.
6247 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6249 2017-11-22 Olivier Hainque <hainque@adacore.com>
6251 * config/vxworks.c (vxworks_override_options): Pick default
6252 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6253 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6254 DWARF_GNAT_ENCODINGS_DEFAULT.
6255 * config/vxworksae.h: Likewise.
6257 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6259 PR tree-optimization/83104
6260 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6263 2017-11-22 Marek Polacek <polacek@redhat.com>
6264 H.J. Lu <hongjiu.lu@intel.com>
6265 Jason Merrill <jason@redhat.com>
6270 * calls.c (initialize_argument_information): Call
6271 warn_parameter_passing_abi target hook.
6272 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6273 argument onto stack.
6274 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6275 * common.opt: Update -fabi-version description.
6276 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6277 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6279 (ix86_is_empty_record): New function.
6280 (ix86_warn_parameter_passing_abi): New function.
6281 (TARGET_EMPTY_RECORD_P): Redefine.
6282 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6283 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6284 * doc/tm.texi: Regenerated.
6285 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6286 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6287 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6288 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6289 instead of int_size_in_bytes.
6290 * expr.c (copy_blkmode_to_reg): Likewise.
6291 * function.c (aggregate_value_p): Return 0 for empty types.
6292 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6294 (locate_and_pad_parm): Call arg size_in_bytes instead
6296 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6297 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6298 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6300 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6301 (std_gimplify_va_arg_expr): Skip empty records. Call
6302 arg_size_in_bytes instead size_in_bytes.
6303 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6304 * tree-core.h (tree_type_common): Add empty_flag.
6305 (tree_decl_common): Update comments.
6306 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6308 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6309 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6311 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6312 * tree.c (default_is_empty_type): New function.
6313 (default_is_empty_record): New function.
6314 (arg_int_size_in_bytes): New function.
6315 (arg_size_in_bytes): New function.
6316 (get_ultimate_context): New function.
6317 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6318 TRANSLATION_UNIT_WARN_EMPTY_P.
6319 (default_is_empty_record, arg_int_size_in_bytes,
6320 arg_size_in_bytes, get_ultimate_context): Declare.
6322 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6324 * config/arm/arm.c (cmse_clear_registers): New function.
6325 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6326 code by call to cmse_clear_registers.
6327 (cmse_nonsecure_entry_clear_before_return): Likewise.
6329 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6331 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6332 (vdot_s32, vdotq_s32): New.
6333 (vdot_lane_u32, vdotq_lane_u32): New.
6334 (vdot_lane_s32, vdotq_lane_s32): New.
6337 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6340 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6341 for unsigned values with fewer HWIs than the precision.
6342 (test_overflow): New function.
6343 (wide_int_cc_tests): Call it.
6345 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6347 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6348 has the same mode class as Pmode.
6350 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6352 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6353 padding_bits_to_clear_ptr.
6354 (cmse_nonsecure_entry_clear_before_return): Likewise.
6356 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6358 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6359 auto_sbitap instead of integer bitfield to control register needing
6362 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6364 PR tree-optimization/83044
6365 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6366 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6367 that need the upper bound. Subtract offset from
6368 get_addr_base_and_unit_offset only if positive and subtract it
6369 before division by eltsize rather than after it.
6372 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6373 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6374 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6378 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6379 expand_binop with *mul_widen_optab, make sure at least one of the
6380 operands doesn't have VOIDmode.
6383 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6385 PR rtl-optimization/82044
6386 PR tree-optimization/82042
6387 * dse.c (record_store): Check for overflow.
6388 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6389 add_wild_read instead of clear_rhs_from_active_local_stores on
6390 overflow and log it into dump_file.
6392 2017-11-22 Richard Biener <rguenther@suse.de>
6394 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6395 fake edges to exit when looking for a place to insert.
6396 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6398 (insert_into_preds_of_block): Commit edge insertion immediately,
6399 assert that doesn't require new BBs.
6400 (fini_pre): Release expressions.
6401 (pass_pre::execute): Shuffle things around a bit, if the fn
6402 is too large do not compute AVAIL either as this is really the
6405 2017-11-22 Richard Biener <rguenther@suse.de>
6407 PR tree-optimization/83089
6408 * tree-if-conv.c (pass_if_conversion::execute): If anything
6409 changed reset SCEV and free the number of iteration estimates.
6411 2017-11-21 Martin Sebor <msebor@redhat.com>
6413 PR tree-optimization/82945
6414 * calls.h (warn_nonstring_bound): Remove unused function.
6416 2017-11-21 Martin Sebor <msebor@redhat.com>
6418 PR tree-optimization/82945
6419 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6420 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6421 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6423 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6424 * calls.h (get_attr_nonstring_decl): Declare new function.
6425 * doc/extend.texi (attribute nonstring): Update.
6426 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6427 get_attr_nonstring_decl and handle it.
6428 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6429 detection of nul-termination.
6430 (strlen_to_stridx): Change to a pointer.
6431 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6432 (pass_strlen::execute): Same.
6434 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6436 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6437 for the new option -mprefer-vector-width=[none|128|256|512].
6438 * config/i386/i386.c (ix86_target_string): remove old style options
6439 -mprefer-avx256 and make -mprefer-avx128 as alias.
6440 (ix86_option_override_internal): Apply defaults for the
6441 -mprefer-vector-width=[128|256] option.
6442 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6443 Implement macros to work with -mprefer-vector-width=.
6444 * config/i386/i386.opt: Implemented option
6445 -mprefer-vector-width=[none|128|256|512].
6446 * doc/invoke.texi: Documentation for
6447 -mprefer-vector-width=[none|128|256|512].
6449 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6451 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6453 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6455 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6457 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6459 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6461 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6462 * expr.c (expand_expr_real_2): Likewise.
6463 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6464 fold_binary_loc): Likewise.
6465 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6466 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6467 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6468 MINUS_EXPR transformations.
6469 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6470 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6471 * tree-inline.c (estimate_operator_cost): Likewise.
6472 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6473 op_symbol_code): Likewise.
6474 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6475 * vr-values.c (extract_range_from_binary_expr): Likewise.
6476 * varasm.c (initializer_constant_valid_p_1): Likewise.
6477 * tree.def: New tree code POINTER_DIFF_EXPR.
6479 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6481 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6482 integer suffix to movbe mnemonic.
6483 (*bswaphi2_movbe): Ditto.
6484 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6486 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6489 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6490 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6491 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6493 2017-11-21 Martin Liska <mliska@suse.cz>
6495 * tree-inline.c (expand_call_inline): Remove not needed
6498 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6499 Jakub Jelinek <jakub@redhat.com>
6502 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6503 Remove static keyword from f variable.
6505 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6507 PR tree-optimization/83086
6508 * gimple-ssa-store-merging.c
6509 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6510 rather than n.base_addr.
6512 2017-11-21 Martin Liska <mliska@suse.cz>
6514 PR rtl-optimization/82044
6515 PR tree-optimization/82042
6516 * dse.c (check_mem_read_rtx): Check for overflow.
6518 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6520 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6523 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6525 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6526 and its corresponding call arg location note.
6528 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6529 Andrew Burgess <andrew.burgess@embecosm.com>
6531 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6533 (arc_return_slot_offset): Likewise.
6534 (arc_eh_return_address_location): New declaration.
6535 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6536 (MUST_SAVE_REGISTER): Add exception handler case.
6537 (MUST_SAVE_RETURN_ADDR): Likewise.
6538 (arc_frame_pointer_required): Likewise.
6539 (arc_frame_pointer_needed): New function.
6540 (arc_compute_frame_size): Changed.
6541 (arc_expand_prologue): Likewise.
6542 (arc_expand_epilogue): Likewise.
6543 (arc_initial_elimination_offset): Likewise.
6544 (arc_return_slot_offset): Delete.
6545 (arc_eh_return_address_location): New function.
6546 (arc_builtin_setjmp_frame_value): Likewise.
6547 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6548 (EH_RETURN_STACKADJ_RTX): Define.
6549 (EH_RETURN_HANDLER_RTX): Likewise.
6550 * config/arc/arc.md (eh_return): Delete.
6552 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6554 * print-tree.h (debug_vec_tree): Remove prototype.
6555 * gdbinit.in (pvt): Remove macro.
6557 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6559 PR tree-optimization/83047
6560 * gimple-ssa-store-merging.c
6561 (imm_store_chain_info::output_merged_store): If the loads with the
6562 same vuse are in different basic blocks, for load_gsi pick a load
6563 location that is dominated by the other loads.
6566 * config/i386/i386.c (ix86_memmodel_check): Start
6567 -Winvalid-memory-model diagnostics with lowercase letter.
6570 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6571 set in_cold_section_p to the partition containing loc_list->first.
6572 When seeing loc_list->last_before_switch node, update secname and
6573 perform range_across_switch second partition handling only after that.
6576 * run-rtl-passes.c: Include debug.h.
6577 (run_rtl_passes): Call debug_hooks->assembly_start.
6578 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6582 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6583 OPTAB_DIRECT in calls to expand_simple_binop.
6585 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6588 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6590 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6593 * spellcheck.h (best_match::blithely_get_best_candidate): New
6596 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6598 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6599 with lower case letter.
6601 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6603 * config/i386/i386.md (bswaphi2): New expander.
6604 (*bswaphi2_movbe): New insn pattern.
6605 (bswaphi -> rorhi pepehole2): New peephole pattern.
6607 2017-11-20 Jeff Law <law@redhat.com>
6609 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6610 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6611 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6612 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6613 gimple-ssa-evrp-analyze.h.
6615 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6616 set BB_VISITED here.
6617 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6619 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6621 * tree-predcom.c: Add general comment on Store-Store chains.
6622 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6624 (get_chain_last_ref_at): Rename into...
6625 (get_chain_last_write_at): ...this.
6626 (get_chain_last_write_before_load): New function.
6627 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6628 CT_STORE_STORE when write reference is added.
6629 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6630 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6631 (initialize_root_vars_store_elim_1): Ditto.
6632 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6633 definition is created.
6634 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6635 chain by replacing it with dominant stored value.
6637 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6639 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6641 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6643 * vr-values.c (extract_range_from_binary_expr): Use a full range
6646 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6648 * config/arm/arm.md (R4_REGNUM): Define constant.
6649 (nonsecure_call_internal): Remove r4 clobber.
6650 (nonsecure_call_value_internal): Likewise.
6651 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6652 clobber and resequence match_operands.
6653 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6654 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6655 (nonsecure_call_value_reg_thumb2): Likewise.
6657 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6659 PR tree-optimization/78821
6660 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6661 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6662 to the address of the base rather than the base itself.
6663 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6664 (find_bswap_or_nop_finalize): New function.
6665 (find_bswap_or_nop): Use it.
6666 (bswap_replace): Return a tree rather than bool, change first
6667 argument from gimple * to gimple_stmt_iterator, allow inserting
6668 into an empty sequence, allow ins_stmt to be NULL - then emit
6669 all stmts into gsi. Fix up MEM_REF address gimplification.
6670 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6671 (struct store_immediate_info): Add N and INS_STMT non-static
6673 (store_immediate_info::store_immediate_info): Initialize them
6674 from newly added ctor args.
6675 (merged_store_group::apply_stores): Formatting fixes. Sort by
6677 (stmts_may_clobber_ref_p): For stores call also
6678 refs_anti_dependent_p.
6679 (gather_bswap_load_refs): New function.
6680 (imm_store_chain_info::try_coalesce_bswap): New method.
6681 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6682 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6683 (imm_store_chain_info::output_merged_store): Fail if number of
6684 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6685 and NOP_EXPR rhs_code.
6686 (pass_store_merging::process_store): Compute n and ins_stmt, if
6687 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6688 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6690 (pass_store_merging::execute): Calculate dominators.
6692 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6693 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6694 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6695 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6696 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6697 class pass_optimize_bswap, bswap_replace,
6698 pass_optimize_bswap::execute): Moved to ...
6699 * gimple-ssa-store-merging.c: ... this file.
6700 Include optabs-tree.h.
6701 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6702 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6703 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6704 anonymous namespace, remove static keywords.
6705 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6706 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6708 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6711 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6713 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6715 * vec.h (debug_helper): New function.
6716 (DEFINE_DEBUG_VEC): New macro.
6717 * hash-set.h (debug_helper): New function.
6718 (DEFINE_DEBUG_HASH_SET): New macro.
6719 * cfg.c (debug_slim (edge)): New function.
6720 Call DEFINE_DEBUG_VEC for edges.
6721 Call DEFINE_DEBUG_HASH_SET for edges.
6722 * cfghooks.c (debug_slim (basic_block)): New function.
6723 Call DEFINE_DEBUG_VEC for basic blocks.
6724 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6725 * print-tree.c (debug_slim): New function to handle trees.
6726 Call DEFINE_DEBUG_VEC for trees.
6727 Call DEFINE_DEBUG_HASH_SET for trees.
6728 (debug (vec<tree, va_gc>) &): Remove.
6729 (debug (<vec<tree, va_gc>) *): Remove.
6730 * print-rtl.c (debug_slim): New function to handle const_rtx.
6731 Call DEFINE_DEBUG_VEC for rtx_def.
6732 Call DEFINE_DEBUG_VEC for rtx_insn.
6733 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6734 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6735 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6736 (debug (vec<rtx_insn *> *ptr): Remove.
6737 (debug_insn_vector): Remove.
6738 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6740 2017-11-20 Tom de Vries <tom@codesourcery.com>
6742 PR rtl-optimization/82020
6743 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6744 IF_THEN_ELSE condition.
6746 2017-11-19 Jeff Law <law@redhat.com>
6748 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6749 of degenerates resulting from ignoring an edge.
6751 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6754 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6756 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6759 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6760 for uninitialied counts.
6762 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6765 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6767 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6770 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6772 2017-11-19 Tom de Vries <tom@codesourcery.com>
6774 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6777 2017-11-19 Tom de Vries <tom@codesourcery.com>
6779 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6782 2017-11-19 Tom de Vries <tom@codesourcery.com>
6784 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6787 2017-11-19 Tom de Vries <tom@codesourcery.com>
6789 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6791 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6793 2017-11-19 Tom de Vries <tom@codesourcery.com>
6795 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6797 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6798 semicolon after MCORE_EXPORT_NAME call.
6800 2017-11-19 Tom de Vries <tom@codesourcery.com>
6803 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6804 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6806 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6808 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6809 (cgraph_edge::clone): Cleanup updating of profile.
6810 * ipa-cp.c (update_profiling_info): Likewise.
6811 * ipa-inline-transform.c (inline_transform): Likewise.
6812 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6813 * ipa-split.c (execute_split_functions): Do not split when function
6815 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6816 * profile-count.c (profile_count::dump): Add global0.
6817 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6819 (profile_count::to_sreal_scale): Likewise.
6820 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6821 (profile_count::combine_with_ipa_count): New function.
6822 * profile-count.h (profile_guessed_global0adjusted): New.
6823 (profile_count::adjusted_zero): New.
6824 (profile_count::global0adjusted): New.
6825 (profile_count::combine_with_ipa_count): New.
6826 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6827 correct profile of return block of split functions.
6828 (copy_cfg_body): Remove unused profile_count.
6829 (copy_body): Likewise.
6830 (expand_call_inline): Update.
6831 (tree_function_versioning): Update.
6833 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6835 * hash-set.h (hash_set::empty): New.
6836 * tree-ssa-threadbackward.h: Delete.
6837 * tree-ssa-threadbackward.c (class thread_jumps): New.
6838 Move max_threaded_paths into class.
6839 (fsm_find_thread_path): Remove arguments that are now in class.
6840 (profitable_jump_thread_path): Rename to...
6841 (thread_jumps::profitable_jump_thread_path): ...this.
6842 (convert_and_register_jump_thread_path): Rename to...
6843 (thread_jumps::convert_and_register_current_path): ...this.
6844 (check_subpath_and_update_thread_path): Rename to...
6845 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6846 (register_jump_thread_path_if_profitable): Rename to...
6847 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6848 (handle_phi): Rename to...
6849 (thread_jumps::handle_phi): ...this.
6850 (handle_assignment): Rename to...
6851 (thread_jumps::handle_assignment): ...this.
6852 (fsm_find_control_statement_thread_paths): Rename to...
6853 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6854 (find_jump_threads_backwards): Rename to...
6855 (thread_jumps::find_jump_threads_backwards): ...this.
6856 Initialize path local data.
6857 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6858 from within thread_jumps class.
6859 (pass_early_thread_jumps::execute): Same.
6861 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6863 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6865 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6867 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6868 to canonical location.
6870 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6873 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6875 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6877 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6879 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6880 rcrt1.o%s/grcrt1.o%s for -static-pie.
6882 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6884 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6885 ix86_shift_rotate_cost): Break out from ...
6886 (ix86_rtx_costs): ... here.
6887 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6890 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6892 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6893 when entry block was promoted unlikely.
6894 (estimate_bb_frequencies): Increase frequency scale.
6895 * profile-count.h (profile_count): Export precision info.
6897 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6899 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6900 disturbin profile of entry block.
6902 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6904 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6907 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6909 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6910 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6912 2017-11-17 Jeff Law <law@redhat.com>
6914 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6915 from evrp_dom_walker class. Various methods moved into new class.
6916 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6917 (evrp_range_analyzer::enter): New method.
6918 (evrp_range_analyzer::leave): New method.
6919 (evrp_dom_walker): Remove delegators no longer needed by this class.
6920 Replace vr_values data member with evrp_range_analyzer
6922 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6923 method extracted from evrp_dom_walker::before_dom_children.
6924 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6925 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6927 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6928 Add private copy constructor and move assignment operators.
6929 Privatize methods and class data where trivially possible.
6930 (evrp_dom_walker::cleanup): New function, extracted from
6931 execute_early_vrp. Simplify access to class data.
6933 * vr-values.h (get_output_for_vrp): Prototype.
6934 * vr-values.c (get_output_for_vrp): New function extracted from
6935 vrp_visit_assignment_or_call and extract_range_from_stmt.
6936 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6938 2017-11-17 Luis Machado <luis.machado@linaro.org>
6940 * config/aarch64/aarch64.c
6941 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6942 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6943 tune_params::AUTOPREFETCHER_WEAK.
6945 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6948 * config/arm/arm.c (arm_valid_target_attribute_rec):
6949 Parse "arch=" and "+<ext>".
6950 (arm_valid_target_attribute_tree): Re-init global options.
6951 (arm_option_override): Make non-static.
6952 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6953 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6954 (__ARM_FEATURE_CRC32): Support undef.
6955 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6956 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6958 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6960 * gdbinit.in (break-on-diagnostic): New command.
6962 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6964 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6965 return to indirect jump for EH return if control-flow
6966 protection is enabled. Change explicit 'false' argument in
6967 pro_epilogue_adjust_stack with a value of
6969 * config/i386/i386.md (simple_return_indirect_internal):
6970 Remove SImode restriction to support 64-bit.
6972 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6974 * combine.c (added_notes_insn): New.
6975 (try_combine): Handle added_notes_insn like added_links_insn.
6976 Rewrite return value code.
6977 (distribute_notes): Set added_notes_insn to the earliest insn we added
6980 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6982 PR rtl-optimization/82621
6983 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6984 dest of one of those SETs is unused.
6986 2017-11-17 Richard Biener <rguenther@suse.de>
6989 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6990 * tree-pretty-print.c (dump_generic_node): Handle
6991 annot_expr_parallel_kind.
6992 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6993 * gimplify.c (gimple_boolify): Likewise.
6995 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6997 * config.gcc (extra_headers): Add cet.h for x86 targets.
6998 * config/i386/cet.h: New file.
6999 * doc/install.texi: Add --enable-cet/--disable-cet.
7001 2017-11-17 Richard Biener <rguenther@suse.de>
7003 PR tree-optimization/83017
7004 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
7005 (gen_parallel_loop): Properly count iterations.
7006 (parallelize_loops): Handle loop->can_be_parallel independent
7007 of flag_loop_parallelize_all. Make static profitability test match
7009 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
7010 * invoke.texi (parloops-min-per-thread): Document.
7012 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
7014 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
7015 upstreaming review comments.
7017 2017-11-17 Tamar Christina <tamar.christina@arm.com>
7019 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
7020 with fast unaligned access.
7021 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
7023 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7025 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
7026 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
7028 2017-11-17 Richard Biener <rguenther@suse.de>
7030 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
7031 folding of references.
7033 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
7036 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
7037 of replacing call to strncmp with corresponding call to strcmp when
7040 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
7042 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
7043 option prefer-avx256 for skylake-avx512 configuration.
7044 * config/i386/i386.c (ix86_option_override_internal): Ditto.
7045 (get_builtin_code_for_version): Ditto.
7047 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7048 Monk Chiang <sh.chiang04@gmail.com>
7050 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
7051 (FIXED_REGISTERS): Reserve more register numbers.
7052 (CALL_USED_REGISTERS): Likewise.
7053 (REG_ALLOC_ORDER): Likewise.
7054 (REG_CLASS_CONTENTS): Likewise.
7055 (REGISTER_NAMES): Likewise.
7057 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7058 Kito Cheng <kito.cheng@gmail.com>
7060 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
7062 * config/nds32/iterators.md: Add vector mode iterators and attributes.
7064 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
7066 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
7067 parameter list for vec_splats.
7069 2017-11-16 Joseph Myers <joseph@codesourcery.com>
7071 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
7073 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
7075 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
7078 * doc/install.texi: Document incompatibility of Perl >=5.6.26
7079 with the required version of automake 1.11.6.
7081 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
7083 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
7085 (DU_C2_3_power9): Correct reservation combinations.
7086 (FP_DIV_power9, VEC_DIV_power9): New.
7087 (power9-alu): Split out rotate/shift...
7088 (power9-rot): ...to here, correct dispatch resource.
7089 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
7091 (power9-fp): Correct latency.
7092 (power9-sdiv): Add div/sqrt resource.
7093 (power9-ddiv): Correct latency, add div/sqrt resource.
7094 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
7095 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
7097 (power9-qpdiv, power9-qpmul): Adjust resource usage.
7099 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
7101 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
7102 switch statement mapping KF built-ins to TF built-ins if we don't
7103 have the proper ISA 3.0 assembler support.
7105 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7107 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
7108 (lower_emutls_function_body): Do not compute it.
7110 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7112 * ipa-split.c (split_bb_info): Turn time to sreal.
7113 (split_point): Likewise.
7114 (dump_split_point): Likewise.
7115 (fine_split_points): Likewise.
7116 (execute_split_functions): Only zero split_bbs; turn time to sreals.
7118 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7120 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
7123 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7125 * predict.c (combine_predictions_for_bb): Preserve zero predicted
7127 (expensive_function_p): Remove useless assert.
7128 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
7130 2017-11-16 Martin Sebor <msebor@redhat.com>
7132 PR tree-optimization/82588
7133 PR tree-optimization/82583
7134 * tree-vrp.c (check_array_ref): Handle flexible array members,
7135 string literals, and inner indices.
7136 (search_for_addr_array): Add detail to diagnostics.
7138 2017-11-16 Nathan Sidwell <nathan@acm.org>
7142 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
7143 (SET_DECL_ASSEMBLER_NAME): Forward to
7144 overwrite_decl_assembler_name.
7145 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
7146 (overwrite_decl_assembler_name): Declare.
7147 * tree.c (overwrite_decl_assembler_name): New.
7148 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
7149 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
7150 (LANG_HOOKS_INITIALIZER): Add it.
7151 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
7152 * langhooks.c (lhd_set_decl_assembler_name): Use
7153 SET_DECL_ASSEMBLER_NAME.
7154 (lhd_overwrite_decl_assembler_name): Default implementation.
7156 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7157 Jackson Woodruff <jackson.woodruff@arm.com>
7159 PR tree-optimization/71026
7160 * match.pd: Canonicalize constant multiplies in division.
7162 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7164 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7165 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7166 Remove OPT_fomit_frame_pointer entry.
7167 * common/config/arc/arc-common.c: Likewise.
7168 * common/config/arm/arm-common.c: Likewise.
7169 * common/config/avr/avr-common.c: Likewise.
7170 * common/config/c6x/c6x-common.c: Likewise.
7171 * common/config/cr16/cr16-common.c: Likewise.
7172 * common/config/cris/cris-common.c: Likewise.
7173 * common/config/epiphany/epiphany-common.c: Likewise.
7174 * common/config/fr30/fr30-common.c: Likewise.
7175 * common/config/frv/frv-common.c: Likewise.
7176 * common/config/ia64/ia64-common.c: Likewise.
7177 * common/config/iq2000/iq2000-common.c: Likewise.
7178 * common/config/lm32/lm32-common.c: Likewise.
7179 * common/config/m32r/m32r-common.c: Likewise.
7180 * common/config/mcore/mcore-common.c: Likewise.
7181 * common/config/microblaze/microblaze-common.c: Likewise.
7182 * common/config/mips/mips-common.c: Likewise.
7183 * common/config/mmix/mmix-common.c: Likewise.
7184 * common/config/mn10300/mn10300-common.c: Likewise.
7185 * common/config/nios2/nios2-common.c: Likewise.
7186 * common/config/pa/pa-common.c: Likewise.
7187 * common/config/pdp11/pdp11-common.c: Likewise.
7188 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7189 * common/config/riscv/riscv-common.c: Likewise.
7190 * common/config/rs6000/rs6000-common.c: Likewise.
7191 * common/config/rx/rx-common.c: Likewise.
7192 * common/config/s390/s390-common.c: Likewise.
7193 * common/config/sh/sh-common.c: Likewise.
7194 * common/config/sparc/sparc-common.c: Likewise.
7195 * common/config/tilegx/tilegx-common.c: Likewise.
7196 * common/config/tilepro/tilepro-common.c: Likewise.
7197 * common/config/v850/v850-common.c: Likewise.
7198 * common/config/visium/visium-common.c: Likewise.
7199 * common/config/xstormy16/xstormy16-common.c: Likewise.
7200 * common/config/xtensa/xtensa-common.c: Likewise.
7201 * invoke.texi (-fomit-frame-pointer): Update documentation.
7203 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7205 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7207 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7209 * cfg.c (scale_bbs_frequencies_int,
7210 cale_bbs_frequencies_gcov_type): Remove.
7211 * cfg.h (scale_bbs_frequencies_int,
7212 cale_bbs_frequencies_gcov_type): Remove.
7214 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7216 * tree-ssa-loop-manip.c
7217 (scale_dominated_blocks_in_loop): Update to profile counts.
7218 (tree_transform_and_unroll_loop): Likewise.
7220 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7222 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7223 scale_bbs_frequencies_int.
7225 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7227 * final.c (compute_alignments): Use counts rather than frequencies.
7229 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7231 * cfgloopanal.c: Include sreal.h
7232 (average_num_loop_insns): Use counts and sreal for accounting.
7234 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7236 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7239 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7240 Kito Cheng <kito.cheng@gmail.com>
7242 * config/nds32/constraints.md: Provide more constraints.
7243 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7244 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7245 support constraints usage.
7247 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7249 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7251 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7252 Kito Cheng <kito.cheng@gmail.com>
7254 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7255 * config/nds32/nds32.opt: Refine the layout.
7256 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7257 TARGET_EXT_STRING): Support new options.
7258 * config/nds32/nds32.h: Likewise.
7259 * config/nds32/nds32.md: Likewise.
7260 * config/nds32/nds32-predicates.c: Likewise.
7261 * config/nds32/constraints.md: Likewise.
7262 * common/config/nds32/nds32-common.c: Likewise.
7264 2017-11-16 Julia Koval <julia.koval@intel.com>
7267 * config/i386/gfniintrin.h: Add sse check.
7268 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7270 2017-11-16 Julia Koval <julia.koval@intel.com>
7272 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7273 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7274 (ix86_handle_option): Handle -mavx512vbmi2.
7275 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7276 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7277 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7278 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7279 (ix86_valid_target_attribute_inner_p): Ditto.
7280 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7281 * config/i386/i386.opt (mavx512vbmi2): New option.
7282 * doc/invoke.texi: Add new option.
7284 2017-11-16 Julia Koval <julia.koval@intel.com>
7286 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7287 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7288 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7289 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7290 _mm512_gf2p8mul_epi8): New intrinsics.
7291 * config/i386/i386-builtin-types.def
7292 (V64QI_FTYPE_V64QI_V64QI): New type.
7293 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7294 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7295 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7296 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7297 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7298 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7300 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7302 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7303 explicitly as a stream of bytes.
7305 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7307 * config/rs6000/altivec.h (vec_xst_be): New #define.
7308 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7309 and externalize from *altivec_vperm_<mode>_internal.
7310 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7312 (XL_BE_V8HI): Likewise.
7313 (XL_BE_V4SI): Likewise.
7314 (XL_BE_V4SI): Likewise.
7315 (XL_BE_V2DI): Likewise.
7316 (XL_BE_V4SF): Likewise.
7317 (XL_BE_V2DF): Likewise.
7318 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7319 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7320 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7321 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7322 VSX_BUILTIN_VEC_XST_BE.
7323 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7324 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7326 (altivec_init_builtins): Replace conditional calls to def_builtin
7327 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7328 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7329 with unconditional calls. Remove calls to def_builtin for
7330 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7331 __builtin_vec_xst_be.
7332 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7333 to define_expand, and add alternate RTL generation for P8.
7334 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7335 vsx_ld_elemrev_v8hi.
7336 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7337 add alternate RTL generation for P8.
7338 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7339 vsx_ld_elemrev_v16qi.
7340 (vsx_st_elemrev_v8hi): Convert define_insn
7341 to define_expand, and add alternate RTL generation for P8.
7342 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7343 vsx_st_elemrev_v8hi.
7344 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7345 add alternate RTL generation for P8.
7346 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7347 vsx_st_elemrev_v16qi.
7349 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7352 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7353 TARGET_AVX512ER check.
7354 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7355 neither -mzeroupper nor -mno-zeroupper is used and
7356 TARGET_EMIT_VZEROUPPER is set.
7357 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7358 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7360 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7362 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7363 folding of vector compares.
7364 (fold_build_vec_cmp): New helper function.
7365 (fold_compare_helper): New helper function.
7366 (builtin_function_type): Add compare builtins to the list of functions
7367 having unsigned arguments. Cosmetic updates to comment indentation.
7368 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7369 the not+eq combination.
7371 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7373 PR tree-optimization/82726
7374 PR tree-optimization/70754
7375 * tree-predcom.c (order_drefs_by_pos): New function.
7376 (combine_chains): Move code setting has_max_use_after to...
7377 (try_combine_chains): ...here. New parameter. Sort combined chains
7378 according to position information.
7379 (tree_predictive_commoning_loop): Update call to above function.
7380 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7382 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7384 PR tree-optimization/82726
7386 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7388 PR tree-optimization/70754
7389 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7390 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7391 combined stmt before it if not NULL.
7392 (combine_chains): Process refs reversely and compute dominance point
7396 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7398 PR tree-optimization/79663
7399 * tree-predcom.c (combine_chains): Process refs in reverse order
7400 only for ZERO length chains, and add explaining comment.
7402 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7404 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7405 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7406 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7408 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7410 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7412 2017-11-15 Martin Liska <mliska@suse.cz>
7414 * tree-cfg.c (pass_warn_function_return::execute):
7415 Compare warn_return_type for greater than zero.
7417 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7421 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7422 to return true on Xeon and not on Xeon Phi.
7423 (ix86_check_avx256_register): Changed to ...
7424 (ix86_check_avx_upper_register): ... this. Add extra check for
7425 VALID_AVX512F_REG_OR_XI_MODE.
7426 (ix86_avx_u128_mode_needed): Changed
7427 ix86_check_avx256_register to ix86_check_avx_upper_register.
7428 (ix86_check_avx256_stores): Changed to ...
7429 (ix86_check_avx_upper_stores): ... this. Changed
7430 ix86_check_avx256_register to ix86_check_avx_upper_register.
7431 (ix86_avx_u128_mode_after): Changed
7432 avx_reg256_found to avx_upper_reg_found. Changed
7433 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7434 (ix86_avx_u128_mode_entry): Changed
7435 ix86_check_avx256_register to ix86_check_avx_upper_register.
7436 (ix86_avx_u128_mode_exit): Ditto.
7437 * config/i386/i386.h: (host_detect_local_cpu): New define.
7439 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7441 * config/arm/xgene1.md (xgene1): Split into automatons
7442 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7443 (xgene1_f_load): Adjust reservations and/or types.
7444 (xgene1_f_store): Likewise.
7445 (xgene1_load_pair): Likewise.
7446 (xgene1_store_pair): Likewise.
7447 (xgene1_fp_load1): Likewise.
7448 (xgene1_load1): Likewise.
7449 (xgene1_store1): Likewise.
7450 (xgene1_move): Likewise.
7451 (xgene1_alu): Likewise.
7452 (xgene1_simd): Likewise.
7453 (xgene1_bfm): Likewise.
7454 (xgene1_neon_load1): Likewise.
7455 (xgene1_neon_store1): Likewise.
7456 (xgene1_neon_logic): Likewise.
7457 (xgene1_neon_st1): Likewise.
7458 (xgene1_neon_ld1r): Likewise.
7459 (xgene1_alu_cond): Added.
7460 (xgene1_shift_reg): Likwise.
7461 (xgene1_bfx): Likewise.
7462 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7464 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7467 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7469 (can_widen_mult_without_libcall): New function.
7470 (expand_mul_overflow): If only checking unsigned mul overflow,
7471 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7472 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7473 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7474 (expand_DIVMOD): Formatting fix.
7475 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7476 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7477 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7479 PR tree-optimization/82977
7480 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7481 constructed temporary to strlen_to_stridx.put.
7483 2017-11-15 Martin Liska <mliska@suse.cz>
7485 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7486 * configure: Regenerate.
7488 2017-11-15 Martin Liska <mliska@suse.cz>
7491 * config/sh/sh-mem.cc: Use proper probability for
7494 2017-11-14 Jeff Law <law@redhat.com>
7496 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7497 the red zone for stack_clash_protection_final_dynamic_probe targets
7498 when the total dynamic stack size is zero bytes.
7500 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7501 blocks is post order.
7503 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7505 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7506 * final.c (rest_of_clean_state): Set it for the
7507 -fcompare-debug dump.
7508 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7509 class when TDF_COMPARE_DEBUG is set.
7511 * dwarf2out.c (gen_producer_string): Discard
7514 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7517 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7518 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7519 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7520 (__TGMATH_REAL_2_3): Remove macros.
7521 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7522 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7524 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7526 (remquo): Define using __TGMATH_REAL_3.
7528 2017-11-14 Jeff Law <law@redhat.com>
7530 * vr-values.c: New file with contents extracted from tree-vrp.c.
7531 * Makefile.in (OBJS): Add vr-values.o
7532 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7533 (set_value_range, set_and_canonicalize_value_range): Likewise.
7534 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7535 (value_range_constant_singleton, symbolic_range_p): Likewise.
7536 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7537 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7538 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7539 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7540 (find_case_label_range, find_case_label_index): Likewise.
7541 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7542 (range_int_cst_singleton_p, value_inside_range): Likewise.
7543 (get_single_symbol): Likewise.
7544 (switch_update): Move structure definition here.
7545 (to_remove_edges, to_update_switch_stmts): Provide externs.
7546 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7547 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7548 (vrp_val_is_min, set_value_range): Likewise.
7549 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7550 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7551 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7552 (range_is_nonnull, range_int_cst_p): Likewwise.
7553 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7554 (get_single_symbol, operand_less_p): Likewise
7555 (compare_values_warnv, compare_values): Likewise.
7556 (value_inside_range, value_range_constant_singleton): Likewise.
7557 (zero_nonzero_bitgs_from_vr): Likewise.
7558 (extract_range_from_binary_expr_1): Likewise.
7559 (overflow_comparison_p): Likewise.
7560 (to_remove_edges, to_update_switch_stmts): Likewise.
7561 (find_case_label-index, find_case_label_range): Likewise.
7562 (switch_update, set_value_range_to_nonnegative): Remove.
7563 (set_value_range_to_truthvalue): Likewise.
7564 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7565 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7566 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7567 (find_case_label_ranges, test_for_singularity): Likewise.
7568 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7569 (x_vr_values): Move to its remaining use site.
7571 2017-11-10 Jeff Law <law@redhat.com>
7573 * vr-values.h (VR_INITIALIZER): Move #define here.
7574 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7575 * Makefile.in (OBJS): Add tree-evrp.o
7576 * tree-vrp.h (assert_info): Move structure definition here.
7577 (set_value_range_to_varying): Prototype.
7578 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7579 (infer_value_range, register_edge_assert_for): Likewise.
7580 (stmt_interesting_for_vrp): Likewise.
7581 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7582 (set_value_range_to_varying): No longer static.
7583 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7584 (infer_value_range, register_edge_assert_for): Likewise.
7586 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7588 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7589 generate the XXBRD instruction.
7591 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7592 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7593 as being compatible if -mabi=ieeelongdouble.
7594 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7595 to setup float128 built-ins with hardware support.
7596 (BU_FLOAT128_HW_2): Likewise.
7597 (BU_FLOAT128_HW_3): Likewise.
7598 (BU_FLOAT128_HW_VSX_1): Likewise.
7599 (BU_FLOAT128_HW_VSX_2): Likewise.
7600 (scalar_extract_expq): Change float128 built-in functions to
7601 accommodate having both KFmode and TFmode functions. Use the
7602 KFmode variant as the default.
7603 (scalar_extract_sigq): Likewise.
7604 (scalar_test_neg_qp): Likewise.
7605 (scalar_insert_exp_q): Likewise.
7606 (scalar_insert_exp_qp): Likewise.
7607 (scalar_test_data_class_qp): Likewise.
7608 (sqrtf128_round_to_odd): Delete processing the round to odd
7609 built-in functions as special built-in functions, and define them
7610 as float128 built-ins. Use the KFmode variant as the default.
7611 (truncf128_round_to_odd): Likewise.
7612 (addf128_round_to_odd): Likewise.
7613 (subf128_round_to_odd): Likewise.
7614 (mulf128_round_to_odd): Likewise.
7615 (divf128_round_to_odd): Likewise.
7616 (fmaf128_round_to_odd): Likewise.
7617 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7618 support for KFmode and TFmode xststdcqp calls.
7619 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7620 point, switch the built-in handlers for the get/set float128
7621 exponent, get float128 mantissa, float128 test built-ins, and the
7622 float128 round to odd built-in functions. Eliminate creating the
7623 float128 round to odd built-in functions as special built-ins.
7624 (rs6000_init_builtins): Eliminate special creation of the float128
7625 round to odd built-in functions.
7626 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7627 function insns to support both TFmode and KFmode variants.
7628 (xsxsigqp_<mode>): Likewise.
7629 (xsiexpqpf_<mode>): Likewise.
7630 (xsiexpqp_<mode>): Likewise.
7631 (xststdcqp_<mode>): Likewise.
7632 (xststdcnegqp_<mode>): Likewise.
7633 (xststdcqp_<mode>): Likewise.
7635 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7637 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7638 unused path_in_freq_ptr parameter.
7639 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7641 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7643 * ipa-inline.c (edge_badness): Dump sreal frequency.
7644 (compute_inlined_call_time): Match natural implementaiton ...
7645 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7646 forgotten division by CGRAPH_FREQ_BASE.
7648 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7650 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7651 Solaris 11. Update comment.
7652 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7654 * config/sol2.h (STARTFILE_SPEC): Likewise.
7655 * configure: Regenerate.
7657 2017-11-14 Carl Love <cel@us.ibm.com>
7659 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7660 le_ and be_ prefixes to swap* variables. Remove
7661 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7664 2017-11-14 Jason Merrill <jason@redhat.com>
7666 Support GTY((cache)) on hash_map.
7667 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7668 (ggc_cache_remove): Override it instead of ggc_mx.
7669 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7670 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7671 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7672 (simple_cache_map_traits): Override maybe_mx.
7673 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7674 (hash_map): Friend gt_cleare_cache.
7675 (gt_cleare_cache): New.
7676 * tree.h (tree_cache_traits): New hash_map traits class.
7677 (tree_cache_map): New typedef.
7679 2017-11-14 Richard Biener <rguenther@suse.de>
7681 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7682 paramter and handling.
7683 (cleanup_control_flow_bb): Likewise.
7684 (cleanup_control_flow_pre): New helper performing a DFS walk
7685 to call cleanup_control_flow_bb in PRE order.
7686 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7687 via cleanup_control_flow_pre.
7689 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7691 * config/aarch64/aarch64-simd.md
7692 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7693 (*aarch64_simd_bsl<mode>_alt): Likewise.
7694 (aarch64_simd_bsldi_internal): New.
7695 (aarch64_simd_bsldi_alt): Likewise.
7697 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7699 * tracer.c (better_p): Do not compare frequencies.
7700 * reg-stack.c (better_edge): Likewise.
7701 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7704 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7706 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7707 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7708 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7709 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7710 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7711 * predict.c (maybe_hot_frequency_p): Inline to ...
7712 (maybe_hot_count_p): ... here; rewrite to counts.
7713 (counts_to_freqs): Rename to ...
7714 (update_max_bb_count): ... this one.
7715 (expensive_function_p): Use counts.
7716 (estimate_bb_frequencies): Update.
7717 (rebuild_frequencies): Update.
7718 * predict.h (counts_to_freqs): Rename to ...
7719 (update_max_bb_count): ... this one.
7720 * profile.c (compute_branch_probabilities): Add debug info
7721 * tree-inline.c (expand_call_inline): Update debug info.
7722 (optimize_inline_calls): Use update_max_bb_count..
7723 (tree_function_versioning): Use update_max_bb_count..
7724 * value-prof.c (gimple_value_profile_transformations):
7725 Do not use update_max_bb_count.
7727 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7729 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7730 always use frequencies.
7732 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7734 * bb-reorder.c: Remove frequencies from comments.
7735 (better_edge_p): Use profile counts.
7736 (find_traces): Dump profile counts.
7737 (rotate_loop): Use profile counts.
7738 (find_traces_1_round): Likewise.
7739 (connect_better_edge_p): Use counts instead of probabilities for
7741 (copy_bb_p): Drop early check for non-0 frequency.
7742 (sanitize_hot_paths): Update comments.
7744 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7746 * ipa-split.c (struct split_point): Add count.
7747 (consider_split): Do not compute incoming frequency; compute incoming
7748 count and store it to split_point.
7749 (split_function): Set count of the call to split part correctly.
7751 2017-11-13 Carl Love <cel@us.ibm.com>
7753 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7755 2017-11-13 Tom Tromey <tom@tromey.com>
7757 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7759 2017-11-13 Carl Love <cel@us.ibm.com>
7761 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7762 Add support for builtins:
7763 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7764 vector {un,}signed {char,int,short},
7765 vector {un,}signed {char,int,short}) arguments.
7766 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7767 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7768 Add BU_P9V_AV_2 expansions for the builtins.
7769 * config/rs6000/altivec.h (vec_first_match_index,
7770 vec_first_mismatch_index, vec_first_match_or_eos_index,
7771 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7772 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7773 new extern declaration.
7774 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7775 * config/rs6000/vsx.md (first_match_index_<mode>,
7776 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7777 first_mismatch_or_eos_index_<mode>): Add define expand.
7778 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7779 * doc/extend.texi: Update the built-in documenation file for the new
7782 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7784 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7785 into the min/max operations for _Float<N> and _Float<N>X types.
7787 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7790 * dwarf2out.c (do_eh_frame): New static variable.
7791 (dwarf2out_begin_prologue): Set it.
7792 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7794 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7796 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7798 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7800 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7803 * debug.h (dwarf2out_do_eh_frame): Declare.
7804 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7805 (dwarf2out_do_frame): Use it.
7806 (dwarf2out_do_cfi_asm): Likewise.
7807 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7808 (dwarf2out_assembly_start): Likewise.
7809 (dwarf2out_begin_prologue): Fix comment.
7810 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7811 if the target needs either debug or unwind DWARF2 info.
7812 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7813 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7814 -ftrapping-math, -ftrapv and -fwrapv.
7816 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7818 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7819 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7820 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7821 (estimate_edge_size_and_time): Likewise.
7822 (ipa_merge_fn_summary_after_inlining): Likewise.
7823 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7824 (compute_uninlined_call_time): Use sreal_frequency.
7825 (compute_inlined_call_time): Likewise.
7826 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7827 * profile-count.c: Include sreal.h.
7828 (profile_count::to_sreal_scale): New.
7829 * profile-count.h: Forward declare sreal.
7830 (profile_count::to_sreal_scale): Declare.
7832 2017-11-13 Nathan Sidwell <nathan@acm.org>
7834 * diagnostic.c (maybe_line_and_column): New.
7835 (diagnostic_get_location_text): Use it.
7836 (diagnostic_report_current_module): Likewise.
7837 (test_diagnostic_get_location_text): Add tests.
7839 2017-11-13 Luis Machado <luis.machado@linaro.org>
7841 * doc/md.texi (Specifying processor pipeline description): Fix
7842 incorrect latency for the div instruction example.
7844 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7846 PR tree-optimization/78821
7847 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7848 that bit_not_p is the same.
7849 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7850 (split_group): Count precisely bit_not_p bits in each statement.
7851 (invert_op): New function.
7852 (imm_store_chain_info::output_merged_store): Use invert_op to
7853 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7854 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7856 2017-11-13 Martin Liska <mliska@suse.cz>
7858 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7859 (struct source_info): Likewise.
7860 (add_branch_counts): Likewise.
7861 (add_line_counts): Likewise.
7862 (function_summary): Likewise.
7863 (output_intermediate_line): Likewise.
7864 (generate_results): Likewise.
7866 2017-11-13 Martin Liska <mliska@suse.cz>
7868 * gcov.c (struct block_info): Remove typedef for block_t.
7869 (struct line_info): Likewise.
7870 (line_info::has_block): Likewise.
7871 (EXIT_BLOCK): Likewise.
7872 (unblock): Likewise.
7873 (circuit): Likewise.
7874 (get_cycles_count): Likewise.
7875 (process_file): Likewise.
7876 (read_graph_file): Likewise.
7877 (solve_flow_graph): Likewise.
7878 (find_exception_blocks): Likewise.
7879 (add_line_counts): Likewise.
7880 (accumulate_line_info): Likewise.
7881 (output_line_details): Likewise.
7883 2017-11-13 Martin Liska <mliska@suse.cz>
7885 * gcov.c (struct arc_info): Remove typedef for arc_t.
7886 (struct line_info): Likewise.
7887 (add_branch_counts): Likewise.
7888 (output_branch_count): Likewise.
7889 (function_info::~function_info): Likewise.
7890 (circuit): Likewise.
7891 (output_intermediate_line): Likewise.
7892 (read_graph_file): Likewise.
7893 (solve_flow_graph): Likewise.
7894 (find_exception_blocks): Likewise.
7895 (add_line_counts): Likewise.
7896 (accumulate_line_info): Likewise.
7897 (output_line_details): Likewise.
7898 (output_function_details): Likewise.
7900 2017-11-13 Martin Liska <mliska@suse.cz>
7902 * gcov.c (struct function_info): Remove typedef for function_t.
7903 (struct source_info): Likewise.
7904 (source_info::get_functions_at_location): Likewise.
7905 (solve_flow_graph): Likewise.
7906 (find_exception_blocks): Likewise.
7907 (add_line_counts): Likewise.
7908 (output_intermediate_file): Likewise.
7909 (process_file): Likewise.
7910 (generate_results): Likewise.
7911 (release_structures): Likewise.
7912 (read_graph_file): Likewise.
7913 (read_count_file): Likewise.
7914 (accumulate_line_counts): Likewise.
7915 (output_lines): Likewise.
7917 2017-11-13 Martin Liska <mliska@suse.cz>
7919 * gcov.c (function_info::function_info): Remove num_counts
7920 and add vector<gcov_type>.
7921 (function_info::~function_info): Use the vector.
7922 (process_file): Likewise.
7923 (read_graph_file): Likewise.
7924 (read_count_file): Likewise.
7925 (solve_flow_graph): Likewise.
7927 2017-11-13 Martin Liska <mliska@suse.cz>
7929 * gcov.c (function_info::is_artificial): New function.
7930 (process_file): Erase all artificial early.
7931 (generate_results): Skip as all artificial are already
7934 2017-11-13 Martin Liska <mliska@suse.cz>
7936 * gcov.c (read_graph_file): Store to global vector of functions.
7937 (read_count_file): Iterate the vector.
7938 (process_file): Likewise.
7939 (generate_results): Likewise.
7940 (release_structures): Likewise.
7942 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7944 PR tree-optimization/82954
7945 * gimple-ssa-store-merging.c
7946 (imm_store_chain_info::coalesce_immediate_stores): If
7947 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7949 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7951 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7952 Upddate call to ENDIAN_LANE_N.
7953 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7954 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7955 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7956 and use aarch64_endian_lane_rtx.
7957 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7959 2017-11-12 Tom de Vries <tom@codesourcery.com>
7961 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7963 2017-11-12 Tom de Vries <tom@codesourcery.com>
7965 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7967 2017-11-12 Tom de Vries <tom@codesourcery.com>
7969 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7971 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7972 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7973 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7975 2017-11-11 Martin Sebor <msebor@redhat.com>
7978 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7981 * prefic.c (translate_name): Replace strncpy with memcpy to
7982 avoid -Wstringop-truncation.
7984 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7986 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7988 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7990 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7992 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7994 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7997 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7999 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
8000 enable generating XXBRH if the value is in a vector register.
8001 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
8002 value is in a vector register.
8003 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
8004 register to register bswap64's instead of doing the GPR sequence
8005 used on previous machines.
8006 (bswapdi2_xxbrd): New insn.
8007 (bswapdi2_reg): Disallow on ISA 3.0.
8008 (register to register bswap64 splitter): Do not split the insn on
8009 ISA 3.0 systems that use XXBRD.
8011 2017-11-10 Martin Sebor <msebor@redhat.com>
8014 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
8016 (find_subframework_file): Same.
8018 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8020 * auto-profile.c (afdo_indirect_call): Drop frequency.
8021 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
8022 (cgraph_node::create_edge): Drop frequency argument.
8023 (cgraph_node::create_indirect_edge): Drop frequency argument.
8024 (cgraph_edge::make_speculative): Drop frequency arguments.
8025 (cgraph_edge::resolve_speculation): Do not update frequencies
8026 (cgraph_edge::dump_edge_flags): Do not dump frequency.
8027 (cgraph_node::dump): Check consistency in IPA mode.
8028 (cgraph_edge::maybe_hot_p): Use IPA counter.
8029 (cgraph_edge::verify_count_and_frequency): Rename to ...
8030 (cgraph_edge::verify_count): ... this one; drop frequency checking.
8031 (cgraph_node::verify_node): Update.
8032 * cgraph.h (struct cgraph_edge): Drop frequency.
8033 (cgraph_edge::frequency): New function.
8034 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
8036 (cgraph_edge::rebuild_edges): Likewise.
8037 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
8038 (duplicate_thunk_for_node): Do not pass frequency.
8039 (cgraph_node::create_clone): Scale only counts.
8040 (cgraph_node::create_virtual_clone): Do not pass frequency.
8041 (cgraph_node::create_edge_including_clones): Do not pass frequency.
8042 (cgraph_node::create_version_clone): Do not pass frequency.
8043 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
8044 (cgraph_node::expand_thunk): Do not pass frequency.
8045 (cgraph_node::create_wrapper): Do not pass frequency.
8046 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
8048 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
8049 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
8050 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
8051 (gather_caller_stats): Use frequency function.
8052 (ipcp_cloning_candidate_p): Use frequency function.
8053 (ipcp_propagate_stage): Use frequency function.
8054 (get_info_about_necessary_edges): Use frequency function.
8055 (update_profiling_info): Update only IPA profile.
8056 (update_specialized_profile): Use frequency functoin.
8057 (perhaps_add_new_callers): Update only IPA profile.
8058 * ipa-devirt.c (ipa_devirt): Use IPA profile.
8059 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
8060 (dump_ipa_call_summary): Use frequency function.
8061 (estimate_edge_size_and_time): Use frequency function.
8062 (ipa_merge_fn_summary_after_inlining): Use frequency function.
8063 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
8064 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
8065 (update_noncloned_counts): ... ths one; scale counts only.
8066 (clone_inlined_nodes): Do not scale frequency.
8067 (inline_call): Do not pass frequency.
8068 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
8069 (compute_inlined_call_time): Use IPA profile.
8070 (want_inline_small_function_p): Use IPA profile.
8071 (want_inline_self_recursive_call_p): Use IPA profile.
8072 (edge_badness): Use IPA profile.
8073 (lookup_recursive_calls): Use IPA profile.
8074 (recursive_inlining): Do not pass frequency.
8075 (resolve_noninline_speculation): Do not update frequency.
8076 (inline_small_functions): Collect max of IPA profile.
8077 (dump_overall_stats): Dump IPA porfile.
8078 (dump_inline_stats): Dump IPA porfile.
8079 (ipa_inline): Collect IPA stats.
8080 * ipa-inline.h (clone_inlined_nodes): Update prototype.
8081 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
8082 (ipa_propagate_frequency): Use frequency function.
8083 (ipa_profile): Cleanup.
8084 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
8085 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
8086 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
8087 (input_node): Do not stream frequency.
8088 (input_edge): Do not stream frequency.
8089 (merge_profile_summaries): Scale only IPA profiles.
8090 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
8091 * predict.c (drop_profile): Do not recompute frequency.
8092 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
8093 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
8094 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
8095 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
8096 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
8097 * tree-inline.c (copy_bb): Do not scale frequency.
8098 (expand_call_inline): Do not scale frequency.
8099 (tree_function_versioning): Do not scale frequency.
8100 * ubsan.c (ubsan_create_edge): Do not pass frequency.
8102 2017-11-10 Julia Koval <julia.koval@intel.com>
8104 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
8105 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
8106 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
8107 (_mm256_maskz_gf2p8affine_epi64_epi8)
8108 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
8109 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
8110 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
8111 (__builtin_ia32_vgf2p8affineqb_v32qi)
8112 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
8113 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
8115 2017-11-10 Tamar Christina <tamar.christina@arm.com>
8119 (arm_option_override): Refactor.
8120 (arm_option_reconfigure_globals): New.
8121 (arm_options_perform_arch_sanity_checks): New.
8122 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
8124 (arm_options_perform_arch_sanity_checks): Likewise
8126 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
8128 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
8129 (power9-qpmul): New.
8130 * rs6000/rs6000.md ("type" attr): Add qmul.
8131 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
8132 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
8133 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
8135 2017-11-10 Martin Sebor <msebor@redhat.com>
8138 * builtins.c (compute_objsize): Handle arrays that
8139 compute_builtin_object_size likes to fail for. Make extern.
8140 * builtins.h (compute_objsize): Declare.
8141 (check_strncpy_sizes): New function.
8142 (expand_builtin_strncpy): Call check_strncpy_sizes.
8143 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
8144 -Wstringop-truncation.
8145 (gimple_fold_builtin_strncat): Same.
8146 * gimple.c (gimple_build_call_from_tree): Set call location.
8147 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
8148 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
8149 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
8150 (handle_builtin_strlen): Use strlen_to_stridx.
8151 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
8153 Use strlen_to_stridx.
8154 (pass_strlen::execute): Release strlen_to_stridx.
8155 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
8156 (-Wstringop-truncation): Document new option.
8158 2017-11-10 Martin Liska <mliska@suse.cz>
8160 PR gcov-profile/82702
8161 * gcov.c (main): Handle intermediate files in a different
8163 (get_gcov_intermediate_filename): New function.
8164 (output_gcov_file): Remove support of intermediate files.
8165 (generate_results): Allocate intermediate file.
8166 (release_structures): Clean-up properly fn_end.
8167 (output_intermediate_file): Start iterating with line 1.
8169 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8171 PR tree-optimization/82929
8172 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8173 ops_swapped_p non-static data member.
8174 (store_immediate_info::store_immediate_info): Clear it.
8175 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8176 ops set ops_swapped_p.
8177 (count_multiple_uses): Handle ops_swapped_p.
8179 2017-11-10 Martin Liska <mliska@suse.cz>
8181 * coverage.c (coverage_init): Stream information about
8182 support of has_unexecuted_blocks.
8183 * doc/gcov.texi: Document that.
8184 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8185 * gcov.c (read_graph_file): Likewise.
8186 (output_line_beginning): Fix a small issue with
8189 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8191 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8192 reference of trivial component.
8194 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8197 * gimple-ssa-store-merging.c
8198 (pass_store_merging::terminate_all_aliasing_chains): For
8199 gimple_store_p stmts also call refs_output_dependent_p.
8201 PR rtl-optimization/82913
8202 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8205 2017-11-09 Jeff Law <law@redhat.com>
8207 * vr-values.h: New file with vr_values class.
8208 * tree-vrp.c: Include vr-values.h
8209 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8210 data objects into the vr_values class.
8211 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8212 (get_value_range): Make it a member function within vr_values class.
8213 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8214 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8215 (op_with_constant_singleton_value_range): Likewise.
8216 (extract_range_for_var_from_comparison_expr): Likewise.
8217 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8218 (extract_range_from_binary_expr): Likewise.
8219 (extract_range_from_unary_expr): Likewise.
8220 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8221 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8222 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8223 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8224 (compare_name_with_value, compare_names): Likewise.
8225 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8226 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8227 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8228 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8229 (extract_range_from_phi_node): Likewise.
8230 (simplify_truth_ops_using_ranges): Likewise.
8231 (simplify_div_or_mod_using_ranges): Likewise.
8232 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8233 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8234 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8235 (simplify_float_conversion_using_ranges): Likewise.
8236 (simplify_internal_call_using_ranges): Likewise.
8237 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8238 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8239 poitner via x_vr_values for calls into gimple folder.
8240 (vrp_initialize_lattice): Make this the vr_values ctor.
8241 (vrp_free_lattice): Make this the vr_values dtor.
8242 (set_vr_value): New function.
8243 (class vrp_prop): Add vr_values data member. Add various member
8244 functions as well as member functions that delegate to vr_values.
8245 (check_array_ref): Make a member function within vrp_prop class.
8246 (search_for_addr_array, vrp_initialize): Likewise.
8247 (vrp_finalize): Likewise. Revamp to avoid direct access to
8248 vr_value, values_propagated, etc.
8249 (check_array_bounds): Extract vrp_prop class instance pointer from
8250 walk info structure. Use it to call member functions.
8251 (check_all_array_refs): Make a member function within vrp_prop class.
8252 Smuggle class instance pointer via walk info structure.
8253 (x_vr_values): New local static.
8254 (vrp_valueize): Use x_vr_values to get class instance.
8255 (vr_valueize_1): Likewise.
8256 (class vrp_folder): Add vr_values data member. Add various member
8257 functions as well as member functions that delegate to vr_values.
8258 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8259 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8260 class instance from vr_values. Use it to call member functions.
8261 (vrp_dom_walker): Add vr_values data member.
8262 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8263 instance via x_vr_values.
8264 (identify_jump_threads): Accept vr_values as argument. Store
8265 it into the walker structure.
8266 (evrp_dom_walker): Add vr_values class data member. Add various
8268 (evrp_dom_walker::try_find_new_range): Use vr_values data
8269 member to access the memory allocator.
8270 (evrp_dom_walker::before_dom_children): Store vr_values class
8271 instance into the vrp_folder class.
8272 (evrp_dom_walker::push_value_range): Rework to avoid direct
8273 access to num_vr_values and vr_value.
8274 (evrp_dom_walker::pop_value_range): Likewise.
8275 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8276 Use vr_values to get to dump_all_value_ranges member function.
8277 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8278 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8279 Pass vr_values class instance down to identify_jump_threads.
8280 Remove call to vrp_free_lattice.
8281 (debug_all_value_ranges): Remove.
8283 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8284 (vrp_folder): Likewise.
8286 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8287 Get it from the existing bitmap instead.
8288 (vrp_intersect_ranges_1): Likewise.
8290 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8292 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8294 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8295 set bit_not_p to it.
8296 (imm_store_chain_info::coalesce_immediate_stores): Break group
8297 if bit_not_p is different.
8298 (count_multiple_uses, split_group,
8299 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8300 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8301 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8302 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8303 to store_immediate_info ctor.
8305 2017-11-09 Jim Wilson <jimw@sifive.com>
8307 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8308 (scan_prog_file): Likewise.
8310 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8312 * bb-reorder.c (max_entry_frequency): Remove.
8313 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8314 connect_traces, push_to_next_round_p): Remove prototypes.
8315 (find_traces_1_round): Use counts only.
8316 (push_to_next_round_p): Likewise.
8317 (find_traces): Likewise.
8318 (rotate_loop): Likewise.
8319 (find_traces_1_round): Likewise.
8320 (connect_traces): Likewise.
8321 (edge_order): Likewise.
8323 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8325 * config/arm/arm.c (output_return_instruction): Add comments to
8326 indicate requirement for cmse_nonsecure_entry return to account
8327 for the size of clearing instruction output here.
8328 (thumb_exit): Likewise.
8329 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8330 return in hardfloat mode.
8332 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8334 * config/rs6000/rs6000.c (machine_function): Add a bool,
8335 "toc_is_wrapped_separately".
8336 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8337 if it wasn't explicitly set or unset, we are optimizing for speed, and
8338 doing separate shrink-wrapping.
8339 (rs6000_get_separate_components): Enable the TOC component if
8340 saving the TOC register in the prologue.
8341 (rs6000_components_for_bb): Handle the TOC component.
8342 (rs6000_emit_prologue_components): Store the TOC register where needed.
8343 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8344 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8346 2017-11-09 Martin Jambor <mjambor@suse.cz>
8348 * ipa-param-manipulation.c: New file.
8349 * ipa-param-manipulation.h: Likewise.
8350 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8351 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8352 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8353 (ipa_parm_adjustment): Likewise.
8354 (ipa_parm_adjustment_vec): Likewise.
8355 (ipa_get_vector_of_formal_parms): Moved declaration to
8356 ipa-param-manipulation.h.
8357 (ipa_get_vector_of_formal_parm_types): Likewise.
8358 (ipa_modify_formal_parameters): Likewise.
8359 (ipa_modify_call_arguments): Likewise.
8360 (ipa_combine_adjustments): Likewise.
8361 (ipa_dump_param_adjustments): Likewise.
8362 (ipa_modify_expr): Likewise.
8363 (ipa_get_adjustment_candidate): Likewise.
8364 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8365 ipa-param-manipulation.c.
8366 (ipa_get_vector_of_formal_parm_types): Likewise.
8367 (ipa_modify_formal_parameters): Likewise.
8368 (ipa_modify_call_arguments): Likewise.
8369 (ipa_modify_expr): Likewise.
8370 (get_ssa_base_param): Likewise.
8371 (ipa_get_adjustment_candidate): Likewise.
8372 (index_in_adjustments_multiple_times_p): Likewise.
8373 (ipa_combine_adjustments): Likewise.
8374 (ipa_dump_param_adjustments): Likewise.
8375 * tree-sra.c: Also include ipa-param-manipulation.h
8376 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8379 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8380 Alan Hayward <alan.hayward@arm.com>
8381 David Sherwood <david.sherwood@arm.com>
8383 * doc/sourcebuild.texi (vect_masked_store): Document.
8385 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8386 Alan Hayward <alan.hayward@arm.com>
8387 David Sherwood <david.sherwood@arm.com>
8389 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8391 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8392 Alan Hayward <alan.hayward@arm.com>
8393 David Sherwood <david.sherwood@arm.com>
8395 * doc/sourcebuild.texi (vect_variable_length): Document.
8397 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8398 Alan Hayward <alan.hayward@arm.com>
8399 David Sherwood <david.sherwood@arm.com>
8401 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8403 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8404 Alan Hayward <alan.hayward@arm.com>
8405 David Sherwood <david.sherwood@arm.com>
8407 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8409 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8410 Alan Hayward <alan.hayward@arm.com>
8411 David Sherwood <david.sherwood@arm.com>
8413 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8414 previously undocumented selectors.
8415 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8417 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8419 * doc/rtl.texi (const_vector): Say that elements can be
8420 const_wide_ints too.
8421 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8422 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8423 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8424 * optabs.c (expand_vector_broadcast): Likewise.
8426 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8427 Alan Hayward <alan.hayward@arm.com>
8428 David Sherwood <david.sherwood@arm.com>
8430 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8431 scaled index even if the unscaled address was invalid.
8432 Don't increase the complexity of using a scale in that case.
8434 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8435 Alan Hayward <alan.hayward@arm.com>
8436 David Sherwood <david.sherwood@arm.com>
8438 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8439 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8440 UNITS_PER_WORD bytes.
8441 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8442 based on REGMODE_NATURAL_SIZE of the inner mode.
8443 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8444 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8445 * expmed.c (lowpart_bit_field_p): Divide the value up into
8446 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8447 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8448 whether something is likely to occupy more than one register.
8450 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8453 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8454 profile_count::adjust_for_ipa_scaling.
8455 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8456 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8458 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8460 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8462 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8463 (split_group): Add total_orig and total_new arguments, estimate the
8464 number of statements related to the store group without store merging
8465 and with store merging.
8466 (imm_store_chain_info::output_merged_store): Adjust split_group
8467 callers, punt if estimated number of statements with store merging
8468 is not smaller than estimated number of statements without it.
8470 (handled_load): Remove has_single_use checks.
8471 (pass_store_merging::process_store): Likewise.
8473 2017-11-09 Richard Biener <rguenther@suse.de>
8475 PR tree-optimization/82902
8476 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8478 2017-11-09 Martin Liska <mliska@suse.cz>
8481 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8484 2017-11-09 Martin Liska <mliska@suse.cz>
8486 PR tree-optimization/82669
8487 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8489 2017-11-09 Martin Liska <mliska@suse.cz>
8491 PR gcov-profile/48463
8492 * coverage.c (coverage_begin_function): Output also end locus
8493 of a function and information whether the function is
8495 * gcov-dump.c (tag_function): Parse and print the information.
8496 * gcov.c (INCLUDE_MAP): Add include.
8497 (INCLUDE_SET): Likewise.
8498 (struct line_info): Move earlier in the source file because
8499 of vector<line_info> in function_info structure.
8500 (line_info::line_info): Likewise.
8501 (line_info::has_block): Likewise.
8502 (struct source_info): Add new member index.
8503 (source_info::get_functions_at_location): New function.
8504 (function_info::group_line_p): New function.
8505 (output_intermediate_line): New function.
8506 (output_intermediate_file): Use the mentioned function.
8507 (struct function_start): New.
8508 (struct function_start_pair_hash): Likewise.
8509 (process_file): Add code that identifies group functions.
8510 Assign lines either to global or function scope.
8511 (generate_results): Skip artificial functions.
8512 (find_source): Assign index for each source file.
8513 (read_graph_file): Read new flag artificial and end_line.
8514 (add_line_counts): Assign it either to global of function scope.
8515 (accumulate_line_counts): Isolate core of the function to
8516 accumulate_line_info and call it for both function and global
8518 (accumulate_line_info): New function.
8519 (output_line_beginning): Fix GNU coding style.
8520 (print_source_line): New function.
8521 (output_line_details): Likewise.
8522 (output_function_details): Likewise.
8523 (output_lines): Iterate both source (global) scope and function
8525 (struct function_line_start_cmp): New class.
8526 * doc/gcov.texi: Reflect changes in documentation.
8528 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8531 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8532 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8533 and similarly for not instead of neg.
8535 2017-11-08 Andi Kleen <ak@linux.intel.com>
8537 * config/i386/i386.opt: Add -mforce-indirect-call.
8538 * config/i386/predicates.md: Check for flag_force_indirect_call.
8539 * doc/invoke.texi: Document -mforce-indirect-call
8541 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8543 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8545 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8546 config/riscv/riscv.c (predict.h): New include.
8547 (riscv_slow_unaligned_access_p): No longer static.
8548 (riscv_block_move_straight): Add require.
8549 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8551 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8554 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8555 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8556 nonimmediate_operand predicate for operand 1 instead of
8559 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8561 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8563 * config/aarch64/constraints.md (Uml): New constraint.
8564 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8567 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8569 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8570 of two vec_duplicates into a vec_concat.
8572 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8574 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8575 Simplify vec_merge of vec_duplicate and vec_concat.
8576 * config/aarch64/constraints.md (Utq): New constraint.
8577 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8580 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8582 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8583 Simplify vec_merge of vec_duplicate and const_vector.
8584 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8586 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8587 mode iterator. Update predicate on operand 1 to
8588 handle non-const_vec constants. Delete constraints.
8589 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8591 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8593 PR tree-optimization/78821
8594 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8596 (store_operand_info::store_operand_info): Initialize it to false.
8597 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8598 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8599 store in the group, and if chain_info is non-NULL, to ignore altogether
8601 (compatible_load_p): Fail if bit_not_p does not match.
8602 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8603 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8604 (pass_store_merging::process_store): Adjust
8605 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8606 call terminate_all_aliasing_chains newly when adding a store into
8607 a chain with non-NULL chain_info.
8609 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8611 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8613 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8615 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8617 (aarch64_layout_frame): Initialise emit_frame_chain.
8618 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8619 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8621 2017-11-08 Martin Liska <mliska@suse.cz>
8623 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8624 of gimple_call_internal_p.
8626 2017-11-07 Tom de Vries <tom@codesourcery.com>
8628 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8630 2017-11-07 Tom de Vries <tom@codesourcery.com>
8632 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8635 2017-11-08 Martin Liska <mliska@suse.cz>
8638 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8640 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8642 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8644 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8647 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8648 (*cmp<mode>_ccz_1): New insn with $k alternative.
8651 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8652 RTX_COMM_COMPARE as commutative as well.
8653 (ix86_binary_operator_ok): Formatting fix.
8654 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8655 *<code><mode>3<mask_name><round_saeonly_name>,
8656 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8657 *<s>mul<mode>3_highpart<mask_name>,
8658 *vec_widen_umult_even_v16si<mask_name>,
8659 *vec_widen_umult_even_v8si<mask_name>,
8660 *vec_widen_umult_even_v4si<mask_name>,
8661 *vec_widen_smult_even_v16si<mask_name>,
8662 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8663 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8664 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8665 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8666 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8667 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8668 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8669 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8670 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8671 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8672 ix86_binary_operator_ok. Formatting fixes.
8673 (*<plusminus_insn><mode>3<mask_name><round_name>,
8674 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8677 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8679 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8680 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8681 eq and ne if TARGET_ISEL.
8682 (cmp): New code_iterator.
8683 (UNS, UNSU_, UNSIK): New code_attrs.
8684 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8685 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8687 ("eq<mode>3"): ... this.
8688 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8690 ("ne<mode>3"): ... this.
8692 2017-11-07 Julia Koval <julia.koval@intel.com>
8695 * common/config/i386/i386-common.c
8696 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8697 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8698 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8699 * config/i386/i386.opt: Ditto.
8700 * config/i386/i386.c (ix86_target_string): Ditto.
8701 (ix86_option_override_internal): Ditto.
8702 (ix86_init_mpx_builtins): Move MPX to args2.
8703 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8704 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8705 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8706 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8707 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8708 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8709 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8711 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8714 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8715 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8716 (zero-extendsidi peephole2): Remove peephole.
8718 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8721 * stor-layout.c: Include attribs.h.
8722 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8723 explicit lookup of "aligned" attribute.
8725 2017-11-07 Andrew Waterman <andrew@sifive.com>
8727 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8728 (riscv_expand_block_move): Likewise.
8729 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8731 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8732 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8733 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8734 (riscv_adjust_block_mem): Likewise.
8735 (riscv_block_move_loop): Likewise.
8736 (riscv_expand_block_move): Likewise.
8737 * config/riscv/riscv.md (movmemsi): New pattern.
8739 2017-11-07 Michael Clark <michaeljclark@mac.com>
8741 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8742 (MUSL_DYNAMIC_LINKER): Likewise.
8744 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8746 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8747 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8750 2017-11-07 Richard Biener <rguenther@suse.de>
8752 * match.pd: Fix build.
8754 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8755 Jackson Woodruff <jackson.woodruff@arm.com>
8757 PR tree-optimization/71026
8758 * match.pd: Canonicalize negate in division.
8760 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8763 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8765 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8767 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8768 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8770 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8772 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8773 non-scalar integral types.
8774 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8775 (-(A-B), -(~A)): New transformations.
8777 2017-11-07 Tom de Vries <tom@codesourcery.com>
8779 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8780 semicolon after "do {} while (0)".
8781 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8782 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8783 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8784 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8785 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8787 2017-11-07 Tom de Vries <tom@codesourcery.com>
8789 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8790 after "do {} while (0)".
8791 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8792 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8793 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8794 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8795 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8797 2017-11-07 Tom de Vries <tom@codesourcery.com>
8800 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8802 (arm_rtx_costs_internal): Add missing semicolon after
8803 HANDLE_NARROW_SHIFT_ARITH call.
8805 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8807 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8808 disable isel if it was not set explicitly.
8810 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8812 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8813 (add_type_attribute) likewise.
8815 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8817 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8818 of struct ix86_frame.
8819 (ix86_initial_elimination_offset): Likewise.
8820 (ix86_expand_split_stack_prologue): Likewise.
8822 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8824 * tree-vrp.h (enum value_range_type): Update stale comment.
8826 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8827 Alan Hayward <alan.hayward@arm.com>
8828 David Sherwood <david.sherwood@arm.com>
8830 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8831 (aarch64_expand_vec_perm_const): Take the number of units too.
8832 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8833 (aarch64_expand_vec_perm_const): Likewise.
8834 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8835 (vec_perm<mode>): Update accordingly.
8837 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8838 Alan Hayward <alan.hayward@arm.com>
8839 David Sherwood <david.sherwood@arm.com>
8841 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8842 Take the number of units too.
8843 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8844 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8845 but check for a vector mode before rather than after the call.
8846 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8847 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8848 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8849 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8850 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8851 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8852 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8853 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8854 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8855 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8856 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8857 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8858 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8859 (aarch64_sqdmull2_n<mode>): Update accordingly.
8861 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8862 Alan Hayward <alan.hayward@arm.com>
8863 David Sherwood <david.sherwood@arm.com>
8865 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8866 the number of units too.
8867 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8868 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8869 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8870 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8871 (vec_store_lanesxi<mode>): Update accordingly.
8873 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8874 Alan Hayward <alan.hayward@arm.com>
8875 David Sherwood <david.sherwood@arm.com>
8877 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8878 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8879 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8880 of units rather than the mode.
8881 * config/aarch64/iterators.md (nunits): New mode attribute.
8882 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8883 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8884 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8885 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8886 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8887 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8888 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8889 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8890 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8891 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8892 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8893 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8894 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8895 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8896 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8897 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8898 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8899 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8900 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8901 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8902 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8903 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8904 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8905 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8906 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8907 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8908 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8909 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8910 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8911 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8912 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8913 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8914 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8915 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8916 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8917 (aarch64_simd_vec_setv2di): Likewise.
8919 2017-11-06 Carl Love <cel@us.ibm.com>
8921 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8923 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8924 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8926 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8928 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8929 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8930 Add power 8 macro expansions.
8931 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8932 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8933 power 8 instructions. (VSX_XXBR): Add iterator.
8935 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8937 * config/arm/arm.md (predicable_short_it): Change default to "no",
8938 improve documentation, remove uses that are identical to the default.
8939 (enabled_for_depr_it): Rename to enabled_for_short_it.
8940 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8941 Remove default uses.
8942 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8943 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8944 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8945 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8947 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8950 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8951 float128 helper macros, which are no longer used after deleting
8952 the old 'q' built-in functions, and moving the round to odd
8953 built-in functions to being special built-in functions.
8954 (BU_FLOAT128_2): Likewise.
8955 (BU_FLOAT128_1_HW): Likewise.
8956 (BU_FLOAT128_2_HW): Likewise.
8957 (BU_FLOAT128_3_HW): Likewise.
8958 (FABSQ): Delete old 'q' built-in functions.
8959 (COPYSIGNQ): Likewise.
8960 (SQRTF128_ODD): Move round to odd built-in functions to be
8961 special built-in functions, so that we can handle
8962 -mabi=ieeelongdouble.
8963 (TRUNCF128_ODD): Likewise.
8964 (ADDF128_ODD): Likewise.
8965 (SUBF128_ODD): Likewise.
8966 (MULF128_ODD): Likewise.
8967 (DIVF128_ODD): Likewise.
8968 (FMAF128_ODD): Likewise.
8969 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8970 built-in names to 'f128'.
8971 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8972 old 'q' built-in functions, as the machine independent code for
8973 'f128' built-in functions handles this.
8974 (rs6000_expand_builtin): Add expansion for float128 round to odd
8975 functions, keying off on -mabi=ieeelongdouble of whether to use
8976 the KFmode or TFmode variant.
8977 (rs6000_init_builtins): Initialize the _Float128 round to odd
8979 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8980 _Float128 'q' built-in functions are now mapped into the new
8981 'f128' built-in functions.
8983 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8985 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8986 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8988 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8991 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8992 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8993 * toplev.c: Include "ipa-fnsummary.h".
8994 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8996 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8998 PR tree-optimization/82838
8999 * gimple-ssa-store-merging.c
9000 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
9001 on a separate gimple_seq which is then appended to seq.
9003 2017-11-06 Jeff Law <law@redhat.com>
9006 * config/i386/i386.c (PROBE_INTERVAL): Remove.
9007 (get_probe_interval): New functions.
9008 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
9009 (ix86_adjust_stack_and_probe): Likewise.
9010 (output_adjust_stack_and_probe): Likewise.
9011 (ix86_emit_probe_stack_range): Likewise.
9012 (ix86_expand_prologue): Likewise.
9014 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9016 PR tree-optimization/82816
9017 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
9018 if the modes of the two types are the same.
9019 (convert_plusminus_to_widen): Likewise.
9021 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9023 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
9025 (usadv16qi): New define_expand.
9026 (usadv8hi): New define_expand.
9028 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
9031 * ipa-inline-transform.c (update_noncloned_frequencies): Always
9033 (inline_transform): Likewise.
9034 * predict.c (counts_to_freqs): Remove useless conditional.
9035 * profile-count.h (profile_count::apply_scale): Move sanity check.
9036 * tree-inline.c (copy_bb): Always scale.
9037 (copy_cfg_body): Likewise.
9039 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
9042 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
9044 (*cmp_ite0): Add enabled_for_depr_it attribute.
9045 (*cmp_ite1): Likewise.
9047 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
9049 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
9052 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9054 * tree-vrp.c (vrp_int_const_binop): Return true on success and
9055 return the value by pointer.
9056 (extract_range_from_multiplicative_op_1): Update accordingly.
9057 Return as soon as an operation fails.
9059 2017-11-05 Tom de Vries <tom@codesourcery.com>
9062 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
9063 (DEF_SANITIZER_BUILTIN): ... here.
9064 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
9065 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
9067 2017-11-05 Tom de Vries <tom@codesourcery.com>
9070 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
9072 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
9073 ASM_OUTPUT_BEFORE_CASE_LABEL call.
9074 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
9076 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9077 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9078 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9080 2017-11-05 Tom de Vries <tom@codesourcery.com>
9083 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
9086 2017-11-04 Michael Clark <michaeljclark@mac.com>
9088 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
9089 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
9091 (*addsi3_extended): Likewise.
9092 (*addsi3_extended2): Likewise.
9093 (<optab>si3): Likewise.
9094 (<optab>di3): Likewise.
9095 (<optab><mode>3): Likewise.
9096 (<*optabe>si3_internal): Likewise.
9097 (zero_extendqi<SUPERQI:mode>2): Likewise.
9098 (*add<mode>hi3): Likewise.
9099 (*xor<mode>hi3): Likewise.
9100 (<optab>di3): Likewise.
9101 (*<optab>si3_extend): Likewise.
9102 (*sge<u>_<X:mode><GPR:mode>): Likewise.
9103 (*slt<u>_<X:mode><GPR:mode>): Likewise.
9104 (*sle<u>_<X:mode><GPR:mode>): Likewise.
9106 2017-11-04 Andrew Waterman <andrew@sifive.com>
9108 * config/riscv/riscv.c (riscv_option_override): Conditionally set
9109 TARGET_STRICT_ALIGN based upon -mtune argument.
9111 2017-11-04 Andrew Waterman <andrew@sifive.com>
9113 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
9115 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
9117 * config/i386/i386.c (choose_basereg): Use optional scratch
9118 register and add assertion.
9119 (x86_emit_outlined_ms2sysv_save): Use scratch register when
9120 needed, and don't allocate stack.
9121 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
9122 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
9123 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
9125 2017-11-03 Jeff Law <law@redhat.com>
9127 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
9128 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
9129 to probe at the start of a noreturn function.
9131 2017-11-03 Jakub Jelinek <jakub@redhat.com>
9133 PR tree-optimization/78821
9134 * gimple-ssa-store-merging.c: Update the file comment.
9135 (MAX_STORE_ALIAS_CHECKS): Define.
9136 (struct store_operand_info): New type.
9137 (store_operand_info::store_operand_info): New constructor.
9138 (struct store_immediate_info): Add rhs_code and ops data members.
9139 (store_immediate_info::store_immediate_info): Add rhscode, op0r
9140 and op1r arguments to the ctor, initialize corresponding data members.
9141 (struct merged_store_group): Add load_align_base and load_align
9143 (merged_store_group::merged_store_group): Initialize them.
9144 (merged_store_group::do_merge): Update them.
9145 (merged_store_group::apply_stores): Pick the constant for
9146 encode_tree_to_bitpos from one of the two operands, or skip
9147 encode_tree_to_bitpos if neither operand is a constant.
9148 (class pass_store_merging): Add process_store method decl. Remove
9149 bool argument from terminate_all_aliasing_chains method decl.
9150 (pass_store_merging::terminate_all_aliasing_chains): Remove
9151 var_offset_p argument and corresponding handling.
9152 (stmts_may_clobber_ref_p): New function.
9153 (compatible_load_p): New function.
9154 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
9155 if there is overlap and rhs_code is not INTEGER_CST. For
9156 non-overlapping stores terminate group if rhs is not mergeable.
9157 (get_alias_type_for_stmts): Change first argument from
9158 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
9159 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
9160 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9162 (get_location_for_stmts): Change first argument from
9163 auto_vec<gimple *> & to vec<gimple *> &.
9164 (struct split_store): Remove orig_stmts data member, add orig_stores.
9165 (split_store::split_store): Create orig_stores rather than orig_stmts.
9166 (find_constituent_stmts): Renamed to ...
9167 (find_constituent_stores): ... this. Change second argument from
9168 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9169 to info structures rather than the statements.
9170 (split_group): Rename ALLOW_UNALIGNED argument to
9171 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9172 it. Adjust find_constituent_stores caller.
9173 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9174 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9175 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9176 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9177 (mem_valid_for_store_merging): New function.
9178 (handled_load): New function.
9179 (pass_store_merging::process_store): New method.
9180 (pass_store_merging::execute): Use process_store method. Adjust
9181 terminate_all_aliasing_chains caller.
9183 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9185 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9186 Return true for more constants, symbols and label references.
9187 (aarch64_valid_floating_const): Remove unused function.
9189 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9192 * config/aarch64/aarch64.c (aarch64_layout_frame):
9193 Undo forcing of LR at bottom of frame.
9195 2017-11-03 Jeff Law <law@redhat.com>
9198 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9199 for int_registers_saved.
9201 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9202 extracted from tree-ssa-dom.c.
9203 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9204 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9205 (record_equivalences_from_incoming_edge): Add additional argument
9206 to single_pred_edge_ignoring_loop_edges call.
9207 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9208 (uncprop_dom_walker::before_dom_children): Add additional argument
9209 to single_pred_edge_ignoring_loop_edges call.
9210 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9211 single_pred_edge_ignoring_loop_edges rather than open coding.
9212 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9214 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9216 * match.pd (-(-A)): Rewrite.
9218 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9220 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9221 (rs6000_emit_int_cmove): New declaration.
9222 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9223 (rs6000_emit_sISEL): Delete.
9224 (rs6000_emit_int_cmove): Make non-static.
9225 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9226 instead of rs6000_emit_sISEL.
9228 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9230 * asan.c (create_cond_insert_point): Maintain profile.
9231 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9233 * basic-block.h (struct basic_block_def): Remove frequency.
9234 (EDGE_FREQUENCY): Use to_frequency
9235 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9237 (find_traces): Update to use to_frequency.
9238 (find_traces_1_round): Likewise; use only IPA counts.
9239 (bb_to_key): Likewise.
9240 (connect_traces): Use IPA counts only.
9241 (copy_bb_p): Update to use to_frequency.
9242 (fix_up_crossing_landing_pad): Likewise.
9243 (sanitize_hot_paths): Likewise.
9244 * bt-load.c (basic_block_freq): Likewise.
9245 * cfg.c (init_flow): Set count_max to uninitialized.
9246 (check_bb_profile): Remove frequencies; check counts.
9247 (dump_bb_info): Do not dump frequencies.
9248 (update_bb_profile_for_threading): Update counts only.
9249 (scale_bbs_frequencies_int): Likewise.
9250 (MAX_SAFE_MULTIPLIER): Remove.
9251 (scale_bbs_frequencies_gcov_type): Update counts only.
9252 (scale_bbs_frequencies_profile_count): Update counts only.
9253 (scale_bbs_frequencies): Update counts only.
9254 * cfg.h (struct control_flow_graph): Add count-max.
9255 (update_bb_profile_for_threading): Update prototype.
9256 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9257 (find_many_sub_basic_blocks): Likewise.
9258 * cfgcleanup.c (try_forward_edges): Likewise.
9259 (try_crossjump_to_edge): Likewise.
9260 * cfgexpand.c (expand_gimple_cond): Likewise.
9261 (expand_gimple_tailcall): Likewise.
9262 (construct_init_block): Likewise.
9263 (construct_exit_block): Likewise.
9264 * cfghooks.c (verify_flow_info): Check consistency of counts.
9265 (dump_bb_for_graph): Do not dump frequencies.
9266 (split_block_1): Do not update frequencies.
9267 (split_edge): Do not update frequencies.
9268 (make_forwarder_block): Do not update frequencies.
9269 (duplicate_block): Do not update frequencies.
9270 (account_profile_record): Do not update frequencies.
9271 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9272 for global heuristics.
9273 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9274 (expected_loop_iterations_unbounded): Use counts only.
9275 * cfgloopmanip.c (scale_loop_profile): Simplify.
9276 (create_empty_loop_on_edge): Simplify
9278 (duplicate_loop_to_header_edge): Simplify
9279 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9280 (update_br_prob_note): Take care of removing note when profile
9282 (relink_block_chain): Do not dump frequency.
9283 (rtl_account_profile_record): Use to_frequency.
9284 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9285 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9286 (cgraph_update_edges_for_call_stmt_node): Likewise.
9287 (cgraph_edge::verify_count_and_frequency): Update.
9288 (cgraph_node::verify_node): Temporarily disable frequency verification.
9289 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9290 to_cgraph_frequency.
9291 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9292 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9294 (cgraph_node::expand_thunk): Update profile.
9295 * except.c (dw2_build_landing_pads): Do not update frequency.
9296 * final.c (compute_alignments): Use to_frequency.
9297 (dump_basic_block_info): Do not dump frequency.
9298 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9299 (dump_gimple_bb_header): Do not dump frequency.
9300 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9302 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9303 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9304 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9305 (init_before_recovery): Do not update frequency.
9306 (sched_create_recovery_edges): Do not update frequency.
9307 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9308 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9309 (ipa_cp_c_finalize): Set max_count to uninitialized.
9310 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9311 (param_change_prob): Use counts.
9312 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9314 * ipa-split.c (consider_split): Use to_frequency.
9315 (split_function): Use to_frequency.
9316 * ira-build.c (loop_compare_func): Likewise.
9317 (mark_loops_for_removal): Likewise.
9318 (mark_all_loops_for_removal): Likewise.
9319 * loop-doloop.c (doloop_modify): Do not update frequency.
9320 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9322 * lto-streamer-in.c (input_function): Update count_max.
9323 * omp-expand.c (expand_omp_taskreg): Update count_max.
9324 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9325 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9326 (maybe_hot_count_p): Use ipa counts only.
9327 (maybe_hot_bb_p): Simplify.
9328 (maybe_hot_edge_p): Simplify.
9329 (probably_never_executed): Do not take frequency argument.
9330 (probably_never_executed_bb_p): Do not pass frequency.
9331 (probably_never_executed_edge_p): Likewise.
9332 (combine_predictions_for_bb): Check that profile is nonzero.
9333 (propagate_freq): Do not set frequency.
9334 (drop_profile): Simplify.
9335 (counts_to_freqs): Simplify.
9336 (expensive_function_p): Use to_frequency.
9337 (propagate_unlikely_bbs_forward): Simplify.
9338 (determine_unlikely_bbs): Simplify.
9339 (estimate_bb_frequencies): Add hack to silence graphite issues.
9340 (compute_function_frequency): Use ipa counts.
9341 (pass_profile::execute): Update.
9342 (rebuild_frequencies): Use counts only.
9343 (force_edge_cold): Use counts only.
9344 * profile-count.c (profile_count::dump): Dump new count types.
9345 (profile_count::differs_from_p): Check compatiblity.
9346 (profile_count::to_frequency): New function.
9347 (profile_count::to_cgraph_frequency): New function.
9348 * profile-count.h (struct function): Declare.
9349 (enum profile_quality): Add profile_guessed_local and
9350 profile_guessed_global0.
9351 (class profile_proability): Decrease number of bits to 29;
9352 update from_reg_br_prob_note and to_reg_br_prob_note.
9353 (class profile_count: Update comment; decrease number of bits
9354 to 61. Check compatibility.
9355 (profile_count::compatible_p): New private member function.
9356 (profile_count::ipa_p): New member function.
9357 (profile_count::operator<): Handle global zero correctly.
9358 (profile_count::operator>): Handle global zero correctly.
9359 (profile_count::operator<=): Handle global zero correctly.
9360 (profile_count::operator>=): Handle global zero correctly.
9361 (profile_count::nonzero_p): New member function.
9362 (profile_count::force_nonzero): New member function.
9363 (profile_count::max): New member function.
9364 (profile_count::apply_scale): Handle IPA scalling.
9365 (profile_count::guessed_local): New member function.
9366 (profile_count::global0): New member function.
9367 (profile_count::ipa): New member function.
9368 (profile_count::to_frequency): Declare.
9369 (profile_count::to_cgraph_frequency): Declare.
9370 * profile.c (OVERLAP_BASE): Delete.
9371 (compute_frequency_overlap): Delete.
9372 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9373 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9374 * sched-ebb.c (rank): Use counts only.
9375 * shrink-wrap.c (handle_simple_exit): Use counts only.
9376 (try_shrink_wrapping): Use counts only.
9377 (place_prologue_for_one_component): Use counts only.
9378 * tracer.c (find_best_predecessor): Use to_frequency.
9379 (find_trace): Use to_frequency.
9380 (tail_duplicate): Use to_frequency.
9381 * trans-mem.c (expand_transaction): Do not update frequency.
9382 * tree-call-cdce.c: Do not update frequency.
9383 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9384 (gimple_merge_blocks): Likewise.
9385 (gimple_split_edge): Likewise.
9386 (gimple_duplicate_sese_region): Likewise.
9387 (gimple_duplicate_sese_tail): Likewise.
9388 (move_sese_region_to_fn): Likewise.
9389 (gimple_account_profile_record): Likewise.
9390 (insert_cond_bb): Likewise.
9391 * tree-complex.c (expand_complex_div_wide): Likewise.
9392 * tree-eh.c (lower_resx): Update profile.
9393 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9395 (initialize_cfun): Do not initialize frequencies
9396 (freqs_to_counts): Delete.
9397 (copy_cfg_body): Ignore count parameter.
9398 (copy_body): Update.
9399 (expand_call_inline): Update count_max.
9400 (optimize_inline_calls): Update count_max.
9401 (tree_function_versioning): Update count_max.
9402 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9403 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9405 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9406 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9407 (try_peel_loop): Likewise.
9408 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9410 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9411 (tree_transform_and_unroll_loop): Do not use frequencies
9412 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9413 Use reliable prediction only.
9414 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9415 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9416 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9417 probability scaling.
9418 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9419 not update frequency
9420 (any_remaining_duplicated_blocks): Likewise.
9421 (update_profile): Likewise.
9422 (estimated_freqs_path): Delete.
9423 (freqs_to_counts_path): Delete.
9424 (clear_counts_path): Delete.
9425 (ssa_fix_duplicate_block_edges): Likewise.
9426 (duplicate_thread_path): Likewise.
9427 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9428 * tree-tailcall.c (decrease_profile): Do not update frequency.
9429 (eliminate_tail_call): Likewise.
9430 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9431 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9432 (optimize_mask_stores): Likewise.
9433 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9434 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9435 (ubsan_expand_ptr_ifn): Update profile.
9436 * value-prof.c (gimple_ic): Simplify.
9437 * value-prof.h (gimple_ic): Update prototype.
9438 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9439 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9441 (want_inline_self_recursive_call_p): Likewise.
9442 (resolve_noninline_speculation): Only cummulate defined counts.
9443 (inline_small_functions): Use nonzero_p.
9444 (ipa_inline): Do not access freed node.
9446 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9448 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9449 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9451 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9453 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9454 non-legitimate address.
9456 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9458 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9459 (*lt0_<mode>di, *lt0_<mode>si): New.
9461 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9463 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9464 TARGET_PAIRED_FLOAT.
9466 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9467 Jim Wilson <jim.wilson@linaro.org>
9469 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9470 * config/aarch64/aarch64-tune.md: Regenerated.
9471 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9472 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9474 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9476 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9477 (arc_expand_prologue): Restore blink for millicode.
9478 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9480 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9483 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9484 gen_vec_duplicate after forcing the scalar into a register.
9486 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9488 * combine (try_combine): Print the insns input to try_combine to the
9491 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9494 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9495 Remove second argument from aarch64_process_target_attr call.
9496 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9498 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9500 (aarch64_handle_attr_arch): Remove second argument.
9501 (aarch64_handle_attr_cpu): Ditto.
9502 (aarch64_handle_attr_tune): Ditto.
9503 (aarch64_handle_attr_isa_flags): Ditto.
9504 (aarch64_process_one_target_attr): Ditto.
9505 (aarch64_process_target_attr): Ditto.
9506 (aarch64_option_valid_attribute_p): Remove second argument.
9507 on aarch64_process_target_attr call.
9509 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9511 * diagnostic.c: Include "selftest-diagnostic.h".
9512 (selftest::assert_location_text): New function.
9513 (selftest::test_diagnostic_get_location_text): New function.
9514 (selftest::diagnostic_c_tests): Call it.
9516 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9518 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9519 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9520 (class selftest::test_diagnostic_context): Move to...
9521 * selftest-diagnostic.c: New file.
9522 * selftest-diagnostic.h: New file.
9524 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9526 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9527 offset range for FT32B.
9528 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9529 * config/ft32/ft32.md: Add TARGET_NOPM.
9530 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9531 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9533 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9535 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9537 2017-11-02 Jeff Law <law@redhat.com>
9539 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9540 virtual keyword on FINAL OVERRIDE members.
9542 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9543 virtuals together. Add virtual destructor.
9544 (substitute_and_fold_engine): Similarly.
9546 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9548 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9550 2017-11-02 Richard Biener <rguenther@suse.de>
9552 PR tree-optimization/82795
9553 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9555 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9557 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9559 (gcc_GAS_CHECK_FEATURE): Remove.
9560 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9561 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9562 * configure: Regenerate.
9564 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9566 * config/arc/arc.c (hwloop_optimize): Account for empty
9569 2017-11-02 Richard Biener <rguenther@suse.de>
9572 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9573 Truncate ARRAY_REF index and element size.
9575 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9577 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9579 2017-11-01 Jeff Law <law@redhat.com>
9581 * tree-ssa-ccp.c (ccp_folder): New class derived from
9582 substitute_and_fold_engine.
9583 (ccp_folder::get_value): New member function.
9584 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9585 (ccp_fold_stmt): Remove prototype.
9586 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9587 * tree-ssa-copy.c (copy_folder): New class derived from
9588 substitute_and_fold_engine.
9589 (copy_folder::get_value): Renamed from get_value.
9590 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9591 * tree-vrp.c (vrp_folder): New class derived from
9592 substitute_and_fold_engine.
9593 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9594 (vrp_folder::get_value): New member function.
9595 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9596 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9597 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9598 provide a class interface to folder/substitute routines.
9599 (ssa_prop_fold_stmt_fn): Remove typedef.
9600 (ssa_prop_get_value_fn): Likewise.
9601 (subsitute_and_fold): Remove prototype.
9602 (replace_uses_in): Likewise.
9603 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9604 Renamed from replace_uses_in. Call the virtual member function
9605 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9606 (substitute_and_fold_dom_walker): Remove initialization of
9607 data member entries for calbacks. Add substitute_and_fold_engine
9608 member and initialize it.
9609 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9610 member functions for get_value, replace_phi_args_in c
9611 replace_uses_in, and fold_stmt calls.
9612 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9613 substitute_and_fold. Remove assert. Update ctor call.
9615 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9616 (ssa_prop_visit_phi_fn): Likewise.
9617 (class ssa_propagation_engine): New class to provide an interface
9619 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9621 (ssa_prop_visit_phi): Likewise.
9622 (ssa_propagation_engine::simulate_stmt): Moved into class.
9623 Call visit_phi/visit_stmt from the class rather than via
9624 file scoped static variables.
9625 (ssa_propagation_engine::simulate_block): Moved into class.
9626 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9627 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9628 set file scoped statics for the visit_stmt/visit_phi callbacks.
9629 * tree-complex.c (complex_propagate): New class derived from
9630 ssa_propagation_engine.
9631 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9632 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9633 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9635 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9636 ssa_propagation_engine.
9637 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9638 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9639 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9640 * tree-ssa-copy.c (copy_prop): New class derived from
9641 ssa_propagation_engine.
9642 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9643 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9644 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9645 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9646 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9647 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9648 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9650 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9652 PR rtl-optimization/82778
9653 PR rtl-optimization/82597
9654 * compare-elim.c (struct comparison): Add in_a_setter field.
9655 (find_comparison_dom_walker::before_dom_children): Remove killed
9656 bitmap and df_simulate_find_defs call, instead walk the defs.
9657 Compute last_setter and initialize in_a_setter. Merge definitions
9658 with first initialization for a few variables.
9659 (try_validate_parallel): Use insn_invalid_p instead of
9660 recog_memoized. Return insn rather than just the pattern.
9661 (try_merge_compare): Fix up comment. Don't uselessly test if
9662 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9664 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9667 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9668 Alan Hayward <alan.hayward@arm.com>
9669 David Sherwood <david.sherwood@arm.com>
9671 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9672 aarch64_hard_regno_nregs to get the number of registers
9675 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9676 Alan Hayward <alan.hayward@arm.com>
9677 David Sherwood <david.sherwood@arm.com>
9679 * config/aarch64/constraints.md (Upl): Rename to...
9681 * config/aarch64/aarch64.md
9682 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9685 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9686 Alan Hayward <alan.hayward@arm.com>
9687 David Sherwood <david.sherwood@arm.com>
9689 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9690 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9693 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9694 Alan Hayward <alan.hayward@arm.com>
9695 David Sherwood <david.sherwood@arm.com>
9697 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9698 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9699 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9701 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9702 of a single element.
9703 * config/aarch64/iterators.md: Add a comment above the permute
9704 unspecs to say that they are generated directly by
9705 aarch64_expand_vec_perm_const.
9706 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9708 2017-11-01 Nathan Sidwell <nathan@acm.org>
9710 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9712 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9714 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9715 and medany code models, and describe what they do.
9717 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9719 Revert accidental duplicate:
9721 * combine.c (can_change_dest_mode): Reject changes in
9722 REGMODE_NATURAL_SIZE.
9724 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9726 PR rtl-optimization/64682
9727 PR rtl-optimization/69567
9728 PR rtl-optimization/69737
9729 PR rtl-optimization/82683
9730 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9731 register mentioned in the note, drop the note, unless it came from I3,
9732 in which case it should go to I3 again.
9734 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9736 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9737 and return false if not.
9738 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9740 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9742 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9743 range is known to be empty.
9745 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9746 Alan Hayward <alan.hayward@arm.com>
9747 David Sherwood <david.sherwood@arm.com>
9749 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9750 and CONST_VECTOR_NUNITS instead of computing the number of units from
9751 the byte sizes of the vector and element.
9752 (simplify_binary_operation_1): Likewise.
9753 (simplify_const_binary_operation): Likewise.
9754 (simplify_ternary_operation): Likewise.
9756 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9757 Alan Hayward <alan.hayward@arm.com>
9758 David Sherwood <david.sherwood@arm.com>
9760 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9761 (int_mem_offset): ...this new function.
9762 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9763 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9764 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9767 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9768 Alan Hayward <alan.hayward@arm.com>
9769 David Sherwood <david.sherwood@arm.com>
9771 * lower-subreg.c (interesting_mode_p): New function.
9772 (compute_costs, find_decomposable_subregs, decompose_register)
9773 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9774 (resolve_clobber, dump_choices): Use it.
9776 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9777 Alan Hayward <alan.hayward@arm.com>
9778 David Sherwood <david.sherwood@arm.com>
9780 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9782 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9783 Alan Hayward <alan.hayward@arm.com>
9784 David Sherwood <david.sherwood@arm.com>
9786 * alias.c (find_base_value, find_base_term): Only process integer
9787 truncations. Check the precision rather than the size.
9789 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9790 Alan Hayward <alan.hayward@arm.com>
9791 David Sherwood <david.sherwood@arm.com>
9793 * machmode.h (is_narrower_int_mode): New function
9794 * optabs.c (expand_float, expand_fix): Use it.
9795 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9797 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9798 Alan Hayward <alan.hayward@arm.com>
9799 David Sherwood <david.sherwood@arm.com>
9801 * rtl.h (narrower_subreg_mode): New function.
9802 * ira-color.c (update_costs_from_allocno): Use it.
9804 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9805 Alan Hayward <alan.hayward@arm.com>
9806 David Sherwood <david.sherwood@arm.com>
9808 * optabs-query.h (convert_optab_p): New function, split out from...
9809 (convert_optab_handler): ...here.
9810 (widening_optab_handler): Delete.
9811 (find_widening_optab_handler): Remove permit_non_widening parameter.
9812 (find_widening_optab_handler_and_mode): Likewise. Provide an
9813 override that operates on mode class wrappers.
9814 * optabs-query.c (widening_optab_handler): Delete.
9815 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9816 parameter. Assert that the two modes are the same class and that
9817 the "from" mode is narrower than the "to" mode. Use
9818 convert_optab_handler instead of widening_optab_handler.
9819 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9820 instead of widening_optab_handler.
9821 * expr.c (expand_expr_real_2): Update calls to
9822 find_widening_optab_handler.
9823 * optabs.c (expand_widen_pattern_expr): Likewise.
9824 (expand_binop_directly): Take the insn_code as a parameter.
9825 (expand_binop): Only call find_widening_optab_handler for
9826 conversion optabs; use optab_handler otherwise. Update calls
9827 to find_widening_optab_handler and expand_binop_directly.
9828 Use convert_optab_handler instead of widening_optab_handler.
9829 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9830 find_widening_optab_handler and use scalar_mode rather than
9832 (convert_plusminus_to_widen): Likewise.
9834 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9835 Alan Hayward <alan.hayward@arm.com>
9836 David Sherwood <david.sherwood@arm.com>
9838 * machmode.h (fixed_size_mode): New class.
9839 * rtl.h (get_pool_mode): Return fixed_size_mode.
9840 * gengtype.c (main): Add fixed_size_mode.
9841 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9842 (get_raw_arg_mode): Likewise.
9843 * doc/tm.texi: Regenerate.
9844 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9845 * targhooks.c (default_get_reg_raw_mode): Likewise.
9846 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9847 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9848 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9849 (msp430_get_raw_result_mode): Likewise.
9850 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9851 * dbxout.c (dbxout_parms): Require fixed-size modes.
9852 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9853 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9854 * omp-low.c (lower_oacc_reductions): Likewise.
9855 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9856 (simplify_subreg): Update accordingly.
9857 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9858 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9859 that aren't fixed-size.
9860 (get_pool_mode): Return a fixed_size_mode.
9861 (output_constant_pool_2): Take a fixed_size_mode.
9863 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9864 Alan Hayward <alan.hayward@arm.com>
9865 David Sherwood <david.sherwood@arm.com>
9867 * doc/rtl.texi (vec_series): Document.
9868 (const): Say that the operand can be a vec_series.
9869 * rtl.def (VEC_SERIES): New rtx code.
9870 * rtl.h (const_vec_series_p_1): Declare.
9871 (const_vec_series_p): New function.
9872 * emit-rtl.h (gen_const_vec_series): Declare.
9873 (gen_vec_series): Likewise.
9874 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9875 (gen_vec_series): Likewise.
9876 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9877 * simplify-rtx.c (simplify_unary_operation): Handle negations
9879 (simplify_binary_operation_series): New function.
9880 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9881 (test_vector_ops_series): New function.
9882 (test_vector_ops): Call it.
9883 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9884 gen_const_vec_series.
9885 (altivec_lvsr): Likewise.
9886 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9888 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9889 Alan Hayward <alan.hayward@arm.com>
9890 David Sherwood <david.sherwood@arm.com>
9892 * doc/rtl.texi (const): Update description of address constants.
9893 Say that vector constants are allowed too.
9894 * common.md (E, F): Use CONSTANT_P instead of checking for
9896 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9897 checking for CONST_VECTOR.
9898 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9900 * expr.c (expand_expr_real_2): Check for vector modes instead
9901 of checking for CONST_VECTOR.
9902 * rtl.h (const_vec_p): New function.
9903 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9904 (unwrap_const_vec_duplicate): Handle them here too.
9906 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9907 David Malcolm <dmalcolm@redhat.com>
9908 Alan Hayward <alan.hayward@arm.com>
9909 David Sherwood <david.sherwood@arm.com>
9911 * rtl.h (vec_duplicate_p): New function.
9912 * selftest-rtl.c (assert_rtx_eq_at): New function.
9913 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9914 (assert_rtx_eq_at): Declare.
9915 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9916 * selftest-run-tests.c (selftest::run_tests): Call it.
9917 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9918 (simplify_unary_operation_1): Recursively handle vector duplicates.
9919 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9921 (simplify_subreg): Handle subregs of vector duplicates.
9922 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9923 (selftest::simplify_rtx_c_tests): New functions.
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 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9930 (gen_vec_duplicate): Likewise.
9931 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9933 (gen_const_vector): ...here.
9934 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9935 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9936 whose elements are all equal.
9937 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9938 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9939 (simplify_relational_operation): Likewise.
9940 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9942 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9943 (aarch64_expand_vector_init): Likewise.
9944 * config/arm/arm.c (neon_vdup_constant): Likewise.
9945 (neon_expand_vector_init): Likewise.
9946 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9947 (arm_block_set_unaligned_vect): Likewise.
9948 (arm_block_set_aligned_vect): Likewise.
9949 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9950 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9951 (expand_vec_perm_even_odd_pack): Likewise.
9952 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9953 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9954 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9955 gen_const_vec_duplicate.
9956 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9957 * config/mips/mips.c (mips_gen_const_int_vector): Use
9958 gen_const_vec_duplicate.
9959 (mips_expand_vector_init): Use CONST0_RTX.
9960 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9961 (define_split): Use gen_const_vec_duplicate.
9962 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9963 (define_split): Use gen_const_vec_duplicate.
9964 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9965 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9966 * config/spu/spu.c (spu_const): Likewise.
9968 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9969 Alan Hayward <alan.hayward@arm.com>
9970 David Sherwood <david.sherwood@arm.com>
9972 * combine.c (can_change_dest_mode): Reject changes in
9973 REGMODE_NATURAL_SIZE.
9975 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9977 * configure.ac (--enable-libssp): New.
9978 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9979 trying to determine target-specific default. Adjust indentation.
9980 * configure: Regenerated.
9981 * doc/install.texi (Configuration): Expand --disable-libssp
9984 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9986 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9989 2017-10-31 Martin Jambor <mjambor@suse.cz>
9992 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9994 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9996 * auto-profile.c (autofdo_source_profile::read): Use
9997 UNKNOWN_LOCATION rather than 0.
9998 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9999 (warning_at): ...this overload.
10000 (warning_at_rich_loc_n): Rename to...
10001 (warning_n): ...this overload.
10002 (error_at_rich_loc): Rename to...
10003 (error_at): ...this overload.
10004 (pedwarn_at_rich_loc): Rename to...
10005 (pedwarn): ...this overload.
10006 (permerror_at_rich_loc): Rename to...
10007 (permerror): ...this overload.
10008 (inform_at_rich_loc): Rename to...
10009 (inform): ...this overload.
10010 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
10011 (diagnostic_n_impl_richloc): Rename to...
10012 (diagnostic_n_impl): ...this rich_location *-based decl.
10013 (inform_at_rich_loc): Rename to...
10014 (inform): ...this, and add an assertion.
10015 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
10016 (warning_at_rich_loc): Rename to...
10017 (warning_at): ...this, and add an assertion.
10018 (warning_at_rich_loc_n): Rename to...
10019 (warning_n): ...this, and add an assertion.
10020 (warning_n): Update location_t-based implementation for removal of
10021 location_t-based diagnostic_n_impl.
10022 (pedwarn_at_rich_loc): Rename to...
10023 (pedwarn): ...this, and add an assertion.
10024 (permerror_at_rich_loc): Rename to...
10025 (permerror): ...this, and add an assertion.
10026 (error_n): Update for removal of location_t-based diagnostic_n_impl.
10027 (error_at_rich_loc): Rename to...
10028 (error_at): ...this, and add an assertion.
10029 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
10030 (driver::do_spec_on_infiles): Likewise.
10031 * substring-locations.c (format_warning_va): Update for renaming
10032 of inform_at_rich_loc.
10034 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
10036 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
10037 _Float<N>X built-in functions so that the variant without the
10038 "__builtin_" prefix is only enabled for the GNU C and Objective C
10039 languages when they are in non-strict ANSI/ISO mode.
10040 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
10041 * target.def (floatn_builtin_p): Add a target hook to control
10042 whether _Float<N> and _Float<N>X built-in functions without the
10043 "__builtin_" prefix are enabled, and return true for C and
10044 Objective C in the default hook. Include langhooks.h in
10046 * targhooks.h (default_floatn_builtin_p): Likewise.
10047 * targhooks.c (default_floatn_builtin_p): Likewise.
10048 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
10049 floatn_builtin_p target hook.
10050 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
10052 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
10053 Eric Botcazou <ebotcazou@adacore.com>
10055 PR rtl-optimization/81803
10056 * lra-constraints.c (curr_insn_transform): Also reload the whole
10057 register for a strict subreg no wider than a word if this is for
10058 a WORD_REGISTER_OPERATIONS target.
10060 2017-10-31 Jason Merrill <jason@redhat.com>
10062 * gdbinit.in: Skip over inlines from timevar.h.
10064 2017-10-31 Martin Liska <mliska@suse.cz>
10066 * doc/gcov.texi: Document new option.
10067 * gcov.c (print_usage): Likewise print it.
10068 (process_args): Support the argument.
10069 (format_count): New function.
10070 (format_gcov): Use the function.
10072 2017-10-31 Martin Liska <mliska@suse.cz>
10074 * gcov.c (struct name_map): do not use typedef.
10075 Define operator== and operator<.
10076 (name_search): Remove.
10077 (name_sort): Remove.
10078 (main): Do not allocate names.
10079 (process_file): Add vertical space.
10080 (generate_results): Use std::find.
10081 (release_structures): Do not release memory.
10082 (find_source): Use std::find.
10084 2017-10-31 Martin Liska <mliska@suse.cz>
10086 * gcov.c (struct line_info): Remove it's typedef.
10087 (line_info::line_info): Add proper ctor.
10088 (line_info::has_block): Do not use a typedef.
10089 (struct source_info): Do not use typedef.
10090 (circuit): Likewise.
10091 (get_cycles_count): Likewise.
10092 (output_intermediate_file): Iterate via vector iterator.
10093 (add_line_counts): Use std::vector methods.
10094 (accumulate_line_counts): Likewise.
10095 (output_lines): Likewise.
10097 2017-10-31 Martin Liska <mliska@suse.cz>
10099 * gcov.c (struct source_info): Remove typedef.
10100 (source_info::source_info): Add proper ctor.
10101 (accumulate_line_counts): Use struct, not it's typedef.
10102 (output_gcov_file): Likewise.
10103 (output_lines): Likewise.
10104 (main): Do not allocate an array.
10105 (output_intermediate_file): Use size of vector container.
10106 (process_file): Resize the vector.
10107 (generate_results): Do not preallocate, use newly added vector
10109 (release_structures): Do not release sources.
10110 (find_source): Use vector methods.
10111 (add_line_counts): Do not use typedef.
10113 2017-10-31 Martin Liska <mliska@suse.cz>
10115 * doc/gcov.texi: Document that.
10116 * gcov.c (add_line_counts): Mark lines with a non-executed
10118 (output_line_beginning): Handle such lines.
10119 (output_lines): Pass new argument.
10120 (output_intermediate_file): Print it in intermediate format.
10122 2017-10-31 Martin Liska <mliska@suse.cz>
10124 * color-macros.h: New file.
10125 * diagnostic-color.c: Factor out color related to macros to
10127 * doc/gcov.texi: Document -k option.
10128 * gcov.c (INCLUDE_STRING): Include string.h.
10129 (print_usage): Add -k option.
10130 (process_args): Parse it.
10131 (pad_count_string): New function.
10132 (output_line_beginning): Likewise.
10133 (DEFAULT_LINE_START): New macro.
10134 (output_lines): Support color output.
10136 2017-10-31 Martin Liska <mliska@suse.cz>
10138 PR gcov-profile/82633
10139 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
10140 their interaction with GCOV infrastructure.
10141 * configure.ac: Add -fkeep-{inline,static}-functions to
10143 * configure: Regenerate.
10145 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
10148 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
10150 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
10153 * config/rs6000/rs6000.md (allocate_stack): Force update interval
10154 into a register if it does not fit into an immediate offset field.
10156 2017-10-31 Olivier Hainque <hainque@adacore.com>
10158 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
10160 2017-10-31 Julia Koval <julia.koval@intel.com>
10162 * config.gcc: Add gfniintrin.h.
10163 * config/i386/gfniintrin.h: New.
10164 * config/i386/i386-builtin-types.def
10165 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10166 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10167 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10168 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10169 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10170 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10171 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10172 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10173 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10174 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10175 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10176 * config/i386/immintrin.h: Include gfniintrin.h.
10177 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10179 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10181 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10183 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10185 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10186 (arm_ashldi3_1bit): Remove pattern.
10187 (ashrdi3): Remove shift by 1 expansion.
10188 (arm_ashrdi3_1bit): Remove pattern.
10189 (lshrdi3): Remove shift by 1 expansion.
10190 (arm_lshrdi3_1bit): Remove pattern.
10191 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10192 cost of ashldi3 by 1.
10193 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10194 (<shift>di3_neon): Likewise.
10196 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10198 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10199 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10200 and (*aarch64_simd_mov<VQ:mode>).
10201 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10202 pattern alternative.
10203 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10204 attributes to match pattern alternative.
10206 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10208 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10209 * config/rs6000/emmintrin.h: New file.
10210 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10212 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10214 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10215 * (movdi_vfp_cortexa8): Remove pattern.
10217 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10219 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10221 (Specific, alpha*-dec-osf5.1): Remove.
10222 (Specific, mips-sgi-irix5): Remove.
10223 (Specific, mips-sgi-irix6): Remove.
10225 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10227 PR middle-end/22141
10228 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10229 arguments to clear_bit_region_be.
10231 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10233 * gimplify.c: Include memmodel.h.
10235 2017-10-30 Martin Jambor <mjambor@suse.cz>
10237 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10238 condition whether loop should be executed at all.
10240 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10242 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10243 gimple folding of vec_madd() intrinsics.
10244 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10245 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10246 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10248 2017-10-30 Richard Biener <rguenther@suse.de>
10250 PR tree-optimization/82762
10252 2017-10-23 Richard Biener <rguenther@suse.de>
10254 PR tree-optimization/82129
10256 2017-08-01 Richard Biener <rguenther@suse.de>
10258 PR tree-optimization/81181
10259 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10260 (compute_antic): ... end of iteration here.
10262 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10264 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10265 -std=iso9899:2017 and -std=gnu17.
10266 * doc/standards.texi (C Language): Document C17 support.
10267 * doc/cpp.texi (Overview): Mention -std=c17.
10268 (Standard Predefined Macros): Document C11 and C17 values of
10269 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10270 * doc/extend.texi (Inline): Do not list individual options for
10271 standards newer than C99.
10272 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10274 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10277 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10279 * asan.c (asan_finish_file): Align asan globals array by shadow
10282 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10284 PR middle-end/22141
10285 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10286 (struct store_immediate_info): Add bitregion_start and bitregion_end
10288 (store_immediate_info::store_immediate_info): Add brs and bre
10289 arguments and initialize bitregion_{start,end} from those.
10290 (struct merged_store_group): Add bitregion_start, bitregion_end,
10291 align_base and mask fields. Drop unnecessary struct keyword from
10292 struct store_immediate_info. Add do_merge method.
10293 (clear_bit_region_be): Use memset instead of loop storing zeros.
10294 (merged_store_group::do_merge): New method.
10295 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10296 stores as long as the surrounding bitregions have no gaps.
10297 (merged_store_group::merge_overlapping): Use do_merge.
10298 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10299 is byte aligned, rather than requiring that start and width are
10300 byte aligned. Drop unnecessary struct keyword from
10301 struct store_immediate_info. Allocate and populate also mask array.
10302 Make start of the arrays relative to bitregion_start rather than
10303 start and size them according to bitregion_{end,start} difference.
10304 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10305 struct store_immediate_info.
10306 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10307 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10308 struct keyword from struct store_immediate_info.
10309 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10310 between stores as long as the surrounding bitregions have no gaps.
10312 (struct split_store): Add orig non-static data member.
10313 (split_store::split_store): Initialize orig to false.
10314 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10315 if there is exactly a single original stmt. Change stmts argument
10316 to pointer from reference, if NULL, don't push anything to it. Add
10317 first argument, use it to optimize skipping over orig stmts that
10318 are known to be before bitpos already. Simplify.
10319 (split_group): Return unsigned int count how many stores are or
10320 would be needed rather than a bool. Add allow_unaligned argument.
10321 Change split_stores argument from reference to pointer, if NULL,
10322 only do a dry run computing how many stores would be produced.
10323 Rewritten algorithm to use both alignment and misalign if
10324 !allow_unaligned and handle bitfield stores with gaps.
10325 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10326 from bitregion_start instead of start. Compute allow_unaligned
10327 here, if true, do 2 split_group dry runs to compute which one
10328 produces fewer stores and prefer aligned if equal. Punt if
10329 new count is bigger or equal than original before emitting any
10330 statements, rather than during that. Remove no longer needed
10331 new_ssa_names tracking. Replace num_stmts with
10332 split_stores.length (). Use 32-bit stack allocated entries
10333 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10334 if possible. Handle bitfields with gaps.
10335 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10336 Compute bitregion_{start,end} for the stores and construct
10337 store_immediate_info with that. Formatting fixes.
10339 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10342 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10343 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10345 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10347 * gimplify.c: Include tm_p.h.
10349 * common.opt (gcoff): Re-add as ignored option.
10350 (gcoff1, gcoff2, gcoff3): Likewise.
10352 * Makefile.in (OBJS): Delete sdbout.o.
10353 (GTFILES): Delete $(srcdir)/sdbout.c.
10354 * debug.h: Delete sdb_debug_hooks.
10355 * final.c: Delete sdbout.h include.
10356 (final_scan_insn): Delete SDB_DEBUG check.
10357 (rest_of_clean_state): Likewise.
10358 * output.h: Delete sdb_begin_function_line.
10359 * sdbout.c: Delete.
10360 * sdbout.h: Delete.
10361 * toplev.c: Delete sdbout.h include.
10362 (process_options): Delete SDB_DEBUG check.
10363 * tree-core.h (tree_type_common): Delete pointer field of
10365 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10366 TYPE_SYMTAB_POINTER.
10367 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10368 (TYPE_SYMTAB_IS_POINTER): Delete.
10369 (TYPE_SYMTAB_IS_DIE): Renumber.
10370 * xcoffout.c: Refer to former sdbout.c file.
10371 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10373 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10374 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10375 (Debugging Options): Delete -gcoff.
10376 (-gstabs): Delete SDB reference.
10378 (-gcoff@var{level}): Delete.
10379 * doc/passes.texi (Debugging information output): Delete SDB and
10380 sdbout.c references.
10381 * doc/tm.texi: Regenerate.
10382 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10383 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10385 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10386 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10387 (SDB_DEBUGGING_INFO): Delete.
10388 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10389 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10390 * target.def (output_source_filename): Delete COFF reference.
10392 * common.opt (gcoff): Delete.
10393 (gxcoff+): Update Negative chain.
10394 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10396 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10397 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10398 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10399 and SDB references.
10400 (expand_function_start): Change sdb reference to past tense.
10401 (expand_function_end): Change sdb reference to past tense.
10402 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10403 * opts.c (debug_type_names): Delete coff entry.
10404 (common_handle_option): Delete OPT_gcoff case.
10405 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10407 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10408 * config/cris/cris.h: Delete SDB reference in comment.
10409 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10410 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10411 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10412 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10414 (ix86_expand_prologue): Likewise.
10415 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10416 * config/ia64/ia64.h: Likewise.
10417 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10418 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10420 * config/mmix/mmix.h: Likewise.
10421 * config/nds32/nds32.c: Likewise.
10422 * config/stormy/storym16.h: Likewise.
10423 * config/visium/visium.h: Likewise.
10424 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10426 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10428 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10429 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10430 FRAME_POINTER_REGNUM point at high end of local var area.
10432 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10434 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10435 Move comment around. Do not reset best_edge for a copiable
10436 destination if the copy would cause a partition change.
10437 (better_edge_p): Remove redundant check.
10439 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10441 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10443 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10445 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10446 for math functions that have _Float<N> and _Float<N>X variants.
10447 (mathfn_built_in_2): Add support for math functions that have
10448 _Float<N> and _Float<N>X variants.
10449 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10450 (expand_builtin_mathfn_ternary): Add support for fma with
10451 _Float<N> and _Float<N>X variants.
10452 (expand_builtin): Likewise.
10453 (fold_builtin_3): Likewise.
10454 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10455 create math function _Float<N> and _Float<N>X variants as external
10457 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10458 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10459 the __builtin_ prefix and if not strict ansi, without the prefix.
10460 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10461 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10462 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10463 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10464 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10465 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10466 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10467 function signatures for fma _Float<N> and _Float<N>X variants.
10468 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10469 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10470 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10471 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10472 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10473 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10474 * gencfn-macros.c (print_case_cfn): Add support for math functions
10475 that have _Float<N> and _Float<N>X variants.
10476 (print_define_operator_list): Likewise.
10477 (fltfn_suffixes): Likewise.
10479 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10480 for math functions that have _Float<N> and _Float<N>X variants.
10481 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10482 and _Float<N>X variants.
10483 (COPYSIGN): Likewise.
10486 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10487 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10489 (integer_valued_read_call_p): Likewise.
10490 * fold-const-call.c (fold_const_call_ss): Likewise.
10491 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10492 _Float<N> and _Float<N>X variants.
10493 (fold_const_call_ssss): Add support for fma _Float<N> and
10494 _Float<N>X variants.
10495 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10496 support for copysign and fma _Float<N> and _Float<N>X variants.
10497 (backprop::process_builtin_call_use): Likewise.
10498 * tree-call-cdce.c (can_test_argument_range); Add support for
10499 sqrt _Float<N> and _Float<N>X variants.
10500 (edom_only_function): Likewise.
10501 (get_no_error_domain): Likewise.
10502 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10503 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10504 copysign _Float<N> and _Float<N>X variants.
10505 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10506 handled by machine independent code.
10507 (FMAF128): Likewise.
10508 * doc/cpp.texi (Common Predefined Macros): Document defining
10509 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10510 fma _Float<N> and _Float<N>X variants.
10512 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10515 * config/i386/i386-modes.def (CCFPU): Remove definition.
10516 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10517 (ix86_cc_modes_compatible): Ditto.
10518 (ix86_expand_carry_flag_compare): Ditto.
10519 (ix86_expand_int_movcc): Ditto.
10520 (ix86_expand_int_addcc): Ditto.
10521 (ix86_reverse_condition): Ditto.
10522 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10523 Return true/false for unordered/ordered fp comparisons.
10524 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10525 (ix86_prepare_fp_compare_args): Update for rename.
10526 (ix86_expand_fp_compare): Update for rename. Generate unordered
10527 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10528 (ix86_expand_sse_compare_and_jump): Ditto.
10529 * config/i386/predicates.md (fcmov_comparison_operator):
10530 Remove CCFPU mode handling.
10531 (ix86_comparison_operator): Ditto.
10532 (ix86_carry_flag_operator): Ditto.
10533 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10534 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10535 (*cmpu<mode>_cc_i387): Ditto.
10536 (FPCMP): Remove mode iterator.
10537 (unord): Remove mode attribute.
10538 (unord_subst): New define_subst transformation
10539 (unord): New define_subst attribute.
10540 (unordered): Ditto.
10541 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10542 (*cmpi<unord>xf_i387): Ditto.
10543 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10544 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10545 using unord_subst transformation.
10546 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10547 (round_saeonly): Also handle CCFP mode.
10548 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10549 Remove UNSPEC_SAHF unspec handling.
10551 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10553 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10555 2017-10-27 Jeff Law <law@redhat.com>
10557 * gimple-ssa-sprintf.c: Include domwalk.h.
10558 (class sprintf_dom_walker): New class, derived from dom_walker.
10559 (sprintf_dom_walker::before_dom_children): New function.
10560 (struct call_info): Moved into sprintf_dom_walker class
10561 (compute_formath_length, handle_gimple_call): Likewise.
10562 (sprintf_length::execute): Call the dominator walker rather
10563 than walking the statements.
10565 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10566 gimple statement locations.
10567 (check_array_bounds): Corresponding changes. Get the statement's
10568 location directly from wi->stmt.
10570 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10573 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10575 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10577 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10578 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10580 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10583 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10584 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10585 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10586 maybe_get_pool_constant.
10587 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10590 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10592 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10593 2.26 caveat. Update gas and gld versions.
10594 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10597 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10599 * cgraph.h (set_malloc_flag): Declare.
10600 * cgraph.c (set_malloc_flag_1): New function.
10601 (set_malloc_flag): Likewise.
10602 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10603 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10605 (read_ipa_call_summary): Add support for reading is_return_callee.
10606 (write_ipa_call_summary): Stream is_return_callee.
10607 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10608 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10609 ipa-prop.h, ipa-fnsummary.h.
10610 (pure_const_names): Change to static.
10611 (malloc_state_e): Define.
10612 (malloc_state_names): Define.
10613 (funct_state_d): Add field malloc_state.
10614 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10615 (check_retval_uses): New function.
10616 (malloc_candidate_p): Likewise.
10617 (analyze_function): Add support for malloc attribute.
10618 (pure_const_write_summary): Stream malloc_state.
10619 (pure_const_read_summary): Add support for reading malloc_state.
10620 (dump_malloc_lattice): New function.
10621 (propagate_malloc): New function.
10622 (warn_function_malloc): New function.
10623 (ipa_pure_const::execute): Call propagate_malloc and
10624 ipa_free_fn_summary.
10625 (pass_local_pure_const::execute): Add support for malloc attribute.
10626 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10627 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10629 2017-10-27 Martin Liska <mliska@suse.cz>
10631 PR gcov-profile/82457
10632 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10633 for fork-like functions to be properly instrumented.
10635 2017-10-27 Richard Biener <rguenther@suse.de>
10637 PR middle-end/81659
10638 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10639 info when we redirected EH.
10641 2017-10-26 Michael Collison <michael.collison@arm.com>
10643 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10645 (<optab>_trunchf<GPI:mode>2: New pattern.
10646 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10647 * config/aarch64/iterators.md (wv): New mode attribute.
10648 (vf, VF): New mode attributes.
10649 (vgp, VGP): New mode attributes.
10650 (s): Update attribute with SImode and DImode prefixes.
10652 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10654 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10655 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10656 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10657 (nios2_option_overide): Initialize it. Don't allow R0-relative
10658 addressing with PIC.
10659 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10660 (nios2_symbolic_constant_p): Likewise.
10661 (nios2_legitimate_address_p): Likewise.
10662 (nios2_r0rel_section_name_p): New.
10663 (nios2_symbol_ref_in_r0rel_data_p): New.
10664 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10665 (r0rel_constant_p): New.
10666 (nios2_print_operand_address): Handle r0rel_constant_p.
10667 (nios2_cdx_narrow_form_p): Likewise.
10668 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10669 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10670 (Nios II Options): Document -mr0rel-sec.
10672 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10674 * config/nios2/nios2.c: Include xregex.h.
10675 (nios2_gprel_sec_regex): New.
10676 (nios2_option_overide): Initialize it. Don't allow GP-relative
10677 addressing with PIC.
10678 (nios2_small_section_name_p): Check for regex match.
10679 * config/nios2/nios2.opt (mgprel-sec=): New option.
10680 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10681 (Nios II Options): Document -mgprel-sec.
10683 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10685 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10687 2017-10-26 Tom de Vries <tom@codesourcery.com>
10689 PR tree-optimization/82707
10690 * gimple.c (gimple_copy): Fix unsharing of
10691 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10693 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10695 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10696 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10697 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10698 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10699 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10700 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10701 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10702 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10703 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10704 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10705 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10706 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10707 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10708 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10709 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10710 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10711 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10713 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10715 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10717 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10718 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10719 warning to rs6000.c. Remove the Undocumented flag, since it has
10721 (-mabi=ibmlongdouble): Likewise.
10722 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10723 already set, set the default format for long double.
10724 (rs6000_debug_reg_global): Print whether long double is IBM or
10726 (rs6000_option_override_internal): Rework setting long double
10727 format. Only warn if the user is changing the long double default
10728 and they did not use -Wno-psabi.
10729 * doc/invoke.texi (PowerPC options): Update the documentation for
10730 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10732 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10733 Alan Hayward <alan.hayward@arm.com>
10734 David Sherwood <david.sherwood@arm.com>
10736 * rtl.h (wider_subreg_mode): New function.
10737 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10738 rather than an unsigned int *.
10739 * ira-color.c (regno_max_ref_width): Replace with...
10740 (regno_max_ref_mode): ...this new variable.
10741 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10742 Use wider_subreg_mode.
10743 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10744 rather than an unsigned int *.
10745 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10746 (process_alt_operands): Likewise.
10747 (invariant_p): Likewise.
10748 * lra-spills.c (assign_mem_slot): Likewise.
10749 (add_pseudo_to_slot): Likewise.
10750 * lra.c (collect_non_operand_hard_regs): Likewise.
10751 (add_regs_to_insn_regno_info): Likewise.
10752 * reload1.c (regno_max_ref_width): Replace with...
10753 (regno_max_ref_mode): ...this new variable.
10754 (reload): Update accordingly. Update call to
10755 ira_sort_regnos_for_alter_reg.
10756 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10757 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10758 (scan_paradoxical_subregs): Likewise.
10760 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10762 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10763 (aarch64_frame): Add emit_frame_chain boolean.
10764 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10765 Move eh_return case to aarch64_layout_frame.
10766 (aarch64_layout_frame): Initialize emit_frame_chain.
10767 (aarch64_expand_prologue): Use emit_frame_chain.
10769 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10771 * config/aarch64/aarch64.c (aarch64_layout_frame):
10772 Ensure LR is always stored at the bottom of the callee-saves.
10773 Remove rarely used frame layout which saves callee-saves at top of
10774 frame, so the store of LR can be used as a valid probe in all cases.
10776 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10778 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10779 Improve unaligned TImode/TFmode base/offset split.
10781 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10782 Alan Hayward <alan.hayward@arm.com>
10783 David Sherwood <david.sherwood@arm.com>
10785 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10786 * combine.c (find_single_use_1): Likewise.
10787 (expand_field_assignment): Likewise.
10788 (move_deaths): Likewise.
10789 * lra-constraints.c (simplify_operand_subreg): Likewise.
10790 (curr_insn_transform): Likewise.
10791 * lra.c (collect_non_operand_hard_regs): Likewise.
10792 (add_regs_to_insn_regno_info): Likewise.
10793 * rtlanal.c (reg_referenced_p): Likewise.
10794 (covers_regno_no_parallel_p): Likewise.
10796 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10798 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10799 Don't print any bits outside the precision of the value.
10800 * wide-int.cc (test_printing): Add some new tests.
10802 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10804 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10805 supports -xbrace_comment option.
10806 * configure: Regenerate.
10807 * config.in: Regenerate.
10808 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10809 (ASM_CPU_SPEC): Use it.
10811 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10813 * target.def (static_rtx_alignment): New hook.
10814 * targhooks.h (default_static_rtx_alignment): Declare.
10815 * targhooks.c (default_static_rtx_alignment): New function.
10816 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10817 * doc/tm.texi: Regenerate.
10818 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10819 instead of targetm.constant_alignment. Remove call to
10820 set_mem_attributes.
10821 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10822 (cris_preferred_mininum_alignment): New function, split out from...
10823 (cris_constant_alignment): ...here.
10824 (cris_static_rtx_alignment): New function.
10825 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10827 (ix86_constant_alignment): ...here.
10828 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10829 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10830 (mmix_static_rtx_alignment): New function.
10831 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10832 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10834 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10837 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10838 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10840 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10842 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10843 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10844 mark as a sign-extending load.
10845 (local_pic_load_u): Define.
10847 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10849 PR middle-end/82062
10850 * fold-const.c (operand_equal_for_comparison_p): Also return true
10851 if ARG0 is a simple variant of ARG1 with narrower precision.
10852 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10854 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10856 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10858 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10859 scatter_static, scatter_per_elt.
10860 * x86-tune-costs.h: Add new cost entries.
10862 2017-10-25 Richard Biener <rguenther@suse.de>
10864 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10865 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10866 class pass_fre): Move to ...
10867 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10868 class pass_fre): ... here and adjust for statistics.
10870 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10873 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10874 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10875 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10876 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10879 2017-10-25 Richard Biener <rguenther@suse.de>
10881 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10882 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10883 eliminate_push_avail, eliminate_insert): Move inside...
10884 (class eliminate_dom_walker): ... this class in preparation
10886 (fini_eliminate): Remove by merging with ...
10887 (eliminate): ... this function. Adjust for class changes.
10888 (pass_pre::execute): Remove fini_eliminate call.
10889 (pass_fre::execute): Likewise.
10891 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10894 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10895 (VPERMI2, VPERMI2I): New mode iterators.
10896 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10897 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10899 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10900 mode iterator. Remove 3 old define_insn patterns.
10901 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10902 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10903 VPERMI2 mode iterator, remove the other two expanders.
10904 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10905 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10906 instructions, remove the other two patterns.
10907 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10908 mode iterator, remove the other two patterns.
10909 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10910 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10911 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10912 and adjust argument order accordingly.
10913 (ix86_expand_vec_perm): Adjust caller.
10914 (expand_vec_perm_1): Likewise.
10915 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10916 (expand_vec_perm_vpermt2_vpshub2): ... this.
10917 (ix86_expand_vec_perm_const_1): Adjust caller.
10918 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10921 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10922 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10923 (vec_shl_<mode>): Remove unused expander.
10924 (avx512bw_<shift_insn><mode>3): New define_insn.
10925 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10926 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10928 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10931 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10934 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10936 PR rtl-optimization/82396
10937 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10938 (autopref_multipass_init): Simplify initialization.
10939 (autopref_rank_data): Simplify sort order.
10940 * gcc/sched-int.h (autopref_multipass_data_): Remove
10941 multi_mem_insn_p, min_offset and max_offset.
10943 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10945 PR middle-end/60580
10946 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10947 Check special value of flag_omit_frame_pointer.
10948 (aarch64_can_eliminate): Likewise.
10949 (aarch64_override_options_after_change_1): Simplify handling of
10950 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10952 2017-10-24 Richard Biener <rguenther@suse.de>
10954 PR tree-optimization/82697
10955 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10956 zero for conditional load and unconditional store.
10958 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10960 * doc/install.texi: Document bootstrap-cet.
10962 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10965 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10966 ENDBR instruction at function entrance if function is only
10969 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10972 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10973 patterns to better describe from which operation the CF is computed.
10974 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10975 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10976 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10977 is 0, use _0 suffixed expanders instead of emitting a comparison
10980 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10982 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10983 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10985 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10987 PR middle-end/82569
10988 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10989 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10990 * loop-iv.c (iv_get_reaching_def): Likewise.
10991 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10992 variable is promoted and the partition contains undefined values.
10994 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10996 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10998 (nios2_address_cost): Define.
10999 (nios2_legitimize_address): Recognize (exp + constant) directly.
11000 (TARGET_ADDRESS_COST): Define.
11002 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11004 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
11005 (nios2_symbolic_memory_operand_p): Declare.
11006 (nios2_split_large_constant): Declare.
11007 (nios2_split_symbolic_memory_operand): Declare.
11008 * config/nios2/nios2.c: Adjust includes.
11009 (nios2_symbolic_constant_allowed): New.
11010 (nios2_symbolic_constant_p): New.
11011 (nios2_plus_symbolic_constant_p): New.
11012 (nios2_valid_addr_expr_p): Recognize addresses involving
11013 symbolic constants.
11014 (nios2_legitimate_address_p): Likewise, also LO_SUM.
11015 (nios2_symbolic_memory_operand_p): New.
11016 (nios2_large_constant_p): New.
11017 (nios2_split_large_constant): New.
11018 (nios2_split_plus_large_constant): New.
11019 (nios2_split_symbolic_memory_operand): New.
11020 (nios2_legitimize_address): Code refactoring. Handle addresses
11021 involving symbolic constants.
11022 (nios2_emit_move_sequence): Likewise.
11023 (nios2_print_operand): Improve error output.
11024 (nios2_print_operand_address): Handle LO_SUM.
11025 (nios2_cdx_narrow_form_p): Likewise.
11026 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
11027 operands involving symbolic constants.
11028 (movhi_internal, movsi_internal): Likewise.
11029 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
11030 (extendhisi2, extendqi<mode>2): Likewise.
11032 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11034 * tree-pass.h (PROP_rtl_split_insns): Define.
11035 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
11037 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11039 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
11041 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11044 * target.def (const_not_ok_for_debug_p): Default to
11045 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
11046 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
11047 * targhooks.c (default_const_not_ok_for_debug_p): New function.
11048 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
11049 which targetm.const_not_ok_for_debug_p returned true.
11050 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
11052 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
11054 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
11055 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
11056 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
11058 (ix86_const_not_ok_for_debug_p): New function.
11059 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
11060 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
11062 2017-10-23 David Malcolm <dmalcolm@redhat.com>
11065 * system.h: Conditionally include "unique-ptr.h" if
11066 INCLUDE_UNIQUE_PTR is defined.
11067 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
11068 of defining INCLUDE_UNIQUE_PTR before including "system.h".
11070 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
11072 * config/rl78/rl78.md: New define_expand "subdi3".
11074 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
11077 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
11078 DF_REF_INSN if DF_REF_INSN_INFO is false.
11080 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
11082 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
11083 xmm_move instead of sse_move.
11084 (sse_store_index): New function.
11085 (ix86_register_move_cost): Be more sensible about mismatch stall;
11086 model AVX moves correctly; make difference between sse->integer and
11088 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
11089 moves; make difference between SSE and AVX.
11090 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
11091 and zmm_move. Increase size of sse load and store tables;
11092 add unaligned load and store tables; add ssemmx_to_integer.
11093 * x86-tune-costs.h: Update all entries according to real
11094 move latencies from Agner Fog's manual and chip documentation.
11096 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11099 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
11100 * config/i386/constraints.md (Wf): New constraint.
11101 * config/i386/i386.md (UNSPEC_SBB): New unspec.
11102 (cmp<dwi>_doubleword): Removed.
11103 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
11104 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
11105 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
11106 expand with cmp<dwi>_doubleword. For LTU and GEU use
11107 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
11109 * common.opt (gcolumn-info): Enable by default.
11110 * doc/invoke.texi (gcolumn-info): Document new default.
11112 2017-10-23 Richard Biener <rguenther@suse.de>
11114 PR tree-optimization/82672
11115 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
11116 Fold the stmt if we propagated into it.
11118 2017-10-23 Richard Biener <rguenther@suse.de>
11120 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
11121 (bitmap_remove_expr_from_set): ... this. All callers call this
11122 for non-constant values.
11123 (bitmap_set_subtract): Rename to...
11124 (bitmap_set_subtract_expressions): ... this. Adjust and
11126 (bitmap_set_contains_value): Remove superfluous check.
11127 (bitmap_set_replace_value): Inline into single caller ...
11128 (bitmap_value_replace_in_set): ... here and simplify.
11129 (dependent_clean): Merge into ...
11130 (clean): ... this using an overload. Adjust.
11131 (prune_clobbered_mems): Adjust.
11132 (compute_antic_aux): Likewise.
11133 (compute_partial_antic_aux): Likewise.
11135 2017-10-23 Richard Biener <rguenther@suse.de>
11137 PR tree-optimization/82129
11139 2017-08-01 Richard Biener <rguenther@suse.de>
11141 PR tree-optimization/81181
11142 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11143 (compute_antic): ... end of iteration here.
11145 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11147 * target.def (starting_frame_offset): New hook.
11148 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
11149 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
11150 * doc/tm.texi.in: Regenerate.
11151 * hooks.h (hook_hwi_void_0): Declare.
11152 * hooks.c (hook_hwi_void_0): New function.
11153 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
11154 STARTING_FRAME_OFFSET.
11155 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
11156 * reload1.c (reload): Likewise.
11157 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
11158 instead of STARTING_FRAME_OFFSET.
11159 * function.c (try_fit_stack_local): Likewise.
11160 (assign_stack_local_1): Likewise
11161 (instantiate_virtual_regs): Likewise.
11162 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11163 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11164 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11165 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11166 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11167 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11168 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11169 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11170 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11171 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11172 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11173 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11174 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11175 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11176 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11177 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11178 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11179 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11180 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11181 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11182 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11183 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11184 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11185 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11186 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11187 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11188 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11189 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11190 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11191 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11192 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11193 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11194 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11195 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11196 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11197 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11198 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11199 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11200 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11201 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11202 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11203 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11204 return a HOST_WIDE_INT.
11205 (avr_builtin_setjmp_frame_value): Use it instead of
11206 STARTING_FRAME_OFFSET.
11207 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11208 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11209 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11211 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11212 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11213 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11214 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11215 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11216 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11217 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11218 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11219 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11220 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11221 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11222 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11224 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11225 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11226 * config/mips/mips.c (mips_compute_frame_info): Refer to
11227 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11228 (mips_starting_frame_offset): New function.
11229 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11230 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11231 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11232 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11233 and return a HOST_WIDE_INT.
11234 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11235 (mmix_initial_elimination_offset): Refer to
11236 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11237 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11238 * config/pa/pa.c (pa_starting_frame_offset): New function.
11239 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11240 (pa_expand_prologue): Likewise.
11241 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11242 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11243 !FRAME_GROWS_DOWNWARD handling to...
11244 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11245 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11246 !FRAME_GROWS_DOWNWARD handling to...
11247 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11248 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11249 !FRAME_GROWS_DOWNWARD handling to...
11250 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11251 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11253 (rs6000_starting_frame_offset): New function.
11254 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11255 !FRAME_GROWS_DOWNWARD handling to...
11256 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11257 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11258 !FRAME_GROWS_DOWNWARD handling to...
11259 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11260 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11261 !FRAME_GROWS_DOWNWARD handling to...
11262 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11263 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11264 (rs6000_starting_frame_offset): New function.
11265 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11266 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11267 * config/vax/vax.c (vax_starting_frame_offset): New function.
11268 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11269 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11270 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11271 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11272 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11273 * system.h (STARTING_FRAME_OFFSET): Poison.
11275 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11277 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11278 SCALAR_TYPE_MODE instead of TYPE_MODE.
11280 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11281 Alan Hayward <alan.hayward@arm.com>
11282 David Sherwood <david.sherwood@arm.com>
11284 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11286 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11287 Alan Hayward <alan.hayward@arm.com>
11288 David Sherwood <david.sherwood@arm.com>
11290 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11292 2017-10-23 Richard Biener <rguenther@suse.de>
11294 PR tree-optimization/82129
11295 * tree-ssa-pre.c (bitmap_set_and): Remove.
11296 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11297 canonicalizing expressions in the set to those with lowest
11298 ID rather than taking that from the first edge.
11300 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11302 * combine.c (rtx_equal_for_field_assignment_p): Use
11303 byte_lowpart_offset.
11305 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11306 Alan Hayward <alan.hayward@arm.com>
11307 David Sherwood <david.sherwood@arm.com>
11309 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11310 to a SUBREG_PROMOTED_VAR.
11312 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11313 Alan Hayward <alan.hayward@arm.com>
11314 David Sherwood <david.sherwood@arm.com>
11316 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11317 (expand_debug_source_expr): Likewise.
11318 * combine.c (combine_simplify_rtx): Likewise.
11319 * cse.c (fold_rtx): Likewise.
11320 * optabs.c (expand_float): Likewise.
11321 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11322 (simplify_binary_operation_1): Likewise.
11324 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11325 Alan Hayward <alan.hayward@arm.com>
11326 David Sherwood <david.sherwood@arm.com>
11328 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11329 (record_promoted_value): Likewise.
11330 * expr.c (expand_expr_real_2): Likewise.
11331 * ree.c (update_reg_equal_equiv_notes): Likewise.
11332 (combine_set_extension): Likewise.
11333 * rtlanal.c (low_bitmask_len): Likewise.
11334 * simplify-rtx.c (neg_const_int): Likewise.
11335 (simplify_binary_operation_1): Likewise.
11337 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11338 Alan Hayward <alan.hayward@arm.com>
11339 David Sherwood <david.sherwood@arm.com>
11341 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11342 * regcprop.c (maybe_mode_change): Likewise.
11343 * reload1.c (alter_reg): Likewise.
11345 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11347 * inchash.h (inchash::hash::add_wide_int): New function.
11348 * lto-streamer-out.c (hash_tree): Use it.
11350 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11352 * inchash.h (inchash::hash::add_wide_int): Rename to...
11353 (inchash::hash::add_hwi): ...this.
11354 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11355 (polymorphic_call_target_hasher::hash): Likewise.
11356 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11357 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11358 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11359 * lto-streamer-out.c (hash_tree): Likewise.
11360 * optc-save-gen.awk: Likewise.
11361 * tree.c (add_expr): Likewise.
11363 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11366 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11367 for ordered inequality comparisons even with TARGET_IEEE_FP.
11369 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11372 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11373 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11374 Expand with cmp<dwi>_doubleword.
11376 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11378 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11379 List CET intrinsics.
11380 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11381 specific to -fcf-protection option.
11383 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11385 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11386 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11387 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11388 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11389 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11390 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11391 (extra_objs): Add cet.o for Linux/x86 targets.
11392 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11393 * config/i386/cet.c: New file.
11394 * config/i386/cetintrin.h: Likewise.
11395 * config/i386/t-cet: Likewise.
11396 * config/i386/cpuid.h (bit_SHSTK): New.
11397 (bit_IBT): Likewise.
11398 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11399 pass IBT and SHSTK bits.
11400 * config/i386/i386-builtin-types.def
11401 (VOID_FTYPE_UNSIGNED_PVOID): New.
11402 (VOID_FTYPE_UINT64_PVOID): Likewise.
11403 * config/i386/i386-builtin.def: Add CET intrinsics.
11404 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11405 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11406 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11407 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11409 * config/i386/i386.c (rest_of_insert_endbranch): New.
11410 (pass_data_insert_endbranch): Likewise.
11411 (pass_insert_endbranch): Likewise.
11412 (make_pass_insert_endbranch): Likewise.
11413 (ix86_notrack_prefixed_insn_p): Likewise.
11414 (ix86_target_string): Add -mibt, -mshstk flags.
11415 (ix86_option_override_internal): Add flag_cf_protection
11417 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11418 (ix86_print_operand): Add 'notrack' prefix output.
11419 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11420 (ix86_expand_builtin): Expand CET intrinsics.
11421 (x86_output_mi_thunk): Add 'endbranch' instruction.
11422 * config/i386/i386.h (TARGET_IBT): New.
11423 (TARGET_IBT_P): Likewise.
11424 (TARGET_SHSTK): Likewise.
11425 (TARGET_SHSTK_P): Likewise.
11426 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11427 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11428 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11429 (builtin_setjmp_setup): New pattern.
11430 (builtin_longjmp): Likewise.
11431 (rdssp<mode>): Likewise.
11432 (incssp<mode>): Likewise.
11433 (saveprevssp): Likewise.
11434 (rstorssp): Likewise.
11435 (wrss<mode>): Likewise.
11436 (wruss<mode>): Likewise.
11437 (setssbsy): Likewise.
11438 (clrssbsy): Likewise.
11439 (nop_endbr): Likewise.
11440 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11442 * config/i386/immintrin.h: Include <cetintrin.h>.
11443 * config/i386/linux-common.h
11444 (file_end_indicate_exec_stack_and_cet): New prototype.
11445 (TARGET_ASM_FILE_END): New.
11447 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11449 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11450 latencies instead of having separate table; make difference between
11451 integer and float costs.
11452 * i386.h (processor_costs): Remove scalar_stmt_cost,
11453 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11454 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11456 * x86-tune-costs.h: Remove entries which has been removed in
11457 procesor_costs from all tables; make cond_taken_branch_cost
11458 and cond_not_taken_branch_cost COST_N_INSNS based.
11460 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11462 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11464 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11466 * config/i386/i386.md (isa): Remove fma_avx512f.
11467 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11468 <avx512>_fmadd_<mode>_mask3<round_name>,
11469 <avx512>_fmsub_<mode>_mask<round_name>,
11470 <avx512>_fmsub_<mode>_mask3<round_name>,
11471 <avx512>_fnmadd_<mode>_mask<round_name>,
11472 <avx512>_fnmadd_<mode>_mask3<round_name>,
11473 <avx512>_fnmsub_<mode>_mask<round_name>,
11474 <avx512>_fnmsub_<mode>_mask3<round_name>,
11475 <avx512>_fmaddsub_<mode>_mask<round_name>,
11476 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11477 <avx512>_fmsubadd_<mode>_mask<round_name>,
11478 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11479 (*vec_widen_umult_even_v16si<mask_name>,
11480 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11481 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11483 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11485 * extend.texi: Add 'nocf_check' documentation.
11486 * gimple.texi: Add second parameter to
11487 gimple_build_call_from_tree.
11488 * invoke.texi: Add -fcf-protection documentation.
11489 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11491 2017-10-20 Richard Biener <rguenther@suse.de>
11493 PR tree-optimization/82473
11494 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11495 the largest input type.
11497 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11499 * c-attribs.c (handle_nocf_check_attribute): New function.
11500 (c_common_attribute_table): Add 'nocf_check' handling.
11501 * gimple-parser.c: Add second argument NULL to
11502 gimple_build_call_from_tree.
11503 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11504 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11506 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11507 * common.opt: Add fcf-protection flag.
11508 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11509 * flag-types.h: Add enum cf_protection_level.
11510 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11511 Add 'nocf_check' attribute propagation to gimple call.
11512 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11513 (gimple_build_call_from_tree): Update prototype.
11514 (gimple_call_nocf_check_p): New function.
11515 (gimple_call_set_nocf_check): Likewise.
11516 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11517 * ipa-icf.c: Add nocf_check attribute in statement hash.
11518 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11519 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11520 * toplev.c (process_options): Add flag_cf_protection handling.
11522 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11524 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11526 2017-10-20 Richard Biener <rguenther@suse.de>
11528 PR tree-optimization/82603
11529 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11530 remove false predicated stores.
11532 2017-10-20 Richard Biener <rguenther@suse.de>
11534 * graphite-isl-ast-to-gimple.c
11535 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11536 Remove return value and simplify, dump copied stmt after lhs
11538 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11539 Reduce dump verbosity.
11540 (gsi_insert_earliest): Likewise.
11541 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11542 * graphite.c (print_global_statistics): Adjust dumping.
11543 (print_graphite_scop_statistics): Likewise.
11544 (print_graphite_statistics): Do not dump loops here.
11545 (graphite_transform_loops): But here.
11547 2017-10-20 Nicolas Roche <roche@adacore.com>
11549 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11550 * configure: Regenerate.
11552 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11555 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11556 functions when optimizing replace GIMPLE_RETURN stmts with
11557 calls to __builtin_unreachable ().
11560 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11561 for -fsanitize=thread link of executables.
11562 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11563 link of executables.
11566 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11567 New mode iterators.
11568 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11569 define_insns for logical vector shifts to use VI248_AVX512BW
11570 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11571 condition, useless isa and prefix attributes. Change the first
11572 2 of these define_insns to ...
11573 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11574 define_insn for avx512vl.
11575 (<shift_insn><mode>3): ... and this, new define_insn without
11576 masking for non-avx512vl.
11579 * config/i386/sse.md (*andnot<mode>3,
11580 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11581 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11582 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11583 not applied use empty suffix even for TARGET_AVX512VL.
11584 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11585 is applied, supply evex,evex or evex,evex,evex instead of just
11588 2017-10-20 Julia Koval <julia.koval@intel.com>
11590 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11591 (OPTION_MASK_ISA_GFNI_UNSET): New.
11592 (ix86_handle_option): Handle OPT_mgfni.
11593 * config/i386/cpuid.h (bit_GFNI): New.
11594 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11595 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11596 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11597 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11598 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11599 * config/i386/i386.opt: Add mgfni.
11601 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11603 * config/msp430/msp430.c (msp430_option_override): Disable
11604 -fdelete-null-pointer-checks.
11605 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11607 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11609 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11610 of x87 and SSE instructions.
11612 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11614 * asan.c (create_cond_insert_point): Do not update edge count.
11615 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11616 (afdo_propagate_circuit): Likewise.
11617 (afdo_calculate_branch_prob): Likewise.
11618 (afdo_annotate_cfg): Likewise.
11619 * basic-block.h (struct edge_def): Remove count.
11620 (edge_def::count): New accessor.
11621 * bb-reorder.c (rotate_loop): Update.
11622 (find_traces_1_round): Update.
11623 (connect_traces): Update.
11624 (sanitize_hot_paths): Update.
11625 * cfg.c (unchecked_make_edge): Update.
11626 (make_single_succ_edge): Update.
11627 (check_bb_profile): Update.
11628 (dump_edge_info): Update.
11629 (update_bb_profile_for_threading): Update.
11630 (scale_bbs_frequencies_int): Update.
11631 (scale_bbs_frequencies_gcov_type): Update.
11632 (scale_bbs_frequencies_profile_count): Update.
11633 (scale_bbs_frequencies): Update.
11634 * cfganal.c (connect_infinite_loops_to_exit): Update.
11635 * cfgbuild.c (compute_outgoing_frequencies): Update.
11636 (find_many_sub_basic_blocks): Update.
11637 * cfgcleanup.c (try_forward_edges): Update.
11638 (try_crossjump_to_edge): Update
11639 * cfgexpand.c (expand_gimple_cond): Update
11640 (expand_gimple_tailcall): Update
11641 (construct_exit_block): Update
11642 * cfghooks.c (verify_flow_info): Update
11643 (redirect_edge_succ_nodup): Update
11644 (split_edge): Update
11645 (make_forwarder_block): Update
11646 (duplicate_block): Update
11647 (account_profile_record): Update
11648 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11649 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11650 * cfgloopmanip.c (scale_loop_profile): Update.
11652 (lv_adjust_loop_entry_edge): Update.
11653 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11654 (force_nonfallthru_and_redirect): Update.
11655 (purge_dead_edges): Update.
11656 (rtl_flow_call_edges_add): Update.
11657 * cgraphunit.c (init_lowered_empty_function): Update.
11658 (cgraph_node::expand_thunk): Update.
11659 * gimple-pretty-print.c (dump_probability): Update.
11660 (dump_edge_probability): Update.
11661 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11662 * haifa-sched.c (sched_create_recovery_edges): Update.
11663 * hsa-gen.c (convert_switch_statements): Update.
11664 * ifcvt.c (dead_or_predicable): Update.
11665 * ipa-inline-transform.c (inline_transform): Update.
11666 * ipa-split.c (split_function): Update.
11667 * ipa-utils.c (ipa_merge_profiles): Update.
11668 * loop-doloop.c (add_test): Update.
11669 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11670 * lto-streamer-in.c (input_cfg): Update.
11671 (input_function): Update.
11672 * lto-streamer-out.c (output_cfg): Update.
11673 * modulo-sched.c (sms_schedule): Update.
11674 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11675 * predict.c (maybe_hot_edge_p): Update.
11676 (unlikely_executed_edge_p): Update.
11677 (probably_never_executed_edge_p): Update.
11678 (dump_prediction): Update.
11679 (drop_profile): Update.
11680 (propagate_unlikely_bbs_forward): Update.
11681 (determine_unlikely_bbs): Update.
11682 (force_edge_cold): Update.
11683 * profile.c (compute_branch_probabilities): Update.
11684 * reg-stack.c (better_edge): Update.
11685 * shrink-wrap.c (handle_simple_exit): Update.
11686 * tracer.c (better_p): Update.
11687 * trans-mem.c (expand_transaction): Update.
11688 (split_bb_make_tm_edge): Update.
11689 * tree-call-cdce.c: Update.
11690 * tree-cfg.c (gimple_find_sub_bbs): Update.
11691 (gimple_split_edge): Update.
11692 (gimple_duplicate_sese_region): Update.
11693 (gimple_duplicate_sese_tail): Update.
11694 (gimple_flow_call_edges_add): Update.
11695 (insert_cond_bb): Update.
11696 (execute_fixup_cfg): Update.
11697 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11698 * tree-complex.c (expand_complex_div_wide): Update.
11699 * tree-eh.c (lower_resx): Update.
11700 (unsplit_eh): Update.
11701 (cleanup_empty_eh_move_lp): Update.
11702 * tree-inline.c (copy_edges_for_bb): Update.
11703 (freqs_to_counts): Update.
11704 (copy_cfg_body): Update.
11705 * tree-ssa-dce.c (remove_dead_stmt): Update.
11706 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11707 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11708 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11709 (unloop_loops): Update.
11710 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11711 * tree-ssa-loop-split.c (connect_loops): Update.
11712 (split_loop): Update.
11713 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11714 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11715 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11716 * tree-ssa-reassoc.c (branch_fixup): Update.
11717 * tree-ssa-tail-merge.c (replace_block_by): Update.
11718 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11719 (compute_path_counts): Update.
11720 (update_profile): Update.
11721 (recompute_probabilities): Update.
11722 (update_joiner_offpath_counts): Update.
11723 (estimated_freqs_path): Update.
11724 (freqs_to_counts_path): Update.
11725 (clear_counts_path): Update.
11726 (ssa_fix_duplicate_block_edges): Update.
11727 (duplicate_thread_path): Update.
11728 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11729 (case_bit_test_cmp): Update.
11730 (collect_switch_conv_info): Update.
11731 (gen_inbound_check): Update.
11732 (do_jump_if_equal): Update.
11733 (emit_cmp_and_jump_insns): Update.
11734 * tree-tailcall.c (decrease_profile): Update.
11735 (eliminate_tail_call): Update.
11736 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11737 (vect_do_peeling): Update.
11738 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11739 * ubsan.c (ubsan_expand_null_ifn): Update.
11740 (ubsan_expand_ptr_ifn): Update.
11741 * value-prof.c (gimple_divmod_fixed_value): Update.
11742 (gimple_mod_pow2): Update.
11743 (gimple_mod_subtract): Update.
11744 (gimple_ic): Update.
11745 (gimple_stringop_fixed_value): Update.
11747 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11750 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11752 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11754 PR rtl-optimization/82395
11755 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11756 based on non_spilled_static_chain_regno_p.
11758 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11760 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11761 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11762 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11764 2017-10-19 Martin Sebor <msebor@redhat.com>
11766 PR tree-optimization/82596
11767 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11769 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11771 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11772 (get_mem_refs_of_builtin_call): Likewise.
11773 * builtins.c (expand_builtin_apply): Adjust call to
11774 allocate_dynamic_stack_space.
11775 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11776 the third argument to allocate_dynamic_stack_space, otherwise -1.
11777 (expand_builtin): Deal with all alloca variants.
11778 (is_inexpensive_builtin): Likewise.
11779 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11780 * calls.c (special_function_p): Deal with all alloca variants.
11781 (initialize_argument_information): Adjust call to
11782 allocate_dynamic_stack_space.
11783 (expand_call): Likewise.
11784 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11785 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11786 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11787 use it for the stack usage computation.
11788 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11789 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11790 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11791 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11792 (in_loop_p): Remove first argument and useless check.
11793 (pass_walloca::execute): Remove useless test and adjust call to above.
11794 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11795 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11796 (gimplify_call_expr): Deal with all alloca variants.
11797 * hsa-gen.c (gen_hsa_alloca): Likewise.
11798 (gen_hsa_insns_for_call): Likewise.
11799 * ipa-pure-const.c (special_builtin_state): Likewise.
11800 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11801 * tree-object-size.c (alloc_object_size): Likewise.
11802 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11803 (call_may_clobber_ref_p_1): Likewise.
11804 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11805 (ccp_fold_stmt): Likewise.
11806 (optimize_stack_restore): Likewise.
11807 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11808 (mark_all_reaching_defs_necessary_1): Likewise.
11809 (propagate_necessity): Likewise.
11810 (eliminate_unnecessary_stmts): Likewise.
11811 * tree.c (build_common_builtin_nodes): Build
11812 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11813 (build_alloca_call_expr): New function.
11814 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11815 (CASE_BUILT_IN_ALLOCA): Likewise.
11816 (build_alloca_call_expr): Declare.
11817 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11819 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11822 * dwarf2out.c (new_die_raw): New static inline function.
11823 (new_die): Use it to create the DIE.
11824 (add_AT_external_die_ref): Likewise.
11825 (clone_die): Likewise.
11826 (clone_as_declaration): Likewise.
11827 (dwarf2out_vms_debug_main_pointer): Likewise.
11828 (base_type_die): Likewise. Remove early return for corner cases.
11829 Do not call add_pubtype on the DIE here.
11830 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11831 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11832 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11833 native order exists for base types, attach the DIE manually and call
11834 add_pubtype on it. Do not equate a reverse order DIE to the type.
11836 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11838 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11839 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11841 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11843 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11845 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11846 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11847 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11848 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11849 * opts.c (sanitizer_opts): Add builtin.
11850 * ubsan.c (instrument_builtin): New function.
11851 (pass_ubsan::execute): Call it.
11852 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11853 * doc/invoke.texi: Document -fsanitize=builtin.
11855 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11856 builtins, store max (log2 (align), 0) into uchar field instead of
11857 align into uptr field.
11858 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11859 store uchar 0 field instead of uptr 0 field.
11860 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11861 instead of passing one address of struct with 2 locations pass
11862 two addresses of structs with 1 location each.
11863 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11864 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11865 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11866 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11867 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11868 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11869 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11870 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11872 2017-10-19 Martin Liska <mliska@suse.cz>
11875 * file-find.c (remove_prefix): Remove.
11876 * file-find.h (remove_prefix): Likewise.
11877 * gcc-ar.c: Remove smartness of lookup.
11879 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11881 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11882 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11883 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11885 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11888 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11889 (setcc + and to xor + setcc): New peephole2.
11891 2017-10-19 Tom de Vries <tom@codesourcery.com>
11893 * doc/sourcebuild.texi (Test Directives, Variants of
11894 dg-require-support): Add dg-require-stack-size.
11896 2017-10-19 Martin Liska <mliska@suse.cz>
11899 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11900 that have a large alignment.
11901 (gimplify_target_expr): Likewise.
11903 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11905 PR rtl-optimization/82602
11906 * ira.c (rtx_moveable_p): Return false for volatile asm.
11908 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11911 * config/i386/i386-modes.def (CCGZ): New CC mode.
11912 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11913 * config/i386/predicates.md (ix86_comparison_operator):
11915 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11916 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11917 with double-word subtraction.
11918 (put_condition_code): Handle CCGZmode.
11920 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11922 * wide-int.cc (debug (const wide_int &)): New.
11923 (debug (const wide_int *)): New.
11924 (debug (const widest_int &)): New.
11925 (debug (const widest_int *)): New.
11927 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11929 PR middle-end/82556
11930 * lra-constraints.c (curr_insn_transform): Use non-input operand
11931 instead of output one for matched reload.
11933 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11935 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11936 (tree-ssa-loop-ivopts.h): New header file.
11937 (struct builtin_info): New fields.
11938 (classify_builtin_1): Compute and record base and offset parts for
11939 memset builtin partition by calling strip_offset.
11940 (offset_cmp, fuse_memset_builtins): New functions.
11941 (finalize_partitions): Fuse adjacent memset partitions by calling
11943 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11944 Expose the interface.
11945 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11947 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11949 PR tree-optimization/82574
11950 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11951 that data reference must be executed exactly once per iteration
11952 against the outermost loop in nest.
11953 (classify_partition): Update call to above function.
11955 2017-10-18 Richard Biener <rguenther@suse.de>
11957 PR tree-optimization/82591
11958 * graphite.c (graphite_transform_loops): Move code gen message
11960 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11961 Here. Handle scop_to_isl_ast failing.
11962 (scop_to_isl_ast): Limit the number of ISL operations.
11964 2017-10-18 Richard Biener <rguenther@suse.de>
11966 * graphite-isl-ast-to-gimple.c
11967 (translate_isl_ast_to_gimple::set_rename): Simplify.
11968 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11969 (graphite_copy_stmts_from_block): ... here.
11970 (copy_bb_and_scalar_dependences): Simplify.
11971 (add_parameters_to_ivs_params): Canonicalize.
11972 (generate_entry_out_of_ssa_copies): Simplify.
11973 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11974 by passing in ISL dimension.
11975 (parameter_index_in_region_1): Rename to ...
11976 (parameter_index_in_region): ... this.
11977 (extract_affine): Adjust assert, pass down parameter index.
11978 (add_param_constraints): Use range-info when available.
11979 (build_scop_context): Adjust.
11980 * sese.c (new_sese_info): Adjust.
11981 (free_sese_info): Likewise.
11982 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11983 Remove unused typedefs.
11984 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11986 2017-10-18 Martin Liska <mliska@suse.cz>
11988 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11990 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11992 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11993 (s390_sched_init): Do not reset s390_sched_state if we entered the
11994 current basic block via a fallthru edge and all others are unlikely.
11996 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11998 * config/s390/s390.c (NUM_SIDES): New variable.
11999 (LONGRUNNING_THRESHOLD): New variable.
12000 (LATENCY_FACTOR): New variable.
12001 (s390_sched_score): Decrease score for long-running instructions on
12003 (s390_sched_variable_issue): Perform bookkeeping for long-running
12006 2017-10-18 Richard Biener <rguenther@suse.de>
12008 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
12009 Simplify with removal of the parameter rename map.
12010 (set_rename): Likewise.
12011 (should_copy_to_new_region): Likewise.
12012 (graphite_copy_stmts_from_block): Likewise.
12013 (copy_bb_and_scalar_dependences): Remove initialization of
12014 unused copied_bb_map.
12015 (copy_def): Remove.
12016 (copy_internal_parameters): Likewise.
12017 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
12018 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
12019 Use INTEGRAL_TYPE_P.
12020 (parameter_index_in_region_1): Rename to ...
12021 (assign_parameter_index_in_region): ... this. Assert we have
12022 a parameter we handle.
12023 (scan_tree_for_params): Adjust.
12024 * sese.h (parameter_rename_map_t): Remove.
12025 (struct sese_info_t): Remove unused parameter_rename_map and
12026 copied_bb_map members.
12027 * sese.c (new_sese_info): Adjust.
12028 (free_sese_info): Likewise.
12030 2017-10-18 Martin Liska <mliska@suse.cz>
12033 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
12034 on an abnormal edge.
12036 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
12038 * doc/invoke.texi (ffunction-sections and fdata-sections):
12041 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12043 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
12044 the use statement can throw internally.
12046 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12048 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
12049 any RTX present on the RHS of a SET.
12050 * compare-elim.c (try_eliminate_compare): Restore comment.
12052 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12054 * langhooks.h (struct lang_hooks): Document that tree_size langhook
12055 may be also called on tcc_type nodes.
12056 * langhooks.c (lhd_tree_size): Likewise.
12058 2017-10-17 David Malcolm <dmalcolm@redhat.com>
12060 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
12061 format_warning_at_substring.
12062 (maybe_warn): Convert source_range * param to a location_t. Pass
12063 UNKNOWN_LOCATION rather than NULL to fmtwarn.
12064 (format_directive): Remove code to extract source_ranges and
12065 source_range * in favor of just a location_t.
12066 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
12068 * substring-locations.c (format_warning_va): Convert
12069 source_range * param to a location_t.
12070 (format_warning_at_substring): Likewise.
12071 * substring-locations.h (format_warning_va): Likewise.
12072 (format_warning_at_substring): Likewise.
12074 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
12076 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
12078 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
12079 and scatter/gather ops.
12081 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
12082 vec_gather_load and vec_scatter_store.
12083 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
12084 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
12085 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
12086 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
12087 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
12088 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
12090 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
12092 * reg-stack.c (compare_for_stack_reg): Add bool argument.
12093 Detect FTST instruction and handle its register pops. Only pop
12094 second operand if can_pop_second_op is true.
12095 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
12096 set can_pop_second_op to false in the compare_for_stack_reg call.
12098 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
12099 output_fp_compare for stack register operands.
12100 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
12101 instructions here. Do not emit stack register pops here. Assert
12102 that FCOMPP pops next to top stack register. Rewrite function.
12104 2017-10-17 Nathan Sidwell <nathan@acm.org>
12106 PR middle-end/82577
12107 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
12108 use DECL_ASSEMBLER_NAME_RAW.
12110 PR middle-end/82546
12111 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
12114 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
12115 Wilco Dijkstra <wilco.dijkstra@arm.com>
12117 * builtins.c (expand_builtin_update_setjmp_buf): Add a
12118 converstion to Pmode from the buf_addr.
12120 2017-10-17 Richard Biener <rguenther@suse.de>
12122 * graphite-dependences.c (scop_get_reads_and_writes): Change
12123 output parameters to references.
12125 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
12127 PR 71026/tree-optimization
12128 * fold-const.c (distribute_real_division): Removed.
12129 (fold_binary_loc): Remove calls to distribute_real_divison.
12131 2017-10-17 Richard Biener <rguenther@suse.de>
12133 * graphite-scop-detection.c
12134 (scop_detection::stmt_has_simple_data_refs_p): Always use
12135 the full nest as region.
12136 (try_generate_gimple_bb): Likewise.
12137 * sese.c (scalar_evolution_in_region): Simplify now that
12138 SCEV can handle instantiation in regions.
12139 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
12140 in the non-loop part of a function if requested.
12142 2017-10-17 Richard Biener <rguenther@suse.de>
12144 PR tree-optimization/82563
12145 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
12147 (graphite_regenerate_ast_isl): Call it.
12148 * graphite-scop-detection.c (build_scops): Remove entry edge split.
12150 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12152 PR tree-optimization/82549
12153 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
12154 Formatting fixes. Instead of calling make_bit_field_ref with negative
12157 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
12159 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
12160 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12161 _mm_maskz_reduce_ss): New.
12162 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12163 __builtin_ia32_reducess_mask): Ditto..
12164 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12165 * config/i386/sse.md (reduces<mode>): Renamed to ...
12166 (reduces<mode><mask_scalar_name>): ... this.
12167 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12169 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12170 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12172 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12174 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12175 * selftest-run-tests.c (selftest::run_tests): Call
12176 selftest::unique_ptr_tests_cc_tests.
12177 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12178 * unique-ptr-tests.cc: New file.
12180 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12183 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12185 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12186 (make_hard_regno_born, make_hard_regno_dead): Update
12187 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12189 2017-10-16 Jeff Law <law@redhat.com>
12191 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12193 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12195 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12197 2017-10-16 Olivier Hainque <hainque@adacore.com>
12199 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12200 with_cpu if we were configured for an e500v2 target cpu name.
12202 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12204 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12205 * doc/invoke.texi: Document +nodsp as a valid extension for
12208 2017-10-16 Martin Liska <mliska@suse.cz>
12210 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12211 (test_set_range): Likewise.
12212 (test_range_functions): Rename to ...
12213 (test_bit_in_range): ... this.
12214 (sbitmap_c_tests): Add new test.
12216 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12218 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12220 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12221 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12223 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12225 * config/aarch64/aarch64-builtins.c
12226 (aarch64_types_quadopu_lane_qualifiers): New.
12227 (TYPES_QUADOPU_LANE): New.
12228 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12229 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12230 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12231 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12232 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12233 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12234 (Vdottype, DOTPROD): New.
12235 (sur): Add SDOT and UDOT.
12237 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12239 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12240 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12241 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12242 Add TARGET_DOTPROD.
12243 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12244 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12245 Enable TARGET_DOTPROD.
12246 (cortex-a75.cortex-a55): Likewise.
12247 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12249 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12251 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12252 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12254 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12256 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12257 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12258 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12259 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12260 * config/arm/types.md (neon_dot, neon_dot_q): New.
12261 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12263 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12265 * config/arm/arm.h (TARGET_DOTPROD): New.
12266 * config/arm/arm.c (arm_arch_dotprod): New.
12267 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12268 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12269 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12270 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12271 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12272 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12273 * doc/invoke.texi (armv8.2-a): Document dotprod
12275 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12277 * i386.c (ix86_vec_cost): New function.
12278 (ix86_rtx_costs): Handle vector operations better.
12279 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12280 * x86-tune-costs.h: Add new costs to all tables.
12282 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12284 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12286 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12287 divsd, sqrtss and sqrtsd
12288 * x86-tune-costs.h: Add new entries to all costs.
12289 (znver1_cost): Fix to match real instruction latencies.
12291 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12292 Michael Collison <michael.collison@arm.com>
12294 * compare-elim.c: Include emit-rtl.h.
12295 (can_merge_compare_into_arith): New function.
12296 (try_validate_parallel): Likewise.
12297 (try_merge_compare): Likewise.
12298 (try_eliminate_compare): Call the above when no previous clobber
12300 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12303 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12305 PR middle-end/62263
12306 PR middle-end/82498
12307 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12308 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12310 PR middle-end/62263
12311 PR middle-end/82498
12312 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12313 to be any operand_equal_p operands. For & (B - 1) require
12314 B to be power of 2. Recognize
12315 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12317 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12320 * optabs.c (expand_memory_blockage): Fix call of
12321 targetm.have_memory_blockage.
12323 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12326 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12327 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12328 objects to extra_objs instead of overwriting it.
12330 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12332 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12333 Use any_fp_register_operand as operand[3] predicate. Simplify
12334 equality test for operands[2] and operands[4] memory location.
12335 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12336 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12337 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12338 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12339 any_fp_register_operand as operand[1] predicate. Simplify
12340 equality test for operands[0] and operands[3] memory location.
12341 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12342 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12343 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12345 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12347 * target-insns.def: Add memory_blockage.
12348 * optabs.c (expand_memory_blockage): New function.
12349 (expand_asm_memory_barrier): Rename ...
12350 (expand_asm_memory_blockage): ... to this.
12351 (expand_mem_thread_fence): Call expand_memory_blockage
12352 instead of expand_asm_memory_barrier.
12353 (expand_mem_singnal_fence): Ditto.
12354 (expand_atomic_load): Ditto.
12355 (expand_atomic_store): Ditto.
12356 * doc/md.texi (Standard Pattern Names For Generation):
12357 Document memory_blockage instruction pattern.
12359 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12361 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12362 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12363 * config/rl78/rl78.md: New define_expand "adddi3".
12365 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12367 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12370 2017-10-13 Jeff Law <law@redhat.com>
12372 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12374 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12377 * internal-fn.c (expand_mul_overflow): If both operands have
12378 the same highpart of -1 or 0 and the topmost bit of lowpart
12379 is different, overflow is if res <= 0 rather than res < 0.
12381 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12383 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12384 TARGET_P9_VECTOR code for unaligned_load case.
12386 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12388 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12390 2017-10-13 Nathan Sidwell <nathan@acm.org>
12392 * tree-core.h (tree_contains_struct): Make bool.
12393 * tree.c (tree_contains_struct): Likewise.
12394 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12395 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12396 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12397 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12398 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12400 2017-10-13 Richard Biener <rguenther@suse.de>
12402 * graphite-isl-ast-to-gimple.c
12403 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12404 parameters and dominance check.
12405 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12406 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12407 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12408 Do not update SSA form here or do intermediate IL verification.
12409 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12410 (graphite_initialize): Remove check on the number of loops in
12411 the function and inline into graphite_transform_loops.
12412 (graphite_finalize): Inline into graphite_transform_loops.
12413 (graphite_transform_loops): Perform SSA update and IL verification
12415 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12417 2017-10-13 Richard Biener <rguenther@suse.de>
12419 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12420 graphite_expression_type_precision): Avoid global constructor
12422 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12423 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12424 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12425 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12426 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12427 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12429 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12432 * config/i386/i386.h (ix86_red_zone_size): New.
12433 * config/i386/i386.md (push peephole2s): Replace
12434 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12436 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12437 Alan Hayward <alan.hayward@arm.com>
12438 David Sherwood <david.sherwood@arm.com>
12440 * combine.c (can_change_dest_mode): Reject changes in
12441 REGMODE_NATURAL_SIZE.
12443 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12444 Alan Hayward <alan.hayward@arm.com>
12445 David Sherwood <david.sherwood@arm.com>
12447 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12448 (expand_debug_source_expr): Likewise.
12449 * combine.c (combine_simplify_rtx): Likewise.
12450 * cse.c (fold_rtx): Likewise.
12451 * fwprop.c (canonicalize_address): Likewise.
12452 * targhooks.c (default_shift_truncation_mask): Likewise.
12454 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12455 Alan Hayward <alan.hayward@arm.com>
12456 David Sherwood <david.sherwood@arm.com>
12458 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12459 (widened_mode): Likewise.
12460 (expand_unop): Likewise.
12461 * ree.c (transform_ifelse): Likewise.
12462 (merge_def_and_ext): Likewise.
12463 (combine_reaching_defs): Likewise.
12464 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12466 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12467 Alan Hayward <alan.hayward@arm.com>
12468 David Sherwood <david.sherwood@arm.com>
12470 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12471 * combine.c (gen_lowpart_for_combine): Likewise.
12472 * dwarf2out.c (rtl_for_decl_location): Likewise.
12473 * final.c (alter_subreg): Likewise.
12474 * rtlhooks.c (gen_lowpart_general): Likewise.
12475 (gen_lowpart_if_possible): Likewise.
12477 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12478 Alan Hayward <alan.hayward@arm.com>
12479 David Sherwood <david.sherwood@arm.com>
12481 * calls.c (expand_call): Use subreg_lowpart_offset.
12482 * cse.c (cse_insn): Likewise.
12483 * regcprop.c (copy_value): Likewise.
12484 (copyprop_hardreg_forward_1): Likewise.
12486 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12489 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12490 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12491 =Q constraints to +Q and into insn condition add check
12492 that operands[0] and operands[1] are equal.
12493 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12494 =Q constraints to +Q and into insn condition add check
12495 that operands[0] is equal to either operands[1] or operands[2].
12498 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12499 instead of handling MINUS_EXPR twice (once for each argument),
12500 canonicalize operand order and handle just once, use rtype where
12501 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12504 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12505 any values of __C while still being pattern recognizable as a simple
12506 rotate instruction.
12508 2017-10-13 Richard Biener <rguenther@suse.de>
12510 PR tree-optimization/82451
12512 2017-10-02 Richard Biener <rguenther@suse.de>
12514 PR tree-optimization/82355
12515 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12516 a mapping for the enclosing loop but avoid generating one for
12517 the loop tree root.
12518 (copy_bb_and_scalar_dependences): Remove premature codegen
12519 error on PHIs in blocks duplicated into multiple places.
12520 * graphite-scop-detection.c
12521 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12522 in the region use it as loop and nest to analyze the DR in.
12523 (try_generate_gimple_bb): Likewise.
12524 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12525 (add_loop_constraints): For blocks in a loop not in the region
12526 create a dimension with a single iteration.
12527 * sese.h (gbb_loop_at_index): Remove assert.
12529 * cfgloop.c (loop_preheader_edge): For the loop tree root
12530 return the single successor of the entry block.
12531 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12532 Reset the SCEV hashtable and niters.
12533 * graphite-scop-detection.c
12534 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12535 assert that we only have POLYNOMIAL_CHREC that vary in loops
12536 contained in the region.
12537 (scop_detection::graphite_can_represent_expr): Adjust.
12538 (scop_detection::stmt_has_simple_data_refs_p): For loops
12539 not in the region set loop to NULL. The nest is now the
12540 entry edge to the region.
12541 (try_generate_gimple_bb): Likewise.
12542 * sese.c (scalar_evolution_in_region): Adjust for
12543 instantiate_scev change.
12544 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12545 Make nest parameter the edge into the region.
12546 (create_data_ref): Likewise.
12547 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12548 entry edge into a region and adjust instantiate_scev calls.
12549 (create_data_ref): Likewise.
12550 (graphite_find_data_references_in_stmt): Likewise.
12551 (find_data_references_in_stmt): Pass the loop preheader edge
12552 from the nest argument.
12553 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12554 parameter the edge into the region.
12555 (instantiate_parameters): Use the loop preheader edge as entry.
12556 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12558 (get_instantiated_value_entry): Make instantiate_below parameter
12559 the edge into the region.
12560 (instantiate_scev_name): Likewise. Adjust dominance checks,
12561 when we cannot use loop-based instantiation instantiate by
12562 walking use-def chains.
12563 (instantiate_scev_poly): Adjust.
12564 (instantiate_scev_binary): Likewise.
12565 (instantiate_scev_convert): Likewise.
12566 (instantiate_scev_not): Likewise.
12567 (instantiate_array_ref): Remove.
12568 (instantiate_scev_3): Likewise.
12569 (instantiate_scev_2): Likewise.
12570 (instantiate_scev_1): Likewise.
12571 (instantiate_scev_r): Do not blindly handle N-operand trees.
12572 Do not instantiate array-refs. Handle all constants and invariants.
12573 (instantiate_scev): Make instantiate_below parameter
12574 the edge into the region.
12575 (resolve_mixers): Use the loop preheader edge for the region
12576 parameter to instantiate_scev_r.
12577 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12579 2017-10-13 Richard Biener <rguenther@suse.de>
12581 PR tree-optimization/82525
12582 * graphite-isl-ast-to-gimple.c
12583 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12585 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12586 Fail code generation when we cannot represent the isl integer.
12587 (binary_op_to_tree): Elide modulo operations that are no-ops
12588 in the type we code generate. Remove now superfluous code
12591 2017-10-13 Richard Biener <rguenther@suse.de>
12593 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12594 (scop_detection::harmful_loop_in_region): Remove premature
12595 IV type restriction.
12596 (scop_detection::graphite_can_represent_scev): We can handle
12597 pointer IVs just fine.
12599 2017-10-13 Alan Modra <amodra@gmail.com>
12601 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12602 "Clobbers and Scratch Registers". Add paragraph on
12603 alternative to clobbers for scratch registers and OpenBLAS
12606 2017-10-13 Alan Modra <amodra@gmail.com>
12608 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12609 example of a memory input for a string of known length. Move
12610 commentary out of table. Add a number of new examples
12611 covering array memory inputs.
12613 2017-10-12 Martin Liska <mliska@suse.cz>
12615 PR tree-optimization/82493
12616 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12617 (test_range_functions): New function.
12618 (sbitmap_c_tests): Likewise.
12619 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12620 * selftest.h (sbitmap_c_tests): New function.
12622 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12624 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12626 * config/rs6000/amo.h: Fix spacing issue.
12628 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12631 * config/i386/i386.md (*ashl<mode>3_mask_1,
12632 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12633 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12636 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12638 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12639 (profile_probability): Set max_probability
12640 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12642 * profile-count.c (profile_probability::differs_from_p): Do not
12643 rely on max_probaiblity == 10000
12645 2017-10-12 Jeff Law <law@redhat.com>
12647 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12650 2017-10-12 Martin Sebor <msebor@redhat.com>
12654 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12655 (handle_alias_pairs): Call it.
12656 * common.opt (-Wattribute-alias): New option.
12657 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12658 * doc/invoke.texi (-Wattribute-alias): Document.
12660 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12663 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12665 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12667 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12668 (make_hard_regno_born, make_hard_regno_dead): Update
12669 bb_killed_pseudos and bb_gen_pseudos.
12671 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12673 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12675 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12677 * config/alpha/alpha.c (alpha_split_conditional_move):
12678 Use std::swap instead of manually swapping.
12679 (alpha_stdarg_optimize_hook): Ditto.
12680 (alpha_canonicalize_comparison): Ditto.
12682 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12684 * tree-loop-distribution.c (struct builtin_info): New struct.
12685 (struct partition): Refactor fields into struct builtin_info.
12686 (partition_free): Free struct builtin_info.
12687 (build_size_arg_loc, build_addr_arg_loc): Delete.
12688 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12689 information from struct builtin_info.
12690 (find_single_drs): New function refactored from classify_partition.
12691 Also moved builtin validity checks to this function.
12692 (compute_access_range, alloc_builtin): New functions.
12693 (classify_builtin_st, classify_builtin_ldst): New functions.
12694 (classify_partition): Refactor code into functions find_single_drs,
12695 classify_builtin_st and classify_builtin_ldst.
12696 (distribute_loop): Don't do runtime alias check when distributing
12698 (find_seed_stmts_for_distribution): New function.
12699 (pass_loop_distribution::execute): Refactor code finding seed
12700 stmts into above function. Support distribution for the innermost
12701 two-level loop nest. Adjust dump information.
12703 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12705 * tree-loop-distribution.c: Adjust the general comment.
12706 (NUM_PARTITION_THRESHOLD): New macro.
12707 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12708 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12709 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12710 in call to build_partition_graph.
12711 (finalize_partitions): New parameter. Make loop distribution more
12712 conservative by fusing more partitions.
12713 (distribute_loop): Don't do runtime alias check in case of loop nest
12715 (find_seed_stmts_for_distribution): New function.
12716 (prepare_perfect_loop_nest): New function.
12717 (pass_loop_distribution::execute): Refactor code finding seed stmts
12718 and loop nest into above functions. Support loop nest distribution.
12719 Adjust dump information accordingly.
12721 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12723 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12724 and set PTYPE_SEQUENTIAL for merged partition.
12726 2017-10-12 Richard Biener <rguenther@suse.de>
12728 PR tree-optimization/69728
12730 2017-09-19 Richard Biener <rguenther@suse.de>
12732 PR tree-optimization/69728
12733 * graphite-sese-to-poly.c (schedule_error): New global.
12734 (add_loop_schedule): Handle empty domain by failing the
12736 (build_original_schedule): Handle schedule_error.
12738 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12739 domain by returning an unchanged schedule.
12741 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12743 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12744 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12746 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12748 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12751 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12754 * expr.c (store_field): Don't optimize away bitsize == 0 store
12755 from CALL_EXPR with addressable return type.
12757 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12759 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12760 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12761 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12762 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12763 TARGET_ISEL instead of TARGET_ISEL<sel>.
12765 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12767 * config/rs6000/rs6000.c
12768 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12770 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12772 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12774 (reg_or_cint_operand): Fix comment.
12775 (reg_or_zero_operand): New predicate.
12776 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12777 * config/rs6000/rs6000.c (output_isel): Delete.
12778 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12779 instead of reg_or_cint_operand. Output instruction directly (not via
12781 (isel_unsigned_<mode>): Ditto.
12782 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12783 gpc_reg_operand. Add an instruction alternative for this. Output
12784 instruction directly.
12785 (*isel_reversed_unsigned_<mode>): Ditto.
12787 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12789 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12790 (TARGET_CANONICALIZE_COMPARISON): Define.
12792 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12795 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12796 Check whether the dest is REG before adding REG_EQUIV note.
12798 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12801 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12803 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12804 (make_hard_regno_born, make_hard_regno_dead): Update
12805 bb_killed_pseudos and bb_gen_pseudos.
12807 2017-10-11 Nathan Sidwell <nathan@acm.org>
12809 * incpath.h (enum incpath_kind): Name enum, prefix values.
12810 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12811 * incpath.c (heads, tails): Use INC_MAX.
12812 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12813 (merge_include_chains, split_quote_chain,
12814 register_include_chains): Update incpath_kind names.
12815 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12816 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12818 (add_framework_path, darwin_register_objc_includes): Likewise.
12819 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12821 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12823 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12824 Do not use float_operator operator predicate.
12825 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12826 * config/i386/predicates.md (float_operator): Remove predicate.
12828 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12830 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12831 (*jccxf_i387): Ditto.
12832 (*jcc<mode>_i387): Ditto.
12833 (*jccu<mode>_i387): Ditto.
12834 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12835 (*jcc_*_i387 splitters): Remove.
12836 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12837 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12838 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12841 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12843 * profile-count.h (slow_safe_scale_64bit): New function.
12844 (safe_scale_64bit): New inline.
12845 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12846 * profile-count.c: Include wide-int.h
12847 (slow_safe_scale_64bit): New.
12849 2017-10-11 Nathan Sidwell <nathan@acm.org>
12851 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12852 HAS_DECL_ASSEMBLER_NAME_P.
12853 * gimple-expr.c (gimple_decl_printable_name: Check
12854 HAS_DECL_ASSEMBLER_NAME_P too.
12855 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12856 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12857 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12858 * passes.c (rest_of_decl_compilation): Check
12859 HAS_DECL_ASSEMBLER_NAME_P too.
12860 * recog.c (verify_changes): Likewise.
12861 * tree-pretty-print.c (dump_decl_name): Likewise.
12862 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12864 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12865 (SET_DECL_ASSEMBLER_NAME): Use it.
12866 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12867 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12868 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12870 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12872 * config.gcc (i386, x86_64): Add extra objects.
12873 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12874 (ix86_min_insn_size): Declare.
12875 (ix86_issue_rate): Declare.
12876 (ix86_adjust_cost): Declare.
12877 (ia32_multipass_dfa_lookahead): Declare.
12878 (ix86_macro_fusion_p): Declare.
12879 (ix86_macro_fusion_pair_p): Declare.
12880 (ix86_bd_has_dispatch): Declare.
12881 (ix86_bd_do_dispatch): Declare.
12882 (ix86_core2i7_init_hooks): Declare.
12883 (ix86_atom_sched_reorder): Declare.
12884 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12885 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12886 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12887 (rip_relative_addr_p): Rename to ...
12888 (ix86_rip_relative_addr_p): ... this one; export.
12889 (memory_address_length): Update.
12890 (ix86_issue_rate): Move to x86-tune-sched.c.
12891 (ix86_flags_dependent): Move to x86-tune-sched.c.
12892 (ix86_agi_dependent): Move to x86-tune-sched.c.
12893 (exact_dependency_1): Move to x86-tune-sched.c.
12894 (exact_store_load_dependency): Move to x86-tune-sched.c.
12895 (ix86_adjust_cost): Move to x86-tune-sched.c.
12896 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12897 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12898 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12899 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12900 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12901 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12902 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12903 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12904 (min_insn_size): Rename to ...
12905 (ix86_min_insn_size): ... this one; export.
12906 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12907 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12908 (core2i7_first_cycle_multipass_backtrack): Move to
12909 x86-tune-sched-core.c.
12910 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12911 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12912 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12913 (ix86_avoid_jump_mispredicts): Update.
12914 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12915 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12916 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12917 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12918 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12919 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12920 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12921 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12922 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12923 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12924 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12925 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12926 (BIG): Move to ix86-tune-sched-bd.c.
12927 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12928 (enum insn_path): Move to ix86-tune-sched-bd.c.
12929 (get_mem_group): Move to ix86-tune-sched-bd.c.
12930 (is_cmp): Move to ix86-tune-sched-bd.c.
12931 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12932 (is_branch): Move to ix86-tune-sched-bd.c.
12933 (is_prefetch): Move to ix86-tune-sched-bd.c.
12934 (init_window): Move to ix86-tune-sched-bd.c.
12935 (allocate_window): Move to ix86-tune-sched-bd.c.
12936 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12937 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12938 (process_end_window): Move to ix86-tune-sched-bd.c.
12939 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12940 (find_constant): Move to ix86-tune-sched-bd.c.
12941 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12942 (has_immediate): Move to ix86-tune-sched-bd.c.
12943 (get_insn_path): Move to ix86-tune-sched-bd.c.
12944 (get_insn_group): Move to ix86-tune-sched-bd.c.
12945 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12946 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12947 (add_insn_window): Move to ix86-tune-sched-bd.c.
12948 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12949 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12950 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12951 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12952 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12953 (do_dispatch): Move to ix86-tune-sched-bd.c.
12954 (has_dispatch): Move to ix86-tune-sched-bd.c.
12955 * config/i386/t-i386: Add new object files.
12956 * config/i386/x86-tune-costs.h: New file.
12957 * config/i386/x86-tune-sched-atom.c: New file.
12958 * config/i386/x86-tune-sched-bd.c: New file.
12959 * config/i386/x86-tune-sched-core.c: New file.
12960 * config/i386/x86-tune-sched.c: New file.
12962 2017-10-11 Liu Hao <lh_mouse@126.com>
12964 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12965 the generic version below instead.
12966 (should_colorize): Recognize Windows consoles as terminals
12968 * pretty-print.c [__MINGW32__] (write_all): New function.
12969 [__MINGW32__] (find_esc_head): Likewise.
12970 [__MINGW32__] (find_esc_terminator): Likewise.
12971 [__MINGW32__] (eat_esc_sequence): Likewise.
12972 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12974 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12977 2017-10-11 Richard Biener <rguenther@suse.de>
12979 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12980 Properly call analyze_scalar_evolution with the loop of the stmt.
12982 2017-10-11 Richard Biener <rguenther@suse.de>
12984 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12985 * tree-core.h (tree_base): Add chrec_var union member.
12986 * tree.h (CHREC_VAR): Remove.
12987 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12988 * tree-chrec.h (build_polynomial_chrec): Adjust.
12989 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12990 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12992 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12994 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12995 * match.pd: ... here.
12996 ((T) X == (T) Y): Relax condition.
12998 2017-10-11 Bin Cheng <bin.cheng@arm.com>
13000 PR tree-optimization/82472
13001 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
13003 (break_alias_scc_partitions): Update postorder number.
13005 2017-10-11 Martin Liska <mliska@suse.cz>
13008 * opts.c (parse_no_sanitize_attribute): Do not use error_value
13010 * opts.h (parse_no_sanitize_attribute): Remove last argument.
13012 2017-10-11 Martin Liska <mliska@suse.cz>
13014 * print-rtl.c (print_insn): Move declaration of idbuf
13015 to same scope as name.
13017 2017-10-11 Martin Liska <mliska@suse.cz>
13022 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13023 volatile arguments.
13025 2017-10-11 Martin Liska <mliska@suse.cz>
13028 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13029 volatile arguments.
13031 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
13033 * config.gcc (default_gnu_indirect_function): Default to yes for
13034 arm*-*-linux* with glibc.
13036 2017-10-11 Richard Biener <rguenther@suse.de>
13038 * tree-scalar-evolution.c (get_scalar_evolution): Handle
13039 default-defs and types we do not want to analyze.
13040 (interpret_loop_phi): Replace unreachable code with an assert.
13041 (compute_scalar_evolution_in_loop): Remove and inline ...
13042 (analyze_scalar_evolution_1): ... here, replacing condition with
13043 what makes the intent clearer. Remove handling of cases
13044 get_scalar_evolution now handles.
13046 2017-10-10 Jim Wilson <wilson@tuliptree.org>
13048 PR rtl-optimization/81434
13049 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
13050 comment for main loop. In sched_group_found if, also add checks for
13051 pass and min_cost_group.
13053 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
13055 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
13056 (rs6000_insn_cost): New function.
13057 * config/rs6000/rs6000.md (cost): New attribute.
13059 2017-10-10 Jakub Jelinek <jakub@redhat.com>
13060 H.J. Lu <hongjiu.lu@intel.com>
13064 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
13065 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
13066 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
13067 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
13068 to OPTION_MASK_ISA_AVX512VL - builtins that have both
13069 OPTION_MASK_ISA_MMX and some other bit set require both
13070 mmx and the ISAs without the mmx bit.
13071 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
13072 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
13073 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
13074 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
13075 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
13076 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
13077 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
13078 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
13079 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
13080 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
13081 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
13082 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
13083 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
13084 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
13085 Add OPTION_MASK_ISA_MMX.
13087 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
13089 * config.gcc (armv7*-*-freebsd*): New target.
13090 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
13092 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
13094 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
13095 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
13098 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
13100 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
13101 a template parameter.
13102 (WIDE_INT_REF_FOR): Update accordingly.
13103 * tree.h (wi::int_traits <const_tree>): Delete.
13104 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
13105 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
13106 (wi::tree_to_wide_ref): New typedef.
13107 (wi::to_wide): New function.
13108 * calls.c (get_size_range): Use wi::to_wide when operating on
13109 trees as wide_ints.
13110 * cgraph.c (cgraph_node::create_thunk): Likewise.
13111 * config/i386/i386.c (ix86_data_alignment): Likewise.
13112 (ix86_local_alignment): Likewise.
13113 * dbxout.c (stabstr_O): Likewise.
13114 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
13115 * expr.c (const_vector_from_tree): Likewise.
13116 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
13117 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
13118 (fold_negate_expr_1, int_const_binop_1, const_binop)
13119 (fold_convert_const_int_from_real, optimize_bit_field_compare)
13120 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
13121 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
13122 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
13123 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
13124 (fold_not_const, round_up_loc): Likewise.
13125 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
13126 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
13127 (alloca_call_type): Likewise.
13128 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
13129 * godump.c (go_output_typedef): Likewise.
13130 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
13131 * internal-fn.c (get_min_precision): Likewise.
13132 * ipa-cp.c (ipcp_store_vr_results): Likewise.
13133 * ipa-polymorphic-call.c
13134 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
13135 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
13136 (ipa_modify_call_arguments): Likewise.
13137 * match.pd: Likewise.
13138 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
13139 * print-tree.c (print_node_brief, print_node): Likewise.
13140 * stmt.c (expand_case): Likewise.
13141 * stor-layout.c (layout_type): Likewise.
13142 * tree-affine.c (tree_to_aff_combination): Likewise.
13143 * tree-cfg.c (group_case_labels_stmt): Likewise.
13144 * tree-data-ref.c (dr_analyze_indices): Likewise.
13145 (prune_runtime_alias_test_list): Likewise.
13146 * tree-dump.c (dequeue_and_dump): Likewise.
13147 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
13148 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
13149 * tree-pretty-print.c (dump_generic_node): Likewise.
13150 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
13151 (simple_iv_with_niters): Likewise.
13152 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
13153 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
13154 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
13155 * tree-ssa-loop-niter.c (split_to_var_and_offset)
13156 (refine_value_range_using_guard, number_of_iterations_ne_max)
13157 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
13158 (get_cst_init_from_scev, record_nonwrapping_iv)
13159 (scev_var_range_cant_overflow): Likewise.
13160 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13161 * tree-ssa-pre.c (compute_avail): Likewise.
13162 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13163 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13164 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13165 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13166 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13167 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13168 (dump_case_nodes, try_switch_expansion): Likewise.
13169 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13170 (vect_do_peeling): Likewise.
13171 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13172 * tree-vect-stmts.c (vectorizable_load): Likewise.
13173 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13174 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13175 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13176 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13177 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13178 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13179 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13180 (evrp_dom_walker::before_dom_children): Likewise.
13181 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13182 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13183 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13184 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13185 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13186 (get_range_pos_neg): Likewise.
13187 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13188 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13189 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13190 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13191 * config/avr/avr.c (avr_fold_builtin): Likewise.
13192 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13193 * config/msp430/msp430.c (msp430_attr): Likewise.
13194 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13195 * config/powerpcspe/powerpcspe-c.c
13196 (altivec_resolve_overloaded_builtin): Likewise.
13197 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13198 (rs6000_expand_ternop_builtin): Likewise.
13199 * config/rs6000/rs6000-c.c
13200 (altivec_resolve_overloaded_builtin): Likewise.
13201 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13202 (rs6000_expand_ternop_builtin): Likewise.
13203 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13205 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13207 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13208 when copying loop nest with only one inner loop.
13210 2017-10-10 Richard Biener <rguenther@suse.de>
13212 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13213 blocks if SCEV is active.
13214 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13216 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13217 (scev_initialize): Assert we are not yet initialized.
13219 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13221 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13222 inner loop's exit stmt by making it always exit the loop, otherwise
13223 we would generate an infinite empty loop.
13225 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13227 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13228 renaming variables in new preheader if it's deleted.
13230 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13232 * tree-loop-distribution.c (struct partition): Remove unused field
13233 loops of the structure.
13234 (partition_alloc, partition_free): Ditto.
13235 (build_rdg_partition_for_vertex): Ditto.
13237 2017-10-09 Jeff Law <law@redhat.com>
13239 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13240 return type to match prototype and documentation.
13242 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13244 * config/rs6000/rs6000.c (processor_costs): Move to ...
13245 * config/rs6000/rs6000.h: ... here.
13246 (rs6000_cost): Declare.
13248 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13250 * except.c (setjmp_fn): New global variable.
13251 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13252 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13253 if DONT_USE_BUILTIN_SETJMP is defined.
13255 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13257 * target.def (insn_cost): New hook.
13258 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13259 * doc/tm.texi: Regenerate.
13260 * rtlanal.c (insn_cost): Use the new hook.
13262 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13264 * combine.c (combine_validate_cost): Compute the new insn_cost,
13265 not just pattern_cost.
13266 (try_combine): Adjust comment.
13268 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13270 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13272 * combine.c (uid_insn_cost): Adjust comment.
13273 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13275 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13276 * dse.c (find_shift_sequence): Ditto.
13277 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13278 (bb_valid_for_noce_process_p): Use pattern_cost.
13279 * rtl.h (insn_rtx_cost): Delete.
13280 (pattern_cost): New prototype.
13281 (insn_cost): New prototype.
13282 * rtlanal.c (insn_rtx_cost): Rename to...
13283 (pattern_cost): ... this.
13286 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13288 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13289 (*jcc<mode>_0_r_i387): Ditto.
13290 (*jccxf_r_i387): Ditto.
13291 (*jcc<mode>_r_i387): Ditto.
13292 (*jccu<mode>_r_i387): Ditto.
13293 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13294 (*jcc): Rename from *jcc_1.
13296 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13298 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13299 deferred rescans after the lvx/stvx recombination pre-pass.
13301 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13303 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13304 memory operation instruction support.
13305 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13306 (rs6000-ibm-aix[789]*): Likewise.
13307 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13308 Document new functions.
13310 2017-10-09 Richard Biener <rguenther@suse.de>
13312 PR tree-optimization/82397
13313 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13314 equality only for semantically equal trees.
13316 2017-10-09 Richard Biener <rguenther@suse.de>
13318 PR tree-optimization/82449
13319 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13320 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13321 allow constant addresses.
13322 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13325 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13327 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13330 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13333 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13336 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13339 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13341 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13344 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13345 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13347 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13349 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13350 (WI_BINARY_PREDICATE_RESULT): Likewise.
13351 (wi::binary_traits::operator_result): New type.
13352 (wi::binary_traits::predicate_result): Likewise.
13353 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13354 (generic_wide_int::operator==, generic_wide_int::operator!=)
13355 (generic_wide_int::operator&, generic_wide_int::and_not)
13356 (generic_wide_int::operator|, generic_wide_int::or_not)
13357 (generic_wide_int::operator^, generic_wide_int::operator+
13358 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13360 (operator~, unary operator-, operator==, operator!=, operator&)
13361 (operator|, operator^, operator+, binary operator-, operator*): New
13363 * expr.c (get_inner_reference): Use wi::bit_and_not.
13364 * fold-const.c (fold_binary_loc): Likewise.
13365 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13366 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13367 (bit_value_binop): Likewise.
13368 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13369 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13370 (extract_range_from_binary_expr_1): Likewise.
13371 (masked_increment): Likewise.
13372 (simplify_bit_ops_using_ranges): Likewise.
13374 2017-10-09 Martin Jambor <mjambor@suse.cz>
13377 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13378 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13379 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13381 (hsa_fixup_mov_insn_type): New function.
13382 (hsa_op_with_type::get_in_type): Use it.
13383 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13384 immediates in an assert.
13385 (hsa_op_with_type::extend_int_to_32bit): New method.
13386 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13387 types. Convert to dest type if necessary.
13388 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13389 (reg_for_gimple_ssa): Pass false as min32int to
13390 hsa_type_for_scalar_tree_type.
13391 (gen_hsa_addr): Fixup type when creating addresable temporary.
13392 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13393 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13394 necessary. Call hsa_fixup_mov_insn_type.
13395 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13396 extend operands and convert to dest type if necessary.
13397 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13398 to dest type if necessary.
13399 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13400 if conversion nt necessary and size matches.
13401 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13402 to dest type if necessary.
13403 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13404 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13406 (gen_hsa_clrsb): Likewise.
13407 (gen_hsa_ffs): Likewise.
13408 (gen_hsa_divmod): Extend operands and convert to dest type if
13410 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13412 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13414 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13415 Remove empty default arguments. Use a brace block as output
13417 (conditional return): Ditto.
13419 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13420 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13422 (group_ending_nop): Ditto.
13423 (doloop_end): Ditto.
13424 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13425 (splitters for those): Ditto.
13427 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13429 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13430 a conditional jump (and the compare for it) so that pc_rtx is the
13432 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13433 for the deleted and renamed ctr<mode>_internal[234] patterns.
13434 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13435 Delete second conditional return pattern.
13436 (ctr<mode>_internal2): Delete this second bdnz pattern.
13437 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13438 (ctr<mode>_internal4): Delete this second bdz pattern.
13440 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13442 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13443 (always_initialized_rtx_for_ssa_name_p): New predicate.
13444 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13445 (finish_out_of_ssa): Free new field of SA.
13446 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13447 * tree-ssa-coalesce.c: Include tree-ssa.h.
13448 (get_parm_default_def_partitions): Remove extern keyword.
13449 (get_undefined_value_partitions): New function.
13450 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13451 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13452 uninitialized bits.
13453 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13455 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13457 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13459 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13461 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13462 for targets that preffer 128bit.
13464 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13466 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13468 2017-10-08 Olivier Hainque <hainque@adacore.com>
13470 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13471 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13473 (thumb_set_return_address): Likewise.
13475 2017-10-08 Olivier Hainque <hainque@adacore.com>
13477 * common/config/arm/arm-common.c (arm_except_unwind_info):
13478 Handle DWARF2_UNWIND_INFO.
13480 2017-10-07 Michael Collison <michael.collison@arm.com>
13482 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13485 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13487 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13488 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13489 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13490 defined, force the creation of a new block for a dispatch label.
13492 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13494 * invoke.texi (Wsuggest-attribute=cold): Document.
13495 * common.opt (Wsuggest-attribute=cold): New
13496 * ipa-pure-const.c (warn_function_cold): New function.
13497 * predict.c (compute_function_frequency): Use it.
13498 * predict.h (warn_function_cold): Declare.
13500 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13502 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13505 2017-10-06 Martin Liska <mliska@suse.cz>
13507 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13508 keyword for member functions.
13509 (struct sanopt_tree_couple): New struct.
13510 (struct sanopt_tree_couple_hash): New function.
13511 (struct sanopt_ctx): Add new hash_map.
13512 (has_dominating_ubsan_ptr_check): New function.
13513 (record_ubsan_ptr_check_stmt): Likewise.
13514 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13515 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13516 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13518 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13521 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13522 aarch64_simd_valid_immediate on CONST_VECTORs.
13523 (aarch64_reg_or_bic_imm): Likewise.
13525 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13527 PR rtl-optimization/82396
13528 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13530 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13532 * graphite-dependences.c (scop_get_reads): Move code to...
13533 (scop_get_must_writes): Move code to...
13534 (scop_get_may_writes): Move code to...
13535 (scop_get_reads_and_writes): ... here.
13536 (scop_get_dependences): Call scop_get_reads_and_writes.
13538 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13540 PR tree-optimization/82434
13541 * fold-const.h (can_native_encode_type_p,
13542 can_native_encode_string_p): Remove.
13543 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13544 don't encode anything, just return what would be otherwise returned.
13545 (native_encode_fixed, native_encode_complex, native_encode_vector):
13547 (native_encode_string): Likewise. Inline by hand
13548 can_native_encode_string_p.
13549 (can_native_encode_type_p): Remove.
13550 (can_native_encode_string_p): Remove.
13551 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13552 STRING_CSTs using can_native_encode_string_p, test all
13553 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13554 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13555 argument from native_encode_expr.
13556 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13557 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13560 2017-10-06 Richard Biener <rguenther@suse.de>
13562 PR tree-optimization/82397
13563 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13564 operand_equal_p but rely on data_ref_compare_tree for detecting
13566 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13567 to match up with dr_group_sort_cmp.
13569 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13572 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13574 * config/s390/s390-builtin-types.def: Regenerate.
13576 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13579 * config/s390/s390-builtin-types.def: Regenerate.
13580 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13581 Change flag from B_VXE to B_VX.
13582 (s390_vec_min_dbl): Remove B_VXE flag.
13584 2017-10-06 Richard Biener <rguenther@suse.de>
13586 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13587 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13588 translate_isl_ast_to_gimple::is_valid_rename,
13589 translate_isl_ast_to_gimple::get_rename,
13590 translate_isl_ast_to_gimple::get_def_bb_for_const,
13591 translate_isl_ast_to_gimple::get_new_name,
13592 translate_isl_ast_to_gimple::collect_all_ssa_names,
13593 translate_isl_ast_to_gimple::copy_loop_phi_args,
13594 translate_isl_ast_to_gimple::collect_all_ssa_names,
13595 translate_isl_ast_to_gimple::copy_loop_phi_args,
13596 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13597 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13598 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13599 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13600 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13601 translate_isl_ast_to_gimple::copy_cond_phi_args,
13602 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13603 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13604 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13605 translate_isl_ast_to_gimple::rename_uses,
13606 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13607 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13608 (set_rename_for_each_def): Likewise.
13609 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13610 here. Handle rewriting SCEV analyzable uses here.
13611 (copy_bb_and_scalar_dependences): Generate code for PHI
13613 (graphite_regenerate_ast_isl): Adjust.
13614 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13615 (add_write, add_read): New functions.
13616 (build_cross_bb_scalars_def): Use it and simplify.
13617 (build_cross_bb_scalars_use): Likewise.
13618 (graphite_find_cross_bb_scalar_vars): Inline into...
13619 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13620 simulating out-of-SSA. Compute liveout and add dependencies.
13621 (build_scops): Force an empty entry block.
13622 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13624 (sese_build_liveouts): Declare.
13625 (sese_trivially_empty_bb_p): Likewise.
13626 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13627 compute liveout and debug_liveout.
13628 (sese_bad_liveouts_use): Remove.
13629 (sese_reset_debug_liveouts_bb): Likewise.
13630 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13631 (sese_build_liveouts): Build liveout and debug_liveout and store
13633 (new_sese_info): Adjust.
13634 (free_sese_info): Likewise.
13635 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13636 do not build liveout here.
13637 (move_sese_in_condition): Adjust region entry.
13638 (scev_analyzable_p): Match up with chrec_apply requirements.
13639 (sese_trivially_empty_bb_p): New.
13640 * tree-into-ssa.c (get_reaching_def): Properly support generating
13641 default-defs for incremental rewrite of anonymous names.
13643 2017-10-06 Richard Biener <rguenther@suse.de>
13645 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13646 precision do not perform modulo reduction.
13648 2017-10-06 Richard Biener <rguenther@suse.de>
13650 PR tree-optimization/82436
13651 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13652 conservatively choose the vectorization factor when checking
13653 whether we can perform the required load permutation.
13654 (vect_transform_slp_perm_load): Assert when we may not fail.
13656 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13658 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13659 message for incompatible -msdata=* and -mcall-* options.
13661 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13663 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13664 rate for post-reload scheduling.
13666 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13668 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13670 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13672 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13673 to improve monte carlo in scimark.
13675 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13677 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13678 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13679 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13680 pentium4_cost, nocona_cost): Set reassociation width to 1.
13681 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13682 width to 2 for fp operations and 1 otherwise.
13683 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13685 (atom_cost): Set reassociation width to 2.
13686 (slm_cost, generic_cost): Set fp reassociation width
13687 to 2 and 1 otherwise.
13688 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13689 (core_cost): Set fp reassociation width to 4 and vector to 2.
13690 (ix86_reassociation_width): Rewrite using cost table; special case
13691 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13692 and TARGET_AVX128_OPTIMAL.
13693 * config/i386/i386.h (processor_costs): Add
13694 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13695 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13696 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13697 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13698 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13699 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13701 2017-10-05 Nathan Sidwell <nathan@acm.org>
13703 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13705 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13707 * config/arm/arm.c (arm_test_fpu_data): New.
13708 (arm_run_selftests): Call arm_test_fpu_data.
13710 2017-10-04 Nathan Sidwell <nathan@acm.org>
13712 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13714 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13716 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13718 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13719 check type for aarch64_simd_valid_immediate.
13720 (aarch64_output_simd_mov_immediate): Update prototype.
13721 (aarch64_simd_valid_immediate): Update prototype.
13722 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13723 support for ORR-immediate.
13724 (and<mode>3): modified pattern to add support for BIC-immediate.
13725 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13726 now checks for valid immediate for BIC and ORR based on new enum
13728 (aarch64_output_simd_mov_immediate): Function now used to output
13729 BIC/ORR imm as well based on new enum argument.
13730 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13732 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13733 (aarch64_reg_or_bic_imm): Likewise.
13735 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13737 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13738 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13740 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13744 PR rtl-optimization/82396
13745 * haifa-sched.c (autopref_multipass_init): Simplify
13747 (autopref_rank_data): Simplify sort order.
13748 * sched-int.h (autopref_multipass_data_): Remove
13749 multi_mem_insn_p, min_offset and max_offset.
13751 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13753 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13755 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13757 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13758 vect_doubleint_cvt.
13760 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13762 * doc/sourcebuild.texi: Document vect_long_mult.
13764 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13766 PR tree-optimization/82413
13767 * fold-const.c (build_range_check): Use widest_int when comparing
13768 the maximum ETYPE value with HIGH.
13770 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13772 PR rtl-optimization/82396
13773 * haifa-sched.c (autopref_multipass_init): Simplify
13775 (autopref_rank_data): Simplify sort order.
13776 * sched-int.h (autopref_multipass_data_): Remove
13777 multi_mem_insn_p, min_offset and max_offset.
13779 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13781 PR tree-optimization/82381
13782 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13783 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13786 PR tree-optimization/82374
13787 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13788 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13789 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13790 current_function_decl to the new decl.
13792 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13794 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13795 helper macro for IEEE float128 hardware built-in functions.
13796 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13798 (TRUNCF128_ODD): Likewise.
13799 (ADDF128_ODD): Likewise.
13800 (SUBF128_ODD): Likewise.
13801 (MULF128_ODD): Likewise.
13802 (DIVF128_ODD): Likewise.
13803 (FMAF128_ODD): Likewise.
13804 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13805 UNSPEC_TRUNC_ROUND_TO_ODD.
13806 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13807 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13808 floating point round to odd instructions.
13809 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13810 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13811 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13812 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13813 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13814 (trunc<mode>sf2_hw): Change the truncate with round to odd
13815 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13816 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13817 to odd hardware instructions.
13818 (sub<mode>3_odd): Likewise.
13819 (mul<mode>3_odd): Likewise.
13820 (div<mode>3_odd): Likewise.
13821 (sqrt<mode>2_odd): Likewise.
13822 (fma<mode>4_odd): Likewise.
13823 (fms<mode>4_odd): Likewise.
13824 (nfma<mode>4_odd): Likewise.
13825 (nfms<mode>4_odd): Likewise.
13826 (trunc<mode>df2_odd): Change the truncate with round to odd
13827 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13829 * doc/extend.texi (PowerPC built-in functions): Update documentation
13830 for existing IEEE float128-bit built-in functions. Add built-in
13831 functions that generate the IEEE 128-bit floating point round to
13834 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13836 PR rtl-optimization/77729
13837 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13838 to (X&(C1&~C2))|C2 transformations.
13840 2017-10-03 Martin Jambor <mjambor@suse.cz>
13842 PR tree-optimization/82363
13843 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13844 mismatch, mark lacc written regardless of racc.
13846 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13848 PR tree-optimization/82381
13849 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13850 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13851 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13852 fallthrough into reassoc_stmt_dominates_stmt_p.
13855 * combine.c (combine_instructions): Don't combine in unreachable
13858 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13861 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13862 function to not use the have_cpu variable. Do not set cpu_index,
13863 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13864 or the default cpu.
13865 (rs6000_valid_attribute_p): Remove duplicate initializations of
13866 old_optimize and func_optimize.
13867 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13868 (rs6000_activate_target_options): Make global.
13869 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13872 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13874 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13875 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13876 Set *poffset to 0 and *psize and *pmax_size to -1 if
13877 *poffset + *psize overflows in HOST_WIDE_INT.
13879 PR tree-optimization/82387
13880 PR tree-optimization/82388
13881 PR tree-optimization/82389
13882 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13883 instead of live_bytes non-NULL.
13885 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13888 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13891 2017-10-02 Richard Biener <rguenther@suse.de>
13893 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13894 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13895 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13897 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13899 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13900 requested precision matches the type's.
13901 * calls.c (alloc_max_size): Calculate the new candidate size as
13902 a widest_int and use wi::to_widest when comparing it with the
13903 current candidate size.
13904 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13905 zero rather than integer_zero_node.
13906 * match.pd: Check for a no-op conversion before using wi::add
13907 rather than after. Use tree_to_uhwi when summing small shift
13908 counts into an unsigned int.
13910 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13911 Alan Hayward <alan.hayward@arm.com>
13912 David Sherwood <david.sherwood@arm.com>
13915 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13916 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13917 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13918 POINTER_AND_FP_REGS.
13920 2017-10-02 Richard Biener <rguenther@suse.de>
13922 PR tree-optimization/82355
13923 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13924 a mapping for the enclosing loop but avoid generating one for
13925 the loop tree root.
13926 (copy_bb_and_scalar_dependences): Remove premature codegen
13927 error on PHIs in blocks duplicated into multiple places.
13928 * graphite-scop-detection.c
13929 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13930 in the region use it as loop and nest to analyze the DR in.
13931 (try_generate_gimple_bb): Likewise.
13932 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13933 (add_loop_constraints): For blocks in a loop not in the region
13934 create a dimension with a single iteration.
13935 * sese.h (gbb_loop_at_index): Remove assert.
13937 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13939 * omp-expand.c (adjust_context_scope): New function.
13940 (expand_parallel_call): Call adjust_context_scope.
13942 2017-10-01 Jeff Law <law@redhat.com>
13944 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13945 dom_opt_dom_walker class with direct access to private members.
13946 Add comments. Call test_for_singularity.
13947 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13948 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13949 m_dummy_cond anymore.
13950 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13952 (pass_dominator:execute): Build the dummy_cond here and pass it
13953 to the dom_opt_dom_walker ctor.
13954 (test_for_singularity): New function.
13956 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13957 Maya Rashish <coypu@sdf.org>
13959 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13960 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13961 (alpha*-*-netbsd*) Use nbsd_tm_file.
13962 (arm*-*-netbsdelf*) Likewise.
13963 (i[34567]86-*-netbsdelf*) Likewise.
13964 (x86_64-*-netbsd*) Likewise.
13965 (mips*-*-netbsd*) Likewise.
13966 (powerpc-*-netbsd*) Likewise.
13967 (sh*-*-netbsd*) Likewise.
13968 (sparc-*-netbsdelf*) Likewise.
13969 (sparc64-*-netbsd*) Likewise.
13970 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13972 (vax-*-netbsdelf*) Likewise.
13973 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13974 (UINT_FAST8_TYPE) Likewise.
13975 (INT_FAST16_TYPE) Check CHAR_FAST16.
13976 (UINT_FAST16_TYPE) Likewise.
13978 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13981 * config/i386/i386.md
13982 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13983 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13984 *divmodsi4_zext_2): New define_insn_and_split.
13985 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13986 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13987 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13988 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13989 New define_insn_and_split.
13990 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13991 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13992 operands[1] having DImode when mode is SImode.
13994 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13995 always SImode for DIV and MOD in REG_EQUAL notes.
13997 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13999 PR middle-end/82319
14000 * match.pd: Fix handling of NaNs in pattern.
14002 2017-09-29 Jeff Law <law@redhat.com>
14004 * sbitmap.c (bitmap_bit_in_range_p): New function.
14005 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
14006 * tree-ssa-dse.c (live_bytes_read): New function.
14007 (dse_classify_store): Ignore reads of dead bytes.
14009 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
14010 typos and whitespace errors.
14011 * config/i386/predicates.md (address_no_seg_operand): Likewise.
14012 * config/s390/s390.c (s390_emit_prologue): Likewise.
14014 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14017 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
14018 with a symbol for LRA.
14020 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14022 PR rtl-optimization/82338
14023 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
14025 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
14027 * genmodes.c (calc_wider_mode): Suppress qsort macro.
14028 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
14029 (qsort_chk): Declare.
14030 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
14031 (qsort_chk): New function.
14033 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14035 PR tree-optimization/82337
14036 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
14037 phi definition if the PHI result appears in an abnormal PHI.
14038 (find_basis_for_base_expr): Don't record a basis if the LHS of the
14039 basis appears in an abnormal PHI.
14041 2017-09-29 Richard Biener <rguenther@suse.de>
14043 * graphite-isl-ast-to-gimple.c
14044 (translate_isl_ast_to_gimple::set_codegen_error): New function.
14045 (binary_op_to_tree): Use it.
14046 (get_rename_from_scev): Likewise.
14047 (copy_loop_phi_nodes): Likewise.
14048 (copy_bb_and_scalar_dependences): Likewise.
14049 (translate_pending_phi_nodes): Likewise.
14051 2017-09-29 Jakub Jelinek <jakub@redhat.com>
14054 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
14055 for movabsq $(i32 << shift), r64.
14057 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
14059 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
14060 index when encoding %esp as %rsp to avoid 0x67 prefix.
14062 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
14064 * config/i386/i386.md (*movsf_internal, *movdf_internal):
14065 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
14067 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
14069 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
14070 Extensions with more than 16 double VFP registers.
14071 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
14072 to_clear_mask and all code related to it. Replace the remaining
14073 entry by a sbitmap and adapt code accordingly.
14075 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
14077 * brig-builtins.def: Change pure attributes to const.
14079 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14081 * config.gcc (default_gnu_indirect_function): Default to yes for
14082 sparc*-*-linux* with glibc.
14084 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14086 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
14087 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
14088 when creating .init_array and .fini_array sections with priority
14091 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
14094 * config/aarch64/aarch64.c
14095 (aarch64_builtin_support_vector_misalignment): Always return false
14096 when misalignment is unknown.
14098 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
14100 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
14101 this function to return false if the definition used by the swap
14102 instruction is artificial, or if the memory address from which the
14103 constant value is loaded is not represented by a base address held
14104 in a register or if the base address register is a frame or stack
14105 pointer. Additionally, return false if the base address of the
14106 loaded constant is a SYMBOL_REF but is not considered to be a
14108 (replace_swapped_load_constant): New function.
14109 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
14110 loaded constant vector with a load of a swapped constant vector.
14112 2017-09-27 Carl Love <cel@us.ibm.com>
14114 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14115 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14116 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14119 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
14121 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
14122 first, always call autopref_rank_data otherwise.
14124 2017-09-27 Richard Biener <rguenther@suse.de>
14126 * graphite-scop-detection.c (find_scop_parameters): Move
14127 loop bound handling ...
14128 (gather_bbs::before_dom_children): ... here, avoiding the need
14129 to build scop_info->loop_nest.
14130 (record_loop_in_sese): Remove.
14131 * sese.h (sese_info_t::loop_nest): Remove.
14132 * sese.c (new_sese_info): Do not allocate loop_nest.
14133 (free_sese_info): Do not free loop_nest.
14135 2017-09-27 Jakub Jelinek <jakub@redhat.com>
14138 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
14139 lhs from calls if the lhs has addressable type.
14141 2017-09-27 Richard Biener <rguenther@suse.de>
14143 * graphite.h (scop::max_alias_set): New member.
14144 * graphite-scop-detection.c: Remove references to non-existing
14145 --param in comments.
14146 (build_alias_sets): Record the maximum alias set used for drs.
14147 (build_scops): Support zero as unlimited for
14148 --param graphite-max-arrays-per-scop.
14149 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
14150 and inline into ...
14151 (build_poly_sr_1): ... here. Compute alias set based on the
14152 maximum alias set used for drs rather than
14153 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
14155 2017-09-27 Richard Biener <rguenther@suse.de>
14157 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
14158 --param loop-block-tile-size=0 to disable tiling.
14160 2017-09-27 Richard Biener <rguenther@suse.de>
14162 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14163 (graphite-max-nb-scop-params): Document special value zero.
14164 * domwalk.h (dom_walker::STOP): New symbolical constant.
14165 (dom_walker::dom_walker): Add optional parameter for bb to
14167 (dom_walker::~dom_walker): Declare.
14168 (dom_walker::before_dom_children): Document STOP return value.
14169 (dom_walker::m_user_bb_to_rpo): New member.
14170 (dom_walker::m_bb_to_rpo): Likewise.
14171 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14172 mapping here if not provided by the user.
14173 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14174 provided by the user.
14175 (dom_walker::STOP): Define.
14176 (dom_walker::walk): Do not compute bb to RPO mapping here.
14177 Support STOP return value from before_dom_children to stop
14179 * graphite-optimize-isl.c (optimize_isl): If the schedule
14180 is the same still generate code if -fgraphite-identity
14181 or -floop-parallelize-all are given.
14182 * graphite-scop-detection.c: Include cfganal.h.
14183 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14185 (gather_bbs::before_dom_children): Return STOP for BBs
14187 (build_scops): Compute bb to RPO mapping and pass it to
14188 the domwalk. Treat --param graphite-max-nb-scop-params=0
14189 as not limiting the number of params.
14190 * graphite.c (graphite_initialize): Remove limit on the
14191 number of basic-blocks in a function.
14192 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14193 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14194 default value of 10.
14196 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14198 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14199 Adjust code to eliminate needing to do the shift right 32-bits
14200 operation after XSCVDPSPN.
14202 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14204 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14205 ((X / Y) != 0 -> X >= Y): Likewise.
14207 2017-09-26 Carl Love <cel@us.ibm.com>
14209 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14210 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14211 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14212 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14213 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14214 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14215 definitions and overloading.
14216 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14217 statement for P9V_BUILTIN_XST_LEN_R.
14218 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14219 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14220 define_expand and define_insn for the instructions and builtins.
14221 * doc/extend.texi: Update the built-in documentation file for the new
14222 built-in functions.
14223 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14224 define_insn for the instructions
14226 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14229 * gcc/config/netbsd-protos.h: New file.
14230 * gcc/config/netbsd.c: New file.
14231 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14232 * gcc/config/t-netbsd: New file.
14233 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14234 (tmake_file) Add t-netbsd.
14235 (extra_objs) Add netbsd.o.
14237 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14241 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14243 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14245 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14246 sign extension from a vector register to a GPR by doing a 32-bit
14247 direct move and then an EXTSW.
14248 (extendsi<mode>2 splitter): Likewise.
14249 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14250 right or vector extract after doing XSCVDPSPN. Use
14251 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14253 (movdi_from_sf_zero_ext): Likewise.
14254 (reload_gpr_from_vsxsf): Likewise.
14255 (p8_mfvsrd_4_disf): Delete, no longer used.
14256 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14257 then needing to move the SFmode to a GPR to use the XSCVDPSP
14258 instruction instead of FRSP and XSCVDPSPN.
14259 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14260 it is adjacent to the other XSCVSPDP insns.
14261 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14262 SFmode to be in traditional Altivec registers.
14263 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14264 (vsx_xscvspdpn): Likewise.
14265 (vsx_xscvspdpn_scalar): Likewise.
14267 2017-09-26 Martin Jambor <mjambor@suse.cz>
14269 * tree-sra.c (compare_access_positions): Put integral types first,
14270 stabilize sorting of integral types, remove conditions putting
14271 non-full-precision integers last.
14272 (sort_and_splice_var_accesses): Disable scalarization if a
14273 non-integert would be represented by a non-full-precision integer.
14275 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14277 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14278 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14279 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14280 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14281 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14282 conditionals inside the function instead of around it. Call
14283 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14284 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14286 2017-09-26 Richard Biener <rguenther@suse.de>
14288 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14290 (scop_detection::build_scop_breadth): ... this. Removed.
14291 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14292 (scop_detection::harmful_stmt_in_bb): Likewise.
14293 (scop_detection::graphite_can_represent_stmt): Likewise.
14294 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14295 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14296 (scop_detection::can_represent_loop_1): ... this. Removed.
14297 (scop_detection::harmful_loop_in_region): Simplify after inlining
14298 the above and remove more quadraticness.
14299 (build_scops): Adjust.
14300 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14303 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14306 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14307 REGNO (base) == SP_REG if base is a REG.
14309 PR middle-end/35691
14310 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14311 if it is different SSA_NAME.
14312 (optimize_range_tests_cmp_bitwise): New function.
14313 (optimize_range_tests): Call it.
14315 2017-09-26 Richard Biener <rguenther@suse.de>
14317 PR tree-optimization/82321
14318 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14319 for the def being inside the loop.
14321 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14323 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14325 * config/s390/s390-builtins.def: Fix constraint on op4.
14327 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14329 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14330 independent expanders.
14331 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14332 ("vec_ordered", "vec_unordered"): New expanders.
14334 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14336 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14339 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14341 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14342 vec_unpacks_lo_v16qi.
14343 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14345 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14347 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14348 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14349 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14351 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14353 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14355 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14357 ("*vec_slb<mode>"): New insn pattern.
14358 ("vec_shr_<mode>"): New expander.
14359 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14360 and force the shift count operand to V16QImode.
14361 ("vec_srb<mode>"): Set shift count mode to V16QI.
14363 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14365 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14366 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14367 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14369 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14372 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14374 2017-09-26 Richard Biener <rguenther@suse.de>
14376 PR tree-optimization/82320
14377 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14380 2017-09-25 Jeff Law <law@redhat.com>
14382 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14383 prototype for new argument.
14384 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14385 mostly extracted from rs6000_emit_allocate_stack.
14386 (rs6000_emit_probe_stack_range_stack_clash): New function.
14387 (rs6000_emit_allocate_stack): Call
14388 rs6000_emit_probe_stack_range_stack_clash as needed.
14389 (rs6000_emit_probe_stack_range): Add additional argument
14390 to call to gen_probe_stack_range{si,di}.
14391 (output_probe_stack_range): New.
14392 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14393 (output_probe_stack_range_stack_clash): New.
14394 (rs6000_emit_prologue): Emit notes into dump file as requested.
14395 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14396 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14397 Add additional operand and pass it to output_probe_stack_range.
14399 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14401 PR tree-optimization/82163
14402 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14403 (checking_verify_loop_closed_ssa): New parameter.
14404 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14405 (check_loop_closed_ssa_stmt): Delete.
14406 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14407 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14408 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14411 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14413 * brig-builtins.def: Treat HSAIL barrier builtins as
14414 setjmp/longjump style functions.
14416 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14418 * target.def (constant_alignment): New hook.
14419 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14420 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14421 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14422 * doc/tm.texi: Regenerate.
14423 * targhooks.h (default_constant_alignment): Declare.
14424 (constant_alignment_word_strings): Likewise.
14425 * targhooks.c (default_constant_alignment): New function.
14426 (constant_alignment_word_strings): Likewise.
14427 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14428 instead of CONSTANT_ALIGNMENT.
14429 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14430 (force_const_mem): Likewise.
14431 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14432 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14433 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14434 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14435 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14437 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14438 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14439 constant_alignment_word_strings.
14440 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14441 (CONSTANT_ALIGNMENT): Likewise.
14442 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14443 (arm_constant_alignment): New function.
14444 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14445 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14446 constant_alignment_word_strings.
14447 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14448 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14449 constant_alignment_word_strings.
14450 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14451 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14452 (cris_constant_alignment): New function.
14453 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14454 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14455 (epiphany_constant_alignment): New function.
14456 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14457 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14458 constant_alignment_word_strings.
14459 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14460 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14461 constant_alignment_word_strings.
14462 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14463 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14464 constant_alignment_word_strings.
14465 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14466 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14467 * config/i386/i386.c (ix86_constant_alignment): Make static.
14468 Use the same interface as the target hook.
14469 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14470 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14471 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14472 constant_alignment_word_strings.
14473 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14474 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14475 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14476 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14477 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14478 constant_alignment_word_strings.
14479 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14480 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14481 constant_alignment_word_strings.
14482 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14483 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14484 constant_alignment_word_strings.
14485 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14486 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14488 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14489 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14490 * config/mips/mips.c (mips_constant_alignment): New function.
14491 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14492 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14493 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14494 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14495 (mmix_constant_alignment): Make static. Use the same interface
14496 as the target hook.
14497 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14498 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14499 constant_alignment_word_strings.
14500 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14501 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14502 constant_alignment_word_strings.
14503 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14504 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14505 constant_alignment_word_strings.
14506 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14507 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14508 (rs6000_constant_alignment): New function.
14509 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14510 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14511 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14512 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14513 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14514 (rs6000_constant_alignment): New function.
14515 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14516 * config/s390/s390.c (s390_constant_alignment): New function.
14517 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14518 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14519 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14520 constant_alignment_word_strings.
14521 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14522 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14523 (sparc_constant_alignment): New function.
14524 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14525 * config/spu/spu.c (spu_constant_alignment): New function.
14526 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14527 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14528 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14529 constant_alignment_word_strings.
14530 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14531 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14532 constant_alignment_word_strings.
14533 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14534 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14535 constant_alignment_word_strings.
14536 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14537 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14538 (visium_constant_alignment): New function.
14539 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14540 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14541 (xtensa_constant_alignment): New function.
14542 * system.h (CONSTANT_ALIGNMENT): Poison.
14544 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14546 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14547 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14548 (rs6000_builtin_valid_without_lhs): New helper function.
14549 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14550 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14552 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14554 * target.h (vec_perm_indices): Use unsigned short rather than
14556 (auto_vec_perm_indices): Likewise.
14557 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14558 Use unsigned int rather than unsigned char.
14559 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14561 2017-09-25 Richard Biener <rguenther@suse.de>
14563 * cfgloop.h (sort_sibling_loops): Declare.
14564 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14565 (sort_sibling_loops): New function sorting the sibling loop list
14567 * graphite.c (graphite_transform_loops): Sort sibling loops.
14569 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14571 * target.def (vec_perm_const_ok): Change sel parameter to
14573 * optabs-query.c (can_vec_perm_p): Update accordingly.
14574 * doc/tm.texi: Regenerate.
14575 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14576 auto_vec_perm_indices and remove separate nelt field.
14577 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14578 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14579 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14580 (aarch64_expand_vec_perm_const): Update accordingly.
14581 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14582 to vec_perm_indices.
14583 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14584 auto_vec_perm_indices and remove separate nelt field.
14585 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14586 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14587 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14589 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14590 to vec_perm_indices.
14591 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14592 sel to vec_perm_indices.
14593 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14594 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14595 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14597 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14600 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14603 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14604 on the FUNCTION_DECL function context if it has a DIE that is a
14607 2017-09-25 Richard Biener <rguenther@suse.de>
14609 PR tree-optimization/82285
14610 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14613 2017-09-25 Tom de Vries <tom@codesourcery.com>
14617 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14620 2017-09-25 Richard Biener <rguenther@suse.de>
14622 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14623 ISL errors other than isl_error_quota happen. Dump if the
14624 schedule is the same.
14625 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14626 errors instead of aborting inside ISL.
14628 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14631 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14632 of libgcc_eh for m64.
14633 * config/i386/darwin64.h: Likewise.
14635 2017-09-25 Richard Biener <rguenther@suse.de>
14637 PR middle-end/82144
14638 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14639 attribute for incomplete types nor twice for complete ones.
14641 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14644 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14645 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14648 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14651 * config/i386/i386.opt (mprefer-avx256): Use
14652 ix86_target_flags variable.
14653 * config/i386/i386.c (ix86_target_string): Move
14654 -mprefer-avx256 to flag2_opts.
14656 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14658 PR middle-end/35691
14659 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14660 and x != -1 | y != -1 into (x & y) != -1.
14662 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14664 * config.gcc: Add new case statement to set
14665 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14666 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14667 s390x-*-linux* case statements. Added aarch64 to the list of
14668 supported architectures.
14670 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14672 PR tree-optimization/82289
14673 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14674 STMT_VINFO_RELEVANT_P.
14676 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14677 Alan Hayward <alan.hayward@arm.com>
14678 David Sherwood <david.sherwood@arm.com>
14680 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14681 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14682 (extract_range_from_binary_expr_1): Don't call
14683 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14685 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14686 Alan Hayward <alan.hayward@arm.com>
14687 David Sherwood <david.sherwood@arm.com>
14689 * target.def (preferred_vector_alignment): New hook.
14690 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14692 * doc/tm.texi: Regenerate.
14693 * targhooks.h (default_preferred_vector_alignment): Declare.
14694 * targhooks.c (default_preferred_vector_alignment): New function.
14695 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14697 (DR_TARGET_ALIGNMENT): New macro.
14698 (aligned_access_p): Update commentary.
14699 (vect_known_alignment_in_bytes): New function.
14700 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14702 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14703 Calculate the misalignment based on the target alignment rather than
14705 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14706 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14707 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14708 the target alignment, rather than masking the element misalignment
14709 with the number of elements in a vector. Also use the target
14710 alignment when calculating the maximum number of peels.
14711 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14712 instead of TYPE_ALIGN_UNIT.
14713 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14714 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14715 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14716 (vect_create_data_ref_ptr): Likewise.
14717 (vect_setup_realignment): Realign by ANDing with
14718 -DR_TARGET_MISALIGNMENT.
14719 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14720 the number of peels based on DR_TARGET_ALIGNMENT.
14721 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14722 with the guaranteed alignment boundary when deciding whether
14724 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14725 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14726 (ensure_base_align): Remove stmt_info parameter. Get the
14727 target base alignment from DR_TARGET_ALIGNMENT.
14728 (vectorizable_store): Update call accordingly. Interpret
14729 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14731 (vectorizable_load): Likewise.
14733 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14734 Alan Hayward <alan.hayward@arm.com>
14735 David Sherwood <david.sherwood@arm.com>
14737 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14738 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14739 (vect_enhance_data_refs_alignment): Likewise.
14741 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14743 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14744 error. Only quit immediately if parsing is complete.
14745 (BEGIN): Initialize fatal_err and parse_done.
14746 (begin fpu, end fpu): Check number of arguments.
14747 (begin arch, end arch): Likewise.
14748 (begin cpu, end cpu): Likewise.
14749 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14750 (optalias): Likewise.
14752 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14754 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14755 * config/arm/arm-isa.h: Delete. Move definitions to ...
14756 * arm-cpus.in: ... here. Use new feature and fgroup values.
14757 * config/arm/arm.c (arm_option_override): Use lower case for feature
14759 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14760 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14761 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14763 (print_isa_bits_for): New function.
14764 (gen_isa): New function.
14765 (gen_comm_data): Use print_isa_bits_for.
14766 (define feature): New keyword.
14767 (define fgroup): New keyword.
14768 * config/arm/t-arm (TM_H): Remove.
14769 (GTM_H): Add arm-isa.h.
14770 (arm-isa.h): Add rule to generate file.
14771 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14772 case for feature bit names.
14774 2017-09-22 Richard Biener <rguenther@suse.de>
14776 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14778 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14779 print of no dependency loops ...
14780 * graphite.c (graphite_transform_loops): ... here.
14781 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14783 (same_close_phi_node, remove_duplicate_close_phi,
14784 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14785 (canonicalize_loop_closed_ssa): ... here and simplify.
14786 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14787 (optimize_isl): Use dump_printf_loc to tell when we stopped
14788 optimizing because of an ISL timeout.
14790 2017-09-22 Richard Biener <rguenther@suse.de>
14792 PR tree-optimization/82291
14793 * tree-if-conv.c (predicate_mem_writes): Make sure to
14794 remove writes in blocks predicated with false.
14796 2017-09-22 Richard Biener <rguenther@suse.de>
14798 * sese.c: Include cfganal.h.
14799 (if_region_set_false_region): Remove.
14800 (create_if_region_on_edge): Likewise.
14801 (move_sese_in_condition): Re-implement without destroying
14804 2017-09-22 Richard Biener <rguenther@suse.de>
14806 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14807 Verify both BBs contain loop PHI nodes before dispatching to
14808 copy_loop_phi_args.
14809 (graphite_regenerate_ast_isl): Do not recompute dominators,
14810 do not verify three times. Restructure for clarity.
14811 * graphite-scop-detection.c (same_close_phi_node,
14812 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14813 defined_in_loop_p, canonicalize_loop_closed_ssa,
14814 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14815 checking and SSA rewrite, move to ...
14816 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14817 (graphite_initialize): Do not pass in ctx, do not reset the
14818 SCEV cache, compute only dominators.
14819 (graphite_transform_loops): Allocate ISL ctx after
14820 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14821 Maintain post-dominators only around build_scops.
14822 * sese.c (if_region_set_false_region): Make static. Free
14823 and recompute dominators.
14824 (move_sese_in_condition): Assert we don't get called with
14825 post-dominators computed.
14826 * sese.h (if_region_set_false_region): Remove.
14828 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14830 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14831 mode attribute for TARGET_AVX512VL.
14833 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14835 * config/i386/i386.opt (mprefer-avx256): New option.
14836 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14838 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14839 for TARGET_PREFER_AVX256.
14840 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14842 2017-09-21 Jeff Law <law@redhat.com>
14844 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14845 Fix dump output if the only stack space is for pushed registers.
14847 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14849 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14852 2017-09-21 Martin Sebor <msebor@redhat.com>
14855 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14856 code (in C++) or code that triggers warnings.
14858 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14860 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14862 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14864 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14865 * sched-rgn.c: Ditto.
14866 * sel-sched-ir.c: Ditto.
14868 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14870 * toplev.h (set_random_seed): Adjust return type.
14871 * toplev.c (init_local_tick): Move eager initialization of random_seed
14872 to get_random_seed. Adjust comment.
14873 (init_random_seed): Inline to get_random_seed, delete.
14874 (get_random_seed): Initialize random_seed lazily.
14875 (set_random_seed): Do not return previous value.
14876 (print_switch_value): Do not call get_random_seed.
14878 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14880 * cgraph.c (delete_function_version): New, broken out from...
14881 (cgraph_node::delete_function_version): ...here. Rename to
14882 cgraph_node::delete_function_version_by_decl. Update all uses.
14883 (cgraph_node::remove): Call delete_function_version.
14885 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14888 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14889 VAR_DECLs to which addressable non-volatile parameters are mapped
14890 and for id->retvar after the return value assignment. Clear
14891 id->retval and id->retbnd after inlining.
14893 2017-09-21 Richard Biener <rguenther@suse.de>
14895 PR tree-optimization/82276
14896 PR tree-optimization/82244
14897 * tree-vrp.c (build_assert_expr_for): Set
14898 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14900 (remove_range_assertions): Revert earlier change.
14902 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14905 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14907 2017-09-21 Richard Biener <rguenther@suse.de>
14909 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14910 Restore valid IL after code generation errors.
14911 * graphite.c (graphite_transform_loops): Diagnose code
14912 generation issues as MSG_MISSED_OPTIMIZATION and continue
14913 with processing SCOPs.
14915 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14916 Alan Hayward <alan.hayward@arm.com>
14917 David Sherwood <david.sherwood@arm.com>
14919 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14920 rather than choosing between plus_constant and gen_rtx_<CODE>.
14921 * expr.c (emit_push_insn): Likewise.
14922 (expand_expr_real_2): Likewise.
14924 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14925 Alan Hayward <alan.hayward@arm.com>
14926 David Sherwood <david.sherwood@arm.com>
14928 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14930 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14931 Alan Hayward <alan.hayward@arm.com>
14932 David Sherwood <david.sherwood@arm.com>
14934 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14935 calling tree_to_uhwi.
14937 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14938 Alan Hayward <alan.hayward@arm.com>
14939 David Sherwood <david.sherwood@arm.com>
14941 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14942 INTEGER_CST rather than a negative test for ADDR_EXPR.
14944 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14945 Alan Hayward <alan.hayward@arm.com>
14946 David Sherwood <david.sherwood@arm.com>
14948 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14949 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14951 2017-09-21 Richard Biener <rguenther@suse.de>
14953 PR tree-optimization/71351
14954 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14955 graphite_create_new_loop_guard): Remove, fold remaining parts
14957 (translate_isl_ast_node_for): ... here and simplify.
14959 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14962 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14963 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14964 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14965 alternative always use QI mode, for -Os imov (=R,R) alternative
14966 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14969 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14970 Jeff Law <law@redhat.com>
14972 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14973 (allocate_stack_space): New function, partially extracted from
14974 s390_emit_prologue.
14975 (s390_emit_prologue): Track offset to most recent stack probe.
14976 Code to allocate space moved into allocate_stack_space.
14977 Dump actions when no stack is allocated.
14978 (s390_prologue_plus_offset): New function.
14979 (s390_emit_stack_probe): Likewise.
14981 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14983 * common.opt (Wa, Wl, Wp, g, gz=): Add
14985 (gno-column-info): Remove.
14986 (gcolumn-info): Drop RejectNegative.
14987 (gno-): New prefix.
14988 (gno-record-gcc-switches): Remove.
14989 (grecord-gcc-switches): Drop RejectNegative.
14990 (gno-split-dwarf): Remove.
14991 (gsplit-dwarf): Drop RejectNegative.
14992 (gno-strict-dwarf): Remove.
14993 (gstrict-dwarf): Drop RejectNegative.
14994 * config/darwin.opt (gfull, gused): Add RejectNegative.
14995 * dwarf2out.c (gen_producer_string): Drop
14996 gno-record-gcc-switches handler.
14997 * optc-gen.awk: Add g to prefixes with negative forms.
14998 * opts-common.c (remapping_prefix_p): New.
14999 (find_opt): Check it.
15000 (generate_canonical_option): Test g prefix.
15001 (option_map): Add -gno- mapping.
15002 (add_misspelling_candidates): Check remapping_prefix_p.
15004 2017-09-20 Jeff Law <law@redhat.com>
15006 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
15007 thinko in stack clash protection support.
15009 * explow.c (compute_stack_clash_protection_loop_data): Use
15010 CONST_INT_P instead of explicit test. Verify object is a
15011 CONST_INT_P before looking at INTVAL.
15012 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
15013 instead of explicit test.
15015 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
15018 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
15019 address instead of to r1 and r11.
15021 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
15023 * config.gcc: Support "knm".
15024 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
15025 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15027 * config/i386/i386.c (m_KNM): Define.
15028 (processor_target_table): Add "knm".
15030 (ix86_option_override_internal): Add "knm".
15031 (ix86_issue_rate): Add PROCESSOR_KNM.
15032 (ix86_adjust_cost): Ditto.
15033 (ia32_multipass_dfa_lookahead): Ditto.
15034 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
15035 (fold_builtin_cpu): Add M_INTEL_KNM.
15036 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
15037 (processor_type): Add PROCESSOR_KNM.
15038 * config/i386/x86-tune.def: Add m_KNM.
15039 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
15041 2017-09-20 Richard Biener <rguenther@suse.de>
15043 PR tree-optimization/80213
15044 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
15045 are allowed in empty BBs as well.
15046 (canonicalize_loop_closed_ssa): Also look for other complex
15048 (scop_detection::get_sese): Include the loop-closed PHI block
15050 (scop_detection::merge_sese): Remove code adding extra blocks.
15051 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
15052 (build_scops): Assert the final returned scop is invalid.
15054 2017-09-20 Richard Biener <rguenther@suse.de>
15056 PR tree-optimization/82264
15057 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
15059 (vn_phi_lookup): Likewise.
15060 (vn_phi_insert): Likewise.
15062 2017-09-20 Jakub Jelinek <jakub@redhat.com>
15064 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
15065 that fits into uhwi or shwi, add DW_AT_const_value regardless
15066 of early_dwarf without going through RTL, using add_AT_unsigned
15069 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
15070 (DEBUG_LTO_ABBREV_SECTION): Likewise.
15071 (DEBUG_LTO_MACINFO_SECTION): Likewise.
15072 (DEBUG_MACRO_SECTION): Likewise.
15073 (DEBUG_LTO_MACRO_SECTION): Likewise.
15074 (DEBUG_STR_DWO_SECTION): Likewise.
15075 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
15076 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
15077 (DEBUG_LTO_DWO_LINE_SECTION): Define.
15078 (DEBUG_LTO_LINE_STR_SECTION): Define.
15079 (init_sections_and_labels): Initialize debug_line_str_section
15080 variable. Initialize debug_loc_section for -gdwarf-5 to
15081 DEBUG_LOCLISTS_SECTION. Formatting fixes.
15083 2017-09-20 Richard Biener <rguenther@suse.de>
15085 * graphite-sese-to-poly.c (extract_affine): Properly handle
15086 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
15088 2017-09-20 Richard Biener <rguenther@suse.de>
15090 PR tree-optimization/81373
15091 * graphite-scop-detection.c (build_cross_bb_scalars_def):
15092 Force SESE live-out defs to be handled even if they are
15095 2017-09-19 Jeff Law <law@redhat.com>
15097 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
15098 nothing for stack adjustments with REG_STACK_CHECK.
15099 * sched-deps.c (parse_add_or_inc): Reject insns with
15100 REG_STACK_CHECK from dependency breaking.
15101 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
15102 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
15103 * reg-notes.def (STACK_CHECK): New note.
15105 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
15106 (ix86_expand_prologue): Dump stack clash info as needed.
15107 Call ix86_adjust_stack_and_probe_stack_clash as needed.
15109 * function.c (dump_stack_clash_frame_info): New function.
15110 * function.h (dump_stack_clash_frame_info): Prototype.
15111 (enum stack_clash_probes): New enum.
15113 * config/alpha/alpha.c (alpha_expand_prologue): Also check
15114 flag_stack_clash_protection.
15115 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
15116 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
15117 (arm_frame_pointer_required): Likewise.
15118 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
15119 (ia64_expand_prologue): Likewise.
15120 * config/mips/mips.c (mips_expand_prologue): Likewise.
15121 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
15122 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15123 (sparc_flat_expand_prologue): Likewise.
15124 * config/spu/spu.c (spu_expand_prologue): Likewise.
15126 * explow.c: Include "params.h".
15127 (anti_adjust_stack_and_probe_stack_clash): New function.
15128 (get_stack_check_protect): Likewise.
15129 (compute_stack_clash_protection_loop_data): Likewise.
15130 (emit_stack_clash_protection_loop_start): Likewise.
15131 (emit_stack_clash_protection_loop_end): Likewise.
15132 (allocate_dynamic_stack_space): Use get_stack_check_protect.
15133 Use anti_adjust_stack_and_probe_stack_clash.
15134 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
15135 (emit_stack_clash_protection_loop_start): Likewise.
15136 (emit_stack_clash_protection_loop_end): Likewise.
15137 * rtl.h (get_stack_check_protect): Prototype.
15138 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
15139 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
15140 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
15142 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
15144 * doc/tm.texi: Rebuilt.
15145 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
15146 get_stack_check_protect.
15147 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
15148 * config/arm/arm.c (arm_expand_prologue): Likewise.
15149 (arm_frame_pointer_required): Likewise.
15150 * config/i386/i386.c (ix86_expand_prologue): Likewise.
15151 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
15152 * config/mips/mips.c (mips_expand_prologue): Likewise.
15153 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
15154 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
15155 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15156 (sparc_flat_expand_prologue): Likewise.
15158 * common.opt (-fstack-clash-protection): New option.
15159 * flag-types.h (enum stack_check_type): Note difference between
15160 -fstack-check= and -fstack-clash-protection.
15161 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15162 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15163 * toplev.c (process_options): Issue warnings/errors for cases
15164 not handled with -fstack-clash-protection.
15165 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15166 (-fstack-check): Note additional problem with -fstack-check=generic.
15167 Note that -fstack-check is primarily for Ada and refer users
15168 to -fstack-clash-protection for stack-clash-protection.
15169 Document new params for stack clash protection.
15171 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15173 * config/i386/i386.c (ix86_split_long_move): Do not handle
15174 address used for LEA in a special way.
15176 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15178 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15180 2017-09-19 Martin Sebor <msebor@redhat.com>
15183 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15184 of incompatible types.
15186 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15188 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15189 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15190 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15191 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15193 2017-09-19 Richard Biener <rguenther@suse.de>
15195 PR tree-optimization/82244
15196 * tree-vrp.c (remove_range_assertions): Do not propagate
15197 a constant to abnormals but replace the assert with a copy.
15199 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15201 PR rtl-optimization/57878
15202 PR rtl-optimization/68988
15203 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15204 avoidance test involving non_reload_pseudos. Move frequency test
15205 below the general fragmentation avoidance test.
15207 2017-09-19 Richard Biener <rguenther@suse.de>
15209 PR tree-optimization/69728
15210 * graphite-sese-to-poly.c (schedule_error): New global.
15211 (add_loop_schedule): Handle empty domain by failing the
15213 (build_original_schedule): Handle schedule_error.
15215 2017-09-19 Richard Biener <rguenther@suse.de>
15217 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15218 Do not iterate to sibling loops but only to siblings of inner
15221 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15224 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15225 that the registers are different.
15227 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15229 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15230 to processor_model and "amdfam17h" to arch_names_table.
15231 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15233 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15236 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15238 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15239 Alan Hayward <alan.hayward@arm.com>
15240 David Sherwood <david.sherwood@arm.com>
15242 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15244 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15246 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15247 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15248 vect_schedule_slp_instance.
15249 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15250 Update call to vect_slp_analyze_node_operations. Simplify return
15252 (vect_slp_analyze_bb_1): Update call accordingly.
15253 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15254 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15255 (vect_schedule_slp): Update call accordingly.
15257 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15258 Alan Hayward <alan.hayward@arm.com>
15259 David Sherwood <david.sherwood@arm.com>
15261 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15262 with types that aren't in fact scalar.
15264 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15266 * tree-vect-slp.c (vect_record_max_nunits): New function,
15268 (vect_build_slp_tree_1): ...here.
15269 (vect_build_slp_tree_2): Call it for phis too.
15271 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15273 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15274 to vect_get_vec_def_for_operand when getting the mask operand.
15276 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15277 Alan Hayward <alan.hayward@arm.com>
15278 David Sherwood <david.sherwood@arm.com>
15280 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15283 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15284 Alan Hayward <alan.hayward@arm.com>
15285 David Sherwood <david.sherwood@arm.com>
15287 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15288 calculation for vector booleans.
15290 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15291 Alan Hayward <alan.hayward@arm.com>
15292 David Sherwood <david.sherwood@arm.com>
15294 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15296 (vect_transform_stmt): ...here.
15297 (vect_analyze_stmt): Use it instead of calling
15298 vectorizable_live_operation directly.
15300 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15302 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15303 non-SIMT targets in acc vector loops.
15305 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15307 * configure.ac: Add arc and check if assembler supports gdwarf2.
15308 * configure: Regenerate.
15310 2017-09-18 Richard Biener <rguenther@suse.de>
15312 PR tree-optimization/82220
15313 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15314 epilogue niters from the min_profitable_iters compute.
15316 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15319 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15320 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15321 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15323 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15326 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15327 switching to a new text section.
15329 2017-09-18 Richard Biener <rguenther@suse.de>
15331 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15333 (build_alias_set): Reject aliases with no access function.
15335 2017-09-18 Richard Biener <rguenther@suse.de>
15337 PR tree-optimization/79622
15338 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15340 (build_cross_bb_scalars_use): Likewise.
15342 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15344 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15346 2017-09-18 Alan Modra <amodra@gmail.com>
15349 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15350 stack_pointer_rtx for count 0. Update comments. Break up
15351 large rtl expression.
15353 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15355 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15356 Increase to 20 bytes.
15357 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15358 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15359 or avx version of the stub.
15361 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15364 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15365 compute the minimum stack alignment. Also update preferred stack
15366 boundary for leaf functions.
15368 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15370 PR tree-optimization/82228
15371 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15374 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15376 * common/config/nds32/nds32-common.c
15377 (nds32_option_optimization_table): Refine formatting.
15378 (nds32_option_optimization_table): Use -fsched-pressure and
15379 -fomit-frame-pointer for specific optimization level.
15381 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15383 * config/nds32/nds32.c: Refine formatting and comments.
15384 * config/nds32/nds32.h: Likewise.
15385 * config/nds32/nds32.md: Likewise.
15386 * config/nds32/nds32-cost.c: Likewise.
15387 * config/nds32/nds32-isr.c: Likewise.
15388 * config/nds32/nds32-md-auxiliary.c: Likewise.
15389 * config/nds32/nds32-multiple.md: Likewise.
15390 * config/nds32/nds32-predicates.c: Likewise.
15392 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15393 Jakub Jelinek <jakub@redhat.com>
15395 Add support for -std=c++2a.
15396 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15398 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15400 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15403 * doc/extend.texi (Common Function Attributes): Add
15404 references to ARM, AArch64, and S/390 specific attributes.
15405 (Function Specific Option Pragmas): Add AArch64 and S/390
15406 to list of back ends that support the target pragma.
15408 2017-09-15 Nathan Sidwell <nathan@acm.org>
15410 * doc/standards.texi: Fix C++17 description. Update URLs for
15413 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15415 * common.opt (Wcast-align=strict): New warning option.
15416 * doc/invoke.texi: Document -Wcast-align=strict.
15418 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15420 * cgraph.h (cgraph_thunk_info): Add comments.
15421 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15422 assert for VIRTUAL_* arguments stricter.
15424 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15426 PR tree-optimization/71026
15427 * match.pd: Move RDIV patterns from fold-const.c
15428 * fold-const.c (distribute_real_division): Removed.
15429 (fold_binary_loc): Remove calls to distribute_real_divison.
15431 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15433 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15434 c++1z and gnu++1z as deprecated. Change other references to
15435 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15436 Change -Wc++1z-compat to -Wc++17-compat.
15437 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15438 * dwarf2out.c (highest_c_language): Handle C++17.
15439 (gen_compile_unit_die): Likewise.
15441 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15443 PR rtl-optimization/82192
15444 * combine.c (make_extraction): Don't look through non-paradoxical
15445 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15448 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15449 Alan Hayward <alan.hayward@arm.com>
15450 David Sherwood <david.sherwood@arm.com>
15452 * target.def (function_arg_offset): New hook.
15453 * targhooks.h (default_function_arg_offset): Declare.
15454 * targhooks.c (default_function_arg_offset): New function.
15455 * function.c (locate_and_pad_parm): Use
15456 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15457 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15458 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15459 * doc/tm.texi: Regenerate.
15460 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15461 * config/spu/spu.c (spu_function_arg_offset): New function.
15462 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15463 * system.h (FUNCTION_ARG_OFFSET): Poison.
15465 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15466 Alan Hayard <alan.hayward@arm.com>
15467 David Sherwood <david.sherwood@arm.com>
15469 * target.def (truly_noop_truncation): New hook.
15470 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15471 than TRULY_NOOP_TRUNCATION.
15472 * hooks.h (hook_bool_uint_uint_true): Declare.
15473 * hooks.c (hook_bool_uint_uint_true): New function.
15474 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15475 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15476 * doc/tm.texi: Regenerate.
15477 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15478 rather than TRULY_NOOP_TRUNCATION in comments.
15479 (simplify_comparison): Likewise.
15480 (record_truncated_value): Likewise.
15481 * expmed.c (extract_bit_field_1): Likewise.
15482 (extract_split_bit_field): Likewise.
15483 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15484 instead of TRULY_NOOP_TRUNCATION.
15485 * function.c (assign_parm_setup_block): Likewise.
15486 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15487 * rtlhooks.c: Include target.h.
15488 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15489 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15490 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15491 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15492 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15493 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15494 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15495 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15496 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15497 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15498 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15499 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15500 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15501 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15502 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15503 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15504 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15505 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15506 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15507 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15508 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15509 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15510 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15511 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15512 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15513 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15514 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15515 rather than TRULY_NOOP_TRUNCATION in comments.
15516 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15517 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15518 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15519 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15520 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15521 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15522 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15523 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15524 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15525 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15526 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15527 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15528 rather than TRULY_NOOP_TRUNCATION in comments.
15529 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15530 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15531 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15532 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15533 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15534 TRULY_NOOP_TRUNCATION condition.
15535 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15536 (TRULY_NOOP_TRUNCATION): Delete.
15537 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15538 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15539 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15540 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15541 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15542 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15543 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15544 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15545 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15546 rather than TRULY_NOOP_TRUNCATION in comments.
15547 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15548 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15549 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15550 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15551 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15552 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15554 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15557 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15558 (*cmp_ior): Likewise.
15559 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15560 (*ior_scc_scc_cmp): Likewise.
15561 (*and_scc_scc): Likewise.
15562 (*and_scc_scc_cmp): Likewise.
15564 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15565 Alan Hayard <alan.hayward@arm.com>
15566 David Sherwood <david.sherwood@arm.com>
15568 * target.def (can_change_mode_class): New hook.
15569 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15570 (hard_regno_nregs): Likewise.
15571 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15572 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15573 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15574 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15575 (LOAD_EXTEND_OP): Update accordingly.
15576 * doc/tm.texi: Regenerate.
15577 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15578 CANNOT_CHANGE_MODE_CLASS.
15579 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15580 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15581 * combine.c (simplify_set): Update accordingly.
15582 * emit-rtl.c (validate_subreg): Likewise.
15583 * recog.c (general_operand): Likewise.
15584 * regcprop.c (mode_change_ok): Likewise.
15585 * reload1.c (choose_reload_regs): Likewise.
15586 (inherit_piecemeal_p): Likewise.
15587 * rtlanal.c (simplify_subreg_regno): Likewise.
15588 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15589 instead of CANNOT_CHANGE_MODE_CLASS.
15590 (reload_cse_simplify_operands): Likewise.
15591 * reload.c (push_reload): Use targetm.can_change_mode_class
15592 instead of CANNOT_CHANGE_MODE_CLASS.
15593 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15594 REG_CANNOT_CHANGE_MODE_P.
15595 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15596 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15597 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15598 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15599 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15600 (arm_can_change_mode_class): New function.
15601 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15602 than CANNOT_CHANGE_MODE_CLASS in comments.
15603 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15604 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15605 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15606 (ix86_can_change_mode_class): ...this new function, inverting the
15607 sense of the return value.
15608 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15609 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15610 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15611 (ia64_can_change_mode_class): New function.
15612 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15613 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15614 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15615 (m32c_can_change_mode_class): ...this new function, inverting the
15616 sense of the return value.
15617 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15618 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15619 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15620 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15621 (mips_can_change_mode_class): ...this new function, inverting the
15622 sense of the return value.
15623 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15624 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15625 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15626 (msp430_can_change_mode_class): New function.
15627 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15628 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15629 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15630 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15631 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15632 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15633 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15634 (pa_cannot_change_mode_class): Replace with...
15635 (pa_can_change_mode_class): ...this new function, inverting the
15636 sense of the return value.
15637 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15638 than CANNOT_CHANGE_MODE_CLASS in comments.
15639 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15640 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15641 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15642 (pdp11_cannot_change_mode_class): Replace with...
15643 (pdp11_can_change_mode_class): ...this new function, inverting the
15644 sense of the return value.
15645 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15646 * config/powerpcspe/powerpcspe-protos.h
15647 (rs6000_cannot_change_mode_class_ptr): Delete.
15648 * config/powerpcspe/powerpcspe.c
15649 (rs6000_cannot_change_mode_class_ptr): Delete.
15650 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15651 (rs6000_option_override_internal): Assign to
15652 targetm.can_change_mode_class instead of
15653 rs6000_cannot_change_mode_class_ptr.
15654 (rs6000_cannot_change_mode_class): Replace with...
15655 (rs6000_can_change_mode_class): ...this new function, inverting the
15656 sense of the return value.
15657 (rs6000_debug_cannot_change_mode_class): Replace with...
15658 (rs6000_debug_can_change_mode_class): ...this new function.
15659 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15660 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15661 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15662 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15663 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15665 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15666 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15667 (rs6000_option_override_internal): Assign to
15668 targetm.can_change_mode_class instead of
15669 rs6000_cannot_change_mode_class_ptr.
15670 (rs6000_cannot_change_mode_class): Replace with...
15671 (rs6000_can_change_mode_class): ...this new function, inverting the
15672 sense of the return value.
15673 (rs6000_debug_cannot_change_mode_class): Replace with...
15674 (rs6000_debug_can_change_mode_class): ...this new function.
15675 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15676 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15677 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15678 (s390_can_change_mode_class): ...this new function, inverting the
15679 sense of the return value.
15680 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15681 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15682 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15683 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15684 (sh_cannot_change_mode_class): Replace with...
15685 (sh_can_change_mode_class): ...this new function, inverting the
15686 sense of the return value.
15687 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15688 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15689 (sparc_can_change_mode_class): New function.
15690 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15691 * config/spu/spu.c (spu_can_change_mode_class): New function.
15692 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15693 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15694 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15695 (visium_can_change_mode_class): New function.
15696 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15698 2017-09-15 Richard Biener <rguenther@suse.de>
15700 PR tree-optimization/82217
15701 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15702 but not undefined case.
15704 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15707 * postreload.c (reload_cse_simplify_operands): Skip
15708 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15710 2017-09-15 Richard Biener <rguenther@suse.de>
15712 PR tree-optimization/68823
15713 * graphite-scop-detection.c (build_alias_set): If we have a
15714 possible dependence check whether we can handle them by just
15715 looking at the DRs DR_ACCESS_FNs.
15716 (build_scops): If build_alias_set fails, fail the SCOP.
15718 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15720 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15721 to support float128 built-in functions that require the ISA 3.0
15723 (BU_FLOAT128_3_HW): Likewise.
15724 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15725 built-in functions.
15726 (FMAF128): Likewise.
15728 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15729 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15730 floating point instructions.
15731 (rs6000_invalid_builtin): Likewise.
15732 (rs6000_builtin_mask_names): Likewise.
15733 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15734 (RS6000_BTM_FLOAT128_HW): Likewise.
15735 (RS6000_BTM_COMMON): Likewise.
15736 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15738 * doc/extend.texi (RS/6000 built-in functions): Document the
15739 IEEE 128-bit floating point square root and fused multiply-add
15740 built-in functions.
15742 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15744 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15745 reg (r2) isn't in the set of registers defined in the prologue.
15747 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15748 Alan Hayward <alan.hayward@arm.com>
15749 David Sherwood <david.sherwood@arm.com>
15751 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15752 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15753 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15754 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15755 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15757 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15758 max_vectorization_factor.
15759 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15761 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15762 Alan Hayward <alan.hayward@arm.com>
15763 David Sherwood <david.sherwood@arm.com>
15765 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15766 (vect_worthwhile_without_simd_p): Declare.
15767 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15768 (vectorizable_reduction): Use it.
15769 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15770 (vectorizable_operation): Likewise.
15772 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15773 Alan Hayward <alan.hayward@arm.com>
15774 David Sherwood <david.sherwood@arm.com>
15776 * tree-vectorizer.h (vect_get_num_copies): New function.
15777 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15778 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15779 (vectorizable_induction): Likewise.
15780 (vectorizable_live_operation): Likewise.
15781 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15782 (vectorizable_bswap): Likewise.
15783 (vectorizable_call): Likewise.
15784 (vectorizable_conversion): Likewise.
15785 (vectorizable_assignment): Likewise.
15786 (vectorizable_shift): Likewise.
15787 (vectorizable_operation): Likewise.
15788 (vectorizable_store): Likewise.
15789 (vectorizable_load): Likewise.
15790 (vectorizable_condition): Likewise.
15791 (vectorizable_comparison): Likewise.
15792 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15794 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15795 Alan Hayward <alan.hayward@arm.com>
15796 David Sherwood <david.sherwood@arm.com>
15798 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15799 of vect_init_vector.
15801 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15802 Alan Hayward <alan.hayward@arm.com>
15803 David Sherwood <david.sherwood@arm.com>
15805 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15806 an inline wrapper that provides a location.
15807 (gimple_build_vector): Likewise.
15808 * gimple-fold.c (gimple_build_vector_from_val): New function.
15809 (gimple_build_vector): Likewise.
15810 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15811 functions to build the initial value. Always return a gimple value.
15812 (get_initial_defs_for_reduction): Likewise. Only compute
15814 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15815 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15816 (vectorizable_induction): Use gimple_build_vector rather than
15819 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15820 Alan Hayward <alan.hayward@arm.com>
15821 David Sherwood <david.sherwood@arm.com>
15823 * target.h (vec_perm_indices): New typedef.
15824 (auto_vec_perm_indices): Likewise.
15825 * optabs-query.h: Include target.h
15826 (can_vec_perm_p): Take a vec_perm_indices *.
15827 * optabs-query.c (can_vec_perm_p): Likewise.
15828 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15829 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15830 * tree-vect-generic.c (lower_vec_perm): Likewise.
15831 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15832 (vect_grouped_load_supported): Likewise.
15833 (vect_shift_permute_load_chain): Likewise.
15834 (vect_permute_store_chain): Use auto_vec_perm_indices.
15835 (vect_permute_load_chain): Likewise.
15836 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15837 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15838 Update uses of can_vec_perm_p.
15839 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15840 mode with a number of elements. Take a vec_perm_indices *.
15841 (vect_create_epilog_for_reduction): Update accordingly.
15842 Use auto_vec_perm_indices.
15843 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15844 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15845 (vect_transform_slp_perm_load): Likewise.
15846 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15847 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15848 (vect_gen_perm_mask_checked): Likewise.
15849 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15850 (vect_gen_perm_mask_checked): Likewise.
15851 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15852 (vectorizable_store): Likewise.
15853 (vectorizable_load): Likewise.
15854 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15855 (vectorizable_bswap): Likewise.
15857 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15858 Alan Hayward <alan.hayward@arm.com>
15859 David Sherwood <david.sherwood@arm.com>
15861 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15862 * tree.c (build_vector): Likewise.
15863 (build_vector_from_ctor): Update accordingly.
15864 (build_vector_from_val): Likewise.
15865 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15866 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15867 * tree-vect-generic.c (add_rshift): Likewise.
15868 (expand_vector_divmod): Likewise.
15869 (optimize_vector_constructor): Likewise.
15870 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15871 (vect_transform_slp_perm_load): Likewise.
15872 (vect_schedule_slp_instance): Likewise.
15873 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15874 (vectorizable_call): Likewise.
15875 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15876 * expmed.c (make_tree): Likewise.
15877 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15878 a vector passed to build_vector.
15879 (fold_convert_const): Likewise.
15880 (exact_inverse): Likewise.
15881 (fold_ternary_loc): Likewise.
15882 (fold_relational_const): Likewise.
15883 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15884 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15885 (const_unop): Likewise. Store the reduction accumulator in a
15886 variable rather than an array.
15887 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15888 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15889 the new vector, rather than constructing it after the input arrays.
15890 (native_interpret_vector): Use auto_vec<tree> when building
15891 a vector passed to build_vector. Add elements in order.
15892 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15893 auto_vec<tree> when building a vector passed to build_vector.
15894 (vect_create_epilog_for_reduction): Likewise.
15895 (vectorizable_induction): Likewise.
15896 (get_initial_def_for_reduction): Likewise. Fix indentation of
15898 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15900 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15901 passed to build_vector.
15903 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15904 Alan Hayward <alan.hayward@arm.com>
15905 David Sherwood <david.sherwood@arm.com>
15907 * tree-core.h (tree_base::u): Add an "nelts" field.
15908 (tree_vector): Use VECTOR_CST_NELTS as the length.
15909 * tree.c (tree_size): Likewise.
15910 (make_vector): Initialize VECTOR_CST_NELTS.
15911 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15912 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15913 TYPE_VECTOR_SUBPARTS.
15914 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15915 as the number of units, setting it from VECTOR_CST_NELTS.
15916 (const_vector_from_tree): Likewise.
15917 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15918 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15919 (fold_negate_expr_1): Likewise.
15920 (fold_convert_const): Likewise.
15921 (const_binop): Likewise. Differentiate the number of output and
15923 (const_unop): Likewise.
15924 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15925 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15926 in cases that did the opposite.
15928 2017-09-14 Richard Biener <rguenther@suse.de>
15930 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15933 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15935 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15937 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15940 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15941 if and where to split a bb, except for splitting before debug insn
15942 sequences followed by non-label real insn. Delete debug insns
15943 in between basic blocks.
15945 * combine.c (make_compound_operation_int): Formatting fixes.
15947 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15948 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15949 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15950 * config/sol2.h (LINK_EH_SPEC): Likewise.
15951 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15952 * config/s390/linux.h (LINK_SPEC): Likewise.
15953 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15954 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15955 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15956 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15957 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15958 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15959 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15960 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15962 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15964 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15966 (ENDFILE_LINUX_SPEC): Likewise.
15967 (LINK_EH_SPEC): Likewise.
15968 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15969 (LINK_OS_LINUX_SPEC32): Likewise.
15970 (LINK_OS_LINUX_SPEC64): Likewise.
15971 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15972 (LINK_OS_LINUX_SPEC): Likewise.
15974 2017-09-13 Martin Liska <mliska@suse.cz>
15976 PR middle-end/82154
15977 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15978 CASE_HIGH is NULL_TREE.
15980 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15981 Alan Hayward <alan.hayward@arm.com>
15982 David Sherwood <david.sherwood@arm.com>
15984 * target.def (secondary_memory_needed): New hook.
15985 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15986 instead of SECONDARY_MEMORY_NEEDED.
15987 (secondary_memory_needed_mode): Likewise.
15988 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15989 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15990 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15991 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15992 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15993 * doc/tm.texi: Regenerate.
15994 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15995 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15996 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15997 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15998 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15999 * config/i386/i386.c (inline_secondary_memory_needed): Put the
16000 mode argument first and change the reg_class arguments to reg_class_t.
16001 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
16002 Make static. Update the call to inline_secondary_memory_needed.
16003 (ix86_register_move_cost): Update the call to
16004 inline_secondary_memory_needed.
16005 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16006 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
16008 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
16009 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16011 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
16012 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
16013 * config/mips/mips.c (mips_secondary_memory_needed): Make static
16014 and match hook interface. Add comment from mips.h.
16015 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16016 * config/mmix/mmix.md (truncdfsf2): Refer to
16017 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16019 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
16020 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
16021 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16022 (pa_secondary_memory_needed): New function.
16023 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
16024 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
16025 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16026 (pdp11_secondary_memory_needed): Make static and match hook interface.
16027 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
16028 * config/powerpcspe/powerpcspe-protos.h
16029 (rs6000_secondary_memory_needed_ptr): Delete.
16030 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
16032 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16033 (rs6000_option_override_internal): Assign to
16034 targetm.secondary_memory_needed rather than
16035 rs6000_secondary_memory_needed_ptr.
16036 (rs6000_secondary_memory_needed): Match hook interface.
16037 (rs6000_debug_secondary_memory_needed): Likewise.
16038 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
16039 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
16040 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
16041 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16042 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
16043 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
16045 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
16046 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16047 (rs6000_option_override_internal): Assign to
16048 targetm.secondary_memory_needed rather than
16049 rs6000_secondary_memory_needed_ptr.
16050 (rs6000_secondary_memory_needed): Match hook interface.
16051 (rs6000_debug_secondary_memory_needed): Likewise.
16052 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
16053 * config/s390/s390.c (s390_secondary_memory_needed): New function.
16054 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16055 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
16056 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16057 (sparc_secondary_memory_needed): New function.
16058 * lra-constraints.c (check_and_process_move): Refer to
16059 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16061 (curr_insn_transform): Likewise.
16062 (process_alt_operands): Use targetm.secondary_memory_needed
16063 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16064 (check_secondary_memory_needed_p): Likewise.
16065 (choose_split_class): Likewise.
16066 * reload.c: Unconditionally include code that was previously
16067 conditional on SECONDARY_MEMORY_NEEDED.
16068 (push_secondary_reload): Use targetm.secondary_memory_needed
16069 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16070 (push_reload): Likewise.
16071 * reload1.c: Unconditionally include code that was previously
16072 conditional on SECONDARY_MEMORY_NEEDED.
16073 (choose_reload_regs): Use targetm.secondary_memory_needed
16074 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16075 (gen_reload): Likewise.
16076 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
16078 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16079 Alan Hayward <alan.hayward@arm.com>
16080 David Sherwood <david.sherwood@arm.com>
16082 * target.def (secondary_memory_needed_mode): New hook:
16083 * targhooks.c (default_secondary_memory_needed_mode): Declare.
16084 * targhooks.h (default_secondary_memory_needed_mode): New function.
16085 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
16086 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
16087 * doc/tm.texi: Regenerate.
16088 * lra-constraints.c (check_and_process_move): Use
16089 targetm.secondary_memory_needed_mode instead of
16090 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
16091 (curr_insn_transform): Likewise.
16092 * reload.c (get_secondary_mem): Likewise.
16093 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16094 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
16096 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16097 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16098 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
16099 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16100 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
16102 * config/powerpcspe/powerpcspe-protos.h
16103 (rs6000_secondary_memory_needed_mode): Delete.
16104 * config/powerpcspe/powerpcspe.c
16105 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16106 (rs6000_secondary_memory_needed_mode): Make static.
16107 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16108 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
16110 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16112 (rs6000_secondary_memory_needed_mode): Make static.
16113 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16114 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
16115 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16116 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16117 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16119 (sparc_secondary_memory_needed_mode): New function.
16120 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
16122 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
16124 * config/aarch64/constraints.md (Umq): New constraint.
16125 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
16127 (mov<mode>): Update condition.
16129 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16131 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
16132 when bitposition is the same.
16134 2017-09-13 Richard Biener <rguenther@suse.de>
16136 * dwarf2out.c (output_die_symbol): Remove.
16137 (output_die): Do not output a DIEs symbol.
16139 2017-09-13 Richard Biener <rguenther@suse.de>
16141 PR middle-end/82128
16142 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
16143 default-def to avoid breaking iterator update with the weird
16144 interaction with cgraph_update_edges_for_call_stmt_node.
16146 2017-09-13 Richard Biener <rguenther@suse.de>
16148 * tree-cfg.c (verify_gimple_assign_binary): Add verification
16149 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
16150 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
16151 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
16153 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
16155 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
16156 Disable pc relative literal load irrespective of
16157 TARGET_FIX_ERR_A53_84341 for default.
16159 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16161 * config/sparc/sparc.c (output_return): Output the source location of
16162 the insn in the delay slot, if any.
16163 (output_sibcall): Likewise.
16165 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16168 * common.opt (-static-pie): New alias.
16169 (shared): Negate static-pie.
16170 (-no-pie): Update help text.
16172 (static-pie): New option.
16173 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16174 -static-pie support.
16175 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16176 (LINK_EH_SPEC): Likewise.
16177 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16178 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16179 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16180 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16181 (init_gcc_specs): Likewise.
16182 (init_spec): Likewise.
16183 (display_help): Update help message for -pie.
16184 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16187 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16189 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16190 (movdi_aarch64): Likewise.
16191 (movti_aarch64): Likewise.
16193 2017-09-12 Simon Wright <simon@pushface.org>
16196 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16197 calculation of the minor version, always output as 0.
16199 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16202 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16203 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16204 on it early, rather than manual conversion late. For
16205 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16206 instead of performing manual conversion.
16208 2017-09-12 Carl Love <cel@us.ibm.com>
16210 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16211 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16213 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16214 VMULOSW): Add definitions.
16215 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16216 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16217 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16218 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16219 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16221 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16223 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16225 (movti_aarch64): Likewise.
16226 (movdf_aarch64): Likewise.
16227 (movtf_aarch64): Likewise.
16228 (load_pairdi): Likewise.
16229 (store_pairdi): Likewise.
16230 (load_pairdf): Likewise.
16231 (store_pairdf): Likewise.
16232 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16233 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16234 (ldr_got_small_<mode>): Likewise.
16235 (ldr_got_small_28k_<mode>): Likewise.
16236 (ldr_got_tiny): Likewise.
16237 * config/aarch64/iterators.md (ldst_sz): New.
16238 (ldpstp_sz): Likewise.
16239 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16241 (thunderx_load): Split load_8 to load_16.
16242 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16244 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16245 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16246 (xgene1_store_pair): Split store_8 to store_16.
16247 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16248 (falkor_st_0_st_sd): Split store_8 to store_16.
16250 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16252 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16253 and store1/2/3/4 to store_4/8/12/16.
16254 * config/aarch64/aarch64.md: Update for rename.
16255 * config/arm/arm.md: Likewise.: Likewise.
16256 * config/arm/arm.c: Likewise.
16257 * config/arm/thumb1.md: Likewise.
16258 * config/arm/thumb2.md: Likewise.
16259 * config/arm/vfp.md: Likewise.
16260 * config/arm/arm-generic.md: Likewise.
16261 * config/arm/arm1020e.md: Likewise.
16262 * config/arm/arm1026ejs.md: Likewise.
16263 * config/arm/arm1136jfs.md: Likewise.
16264 * config/arm/arm926ejs.md: Likewise.
16265 * config/arm/cortex-a15.md: Likewise.
16266 * config/arm/cortex-a17.md: Likewise.
16267 * config/arm/cortex-a5.md: Likewise.
16268 * config/arm/cortex-a53.md: Likewise.
16269 * config/arm/cortex-a57.md: Likewise.
16270 * config/arm/cortex-a7.md: Likewise.
16271 * config/arm/cortex-a8.md: Likewise.
16272 * config/arm/cortex-a9.md: Likewise.
16273 * config/arm/cortex-m4.md: Likewise.
16274 * config/arm/cortex-m7.md: Likewise.
16275 * config/arm/cortex-r4.md: Likewise.
16276 * config/arm/exynos-m1.md: Likewise.
16277 * config/arm/fa526.md: Likewise.
16278 * config/arm/fa606te.md: Likewise.
16279 * config/arm/fa626te.md: Likewise.
16280 * config/arm/fa726te.md: Likewise.
16281 * config/arm/fmp626.md: Likewise.
16282 * config/arm/iwmmxt.md: Likewise.
16283 * config/arm/ldmstm.md: Likewise.
16284 * config/arm/marvell-pj4.md: Likewise.
16285 * config/arm/xgene1.md: Likewise.
16286 * config/aarch64/thunderx.md: Likewise.
16287 * config/aarch64/thunderx2t99.md: Likewise.
16288 * config/aarch64/falkor.md: Likewise.
16290 2017-09-12 Martin Liska <mliska@suse.cz>
16292 * attribs.c (private_lookup_attribute): New function.
16293 * attribs.h (private_lookup_attribute): Declared here.
16294 (lookup_attribute): Called from this place.
16296 2017-09-12 Richard Biener <rguenther@suse.de>
16298 PR tree-optimization/82157
16299 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16300 stmts with side-effects.
16302 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16303 Alan Hayward <alan.hayward@arm.com>
16304 David Sherwood <david.sherwood@arm.com>
16306 * target.def (hard_regno_nregs): New hook.
16307 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16308 * targhooks.h (default_hard_regno_nregs): Declare.
16309 * targhooks.c (default_hard_regno_nregs): New function.
16310 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16311 (TARGET_HARD_REGNO_NREGS): ...this hook.
16312 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16313 (CLASS_MAX_NREGS): Likewise.
16314 * doc/tm.texi: Regenerate.
16315 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16316 instead of HARD_REGNO_NREGS.
16317 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16318 HARD_REGNO_NREGS in the comment.
16319 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16320 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16321 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16322 Return an unsigned int.
16323 (TARGET_HARD_REGNO_NREGS): Redefine.
16324 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16325 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16326 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16327 (arc_hard_regno_nregs): New function.
16328 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16329 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16330 (arm_hard_regno_nregs): New function.
16331 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16332 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16333 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16334 (TARGET_HARD_REGNO_NREGS): Redefine.
16335 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16336 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16337 (HARD_REGNO_NREGS): Delete.
16338 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16339 (cr16_hard_regno_nregs): New function.
16340 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16341 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16342 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16343 (cris_hard_regno_nregs): New function.
16344 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16345 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16346 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16347 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16348 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16349 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16350 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16351 (frv_hard_regno_nregs): Make static. Take and return an
16353 (frv_class_max_nregs): Remove outdated copy of documentation.
16354 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16355 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16356 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16357 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16358 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16359 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16360 (TARGET_HARD_REGNO_NREGS): Redefine.
16361 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16362 (CLASS_MAX_NREGS): Update comment.
16363 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16364 (ia64_hard_regno_nregs): New function.
16365 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16366 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16367 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16368 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16369 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16371 (m32c_hard_regno_nregs): Likewise. Make static.
16372 (TARGET_HARD_REGNO_NREGS): Redefine.
16373 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16374 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16375 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16376 (m68k_hard_regno_nregs): New function.
16377 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16378 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16379 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16380 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16381 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16382 Take and return an unsigned int.
16383 (TARGET_HARD_REGNO_NREGS): Redefine.
16384 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16385 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16386 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16387 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16388 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16389 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16390 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16391 (msp430_hard_regno_nregs): Make static. Take and return an
16393 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16394 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16395 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16396 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16397 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16398 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16399 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16400 (TARGET_HARD_REGNO_NREGS): Redefine.
16401 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16402 (PA_HARD_REGNO_NREGS): ...this.
16403 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16404 (PA_HARD_REGNO_NREGS): ...this.
16405 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16406 (pa_hard_regno_nregs): New function.
16407 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16408 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16409 (pdp11_hard_regno_nregs): New function.
16410 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16411 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16412 (rs6000_hard_regno_nregs_hook): New function.
16413 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16414 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16415 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16416 Take and return an unsigned int. Move earlier in file.
16417 (TARGET_HARD_REGNO_NREGS): Redefine.
16418 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16419 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16420 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16421 (rl78_hard_regno_nregs): Make static. Take and return an
16423 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16424 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16425 (rs6000_hard_regno_nregs_hook): New function.
16426 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16427 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16428 (TARGET_HARD_REGNO_NREGS): Redefine.
16429 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16430 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16431 instead of HARD_REGNO_NREGS.
16432 (s390_hard_regno_nregs): New function.
16433 (s390_hard_regno_mode_ok): Add comment from s390.h.
16434 (TARGET_HARD_REGNO_NREGS): Redefine.
16435 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16436 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16437 (sh_hard_regno_nregs): New function.
16438 (sh_pass_in_reg_p): Use it.
16439 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16440 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16441 (sparc_hard_regno_nregs): New function.
16442 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16443 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16444 (spu_function_arg_advance): Use it, supplying a valid register number.
16445 (TARGET_HARD_REGNO_NREGS): Redefine.
16446 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16447 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16448 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16449 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16450 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16451 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16452 (CLASS_MAX_NREGS): Remove copy of old documentation.
16453 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16454 (visium_hard_regno_nregs): New function.
16455 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16456 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16457 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16458 xtensa_hard_regno_nregs): New function.
16459 * system.h (HARD_REGNO_NREGS): Poison.
16461 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16463 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16464 hard_regno_nregs instead of HARD_REGNO_NREGS.
16465 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16466 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16467 (c6x_expand_epilogue): Likewise.
16468 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16469 (frv_read_iacc_argument): Likewise.
16470 * config/sh/sh.c: Include regs.h.
16471 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16472 (regs_used): Likewise.
16473 (output_stack_adjust): Likewise.
16474 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16475 * expmed.c: Include regs.h.
16476 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16477 * ree.c: Include regs.h.
16478 (combine_reaching_defs): Use hard_regno_nregs instead of
16480 (add_removable_extension): Likewise.
16482 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16484 * regs.h (hard_regno_nregs): Turn into a function.
16485 (end_hard_regno): Update accordingly.
16486 * caller-save.c (setup_save_areas): Likewise.
16487 (save_call_clobbered_regs): Likewise.
16488 (replace_reg_with_saved_mem): Likewise.
16489 (insert_restore): Likewise.
16490 (insert_save): Likewise.
16491 * combine.c (can_change_dest_mode): Likewise.
16492 (move_deaths): Likewise.
16493 (distribute_notes): Likewise.
16494 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16495 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16496 (rs6000_split_multireg_move): Likewise.
16497 (rs6000_register_move_cost): Likewise.
16498 (rs6000_memory_move_cost): Likewise.
16499 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16500 (rs6000_split_multireg_move): Likewise.
16501 (rs6000_register_move_cost): Likewise.
16502 (rs6000_memory_move_cost): Likewise.
16503 * cselib.c (cselib_reset_table): Likewise.
16504 (cselib_lookup_1): Likewise.
16505 * emit-rtl.c (set_mode_and_regno): Likewise.
16506 * function.c (aggregate_value_p): Likewise.
16507 * ira-color.c (setup_profitable_hard_regs): Likewise.
16508 (check_hard_reg_p): Likewise.
16509 (calculate_saved_nregs): Likewise.
16510 (assign_hard_reg): Likewise.
16511 (improve_allocation): Likewise.
16512 (calculate_spill_cost): Likewise.
16513 * ira-emit.c (modify_move_list): Likewise.
16514 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16515 (ira_hard_reg_in_set_p): Likewise.
16516 * ira.c (setup_reg_mode_hard_regset): Likewise.
16517 (clarify_prohibited_class_mode_regs): Likewise.
16518 (check_allocation): Likewise.
16519 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16520 (lra_setup_reg_renumber): Likewise.
16521 (setup_try_hard_regno_pseudos): Likewise.
16522 (spill_for): Likewise.
16523 (assign_hard_regno): Likewise.
16524 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16525 * lra-constraints.c (in_class_p): Likewise.
16526 (lra_constraint_offset): Likewise.
16527 (simplify_operand_subreg): Likewise.
16528 (lra_constraints): Likewise.
16529 (split_reg): Likewise.
16530 (split_if_necessary): Likewise.
16531 (invariant_p): Likewise.
16532 (inherit_in_ebb): Likewise.
16533 * lra-lives.c (process_bb_lives): Likewise.
16534 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16535 (get_hard_regs): Likewise.
16536 (do_remat): Likewise.
16537 * lra-spills.c (assign_spill_hard_regs): Likewise.
16538 * mode-switching.c (create_pre_exit): Likewise.
16539 * postreload.c (reload_combine_recognize_pattern): Likewise.
16540 * recog.c (peep2_find_free_register): Likewise.
16541 * regcprop.c (kill_value_regno): Likewise.
16542 (set_value_regno): Likewise.
16543 (copy_value): Likewise.
16544 (maybe_mode_change): Likewise.
16545 (find_oldest_value_reg): Likewise.
16546 (copyprop_hardreg_forward_1): Likewise.
16547 * regrename.c (check_new_reg_p): Likewise.
16548 (regrename_do_replace): Likewise.
16549 * reload.c (push_reload): Likewise.
16550 (combine_reloads): Likewise.
16551 (find_dummy_reload): Likewise.
16552 (operands_match_p): Likewise.
16553 (find_reloads): Likewise.
16554 (find_equiv_reg): Likewise.
16555 (reload_adjust_reg_for_mode): Likewise.
16556 * reload1.c (count_pseudo): Likewise.
16557 (count_spilled_pseudo): Likewise.
16558 (find_reg): Likewise.
16559 (clear_reload_reg_in_use): Likewise.
16560 (free_for_value_p): Likewise.
16561 (allocate_reload_reg): Likewise.
16562 (choose_reload_regs): Likewise.
16563 (reload_adjust_reg_for_temp): Likewise.
16564 (emit_reload_insns): Likewise.
16565 (delete_output_reload): Likewise.
16566 * rtlanal.c (subreg_get_info): Likewise.
16567 * sched-deps.c (sched_analyze_reg): Likewise.
16568 * sel-sched.c (init_regs_for_mode): Likewise.
16569 (mark_unavailable_hard_regs): Likewise.
16570 (choose_best_reg_1): Likewise.
16571 (verify_target_availability): Likewise.
16572 * valtrack.c (dead_debug_insert_temp): Likewise.
16573 * var-tracking.c (track_loc_p): Likewise.
16574 (emit_note_insn_var_location): Likewise.
16575 * varasm.c (make_decl_rtl): Likewise.
16576 * reginfo.c (choose_hard_reg_mode): Likewise.
16577 (init_reg_modes_target): Refer directly to
16578 this_target_regs->x_hard_regno_nregs.
16580 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16582 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16583 instead of hard_regno_nregs.
16585 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16587 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16588 end_hard_regno instead of hard_regno_nregs.
16589 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16590 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16591 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16592 * ira-color.c (improve_allocation): Likewise.
16593 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16594 * lra-lives.c (mark_regno_live): Likewise.
16595 (mark_regno_dead): Likewise.
16596 * lra-remat.c (operand_to_remat): Likewise.
16597 * lra.c (collect_non_operand_hard_regs): Likewise.
16598 * postreload.c (reload_combine_note_store): Likewise.
16599 (move2add_valid_value_p): Likewise.
16600 * reload.c (regno_clobbered_p): Likewise.
16602 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16604 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16606 * config/v850/v850.c (v850_reorg): Likewise.
16607 * reload.c (refers_to_regno_for_reload_p): Likewise.
16608 (find_equiv_reg): Likewise.
16609 * reload1.c (reload_reg_reaches_end_p): Likewise.
16611 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16613 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16615 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16616 * config/arm/arm.c (output_move_neon): Likewise.
16617 (arm_attr_length_move_neon): Likewise.
16618 (neon_split_vcombine): Likewise.
16619 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16620 (c6x_mark_reg_written): Likewise.
16621 (c6x_dwarf_register_span): Likewise.
16622 * config/i386/i386.c (ix86_save_reg): Likewise.
16623 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16624 (rws_access_reg): Likewise.
16625 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16626 * mode-switching.c (create_pre_exit): Likewise.
16627 * ree.c (combine_reaching_defs): Likewise.
16628 (add_removable_extension): Likewise.
16629 * regcprop.c (find_oldest_value_reg): Likewise.
16630 (copyprop_hardreg_forward_1): Likewise.
16631 * reload.c (reload_inner_reg_of_subreg): Likewise.
16632 (push_reload): Likewise.
16633 (combine_reloads): Likewise.
16634 (find_dummy_reload): Likewise.
16635 (reload_adjust_reg_for_mode): Likewise.
16636 * reload1.c (find_reload_regs): Likewise.
16637 (forget_old_reloads_1): Likewise.
16638 (reload_reg_free_for_value_p): Likewise.
16639 (reload_adjust_reg_for_temp): Likewise.
16640 (emit_reload_insns): Likewise.
16641 (delete_output_reload): Likewise.
16642 * sel-sched.c (choose_best_reg_1): Likewise.
16643 (choose_best_pseudo_reg): Likewise.
16645 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16646 Alan Hayward <alan.hayward@arm.com>
16647 David Sherwood <david.sherwood@arm.com>
16649 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16650 * target.def (slow_unaligned_access): New hook.
16651 * targhooks.h (default_slow_unaligned_access): Declare.
16652 * targhooks.c (default_slow_unaligned_access): New function.
16653 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16654 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16655 * doc/tm.texi: Regenerate.
16656 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16657 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16658 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16660 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16661 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16663 (rs6000_slow_unaligned_access): New function.
16664 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16665 (expand_block_compare): Likewise.
16666 (expand_strn_compare): Likewise.
16667 (rs6000_rtx_costs): Likewise.
16668 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16669 (riscv_slow_unaligned_access): Likewise.
16670 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16671 (riscv_slow_unaligned_access_p): ...this and make static.
16672 (riscv_option_override): Update accordingly.
16673 (riscv_slow_unaligned_access): New function.
16674 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16675 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16676 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16677 (rs6000_slow_unaligned_access): New function.
16678 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16679 (rs6000_rtx_costs): Likewise.
16680 * config/rs6000/rs6000-string.c (expand_block_compare)
16681 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16682 of SLOW_UNALIGNED_ACCESS.
16683 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16684 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16685 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16686 of SLOW_UNALIGNED_ACCESS.
16687 * expmed.c (simple_mem_bitfield_p): Likewise.
16688 * expr.c (alignment_for_piecewise_move): Likewise.
16689 (emit_group_load_1): Likewise.
16690 (emit_group_store): Likewise.
16691 (copy_blkmode_from_reg): Likewise.
16692 (emit_push_insn): Likewise.
16693 (expand_assignment): Likewise.
16694 (store_field): Likewise.
16695 (expand_expr_real_1): Likewise.
16696 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16697 * lra-constraints.c (simplify_operand_subreg): Likewise.
16698 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16699 * gimple-ssa-store-merging.c: Likewise in block comment at start
16701 * tree-ssa-strlen.c: Include target.h.
16702 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16703 of SLOW_UNALIGNED_ACCESS.
16704 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16706 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16708 PR rtl-optimization/82185
16709 * expmed.c (emit_store_flag_int): Only test tem if it has been
16712 2017-09-12 Richard Biener <rguenther@suse.de>
16714 PR middle-end/82149
16715 * match.pd ((FTYPE) N CMP CST): Fix typo.
16717 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16719 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16721 (mips_near_type_p): Add 'short_call' attribute as a synonym
16723 * doc/extend.texi (short_call): Document new function attribute.
16725 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16728 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16729 assertion check that in the condition.
16730 (get_atomic_generic_size): Likewise. Before testing if parameter
16731 has pointer type, if it has array type, call for C++
16732 default_conversion to perform array-to-pointer conversion.
16734 2017-09-12 Richard Biener <rguenther@suse.de>
16736 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16737 for operations we cannot scalarize.
16739 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16741 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16742 vectors heap vectors. Clean up comments.
16743 Make visited_bbs a reference.
16744 (profitable_jump_thread_path): Make GC
16745 vectors heap vectors. Clean up comments.
16747 (convert_and_register_jump_thread_path): Make GC vectors heap
16749 (check_subpath_and_update_thread_path): Same. Clean up comments.
16750 Make visited_bbs a reference.
16751 (handle_phi): Abstract common code to to
16752 register_jump_thread_path_if_profitable.
16753 Rename VAR_BB to DEF_BB.
16755 Make GC vectors heap vectors.
16756 Make visited_bbs a reference.
16757 (handle_assignment): Same.
16758 (register_jump_thread_path_if_profitable): New.
16759 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16761 Make GC vectors heap vectors. Clean up comments.
16762 Make visited_bbs a reference.
16763 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16764 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16767 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16770 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16771 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16773 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16775 Revert r251800 and r251799.
16777 2017-09-11 Martin Jambor <mjambor@suse.cz>
16780 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16781 arguments in advance.
16782 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16783 use it to find predecessor edges.
16784 (naive_outof_ssa): Collect vector of predecessors.
16786 2017-09-08 Jason Merrill <jason@redhat.com>
16788 PR c++/70029 - ICE with ref-qualifier and -flto
16789 * langhooks.h (struct lang_hooks_for_types): Add
16790 copy_lang_qualifiers.
16791 * attribs.c (build_type_attribute_qual_variant): Use it.
16792 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16794 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16795 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16797 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16800 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16801 (*mulsi3_sp64): New instruction.
16802 (mulsi3): New expander.
16804 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16806 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16808 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16810 * sancov.c: Include memmodel.h.
16812 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16815 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16818 2017-09-07 Carl Love <cel@us.ibm.com>
16820 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16821 the sldi instruction.
16823 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16825 * sancov.c: Include tm_p.h.
16827 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16830 * output.h (switch_to_other_text_partition): New declaration.
16831 * varasm.c (switch_to_other_text_partition): New function.
16832 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16833 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16834 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16835 to the other text partition before emitting LCL label and switch back
16836 after emitting the word after it.
16838 2017-09-07 Richard Biener <rguenther@suse.de>
16840 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16841 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16842 critical edges here, after loop init.
16843 (pass_data_pre): Remove PROP_no_crit_edges flags.
16844 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16845 for valueization of call args to avoid leaking VN_TOP.
16846 (visit_use): Assert we do not visit default defs.
16847 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16848 Use error_mark_node to more easily detect leaking VN_TOP.
16849 All default-defs are varying, not VN_TOP. Mark them visited.
16850 (run_scc_vn): Make code match comment.
16852 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16854 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16855 OPTION_MASK_FLOAT128_KEYWORD.
16856 (POWERPC_MASKS): Likewise.
16857 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16858 support for the -mfloat128-type option, and make -mfloat128
16859 default on PowerPC Linux systems. Define or undefine
16860 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16861 Define __float128 to be __ieee128 if IEEE 128-bit support is
16862 enabled, or undefine it.
16863 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16864 Delete defining __FLOAT128_TYPE__.
16865 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16866 -mfloat128-type option and make -mfloat128 default on PowerPC
16868 (TARGET_FLOAT128_TYPE): Likewise.
16869 (-mfloat128-type): Likewise.
16870 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16871 Delete the -mfloat128-type option and make -mfloat128 default on
16872 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16873 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16874 128-bit floating point is enabled. Change tests from using
16875 -mfloat128-type to -mfloat128.
16876 (rs6000_mangle_type): Use the correct mangling for the __float128
16877 type even if normal long double is restricted to 64-bits.
16878 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16880 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16881 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16882 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16884 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16885 documentation for -mfloat128.
16887 2017-09-06 Olivier Hainque <hainque@adacore.com>
16889 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16891 2017-09-06 Wish Wu <wishwu007@gmail.com>
16892 Jakub Jelinek <jakub@redhat.com>
16894 * asan.c (initialize_sanitizer_builtins): Add
16895 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16896 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16897 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16898 BT_FN_VOID_UINT64_PTR variables.
16899 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16900 (BT_FN_VOID_UINT16_UINT16): Likewise.
16901 (BT_FN_VOID_UINT32_UINT32): Likewise.
16902 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16903 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16904 (BT_FN_VOID_UINT64_PTR): Likewise.
16905 * common.opt (flag_sanitize_coverage): New variable.
16906 (fsanitize-coverage=trace-pc): Remove.
16907 (fsanitize-coverage=): Add.
16908 * flag-types.h (enum sanitize_coverage_code): New enum.
16909 * fold-const.c (fold_range_test): Disable non-short-circuit
16910 optimization if flag_sanitize_coverage.
16911 (fold_truth_andor): Likewise.
16912 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16913 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16914 (coverage_sanitizer_opts): New array.
16915 (get_closest_sanitizer_option): Add OPTS argument, handle also
16916 OPT_fsanitize_coverage_.
16917 (parse_sanitizer_options): Adjusted to also handle
16918 OPT_fsanitize_coverage_.
16919 (common_handle_option): Add OPT_fsanitize_coverage_.
16920 * sancov.c (instrument_comparison, instrument_switch): New function.
16921 (sancov_pass): Add trace-cmp support.
16922 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16923 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16924 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16925 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16926 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16927 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16928 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16929 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16930 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16931 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16933 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16935 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16936 error. Only quit immediately if parsing is complete.
16937 (BEGIN): Initialize fatal_err and parse_done.
16938 (begin fpu, end fpu): Check number of arguments.
16939 (begin arch, end arch): Likewise.
16940 (begin cpu, end cpu): Likewise.
16941 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16942 (optalias): Likewise.
16944 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16946 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16947 * config/arm/arm-isa.h: Delete. Move definitions to ...
16948 * arm-cpus.in: ... here. Use new feature and fgroup values.
16949 * config/arm/arm.c (arm_option_override): Use lower case for feature
16951 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16952 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16953 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16955 (print_isa_bits_for): New function.
16956 (gen_isa): New function.
16957 (gen_comm_data): Use print_isa_bits_for.
16958 (define feature): New keyword.
16959 (define fgroup): New keyword.
16960 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16961 (arm-isa.h): Add rule to generate file.
16962 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16963 case for feature bit names.
16965 2017-09-06 Richard Biener <rguenther@suse.de>
16967 * tree-ssa-pre.c (NECESSARY): Remove.
16968 (create_expression_by_pieces): Do not touch pass-local flags.
16969 (insert_into_preds_of_block): Likewise.
16970 (do_pre_regular_insertion): Likewise.
16971 (eliminate_insert): Likewise.
16972 (eliminate_dom_walker::before_dom_children): Likewise.
16973 (fini_eliminate): Do not look at inserted_exprs.
16974 (mark_operand_necessary): Remove.
16975 (remove_dead_inserted_code): Replace with simple work-list
16976 algorithm based on inserted_exprs and SSA uses.
16977 (pass_pre::execute): Re-order fini_eliminate and
16978 remove_dead_inserted_code.
16980 2017-09-06 Olivier Hainque <hainque@adacore.com>
16982 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16983 for VxWorks 7. Adjust surrounding comments.
16985 2017-09-06 Richard Biener <rguenther@suse.de>
16987 * gimple-ssa-strength-reduction.c
16988 (find_candidates_dom_walker::before_dom_children): Also allow
16991 2017-09-06 Richard Biener <rguenther@suse.de>
16993 PR tree-optimization/82108
16994 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16995 for gap in the non-permutation SLP case.
16997 2017-09-06 Martin Jambor <mjambor@suse.cz>
16999 PR tree-optimization/82078
17000 * tree-sra.c (sort_and_splice_var_accesses): Move call to
17001 add_access_to_work_queue...
17002 (build_accesses_from_assign): ...here.
17003 (propagate_all_subaccesses): Make sure racc is the group
17004 representative, if there is one.
17006 2017-09-06 Jakub Jelinek <jakub@redhat.com>
17008 PR middle-end/82095
17009 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
17012 2017-09-06 Richard Biener <rguenther@suse.de>
17014 * gimple-ssa-strength-reduction.c
17015 (find_candidates_dom_walker::before_doom_children): Use a
17016 type and not a mode check.
17018 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
17021 * config/arm/predicates.md (arm_general_adddi_operand): Create new
17023 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
17025 2017-09-05 Jeff Law <law@redhat.com>
17027 PR tree-optimization/64910
17028 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
17029 cases where we have 3 or more operands.
17031 2017-09-05 Jakub Jelinek <jakub@redhat.com>
17033 PR middle-end/81768
17034 * omp-low.c (lower_omp_for): Recompute tree invariant if
17035 gimple_omp_for_initial/final is ADDR_EXPR.
17037 PR middle-end/81768
17038 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
17039 into gimple val before gimplification fo the COND_EXPR.
17041 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
17043 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
17044 REGION_COPY argument.
17045 (thread_through_all_blocks): Remove unused argument to
17046 duplicate_thread_path.
17048 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17049 Alan Hayward <alan.hayward@arm.com>
17050 David Sherwood <david.sherwood@arm.com>
17052 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
17053 Take a scalar_mode rather than a machine_mode.
17054 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17055 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
17056 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17057 (aarch64_gen_adjusted_ldpstp): Likewise.
17058 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
17060 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17061 Alan Hayward <alan.hayward@arm.com>
17062 David Sherwood <david.sherwood@arm.com>
17064 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
17065 Take a scalar_int_mode instead of a machine_mode.
17066 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17067 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17068 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17069 (aarch64_simd_attr_length_rglist): Delete.
17070 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
17071 a scalar_int_mode instead of a machine_mode.
17072 (aarch64_add_offset): Likewise.
17073 (aarch64_internal_mov_immediate): Likewise
17074 (aarch64_add_constant_internal): Likewise.
17075 (aarch64_add_constant): Likewise.
17076 (aarch64_movw_imm): Likewise.
17077 (aarch64_rtx_arith_op_extract_p): Likewise.
17078 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17079 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17080 Remove assert that the mode isn't a vector.
17081 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17082 (aarch64_expand_mov_immediate): Update calls after above changes.
17083 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
17084 (aarch64_and_bitmask_imm): Check for scalar integer modes.
17085 (aarch64_move_imm): Likewise.
17086 (aarch64_can_const_movi_rtx_p): Likewise.
17087 (aarch64_strip_extend): Likewise.
17088 (aarch64_extr_rtx_p): Likewise.
17089 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
17090 a CONST_INT when the mode parameter is VOIDmode.
17091 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
17093 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17095 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
17096 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17097 (bitwise_type_for_mode): Update accordingly.
17099 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17101 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
17102 * stor-layout.c (mode_for_size_tree): Likewise.
17103 (mode_for_array): Update accordingly.
17104 (layout_decl): Likewise.
17105 (compute_record_mode): Likewise. Only set the mode once.
17107 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17109 * target.def (get_mask_mode): Change return type to opt_mode.
17111 * doc/tm.texi: Regenerate.
17112 * targhooks.h (default_get_mask_mode): Return an opt_mode.
17113 * targhooks.c (default_get_mask_mode): Likewise.
17114 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17115 * optabs-query.c (can_vec_mask_load_store_p): Update use of
17116 targetm.get_mask_mode.
17117 * tree.c (build_truth_vector_type): Likewise.
17119 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17121 * machmode.h (mode_for_vector): Return an opt_mode.
17122 * stor-layout.c (mode_for_vector): Likewise.
17123 (mode_for_int_vector): Update accordingly.
17124 (layout_type): Likewise.
17125 * config/i386/i386.c (emit_memmov): Likewise.
17126 (ix86_expand_set_or_movmem): Likewise.
17127 (ix86_expand_vector_init): Likewise.
17128 (ix86_get_mask_mode): Likewise.
17129 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
17131 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
17132 * expmed.c (extract_bit_field_1): Likewise.
17133 * expr.c (expand_expr_real_2): Likewise.
17134 * optabs-query.c (can_vec_perm_p): Likewise.
17135 (can_vec_mask_load_store_p): Likewise.
17136 * optabs.c (expand_vec_perm): Likewise.
17137 * targhooks.c (default_get_mask_mode): Likewise.
17138 * tree-vect-stmts.c (vectorizable_store): Likewise.
17139 (vectorizable_load): Likewise.
17140 (get_vectype_for_scalar_type_and_size): Likewise.
17142 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17144 * machmode.h (mode_for_int_vector): New function.
17145 * stor-layout.c (mode_for_int_vector): Likewise.
17146 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
17147 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
17148 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17149 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17150 (s390_expand_vcond): Likewise.
17152 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17154 * machmode.h (opt_machine_mode): New type.
17155 (opt_mode<T>): Allow construction from anything that can be
17157 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17158 (mode_for_size): Return an opt_machine_mode.
17159 * stor-layout.c (mode_for_size): Likewise.
17160 (mode_for_size_tree): Update call accordingly.
17161 (bitwise_mode_for_mode): Likewise.
17162 (make_fract_type): Likewise.
17163 (make_accum_type): Likewise.
17164 * caller-save.c (replace_reg_with_saved_mem): Update call
17166 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17167 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17168 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17169 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17170 * expmed.c (extract_bit_field_1): Likewise.
17171 * reload.c (get_secondary_mem): Likewise.
17172 * varasm.c (assemble_integer): Likewise.
17173 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17176 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17178 * machmode.h (decimal_float_mode_for_size): New function.
17179 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17180 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17181 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17182 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17183 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17184 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17186 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17188 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17189 (get_builtin_sync_mode): Likewise.
17190 (expand_ifn_atomic_compare_exchange): Likewise.
17191 (expand_builtin_atomic_clear): Likewise.
17192 (expand_builtin_atomic_test_and_set): Likewise.
17193 (fold_builtin_atomic_always_lock_free): Likewise.
17194 * calls.c (compute_argument_addresses): Likewise.
17195 (emit_library_call_value_1): Likewise.
17196 (store_one_arg): Likewise.
17197 * combine.c (combine_instructions): Likewise.
17198 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17199 * config/arm/arm.c (arm_function_value): Likewise.
17200 (aapcs_allocate_return_reg): Likewise.
17201 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17202 * config/i386/i386.c (construct_container): Likewise.
17203 (ix86_gimplify_va_arg): Likewise.
17204 (ix86_expand_sse_cmp): Likewise.
17205 (emit_memmov): Likewise.
17206 (emit_memset): Likewise.
17207 (expand_small_movmem_or_setmem): Likewise.
17208 (ix86_expand_pextr): Likewise.
17209 (ix86_expand_pinsr): Likewise.
17210 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17211 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17213 * config/mips/mips.c (mips_function_value_1) Likewise.
17214 (mips_block_move_straight): Likewise.
17215 (mips_expand_ins_as_unaligned_store): Likewise.
17216 * config/powerpcspe/powerpcspe.c
17217 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17218 (rs6000_darwin64_record_arg_flush): Likewise.
17219 * config/rs6000/rs6000.c
17220 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17221 (rs6000_darwin64_record_arg_flush): Likewise.
17222 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17223 (sparc_function_value_1): Likewise.
17224 * config/spu/spu.c (adjust_operand): Likewise.
17225 (spu_emit_branch_or_set): Likewise.
17226 (arith_immediate_p): Likewise.
17227 * emit-rtl.c (gen_lowpart_common): Likewise.
17228 * expr.c (expand_expr_real_1): Likewise.
17229 * function.c (assign_parm_setup_block): Likewise.
17230 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17231 * reload1.c (alter_reg): Likewise.
17232 * stor-layout.c (mode_for_vector): Likewise.
17233 (layout_type): Likewise.
17235 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17237 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17238 (spu_convert_move): Likewise.
17239 * lower-subreg.c (resolve_simple_move): Likewise.
17241 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17244 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17245 define_insn to a define_expand.
17246 (altivec_vsum2sws_direct): New define_insn.
17247 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17249 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17251 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17254 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17259 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17260 realign stack if stack alignment needed is less than incoming
17263 2017-09-05 Marek Polacek <polacek@redhat.com>
17266 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17269 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17271 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17273 2017-09-05 Richard Biener <rguenther@suse.de>
17275 PR tree-optimization/82084
17276 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17278 2017-09-05 Richard Biener <rguenther@suse.de>
17280 PR tree-optimization/82102
17281 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17283 2017-09-05 Martin Liska <mliska@suse.cz>
17285 PR tree-optimization/82032
17286 * tree-cfg.c (generate_range_test): New function.
17287 * tree-cfg.h (generate_range_test): Declared here.
17288 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17289 (cleanup_control_expr_graph): Use it.
17290 * tree-switch-conversion.c (try_switch_expansion): Remove
17292 (emit_case_nodes): Use generate_range_test.
17294 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17297 * config/i386/i386.md (*<btsc><mode>_mask): Add
17298 TARGET_USE_BT to insn constraint.
17299 (*btr<mode>_mask): Ditto.
17301 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17303 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17304 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17306 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17309 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17310 TARGET_NEON and TARGET_IWMMXT.
17311 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17312 TARGET_NEON and TARGET_IWMMXT.
17313 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17315 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17317 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17318 (ix86_rewrite_tls_address): Ditto.
17319 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17320 (ix86_rewrite_tls_address_1): Ditto.
17321 (ix86_rewrite_tls_address): Ditto.
17322 * config/i386/predicates.md (tls_address_pattern): New predicate.
17323 * config/i386/i386.md (TLS address splitter): New splitter.
17325 2017-09-04 Richard Biener <rguenther@suse.de>
17327 PR tree-optimization/82084
17328 * fold-const.h (can_native_encode_string_p): Declare.
17329 * fold-const.c (can_native_encode_string_p): Factor out from ...
17330 (native_encode_string): ... here.
17331 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17332 vectorizing stores from constants we later cannot handle.
17334 2017-09-04 Marek Polacek <polacek@redhat.com>
17337 * doc/invoke.texi: Update -Wtautological-compare documentation.
17339 2017-09-04 Jeff Law <law@redhat.com>
17341 PR tree-optimization/64910
17342 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17343 swap the first and last operand if the last is a constant.
17345 2017-09-04 Marek Polacek <polacek@redhat.com>
17348 * convert.c (do_narrow): When sanitizing signed integer overflows,
17349 bail out for signed types.
17350 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17352 2017-09-04 Richard Biener <rguenther@suse.de>
17354 PR tree-optimization/82060
17355 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17356 Move devirtualization after stmt folding and before EH/AB/noreturn
17357 cleanup to get the stmt refs canonicalized. Use a bool instead
17358 of gimple_modified_p since that doesn't work for NOPs. Schedule
17359 NOPs generated by folding for removal.
17361 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17362 Alan Hayward <alan.hayward@arm.com>
17363 David Sherwood <david.sherwood@arm.com>
17365 * coretypes.h (pad_direction): New enum.
17366 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17367 (FUNCTION_ARG_PADDING): Likewise.
17368 * target.def (function_arg_padding): New hook.
17369 * targhooks.h (default_function_arg_padding): Declare.
17370 * targhooks.c (default_function_arg_padding): New function.
17371 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17372 (TARGET_FUNCTION_ARG_PADDING): ...this.
17373 * doc/tm.texi: Regenerate.
17374 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17375 instead of direction.
17376 (compute_argument_addresses): Likewise.
17377 (load_register_parameters): Likewise.
17378 (emit_library_call_value_1): Likewise.
17379 (store_one_arg): Use targetm.calls.function_arg_padding instead
17380 of FUNCTION_ARG_PADDING.
17381 (must_pass_in_stack_var_size_or_pad): Likewise.
17382 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17383 (emit_group_store): Likewise.
17384 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17385 instead of FUNCTION_ARG_PADDING.
17386 (emit_push_insn): Likewise, and propagate enum change throughout
17388 * function.h (direction): Delete.
17389 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17391 * function.c (assign_parm_find_stack_rtl): Likewise.
17392 (assign_parm_setup_block_p): Likewise.
17393 (assign_parm_setup_block): Likewise.
17394 (gimplify_parameters): Likewise.
17395 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17396 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17398 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17399 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17400 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17401 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17402 (aarch64_function_arg_padding): ...this new function.
17403 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17404 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17405 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17406 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17407 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17408 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17409 (arm_pad_arg_upward): Replace with...
17410 (arm_function_arg_padding): ...this new function.
17411 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17413 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17414 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17415 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17416 (ia64_hpux_function_arg_padding): Replace with...
17417 (ia64_function_arg_padding): ...this new function. Use pad_direction
17418 instead of direction. Check for TARGET_HPUX.
17419 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17420 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17421 (iq2000_function_arg_padding): New function.
17422 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17423 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17424 (mips_function_arg_padding): ...this new function.
17425 (mips_pad_reg_upward): Update accordingly.
17426 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17427 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17428 targetm.calls.function_arg_padding.
17429 (FUNCTION_ARG_PADDING): Delete.
17430 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17431 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17432 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17433 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17434 (nios2_block_reg_padding): Return pad_direction instead of direction.
17435 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17436 instead of direction.
17437 (nios2_function_arg_padding): Likewise. Make static.
17438 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17439 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17440 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17441 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17442 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17443 (pa_function_arg_padding): Make static. Return pad_direction instead
17445 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17446 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17447 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17448 instead of direction. Use targetm.calls.function_arg_padding.
17449 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17450 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17451 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17452 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17453 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17455 (function_arg_padding): Rename to...
17456 (rs6000_function_arg_padding): ...this. Make static. Return
17457 pad_direction instead of direction.
17458 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17459 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17460 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17461 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17462 instead of direction. Use targetm.calls.function_arg_padding.
17463 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17464 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17465 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17466 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17467 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17468 (function_arg_padding): Rename to...
17469 (rs6000_function_arg_padding): ...this. Make static. Return
17470 pad_direction instead of direction.
17471 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17472 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17473 * config/s390/s390.c (s390_function_arg_padding): New function.
17474 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17475 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17476 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17477 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17478 (function_arg_padding): Rename to...
17479 (sparc_function_arg_padding): ...this. Make static. Return
17480 pad_direction instead of direction.
17481 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17482 * config/spu/spu.c (spu_function_arg_padding): New function.
17483 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17484 * system.h (FUNCTION_ARG_PADDING): Poison.
17486 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17487 Alan Hayward <alan.hayward@arm.com>
17488 David Sherwood <david.sherwood@arm.com>
17490 * target.def (modes_tieable_p): New hook.
17491 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17492 (TARGET_MODES_TIEABLE_P): ...this.
17493 * doc/tm.texi.in: Regenerate.
17494 * hooks.h (hook_bool_mode_mode_true): Declare.
17495 * hooks.c (hook_bool_mode_mode_true): New function.
17496 * combine.c (subst): Use targetm.modes_tieable_p instead of
17498 * dse.c (find_shift_sequence): Likewise.
17499 * expmed.c (extract_low_bits): Likewise.
17500 * lower-subreg.c: Include target.h.
17501 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17503 * rtlanal.c (rtx_cost): Likewise.
17504 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17505 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17506 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17507 (TARGET_MODES_TIEABLE_P): Redefine.
17508 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17509 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17510 (TARGET_MODES_TIEABLE_P): Redefine.
17511 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17512 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17513 (arc_modes_tieable_p): New function.
17514 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17515 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17516 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17517 (arm_modes_tieable_p): Make static.
17518 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17519 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17520 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17521 (TARGET_MODES_TIEABLE_P): Redefine.
17522 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17523 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17524 (TARGET_MODES_TIEABLE_P): Redefine.
17525 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17526 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17527 (cr16_modes_tieable_p): New function.
17528 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17529 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17530 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17531 (TRULY_NOOP_TRUNCATION): Update comment.
17532 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17533 (TRULY_NOOP_TRUNCATION): Update comment.
17534 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17535 (frv_modes_tieable_p): New function.
17536 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17537 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17538 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17539 (TARGET_MODES_TIEABLE_P): Redefine.
17540 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17541 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17542 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17543 (TARGET_MODES_TIEABLE_P): Redefine.
17544 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17545 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17546 (ia64_modes_tieable_p): New function.
17547 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17548 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17549 (iq2000_modes_tieable_p): New function.
17550 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17551 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17552 (lm32_modes_tieable_p): New function.
17553 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17554 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17555 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17556 (TARGET_MODES_TIEABLE_P): Redefine.
17557 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17558 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17559 (m32r_modes_tieable_p): New function.
17560 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17561 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17562 (m68k_modes_tieable_p): New function.
17563 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17564 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17565 (mcore_modes_tieable_p): New function.
17566 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17567 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17569 (TARGET_MODES_TIEABLE_P): Redefine.
17570 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17571 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17572 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17573 (TARGET_MODES_TIEABLE_P): Redefine.
17574 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17575 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17576 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17577 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17578 (mn10300_modes_tieable_p): ...this and make static.
17579 (TARGET_MODES_TIEABLE_P): Redefine.
17580 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17581 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17582 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17583 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17584 (msp430_modes_tieable_p): Make static.
17585 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17586 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17587 (TARGET_MODES_TIEABLE_P): Redefine.
17588 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17589 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17590 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17591 (TARGET_MODES_TIEABLE_P): Redefine.
17592 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17593 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17594 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17595 (TARGET_MODES_TIEABLE_P): Redefine.
17596 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17597 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17598 (pdp11_modes_tieable_p): New function.
17599 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17600 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17601 (rs6000_modes_tieable_p): New function.
17602 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17603 * config/powerpcspe/powerpcspe.md: Update comment.
17604 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17605 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17606 (TARGET_MODES_TIEABLE_P): Redefine.
17607 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17608 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17609 (rl78_modes_tieable_p): New function.
17610 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17611 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17612 (rs6000_modes_tieable_p): New function.
17613 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17614 * config/rs6000/rs6000.md: Update comment.
17615 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17616 * config/rx/rx.c (rx_modes_tieable_p): New function.
17617 (TARGET_MODES_TIEABLE_P): Redefine.
17618 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17619 * config/s390/s390.c (s390_modes_tieable_p): New function.
17620 (TARGET_MODES_TIEABLE_P): Redefine.
17621 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17622 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17623 (sh_modes_tieable_p): New function.
17624 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17625 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17626 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17627 (sparc_modes_tieable_p): Make static.
17628 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17629 * config/spu/spu.c (spu_modes_tieable_p): New function.
17630 (TARGET_MODES_TIEABLE_P): Redefine.
17631 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17632 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17633 (TARGET_MODES_TIEABLE_P): Redefine.
17634 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17635 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17636 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17637 * config/v850/v850.c (v850_modes_tieable_p): New function.
17638 (TARGET_MODES_TIEABLE_P): Redefine.
17639 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17640 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17641 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17642 (visium_modes_tieable_p): New function.
17643 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17644 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17645 (xtensa_modes_tieable_p): New function.
17646 * system.h (MODES_TIEABLE_P): Poison.
17648 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17649 Alan Hayward <alan.hayward@arm.com>
17650 David Sherwood <david.sherwood@arm.com>
17652 * target.def (hard_regno_mode_ok): New hook.
17653 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17654 (TARGET_HARD_REGNO_MODE_OK): ...this.
17655 * doc/tm.texi.in: Regenerate.
17656 * hooks.h (hook_bool_uint_mode_true): Declare.
17657 * hooks.c (hook_bool_uint_mode_true): New function.
17658 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17659 HARD_REGNO_MODE_OK.
17660 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17661 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17662 instead of HARD_REGNO_MODE_OK.
17663 * caller-save.c: Include target.h.
17664 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17665 HARD_REGNO_MODE_OK.
17666 * combine.c (can_combine_p): Likewise.
17667 (combinable_i3pat): Likewise.
17668 (can_change_dest_mode): Likewise.
17669 * expr.c (init_expr_target): Likewise.
17670 (convert_move): Likewise.
17671 (convert_modes): Likewise.
17672 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17673 (setup_prohibited_mode_move_regs): Likewise.
17674 * ira.h (target_ira): Likewise.
17675 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17676 * lra-constraints.c (process_alt_operands): Likewise.
17677 (split_reg): Likewise.
17678 * recog.c (peep2_find_free_register): Likewise.
17679 * ree.c (combine_reaching_defs): Likewise.
17680 * regcprop.c (maybe_mode_change): Likewise.
17681 * reginfo.c (init_reg_sets_1): Likewise.
17682 (choose_hard_reg_mode): Likewise.
17683 (simplifiable_subregs): Likewise.
17684 * regrename.c (check_new_reg_p): Likewise.
17685 * reload.c (find_valid_class): Likewise.
17686 (find_valid_class_1): Likewise.
17687 (reload_inner_reg_of_subreg): Likewise.
17688 (push_reload): Likewise.
17689 (combine_reloads): Likewise.
17690 (find_dummy_reload): Likewise.
17691 (find_reloads): Likewise.
17692 * reload1.c (find_reg): Likewise.
17693 (set_reload_reg): Likewise.
17694 (allocate_reload_reg): Likewise.
17695 (choose_reload_regs): Likewise.
17696 (reload_adjust_reg_for_temp): Likewise.
17697 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17698 (simplify_subreg_regno): Likewise.
17699 * sel-sched.c (init_regs_for_mode): Likewise.
17700 * varasm.c (make_decl_rtl): Likewise.
17701 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17702 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17703 HARD_REGNO_MODE_OK.
17704 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17705 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17706 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17707 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17708 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17709 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17710 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17711 (arc_mode_class): Delete.
17712 (HARD_REGNO_MODE_OK): Delete.
17713 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17714 (arc_hard_regno_mode_ok): Rename old array to...
17715 (arc_hard_regno_mode_ok_modes): ...this.
17716 (arc_conditional_register_usage): Update accordingly.
17717 (arc_mode_class): Make static.
17718 (arc_hard_regno_mode_ok): New function.
17719 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17720 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17721 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17722 (arm_hard_regno_mode_ok): Make static.
17723 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17724 HARD_REGNO_MODE_OK.
17725 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17726 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17727 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17729 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17730 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17731 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17732 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17733 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17734 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17735 * config/bfin/predicates.md (valid_reg_operand): Use
17736 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17737 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17738 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17739 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17740 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17741 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17742 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17743 (cr16_hard_regno_mode_ok): Make static and return a bool.
17744 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17745 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17746 (cris_hard_regno_mode_ok): New function.
17747 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17748 (epiphany_mode_class): Delete.
17749 (HARD_REGNO_MODE_OK): Delete.
17750 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17751 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17752 (hard_regno_mode_ok): Rename to...
17753 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17754 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17755 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17756 HARD_REGNO_MODE_OK.
17757 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17758 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17759 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17760 (frv_hard_regno_mode_ok): Make static and return a bool.
17761 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17762 HARD_REGNO_MODE_OK.
17763 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17764 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17765 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17766 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17768 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17769 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17770 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17771 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17773 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17774 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17775 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17776 (ia64_hard_regno_mode_ok): New function.
17777 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17778 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17779 (iq2000_hard_regno_mode_ok): New function.
17780 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17781 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17782 (lm32_hard_regno_mode_ok): New function.
17783 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17784 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17785 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17786 instead of HARD_REGNO_MODE_OK.
17787 (m32c_hard_regno_ok): Rename to...
17788 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17789 (m32c_cannot_change_mode_class): Update accordingly.
17790 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17791 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17792 (m32r_mode_class): Delete.
17793 (HARD_REGNO_MODE_OK): Delete.
17794 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17795 (m32r_hard_regno_mode_ok): Rename to...
17796 (m32r_hard_regno_modes): ...this.
17797 (m32r_mode_class): Make static.
17798 (m32r_hard_regno_mode_ok): New function.
17799 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17800 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17801 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17802 (m68k_hard_regno_mode_ok): Make static.
17803 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17804 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17805 (mcore_hard_regno_mode_ok): New function.
17806 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17807 (HARD_REGNO_MODE_OK): Delete.
17808 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17810 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17811 (microblaze_hard_regno_mode_ok): New function.
17812 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17813 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17814 (mips_hard_regno_mode_ok): Delete.
17815 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17816 (mips_hard_regno_mode_ok_p): ...this and make static.
17817 (mips_hard_regno_mode_ok_p): Rename to...
17818 (mips_hard_regno_mode_ok_uncached): ...this.
17819 (mips_hard_regno_mode_ok): New function.
17820 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17821 of HARD_REGNO_MODE_OK.
17822 (mips_option_override): Update after above name changes.
17823 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17824 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17825 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17826 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17827 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17828 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17829 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17830 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17831 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17832 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17833 (msp430_hard_regno_mode_ok): Make static and return a bool.
17834 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17835 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17836 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17838 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17839 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17840 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17841 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17842 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17843 (PA_HARD_REGNO_MODE_OK): ...this
17844 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17845 (PA_HARD_REGNO_MODE_OK): ...this.
17846 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17847 (pa_hard_regno_mode_ok): New function.
17848 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17849 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17850 (pdp11_hard_regno_mode_ok): New function.
17851 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17852 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17854 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17856 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17857 (rs6000_hard_regno_mode_ok): Rename to...
17858 (rs6000_hard_regno_mode_ok_uncached): ...this.
17859 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17860 (rs6000_hard_regno_mode_ok): New function.
17861 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17862 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17863 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17864 (riscv_hard_regno_mode_ok): ...this and make static.
17865 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17866 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17867 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17868 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17869 (rl78_hard_regno_mode_ok): Make static and return bool.
17870 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17871 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17873 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17874 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17875 (rs6000_hard_regno_mode_ok): Rename to...
17876 (rs6000_hard_regno_mode_ok_uncached): ...this.
17877 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17878 (rs6000_hard_regno_mode_ok): New function.
17879 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17880 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17881 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17882 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17883 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17884 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17885 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17886 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17887 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17888 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17889 (sh_hard_regno_mode_ok): Make static.
17890 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17891 instead of HARD_REGNO_MODE_OK.
17892 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17893 (sparc_mode_class): Delete.
17894 (HARD_REGNO_MODE_OK): Delete.
17895 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17896 (hard_regno_mode_classes): Make static.
17897 (sparc_mode_class): Likewise.
17898 (sparc_hard_regno_mode_ok): New function.
17899 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17900 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17901 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17903 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17904 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17905 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17906 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17907 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17908 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17909 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17910 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17911 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17912 (visium_hard_regno_mode_ok): New function.
17913 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17914 instead of HARD_REGNO_MODE_OK.
17915 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17916 (HARD_REGNO_MODE_OK): Delete.
17917 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17918 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17919 (xtensa_option_override): Update accordingly.
17920 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17921 (xtensa_hard_regno_mode_ok): New function.
17922 * system.h (HARD_REGNO_MODE_OK): Poison.
17924 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17925 Alan Hayward <alan.hayward@arm.com>
17926 David Sherwood <david.sherwood@arm.com>
17928 * target.def (hard_regno_call_part_clobbered): New hook.
17929 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17930 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17931 * doc/tm.texi: Regenerate.
17932 * hooks.h (hook_bool_uint_mode_false): Declare.
17933 * hooks.c (hook_bool_uint_mode_false): New function.
17934 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17935 * cselib.c (cselib_process_insn): Use
17936 targetm.hard_regno_call_part_clobbered instead of
17937 HARD_REGNO_CALL_PART_CLOBBERED.
17938 * ira-conflicts.c (ira_build_conflicts): Likewise.
17939 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17940 * lra-constraints.c (need_for_call_save_p): Likewise.
17941 * lra-lives.c: Include target.h.
17942 (check_pseudos_live_through_calls): Use
17943 targetm.hard_regno_call_part_clobbered instead of
17944 HARD_REGNO_CALL_PART_CLOBBERED.
17945 * regcprop.c: Include target.h.
17946 (copyprop_hardreg_forward_1): Use
17947 targetm.hard_regno_call_part_clobbered instead of
17948 HARD_REGNO_CALL_PART_CLOBBERED.
17949 * reginfo.c (choose_hard_reg_mode): Likewise.
17950 * regrename.c (check_new_reg_p): Likewise.
17951 * reload.c (find_equiv_reg): Likewise.
17952 * reload1.c (emit_reload_insns): Likewise.
17953 * sched-deps.c (deps_analyze_insn): Likewise.
17954 * sel-sched.c (init_regs_for_mode): Likewise.
17955 (mark_unavailable_hard_regs): Likewise.
17956 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17957 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17958 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17960 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17961 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17962 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17964 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17966 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17967 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17968 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17970 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17971 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17972 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17974 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17975 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17977 * config/powerpcspe/powerpcspe.c
17978 (rs6000_hard_regno_call_part_clobbered): New function.
17979 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17980 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17981 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17983 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17984 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17985 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17987 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17988 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17989 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17991 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17992 Alan Hayward <alan.hayward@arm.com>
17993 David Sherwood <david.sherwood@arm.com>
17995 * rtl.h (subreg_memory_offset): Declare.
17996 * emit-rtl.c (subreg_memory_offset): New function.
17997 * expmed.c (store_bit_field_1): Use it.
17998 * expr.c (undefined_operand_subword_p): Likewise.
17999 * simplify-rtx.c (simplify_subreg): Likewise.
18001 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
18003 PR rtl-optimization/57448
18006 * optabs.c (expand_atomic_load): Place compiler memory barriers if
18007 using atomic_load pattern.
18008 (expand_atomic_store): Likewise.
18010 2017-09-04 Jakub Jelinek <jakub@redhat.com>
18013 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
18014 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
18015 handling. Use replace_call_with_value with NULL instead of
18016 gsi_replace, unlink_stmt_vdef and release_defs.
18018 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
18021 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
18023 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18026 * rtl.h (emit_library_call_value_1): Declare.
18027 (emit_library_call): Replace declaration with a series of overloads.
18028 Remove the parameter count argument.
18029 (emit_library_call_value): Likewise.
18030 * calls.c (emit_library_call_value_1): Make global. Replace varargs
18031 with an "rtx_mode_t *".
18032 (emit_library_call_value): Delete.
18033 (emit_library_call): Likewise.
18034 * asan.c (asan_emit_stack_protection): Update calls accordingly.
18035 (asan_emit_allocas_unpoison): Likewise.
18036 * builtins.c (expand_builtin_powi): Likewise.
18037 (expand_asan_emit_allocas_unpoison): Likewise.
18038 * cfgexpand.c (expand_main_function): Likewise.
18039 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
18040 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
18041 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
18042 * config/arm/arm.c (arm_trampoline_init): Likewise.
18043 (arm_call_tls_get_addr): Likewise.
18044 (arm_expand_divmod_libfunc): Likewise.
18045 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
18046 (smulsi3_highpart): Likewise.
18047 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
18048 (c6x_expand_compare): Likewise.
18049 (c6x_expand_movmem): Likewise.
18050 * config/frv/frv.c (frv_trampoline_init): Likewise.
18051 * config/i386/i386.c (ix86_trampoline_init): Likewise.
18052 (ix86_expand_divmod_libfunc): Likewise.
18053 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
18054 (ia64_expand_compare): Likewise.
18055 (ia64_profile_hook): Likewise.
18056 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
18057 (nonlocal_goto): Likewise.
18058 (restore_stack_nonlocal): Likewise.
18059 * config/m32r/m32r.c (block_move_call): Likewise.
18060 (m32r_trampoline_init): Likewise.
18061 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
18062 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
18063 (m68k_call_m68k_read_tp): Likewise.
18064 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
18065 (microblaze_expand_divide): Likewise.
18066 * config/mips/mips.h (mips_args): Likewise.
18067 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
18068 (MIPS_ICACHE_SYNC): Likewise.
18069 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
18070 (nios2_trampoline_init): Likewise.
18071 * config/pa/pa.c (hppa_tls_call): Likewise.
18072 (pa_trampoline_init): Likewise.
18073 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
18074 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
18075 (expand_strn_compare): Likewise.
18076 (rs6000_generate_compare): Likewise.
18077 (rs6000_expand_float128_convert): Likewise.
18078 (output_profile_hook): Likewise.
18079 (rs6000_trampoline_init): Likewise.
18080 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
18081 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
18082 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
18083 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
18084 (rs6000_generate_compare): Likewise.
18085 (rs6000_expand_float128_convert): Likewise.
18086 (output_profile_hook): Likewise.
18087 (rs6000_trampoline_init): Likewise.
18088 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
18089 * config/sh/sh.c (sh_trampoline_init): Likewise.
18090 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
18091 (sparc_emit_float_lib_cmp): Likewise.
18092 (sparc32_initialize_trampoline): Likewise.
18093 (sparc64_initialize_trampoline): Likewise.
18094 (sparc_profile_hook): Likewise.
18095 * config/spu/spu.c (ea_load_store): Likewise.
18096 * config/spu/spu.md (floatunssidf2): Likewise.
18097 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
18098 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
18099 * config/visium/visium.c (expand_block_move_4): Likewise.
18100 (expand_block_move_2): Likewise.
18101 (expand_block_move_1): Likewise.
18102 (expand_block_set_4): Likewise.
18103 (expand_block_set_2): Likewise.
18104 (expand_block_set_1): Likewise.
18105 (visium_trampoline_init): Likewise.
18106 (visium_profile_hook): Likewise.
18107 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
18108 (xtensa_setup_frame_addresses): Likewise.
18109 (xtensa_trampoline_init): Likewise.
18110 * except.c (sjlj_emit_function_enter): Likewise.
18111 (sjlj_emit_function_exit): Likewise.
18112 * explow.c (allocate_dynamic_stack_space): Likewise.
18113 (probe_stack_range): Likewise.
18114 * expr.c (convert_mode_scalar): Likewise.
18115 * optabs.c (expand_binop): Likewise.
18116 (expand_twoval_binop_libfunc): Likewise.
18117 (expand_unop): Likewise.
18118 (prepare_cmp_insn): Likewise.
18119 (prepare_float_lib_cmp): Likewise.
18120 (expand_float): Likewise.
18121 (expand_fix): Likewise.
18122 (expand_fixed_convert): Likewise.
18123 (maybe_emit_sync_lock_test_and_set): Likewise.
18124 (expand_atomic_compare_and_swap): Likewise.
18125 (expand_mem_thread_fence): Likewise.
18126 (expand_atomic_fetch_op): Likewise.
18128 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
18130 * doc/generic.texi (OpenACC): Adjust URL.
18131 * doc/invoke.texi (C Dialect Options): Ditto.
18133 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
18135 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
18136 predicate for operand 1. Add (m,<S>) constraint.
18137 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
18138 Prevent memory operand 1 with register operand 2.
18140 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
18142 PR rtl-optimization/82024
18143 * combine.c (try_combine): If the combination result is a PARALLEL,
18144 and we only need to retain the SET in there that would be placed
18145 at I2, check that we can place that at I3 instead, before doing so.
18147 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18150 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
18152 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
18153 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
18156 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
18157 Jeff Law <law@redhat.com>
18159 * varasm.c (bss_initializer_p): Do not put constants into .bss
18160 (categorize_decl_for_section): Handle bss_initializer_p returning
18161 false when DECL_INITIAL is NULL.
18163 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18166 * config/s390/s390.c (s390_can_inline_p): New function.
18168 2017-09-01 Jeff Law <law@redhat.com>
18170 PR tree-optimization/82052
18171 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18172 Always initialize the returned slot after a hash table miss
18173 when INSERT is true.
18175 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18177 * config/s390/s390.md (mem_signal_fence): Remove.
18178 * doc/md.texi (mem_signal_fence): Remove.
18179 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18181 * target-insns.def (mem_signal_fence): Remove.
18183 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18186 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18189 * asan.c (create_odr_indicator): Strip name encoding from assembler
18190 name before appending it after __odr_asan_.
18192 2017-09-01 Martin Liska <mliska@suse.cz>
18194 PR tree-optimization/82059
18195 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18196 frequency only when an edge is redirected.
18198 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18200 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18201 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18202 (arc_conditional_register_usage): Remove ARC600 lp_count
18204 (arc_file_start): Emit Tag_ARC_CPU_variation.
18205 (arc_can_use_doloop_p): New conditions to use ZOLs.
18206 (hwloop_fail): New function.
18207 (hwloop_optimize): Likewise.
18208 (hwloop_pattern_reg): Likewise.
18209 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18210 (arc_reorg_loops): New function, calls reorg_loops.
18211 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18212 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18214 (arc_loop_hazard): Remove function, functionality moved into
18216 (arc_hazard): Remove arc_loop_hazard call.
18217 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18218 into hwloop_optimize.
18219 (arc_label_align): Remove ZOL handling.
18220 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18221 * config/arc/arc.md (doloop_begin): Remove pattern.
18222 (doloop_begin_i): Likewise.
18223 (doloop_end_i): Likewise.
18224 (doloop_fallback): Likewise.
18225 (doloop_fallback_m): Likewise.
18226 (doloop_end): Reimplement expand.
18227 (arc_lp): New pattern for LP instruction.
18228 (loop_end): New pattern.
18229 (loop_fail): Likewise.
18230 (decrement_and_branch_until_zero): Likewise.
18231 * config/arc/arc.opt (mlpc-width): New option.
18232 * doc/invoke.texi (mlpc-width): Document option.
18234 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18236 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18237 (arc_ccfsm_advance): Fix checking for delay slots.
18238 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18240 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18242 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18244 (movhi_insn): Update store instruction constraint which are saving
18245 6-bit short immediates.
18246 (movsi_insn): Consider also short scaled load operations.
18247 (zero_extendhisi2_i): Use Usd constraint instead of T.
18248 (extendhisi2_i): Add q constraint.
18249 (arc_clzsi2): Add type and length attributes.
18250 (arc_ctzsi2): Likewise.
18251 * config/arc/constraints.md (Usc): Update constraint, the
18252 assembler can parse two relocations for a single instruction.
18254 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18256 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18257 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18259 2017-08-31 Olivier Hainque <hainque@adacore.com>
18261 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18262 match as powerpc-wrs-vxworks*.
18264 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18266 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18267 register constraint for by-element operand.
18268 (aarch64_mls_elt_merge<mode>): Likewise.
18270 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18272 * config/arc/arc.c (arc_can_follow_jump): Check for short
18275 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18277 * config.gcc: Use g.opt for arc.
18278 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18279 functionality moved to ...
18280 (legitimate_scaled_address_p): New function, ...here.
18281 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18282 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18283 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18285 (arc_override_options): Handle G option.
18286 (arc_output_pic_addr_const): Correct function definition.
18287 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18288 (arc_decl_anon_ns_mem_p): Delete.
18289 (arc_in_small_data_p): Overhaul this function to take into
18290 consideration the value given via G option.
18291 (arc_rewrite_small_data_1): Renamed and corrected old
18292 arc_rewrite_small_data function.
18293 (arc_rewrite_small_data): New function.
18294 (small_data_pattern): Don't use pic_offset_table_rtx.
18295 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18296 * config/arc/simdext.md (movmisalignv2hi): Use
18297 prepare_move_operands function.
18299 (movmisalign*): Likewise.
18300 * doc/invoke.texi (ARC options): Document -G option.
18302 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18304 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18306 * config/arc/arc.c (arc_print_operand): Output scalled address for
18307 sdata whenever is possible.
18308 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18309 load/stores are available.
18310 (compact_sda_memory_operand): Check for the alignment required by
18311 code density instructions.
18312 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18314 * config/arc/constraints.md (Usd): Update constraint.
18315 (Us0): New constraint.
18316 (Usc): Update constraint.
18318 2017-08-31 Richard Biener <rguenther@suse.de>
18320 PR middle-end/82054
18321 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18322 function only once.
18324 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18326 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18327 Resize type_signature.
18329 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18330 Alan Hayward <alan.hayward@arm.com>
18331 David Sherwood <david.sherwood@arm.com>
18333 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18334 subregs whose inner modes can be stored in GPRs.
18335 (aarch64_classify_index): Likewise.
18337 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18338 Alan Hayward <alan.hayward@arm.com>
18339 David Sherwood <david.sherwood@arm.com>
18341 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18342 (V_INT_EQUIV): ...this.
18343 (v_cmp_result): Rename to...
18344 (v_int_equiv): ...this.
18345 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18346 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18347 (copysign<mode>3): Likewise.
18348 (aarch64_simd_bsl<mode>_internal): Likewise.
18349 (aarch64_simd_bsl<mode>): Likewise.
18350 (vec_cmp<mode><mode>): Likewise.
18351 (vcond<mode><mode>): Likewise.
18352 (vcond<v_cmp_mixed><mode>): Likewise.
18353 (vcondu<mode><v_cmp_mixed>): Likewise.
18354 (aarch64_cm<optab><mode>): Likewise.
18355 (aarch64_cmtst<mode>): Likewise.
18356 (aarch64_fac<optab><mode>): Likewise.
18357 (vec_perm_const<mode>): Likewise.
18358 (vcond_mask_<mode><v_cmp_result>): Rename to...
18359 (vcond_mask_<mode><v_int_equiv>): ...this.
18360 (vec_cmp<mode><v_cmp_result>): Rename to...
18361 (vec_cmp<mode><v_int_equiv>): ...this.
18363 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18364 Alan Hayward <alan.hayward@arm.com>
18365 David Sherwood <david.sherwood@arm.com>
18367 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18369 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18370 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18371 (UNSPEC_LD4_DREG): New unspecs.
18372 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18373 (aarch64_ld2<mode>_dreg_be): Replace with...
18374 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18376 (aarch64_ld3<mode>_dreg_le)
18377 (aarch64_ld3<mode>_dreg_be): Replace with...
18378 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18380 (aarch64_ld4<mode>_dreg_le)
18381 (aarch64_ld4<mode>_dreg_be): Replace with...
18382 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18385 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18387 PR tree-optimization/81987
18388 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18389 insert an initializer in a location not dominated by the stride
18392 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18394 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18395 on the entire header of the finally block in the fallthru case.
18397 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18399 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18401 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18403 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18404 rs6000_emit_move_from_cr and call renamed function.
18405 (rs6000_emit_prologue): Call renamed functions.
18406 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18407 movesi_from_cr, remove volatile CRs.
18409 2017-08-30 Jon Beniston <jon@beniston.com>
18410 Richard Biener <rguenther@suse.de>
18412 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18413 of VECTOR_MODE_P check.
18414 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18415 element vector types.
18417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18419 * df.h (df_read_modify_subreg_p): Remove in favor of...
18420 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18421 const_rtx instead of an rtx.
18422 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18423 * df-problems.c (df_word_lr_mark_ref): Likewise.
18424 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18425 (mark_pseudo_reg_dead): Likewise.
18426 (mark_ref_dead): Likewise.
18427 * reginfo.c (init_subregs_of_mode): Likewise.
18428 * sched-deps.c (sched_analyze_1): Likewise.
18429 * df-scan.c (df_def_record_1): Likewise.
18430 (df_uses_record): Likewise.
18431 (df_read_modify_subreg_p): Remove in favor of...
18432 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18433 const_rtx instead of an rtx.
18435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18436 Alan Hayward <alan.hayward@arm.com>
18437 David Sherwood <david.sherwood@arm.com>
18439 * rtl.h (partial_subreg_p): New function.
18440 * caller-save.c (save_call_clobbered_regs): Use it.
18441 * calls.c (expand_call): Likewise.
18442 * combine.c (combinable_i3pat): Likewise.
18443 (simplify_set): Likewise.
18444 (make_extraction): Likewise.
18445 (make_compound_operation_int): Likewise.
18446 (gen_lowpart_or_truncate): Likewise.
18447 (force_to_mode): Likewise.
18448 (make_field_assignment): Likewise.
18449 (reg_truncated_to_mode): Likewise.
18450 (record_truncated_value): Likewise.
18451 (move_deaths): Likewise.
18452 * cse.c (record_jump_cond): Likewise.
18453 (cse_insn): Likewise.
18454 * cselib.c (cselib_lookup_1): Likewise.
18455 * expmed.c (extract_bit_field_using_extv): Likewise.
18456 * function.c (assign_parm_setup_reg): Likewise.
18457 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18458 * ira-build.c (create_insn_allocnos): Likewise.
18459 * lra-coalesce.c (merge_pseudos): Likewise.
18460 * lra-constraints.c (match_reload): Likewise.
18461 (simplify_operand_subreg): Likewise.
18462 (curr_insn_transform): Likewise.
18463 * lra-lives.c (process_bb_lives): Likewise.
18464 * lra.c (new_insn_reg): Likewise.
18465 (lra_substitute_pseudo): Likewise.
18466 * regcprop.c (mode_change_ok): Likewise.
18467 (maybe_mode_change): Likewise.
18468 (copyprop_hardreg_forward_1): Likewise.
18469 * reload.c (push_reload): Likewise.
18470 (find_reloads): Likewise.
18471 (find_reloads_subreg_address): Likewise.
18472 * reload1.c (alter_reg): Likewise.
18473 (eliminate_regs_1): Likewise.
18474 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18476 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18478 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18479 back to if statements, including unpack.
18481 2017-08-30 Martin Liska <mliska@suse.cz>
18483 PR inline-asm/82001
18484 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18486 (func_checker::compare_asm_inputs_outputs): ... this function.
18487 (func_checker::compare_gimple_asm): Use the function to compare
18488 also ASM constrains.
18489 * ipa-icf-gimple.h: Rename the function.
18491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18492 Alan Hayward <alan.hayward@arm.com>
18493 David Sherwood <david.sherwood@arm.com>
18495 * coretypes.h (complex_mode): New type.
18496 * gdbhooks.py (build_pretty_printer): Handle it.
18497 * machmode.h (complex_mode): New class.
18498 (complex_mode::includes_p): New function.
18499 (is_complex_int_mode): Likewise.
18500 (is_complex_float_mode): Likewise.
18501 * genmodes.c (get_mode_class): Handle complex mode classes.
18502 * function.c (expand_function_end): Use is_complex_int_mode.
18504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18505 Alan Hayward <alan.hayward@arm.com>
18506 David Sherwood <david.sherwood@arm.com>
18508 * coretypes.h (scalar_mode_pod): New typedef.
18509 * gdbhooks.py (build_pretty_printer): Handle it.
18510 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18511 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18512 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18513 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18514 as_a <scalar_mode>.
18516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18517 Alan Hayward <alan.hayward@arm.com>
18518 David Sherwood <david.sherwood@arm.com>
18520 * machmode.h (mode_for_vector): Take a scalar_mode instead
18522 * stor-layout.c (mode_for_vector): Likewise.
18523 * explow.c (promote_mode): Use as_a <scalar_mode>.
18524 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18527 Alan Hayward <alan.hayward@arm.com>
18528 David Sherwood <david.sherwood@arm.com>
18530 * target.def (preferred_simd_mode): Take a scalar_mode
18531 instead of a machine_mode.
18532 * targhooks.h (default_preferred_simd_mode): Likewise.
18533 * targhooks.c (default_preferred_simd_mode): Likewise.
18534 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18535 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18536 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18537 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18538 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18539 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18540 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18541 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18543 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18544 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18545 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18546 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18547 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18548 * doc/tm.texi: Regenerate.
18549 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18553 Alan Hayward <alan.hayward@arm.com>
18554 David Sherwood <david.sherwood@arm.com>
18556 * target.def (scalar_mode_supported_p): Take a scalar_mode
18557 instead of a machine_mode.
18558 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18559 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18560 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18561 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18562 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18563 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18564 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18565 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18566 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18567 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18568 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18569 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18570 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18571 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18573 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18574 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18575 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18576 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18577 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18579 * doc/tm.texi: Regenerate.
18581 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18582 Alan Hayward <alan.hayward@arm.com>
18583 David Sherwood <david.sherwood@arm.com>
18585 * coretypes.h (opt_scalar_mode): New typedef.
18586 * gdbhooks.py (build_pretty_printers): Handle it.
18587 * machmode.h (mode_iterator::get_2xwider): Add overload for
18589 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18591 * expr.c (convert_mode_scalar): Likewise.
18592 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18593 * optabs.c (expand_float): Likewise.
18594 (expand_fix): Likewise.
18595 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18598 Alan Hayward <alan.hayward@arm.com>
18599 David Sherwood <david.sherwood@arm.com>
18601 * optabs.c (expand_float): Explicitly check for scalars before
18602 using a branching expansion.
18603 (expand_fix): Likewise.
18605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18606 Alan Hayward <alan.hayward@arm.com>
18607 David Sherwood <david.sherwood@arm.com>
18609 * expr.c (convert_mode): Split scalar handling out into...
18610 (convert_mode_scalar): ...this new function. Treat the modes
18613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18614 Alan Hayward <alan.hayward@arm.com>
18615 David Sherwood <david.sherwood@arm.com>
18617 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18619 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18622 Alan Hayward <alan.hayward@arm.com>
18623 David Sherwood <david.sherwood@arm.com>
18625 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18626 rather than a machine_mode.
18627 (fixed_from_string): Likewise.
18628 (fixed_convert): Likewise.
18629 (fixed_convert_from_int): Likewise.
18630 (fixed_convert_from_real): Likewise.
18631 (real_convert_from_fixed): Likewise.
18632 * fixed-value.c (fixed_from_double_int): Likewise.
18633 (fixed_from_string): Likewise.
18634 (fixed_convert): Likewise.
18635 (fixed_convert_from_int): Likewise.
18636 (fixed_convert_from_real): Likewise.
18637 (real_convert_from_fixed): Likewise.
18638 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18641 Alan Hayward <alan.hayward@arm.com>
18642 David Sherwood <david.sherwood@arm.com>
18644 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18645 of separate mode class checks. Do not allow vector modes here.
18646 (immed_wide_int_const): Use as_a <scalar_mode>.
18647 * explow.c (trunc_int_for_mode): Likewise.
18648 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18649 (wi::shwi): Likewise.
18650 (wi::min_value): Likewise.
18651 (wi::max_value): Likewise.
18652 * dwarf2out.c (loc_descriptor): Likewise.
18653 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18654 for CONST_WIDE_INT.
18656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18657 Alan Hayward <alan.hayward@arm.com>
18658 David Sherwood <david.sherwood@arm.com>
18660 * tree.h (SCALAR_TYPE_MODE): New macro.
18661 * expr.c (expand_expr_addr_expr_1): Use it.
18662 (expand_expr_real_2): Likewise.
18663 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18664 (fold_convert_const_fixed_from_int): Likewise.
18665 (fold_convert_const_fixed_from_real): Likewise.
18666 (native_encode_fixed): Likewise
18667 (native_encode_complex): Likewise
18668 (native_encode_vector): Likewise.
18669 (native_interpret_fixed): Likewise.
18670 (native_interpret_real): Likewise.
18671 (native_interpret_complex): Likewise.
18672 (native_interpret_vector): Likewise.
18673 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18674 (simd_clone_adjust_argument_types): Likewise.
18675 (simd_clone_init_simd_arrays): Likewise.
18676 (simd_clone_adjust): Likewise.
18677 * stor-layout.c (layout_type): Likewise.
18678 * tree.c (build_minus_one_cst): Likewise.
18679 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18680 * tree-inline.c (estimate_move_cost): Likewise.
18681 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18682 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18683 (vectorizable_reduction): Likewise.
18684 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18685 (vect_recog_mixed_size_cond_pattern): Likewise.
18686 (check_bool_pattern): Likewise.
18687 (adjust_bool_pattern): Likewise.
18688 (search_type_for_mask_1): Likewise.
18689 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18690 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18691 (vectorizable_load): Likewise.
18692 (vectorizable_store): Likewise.
18693 * ubsan.c (ubsan_encode_value): Likewise.
18694 * varasm.c (output_constant): Likewise.
18696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18697 Alan Hayward <alan.hayward@arm.com>
18698 David Sherwood <david.sherwood@arm.com>
18700 * coretypes.h (scalar_mode): New class.
18701 * machmode.h (scalar_mode): Likewise.
18702 (scalar_mode::includes_p): New function.
18703 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18704 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18705 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18706 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18707 * expmed.c (store_bit_field_1): Likewise.
18708 (extract_bit_field_1): Likewise.
18709 * expr.c (write_complex_part): Likewise.
18710 (read_complex_part): Likewise.
18711 (emit_move_complex_push): Likewise.
18712 (expand_expr_real_2): Likewise.
18713 * function.c (assign_parm_setup_reg): Likewise.
18714 (assign_parms_unsplit_complex): Likewise.
18715 * optabs.c (expand_binop): Likewise.
18716 * rtlanal.c (subreg_get_info): Likewise.
18717 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18718 * varasm.c (output_constant_pool_2): Likewise.
18720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18721 Alan Hayward <alan.hayward@arm.com>
18722 David Sherwood <david.sherwood@arm.com>
18724 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18726 (expmed_mult_highpart_optab): Likewise.
18727 (expmed_mult_highpart): Likewise.
18729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18730 Alan Hayward <alan.hayward@arm.com>
18731 David Sherwood <david.sherwood@arm.com>
18733 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18734 instead of a machine_mode.
18735 (builtin_memset_read_str): Likewise.
18736 * builtins.c (c_readstr): Likewise.
18737 (builtin_memcpy_read_str): Likewise.
18738 (builtin_strncpy_read_str): Likewise.
18739 (builtin_memset_read_str): Likewise.
18740 (builtin_memset_gen_str): Likewise.
18741 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18742 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18743 instead of a machine_mode.
18744 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18746 (make_extraction): Likewise.
18747 (try_widen_shift_mode): Take and return scalar_int_modes instead
18749 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18750 a scalar_int_mode instead of a machine_mode.
18751 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18752 (avr_addr_space_pointer_mode): Likewise.
18753 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18754 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18755 (msp430_unwind_word_mode): Likewise.
18756 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18757 (spu_addr_space_pointer_mode): Likewise.
18758 (spu_addr_space_address_mode): Likewise.
18759 (spu_libgcc_cmp_return_mode): Likewise.
18760 (spu_libgcc_shift_count_mode): Likewise.
18761 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18762 (rl78_addr_space_pointer_mode): Likewise.
18763 (fl78_unwind_word_mode): Likewise.
18764 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18766 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18767 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18768 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18769 (mips_valid_pointer_mode): Likewise.
18770 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18771 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18772 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18774 (ft32_addr_space_address_mode): Likewise.
18775 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18776 scalar_int_mode instead of a machine_mode.
18777 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18779 (m32c_addr_space_address_mode): Likewise.
18780 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18781 (rs6000_eh_return_filter_mode): Likewise.
18782 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18783 (rs6000_eh_return_filter_mode): Likewise.
18784 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18785 (s390_libgcc_shift_count_mode): Likewise.
18786 (s390_unwind_word_mode): Likewise.
18787 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18789 * target.def (mode_rep_extended): Likewise.
18790 (valid_pointer_mode): Likewise.
18791 (addr_space.valid_pointer_mode): Likewise.
18792 (eh_return_filter_mode): Return a scalar_int_mode rather than
18794 (libgcc_cmp_return_mode): Likewise.
18795 (libgcc_shift_count_mode): Likewise.
18796 (unwind_word_mode): Likewise.
18797 (addr_space.pointer_mode): Likewise.
18798 (addr_space.address_mode): Likewise.
18799 * doc/tm.texi: Regenerate.
18800 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18802 (do_jump): Use scalar_int_mode for local variables.
18803 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18804 rather than a machine_mode.
18805 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18806 (scompare_loc_descriptor_wide): Likewise.
18807 (scompare_loc_descriptor_narrow): Likewise.
18808 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18810 * except.c (sjlj_emit_dispatch_table): Likewise.
18811 (expand_builtin_eh_copy_values): Likewise.
18812 * explow.c (convert_memory_address_addr_space_1): Likewise.
18813 Take a scalar_int_mode rather than a machine_mode.
18814 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18815 than a machine_mode.
18816 (memory_address_addr_space): Use scalar_int_mode for local variables.
18817 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18818 rather than a machine_mode.
18819 * expmed.c (mask_rtx): Likewise.
18820 (init_expmed_one_conv): Likewise.
18821 (expand_mult_highpart_adjust): Likewise.
18822 (extract_high_half): Likewise.
18823 (expmed_mult_highpart_optab): Likewise.
18824 (expmed_mult_highpart): Likewise.
18825 (expand_smod_pow2): Likewise.
18826 (expand_sdiv_pow2): Likewise.
18827 (emit_store_flag_int): Likewise.
18828 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18830 (extract_low_bits): Likewise.
18831 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18833 * expr.c (pieces_addr::adjust): Likewise.
18834 (can_store_by_pieces): Likewise.
18835 (store_by_pieces): Likewise.
18836 (clear_by_pieces_1): Likewise.
18837 (expand_expr_addr_expr_1): Likewise.
18838 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18839 (expand_expr_real_1): Likewise.
18840 (try_casesi): Likewise.
18841 * final.c (shorten_branches): Likewise.
18842 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18843 type of "mode" to machine_mode.
18844 * internal-fn.c (expand_arith_overflow_result_store): Take a
18845 scalar_int_mode rather than a machine_mode.
18846 (expand_mul_overflow): Use scalar_int_mode for local variables.
18847 * loop-doloop.c (doloop_modify): Likewise.
18848 (doloop_optimize): Likewise.
18849 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18850 than a machine_mode.
18851 (expand_doubleword_shift_condmove): Likewise.
18852 (expand_doubleword_shift): Likewise.
18853 (expand_doubleword_clz): Likewise.
18854 (expand_doubleword_popcount): Likewise.
18855 (expand_doubleword_parity): Likewise.
18856 (expand_absneg_bit): Use scalar_int_mode for local variables.
18857 (prepare_float_lib_cmp): Likewise.
18858 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18859 rather than a machine_mode.
18860 (convert_memory_address_addr_space): Likewise.
18861 (get_mode_bounds): Likewise.
18862 (get_address_mode): Return a scalar_int_mode rather than a
18864 * rtlanal.c (get_address_mode): Likewise.
18865 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18866 than a machine_mode.
18867 * targhooks.c (default_mode_rep_extended): Likewise.
18868 (default_valid_pointer_mode): Likewise.
18869 (default_addr_space_valid_pointer_mode): Likewise.
18870 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18871 than a machine_mode.
18872 (default_libgcc_cmp_return_mode): Likewise.
18873 (default_libgcc_shift_count_mode): Likewise.
18874 (default_unwind_word_mode): Likewise.
18875 (default_addr_space_pointer_mode): Likewise.
18876 (default_addr_space_address_mode): Likewise.
18877 * targhooks.h (default_eh_return_filter_mode): Likewise.
18878 (default_libgcc_cmp_return_mode): Likewise.
18879 (default_libgcc_shift_count_mode): Likewise.
18880 (default_unwind_word_mode): Likewise.
18881 (default_addr_space_pointer_mode): Likewise.
18882 (default_addr_space_address_mode): Likewise.
18883 (default_mode_rep_extended): Take a scalar_int_mode rather than
18885 (default_valid_pointer_mode): Likewise.
18886 (default_addr_space_valid_pointer_mode): Likewise.
18887 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18889 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18890 rather than a machine_mode.
18891 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18892 for local variables.
18893 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18894 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18895 than a machine_mode.
18897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18898 Alan Hayward <alan.hayward@arm.com>
18899 David Sherwood <david.sherwood@arm.com>
18901 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18902 the mode argument to scalar_int_mode.
18903 (do_jump_by_parts_zero_rtx): Likewise.
18904 (do_jump_by_parts_equality_rtx): Likewise.
18905 (do_jump_by_parts_greater): Take a mode argument.
18906 (do_jump_by_parts_equality): Likewise.
18907 (do_jump_1): Update calls accordingly.
18909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18910 Alan Hayward <alan.hayward@arm.com>
18911 David Sherwood <david.sherwood@arm.com>
18913 * is-a.h (safe_dyn_cast): New function.
18914 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18915 (jump_table_for_label): Likewise.
18916 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18917 instead of an rtx_insn *.
18918 (shorten_branches): Use dyn_cast instead of LABEL_P and
18919 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18920 rtx_jump_table_data::get_data_mode.
18921 (final_scan_insn): Likewise.
18923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18924 Alan Hayward <alan.hayward@arm.com>
18925 David Sherwood <david.sherwood@arm.com>
18927 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18928 trying to combine a full-register integer set with a subreg
18931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18932 Alan Hayward <alan.hayward@arm.com>
18933 David Sherwood <david.sherwood@arm.com>
18935 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18936 that is always either address_mode or pointer_mode.
18938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18939 Alan Hayward <alan.hayward@arm.com>
18940 David Sherwood <david.sherwood@arm.com>
18942 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18943 when the two are known to be equal.
18945 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18946 Alan Hayward <alan.hayward@arm.com>
18947 David Sherwood <david.sherwood@arm.com>
18949 * simplify-rtx.c (simplify_const_unary_operation): Use
18950 is_a <scalar_int_mode> instead of checking for a nonzero
18951 precision. Forcibly convert op_mode to a scalar_int_mode
18952 in that case. More clearly differentiate the operand and
18953 result modes and use the former when deciding what the value
18954 of a count-bits operation should be. Use is_int_mode instead
18955 of checking for a MODE_INT. Remove redundant check for whether
18956 this mode has a zero precision.
18958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18959 Alan Hayward <alan.hayward@arm.com>
18960 David Sherwood <david.sherwood@arm.com>
18962 * optabs.c (widen_leading): Change the type of the mode argument
18963 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18964 (widen_bswap): Likewise.
18965 (expand_parity): Likewise.
18966 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18967 (expand_ffs): Likewise.
18968 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18977 (expand_expr_real_1): Update call accordingly.
18979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18980 Alan Hayward <alan.hayward@arm.com>
18981 David Sherwood <david.sherwood@arm.com>
18983 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18984 value_mode arguments. Use scalar_int_mode internally.
18985 (store_bit_field_1): Rename the new integer mode from imode
18986 to op0_mode and use it instead of GET_MODE (op0). Update calls
18987 to store_split_bit_field, store_bit_field_using_insv and
18988 store_fixed_bit_field.
18989 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18990 Use scalar_int_mode internally. Use a bit count rather than a mode
18991 when calculating the largest bit size for get_best_mode.
18992 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18993 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18994 Remove assertion that OP0 has a scalar integer mode.
18995 (store_split_bit_field): Add op0_mode and value_mode arguments.
18996 Update calls to extract_fixed_bit_field.
18997 (extract_bit_field_using_extv): Add an op0_mode argument.
18998 Use scalar_int_mode internally.
18999 (extract_bit_field_1): Rename the new integer mode from imode to
19000 op0_mode and use it instead of GET_MODE (op0). Update calls to
19001 extract_split_bit_field, extract_bit_field_using_extv and
19002 extract_fixed_bit_field.
19003 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
19004 to extract_split_bit_field and extract_fixed_bit_field_1.
19005 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
19006 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
19007 on the target mode.
19008 (extract_split_bit_field): Add an op0_mode argument. Update call
19009 to extract_fixed_bit_field.
19011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19012 Alan Hayward <alan.hayward@arm.com>
19013 David Sherwood <david.sherwood@arm.com>
19015 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
19016 * explow.c (hard_function_value): Likewise.
19017 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
19018 convert_to_mode call outside the loop.
19019 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
19020 for the mode iterator. Require the mode specified by max_pieces
19022 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
19024 (copy_blkmode_to_reg): Likewise.
19025 (set_storage_via_setmem): Likewise.
19026 * optabs.c (prepare_cmp_insn): Likewise.
19027 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19028 * stor-layout.c (finish_bitfield_representative): Likewise.
19030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19031 Alan Hayward <alan.hayward@arm.com>
19032 David Sherwood <david.sherwood@arm.com>
19034 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
19035 * expr.c (convert_move): Use them.
19036 (convert_modes): Likewise.
19037 (store_expr_with_bounds): Likewise.
19039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19040 Alan Hayward <alan.hayward@arm.com>
19041 David Sherwood <david.sherwood@arm.com>
19043 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
19044 parameter for the mode of "x". Remove the "known_x", "known_mode"
19045 and "known_ret" arguments. Change the type of the mode argument
19046 to scalar_int_mode.
19047 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
19048 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
19049 (reg_num_sign_bit_copies_for_combine): Likewise.
19050 * rtlanal.c (nonzero_bits1): Likewise.
19051 (num_sign_bit_copies1): Likewise.
19052 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
19053 (reg_num_sign_bit_copies_general): Likewise.
19054 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
19055 (reg_nonzero_bits_general): Likewise.
19057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19058 Alan Hayward <alan.hayward@arm.com>
19059 David Sherwood <david.sherwood@arm.com>
19061 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
19062 than in subroutines. Return 1 for non-integer modes.
19063 (cached_num_sign_bit_copies): Change the type of the mode parameter
19064 to scalar_int_mode.
19065 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
19066 classes. Handle CONST_INT_P first and then check whether X also
19067 has a scalar integer mode. Check the same thing for inner registers
19068 of a SUBREG and for values that are being extended or truncated.
19070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19071 Alan Hayward <alan.hayward@arm.com>
19072 David Sherwood <david.sherwood@arm.com>
19074 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
19075 in subroutines. Return the mode mask for non-integer modes.
19076 (cached_nonzero_bits): Change the type of the mode parameter
19077 to scalar_int_mode.
19078 (nonzero_bits1): Likewise. Remove early exit for other mode
19079 classes. Handle CONST_INT_P first and then check whether X
19080 also has a scalar integer mode.
19082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19083 Alan Hayward <alan.hayward@arm.com>
19084 David Sherwood <david.sherwood@arm.com>
19086 * expr.c (widest_int_mode_for_size): Make the comment match the code.
19087 Return a scalar_int_mode and assert that the size is greater than
19089 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
19090 (op_by_pieces_d::op_by_pieces_d): Likewise.
19091 (op_by_pieces_d::run): Likewise.
19092 (can_store_by_pieces): Likewise.
19094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19095 Alan Hayward <alan.hayward@arm.com>
19096 David Sherwood <david.sherwood@arm.com>
19098 * combine.c (extract_left_shift): Add a mode argument and update
19100 (make_compound_operation_int): Change the type of the mode parameter
19101 to scalar_int_mode and update the call to extract_left_shift.
19103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19104 Alan Hayward <alan.hayward@arm.com>
19105 David Sherwood <david.sherwood@arm.com>
19107 * combine.c (simplify_and_const_int): Change the type of the mode
19108 parameter to scalar_int_mode.
19109 (simplify_and_const_int_1): Likewise. Update recursive call.
19111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19112 Alan Hayward <alan.hayward@arm.com>
19113 David Sherwood <david.sherwood@arm.com>
19115 * combine.c (simplify_compare_const): Check that the mode is a
19116 scalar_int_mode (rather than VOIDmode) before testing its
19118 (simplify_comparison): Move COMPARISON_P handling out of the
19119 loop and restrict the latter part of the loop to scalar_int_modes.
19120 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
19121 and when considering SUBREG_REGs. Use is_int_mode instead of
19122 checking GET_MODE_CLASS against MODE_INT.
19124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19125 Alan Hayward <alan.hayward@arm.com>
19126 David Sherwood <david.sherwood@arm.com>
19128 * combine.c (try_widen_shift_mode): Move check for equal modes to...
19129 (simplify_shift_const_1): ...here. Use scalar_int_mode for
19130 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
19137 (force_to_mode): ...here. Keep xmode up-to-date and use it
19138 instead of GET_MODE (x).
19140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19141 Alan Hayward <alan.hayward@arm.com>
19142 David Sherwood <david.sherwood@arm.com>
19144 * optabs-query.h (extraction_insn::struct_mode): Change type to
19145 opt_scalar_int_mode and update comment.
19146 (extraction_insn::field_mode): Change type to scalar_int_mode.
19147 (extraction_insn::pos_mode): Likewise.
19148 * combine.c (make_extraction): Update accordingly.
19149 * optabs-query.c (get_traditional_extraction_insn): Likewise.
19150 (get_optab_extraction_insn): Likewise.
19151 * recog.c (simplify_while_replacing): Likewise.
19152 * expmed.c (narrow_bit_field_mem): Change the type of the mode
19153 parameter to opt_scalar_int_mode.
19155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19156 Alan Hayward <alan.hayward@arm.com>
19157 David Sherwood <david.sherwood@arm.com>
19159 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
19160 to a scalar_int_mode instead of a machine_mode.
19161 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19162 (get_best_mode): Return a boolean and use a pointer argument to store
19163 the selected mode. Replace the limit mode parameter with a bit limit.
19164 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19165 for the values returned by bit_field_mode_iterator::next_mode.
19166 (store_bit_field): Update call to get_best_mode.
19167 (store_fixed_bit_field): Likewise.
19168 (extract_fixed_bit_field): Likewise.
19169 * expr.c (optimize_bitfield_assignment_op): Likewise.
19170 * fold-const.c (optimize_bit_field_compare): Likewise.
19171 (fold_truth_andor_1): Likewise.
19172 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19173 Update for new type of m_mode.
19174 (get_best_mode): As above.
19176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19177 Alan Hayward <alan.hayward@arm.com>
19178 David Sherwood <david.sherwood@arm.com>
19180 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19181 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19182 (store_bit_field): Check is_a <scalar_int_mode> before calling
19183 strict_volatile_bitfield_p.
19184 (extract_bit_field): Likewise.
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 * target.def (cstore_mode): Return a scalar_int_mode.
19191 * doc/tm.texi: Regenerate.
19192 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19193 * targhooks.h (default_cstore_mode): Likewise.
19194 * targhooks.c (default_cstore_mode): Likewise, using a forced
19196 * expmed.c (emit_cstore): Expect the target of the cstore to be
19199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19200 Alan Hayward <alan.hayward@arm.com>
19201 David Sherwood <david.sherwood@arm.com>
19203 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19205 (niter_desc): Likewise mode.
19206 (iv_analyze): Add a mode parameter.
19208 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19209 and change its type to scalar_int_mode.
19210 * loop-iv.c: Update commentary at head of file.
19211 (iv_constant): Pass the mode paraeter before the rtx it describes
19212 and change its type to scalar_int_mode. Remove VOIDmode handling.
19213 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19214 (iv_extend): Likewise.
19215 (shorten_into_mode): Likewise.
19216 (iv_add): Use scalar_int_mode.
19217 (iv_mult): Likewise.
19218 (iv_shift): Likewise.
19219 (canonicalize_iv_subregs): Likewise.
19220 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19221 it describes and change its mode to scalar_int_mode. Also change
19222 the type of the returned inner_mode to scalar_int_mode.
19223 (get_biv_step): Likewise, turning outer_mode from a pointer
19224 into a direct parameter. Update call to get_biv_step_1.
19225 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19226 iv_constant and get_biv_step.
19227 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19228 and change its type to scalar_int_mode. Don't initialise iv->mode
19229 to VOIDmode and remove later checks for its still being VOIDmode.
19230 Update calls to iv_analyze_op and iv_analyze_expr. Check
19231 is_a <scalar_int_mode> when changing the mode under consideration.
19232 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19233 Update call to iv_analyze_expr.
19234 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19235 inner register is not also a scalar_int_mode. Update call to
19237 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19238 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19239 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19240 separate mode class checks. Update calls to iv_analyze. Remove
19241 fix-up of VOIDmodes after iv_analyze_biv.
19242 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19243 don't have a scalar_int_mode. Update call to biv_p.
19245 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19246 Alan Hayward <alan.hayward@arm.com>
19247 David Sherwood <david.sherwood@arm.com>
19249 * cfgexpand.c (convert_debug_memory_address): Use
19250 as_a <scalar_int_mode>.
19251 * combine.c (expand_compound_operation): Likewise.
19252 (make_extraction): Likewise.
19253 (change_zero_ext): Likewise.
19254 (simplify_comparison): Likewise.
19255 * cse.c (cse_insn): Likewise.
19256 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19257 (mem_loc_descriptor): Likewise.
19258 (loc_descriptor): Likewise.
19259 * expmed.c (init_expmed_one_mode): Likewise.
19260 (synth_mult): Likewise.
19261 (emit_store_flag_1): Likewise.
19262 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19263 of a comparison with size.
19264 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19265 (reduce_to_bit_field_precision): Likewise.
19266 * function.c (expand_function_end): Likewise.
19267 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19268 * loop-doloop.c (doloop_modify): Likewise.
19269 * optabs.c (expand_binop): Likewise.
19270 (expand_unop): Likewise.
19271 (expand_copysign_absneg): Likewise.
19272 (prepare_cmp_insn): Likewise.
19273 (maybe_legitimize_operand): Likewise.
19274 * recog.c (const_scalar_int_operand): Likewise.
19275 * rtlanal.c (get_address_mode): Likewise.
19276 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19277 (simplify_cond_clz_ctz): Likewise.
19278 * tree-nested.c (get_nl_goto_field): Likewise.
19279 * tree.c (build_vector_type_for_mode): Likewise.
19280 * var-tracking.c (use_narrower_mode): Likewise.
19282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19283 Alan Hayward <alan.hayward@arm.com>
19284 David Sherwood <david.sherwood@arm.com>
19286 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19287 * builtins.c (expand_builtin_signbit): Use it.
19288 * cfgexpand.c (expand_debug_expr): Likewise.
19289 * dojump.c (do_jump): Likewise.
19290 (do_compare_and_jump): Likewise.
19291 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19292 * expmed.c (make_tree): Likewise.
19293 * expr.c (expand_expr_real_2): Likewise.
19294 (expand_expr_real_1): Likewise.
19295 (try_casesi): Likewise.
19296 * fold-const-call.c (fold_const_call_ss): Likewise.
19297 * fold-const.c (unextend): Likewise.
19298 (extract_muldiv_1): Likewise.
19299 (fold_single_bit_test): Likewise.
19300 (native_encode_int): Likewise.
19301 (native_encode_string): Likewise.
19302 (native_interpret_int): Likewise.
19303 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19304 * internal-fn.c (expand_addsub_overflow): Likewise.
19305 (expand_neg_overflow): Likewise.
19306 (expand_mul_overflow): Likewise.
19307 (expand_arith_overflow): Likewise.
19308 * match.pd: Likewise.
19309 * stor-layout.c (layout_type): Likewise.
19310 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19311 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19312 * tree-ssanames.c (get_range_info): Likewise.
19313 * tree-switch-conversion.c (array_value_type) Likewise.
19314 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19315 (vect_recog_divmod_pattern): Likewise.
19316 (vect_recog_mixed_size_cond_pattern): Likewise.
19317 * tree-vrp.c (extract_range_basic): Likewise.
19318 (simplify_float_conversion_using_ranges): Likewise.
19319 * tree.c (int_fits_type_p): Likewise.
19320 * ubsan.c (instrument_bool_enum_load): Likewise.
19321 * varasm.c (mergeable_string_section): Likewise.
19322 (narrowing_initializer_constant_valid_p): Likewise.
19323 (output_constant): Likewise.
19325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19326 Alan Hayward <alan.hayward@arm.com>
19327 David Sherwood <david.sherwood@arm.com>
19329 * machmode.h (NARROWEST_INT_MODE): New macro.
19330 * expr.c (alignment_for_piecewise_move): Use it instead of
19331 GET_CLASS_NARROWEST_MODE (MODE_INT).
19332 (push_block): Likewise.
19333 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19335 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19338 Alan Hayward <alan.hayward@arm.com>
19339 David Sherwood <david.sherwood@arm.com>
19341 * postreload.c (move2add_valid_value_p): Change the type of the
19342 mode parameter to scalar_int_mode.
19343 (move2add_use_add2_insn): Add a mode parameter and use it instead
19345 (move2add_use_add3_insn): Likewise.
19346 (reload_cse_move2add): Update accordingly.
19348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19349 Alan Hayward <alan.hayward@arm.com>
19350 David Sherwood <david.sherwood@arm.com>
19352 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19354 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19355 * optabs.c (expand_unop): Likewise.
19357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19358 Alan Hayward <alan.hayward@arm.com>
19359 David Sherwood <david.sherwood@arm.com>
19361 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19362 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19363 (popcount_loc_descriptor): Likewise.
19364 (bswap_loc_descriptor): Likewise.
19365 (rotate_loc_descriptor): Likewise.
19366 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19367 calling the functions above.
19369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19370 Alan Hayward <alan.hayward@arm.com>
19371 David Sherwood <david.sherwood@arm.com>
19373 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19375 (try_combine): Likewise.
19376 (simplify_if_then_else): Likewise.
19377 * cse.c (cse_insn): Likewise.
19378 * dwarf2out.c (mem_loc_descriptor): Likewise.
19379 * emit-rtl.c (gen_lowpart_common): Likewise.
19380 * simplify-rtx.c (simplify_truncation): Likewise.
19381 (simplify_binary_operation_1): Likewise.
19382 (simplify_const_relational_operation): Likewise.
19383 (simplify_ternary_operation): Likewise.
19384 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19387 Alan Hayward <alan.hayward@arm.com>
19388 David Sherwood <david.sherwood@arm.com>
19390 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19391 * reload.c (push_reload): Likewise.
19392 (find_reloads): Likewise.
19394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19395 Alan Hayward <alan.hayward@arm.com>
19396 David Sherwood <david.sherwood@arm.com>
19398 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19399 (make_compound_operation_int): Likewise.
19400 (change_zero_ext): Likewise.
19401 * expr.c (convert_move): Likewise.
19402 (convert_modes): Likewise.
19403 * fwprop.c (forward_propagate_subreg): Likewise.
19404 * loop-iv.c (get_biv_step_1): Likewise.
19405 * optabs.c (widen_operand): Likewise.
19406 * postreload.c (move2add_valid_value_p): Likewise.
19407 * recog.c (simplify_while_replacing): Likewise.
19408 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19409 (simplify_binary_operation_1): Likewise. Remove redundant
19410 mode equality check.
19412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19413 Alan Hayward <alan.hayward@arm.com>
19414 David Sherwood <david.sherwood@arm.com>
19416 * combine.c (combine_simplify_rtx): Add checks for
19417 is_a <scalar_int_mode>.
19418 (simplify_if_then_else): Likewise.
19419 (make_field_assignment): Likewise.
19420 (simplify_comparison): Likewise.
19421 * ifcvt.c (noce_try_bitop): Likewise.
19422 * loop-invariant.c (canonicalize_address_mult): Likewise.
19423 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19426 Alan Hayward <alan.hayward@arm.com>
19427 David Sherwood <david.sherwood@arm.com>
19429 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19430 is_a <scalar_int_mode> instead of != BLKmode.
19432 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19433 Alan Hayward <alan.hayward@arm.com>
19434 David Sherwood <david.sherwood@arm.com>
19436 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19437 instead of != VOIDmode.
19438 * combine.c (if_then_else_cond): Likewise.
19439 (change_zero_ext): Likewise.
19440 * dwarf2out.c (mem_loc_descriptor): Likewise.
19441 (loc_descriptor): Likewise.
19442 * rtlanal.c (canonicalize_condition): Likewise.
19443 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19446 Alan Hayward <alan.hayward@arm.com>
19447 David Sherwood <david.sherwood@arm.com>
19449 * simplify-rtx.c (simplify_binary_operation_1): Use
19450 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19453 Alan Hayward <alan.hayward@arm.com>
19454 David Sherwood <david.sherwood@arm.com>
19456 * wide-int.h (int_traits<unsigned char>) New class.
19457 (int_traits<unsigned short>) Likewise.
19458 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19459 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19461 * combine.c (set_nonzero_bits_and_sign_copies): Use
19462 is_a <scalar_int_mode>.
19463 (find_split_point): Likewise.
19464 (combine_simplify_rtx): Likewise.
19465 (simplify_logical): Likewise.
19466 (expand_compound_operation): Likewise.
19467 (expand_field_assignment): Likewise.
19468 (make_compound_operation): Likewise.
19469 (extended_count): Likewise.
19470 (change_zero_ext): Likewise.
19471 (simplify_comparison): Likewise.
19472 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19473 (ucompare_loc_descriptor): Likewise.
19474 (minmax_loc_descriptor): Likewise.
19475 (mem_loc_descriptor): Likewise.
19476 (loc_descriptor): Likewise.
19477 * expmed.c (init_expmed_one_mode): Likewise.
19478 * lra-constraints.c (lra_constraint_offset): Likewise.
19479 * optabs.c (prepare_libcall_arg): Likewise.
19480 * postreload.c (move2add_note_store): Likewise.
19481 * reload.c (operands_match_p): Likewise.
19482 * rtl.h (load_extend_op): Likewise.
19483 * rtlhooks.c (gen_lowpart_general): Likewise.
19484 * simplify-rtx.c (simplify_truncation): Likewise.
19485 (simplify_unary_operation_1): Likewise.
19486 (simplify_binary_operation_1): Likewise.
19487 (simplify_const_binary_operation): Likewise.
19488 (simplify_const_relational_operation): Likewise.
19489 (simplify_subreg): Likewise.
19490 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19491 * var-tracking.c (adjust_mems): Likewise.
19492 (prepare_call_arguments): Likewise.
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 * machmode.h (is_int_mode): New fuction.
19499 * combine.c (find_split_point): Use it.
19500 (combine_simplify_rtx): Likewise.
19501 (simplify_if_then_else): Likewise.
19502 (simplify_set): Likewise.
19503 (simplify_shift_const_1): Likewise.
19504 (simplify_comparison): Likewise.
19505 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19506 * cse.c (notreg_cost): Likewise.
19507 (cse_insn): Likewise.
19508 * cselib.c (cselib_lookup_1): Likewise.
19509 * dojump.c (do_jump_1): Likewise.
19510 (do_compare_rtx_and_jump): Likewise.
19511 * dse.c (get_call_args): Likewise.
19512 * dwarf2out.c (rtl_for_decl_init): Likewise.
19513 (native_encode_initializer): Likewise.
19514 * expmed.c (emit_store_flag_1): Likewise.
19515 (emit_store_flag): Likewise.
19516 * expr.c (convert_modes): Likewise.
19517 (store_field): Likewise.
19518 (expand_expr_real_1): Likewise.
19519 * fold-const.c (fold_read_from_constant_string): Likewise.
19520 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19521 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19522 * optabs.c (expand_binop): Likewise.
19523 (expand_unop): Likewise.
19524 (expand_abs_nojump): Likewise.
19525 (expand_one_cmpl_abs_nojump): Likewise.
19526 * simplify-rtx.c (mode_signbit_p): Likewise.
19527 (val_signbit_p): Likewise.
19528 (val_signbit_known_set_p): Likewise.
19529 (val_signbit_known_clear_p): Likewise.
19530 (simplify_relational_operation_1): Likewise.
19531 * tree.c (vector_type_mode): Likewise.
19533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19534 Alan Hayward <alan.hayward@arm.com>
19535 David Sherwood <david.sherwood@arm.com>
19537 * machmode.h (smallest_mode_for_size): Fix formatting.
19538 (smallest_int_mode_for_size): New function.
19539 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19540 instead of smallest_mode_for_size.
19541 * combine.c (make_extraction): Likewise.
19542 * config/arc/arc.c (arc_expand_movmem): Likewise.
19543 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19544 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19545 * config/s390/s390.c (s390_expand_insv): Likewise.
19546 * config/sparc/sparc.c (assign_int_registers): Likewise.
19547 * config/spu/spu.c (spu_function_value): Likewise.
19548 (spu_function_arg): Likewise.
19549 * coverage.c (get_gcov_type): Likewise.
19550 (get_gcov_unsigned_t): Likewise.
19551 * dse.c (find_shift_sequence): Likewise.
19552 * expmed.c (store_bit_field_1): Likewise.
19553 * expr.c (convert_move): Likewise.
19554 (store_field): Likewise.
19555 * internal-fn.c (expand_arith_overflow): Likewise.
19556 * optabs-query.c (get_best_extraction_insn): Likewise.
19557 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19558 * stor-layout.c (layout_type): Likewise.
19559 (initialize_sizetypes): Likewise.
19560 * targhooks.c (default_get_mask_mode): Likewise.
19561 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19564 Alan Hayward <alan.hayward@arm.com>
19565 David Sherwood <david.sherwood@arm.com>
19567 * machmode.h (opt_mode::else_blk): New function.
19568 (int_mode_for_mode): Declare.
19569 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19570 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19572 * cfgexpand.c (expand_debug_expr): Likewise.
19573 * combine.c (gen_lowpart_or_truncate): Likewise.
19574 (gen_lowpart_for_combine): Likewise.
19575 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19576 * config/avr/avr.c (avr_to_int_mode): Likewise.
19577 (avr_out_plus_1): Likewise.
19578 (avr_out_plus): Likewise.
19579 (avr_out_round): Likewise.
19580 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19581 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19582 (s390_expand_vcond): Likewise.
19583 * config/spu/spu.c (spu_split_immediate): Likewise.
19584 (spu_expand_mov): Likewise.
19585 * dse.c (get_stored_val): Likewise.
19586 * expmed.c (store_bit_field_1): Likewise.
19587 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19589 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19590 (extract_low_bits): Likewise.
19591 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19592 handling rather than repeating the check.
19593 (emit_group_store): Likewise.
19594 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19595 * optabs.c (expand_absneg_bit): Likewise.
19596 (expand_copysign_absneg): Likewise.
19597 (expand_copysign_bit): Likewise.
19598 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19599 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19600 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19601 * var-tracking.c (prepare_call_arguments): Likewise.
19602 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19603 int_mode_for_mode instead of mode_for_size.
19604 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19607 Alan Hayward <alan.hayward@arm.com>
19608 David Sherwood <david.sherwood@arm.com>
19610 * machmode.h (int_mode_for_size): New function.
19611 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19612 instead of mode_for_size.
19613 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19615 * combine.c (expand_field_assignment): Use int_mode_for_size
19616 instead of mode_for_size.
19617 (make_extraction): Likewise.
19618 (simplify_shift_const_1): Likewise.
19619 (simplify_comparison): Likewise.
19620 * dojump.c (do_jump): Likewise.
19621 * dwarf2out.c (mem_loc_descriptor): Likewise.
19622 * emit-rtl.c (init_derived_machine_modes): Likewise.
19623 * expmed.c (flip_storage_order): Likewise.
19624 (convert_extracted_bit_field): Likewise.
19625 * expr.c (copy_blkmode_from_reg): Likewise.
19626 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19627 * internal-fn.c (expand_mul_overflow): Likewise.
19628 * lower-subreg.c (simple_move): Likewise.
19629 * optabs-libfuncs.c (init_optabs): Likewise.
19630 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19631 * tree.c (vector_type_mode): Likewise.
19632 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19633 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19634 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19635 * tree-vect-stmts.c (vectorizable_load): Likewise.
19636 (vectorizable_store): Likewise.
19638 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19639 Alan Hayward <alan.hayward@arm.com>
19640 David Sherwood <david.sherwood@arm.com>
19642 * coretypes.h (pod_mode): New type.
19643 (scalar_int_mode_pod): New typedef.
19644 * machmode.h (pod_mode): New class.
19645 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19646 * genmodes.c (emit_mode_int_n): Update accordingly.
19647 * lower-subreg.h (target_lower_subreg): Change type to
19648 scalar_int_mode_pod.
19649 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19650 scalar_int_mode_pod.
19652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19653 Alan Hayward <alan.hayward@arm.com>
19654 David Sherwood <david.sherwood@arm.com>
19656 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19657 machine_mode to scalar_int_mode.
19658 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19659 (rs6000_option_override_internal): Remove cast to int.
19660 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19661 machine_mode to scalar_int_mode.
19662 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19663 (rs6000_option_override_internal): Remove cast to int.
19664 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19665 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19667 * config/s390/s390.c (s390_expand_builtin): Likewise.
19668 * coretypes.h (scalar_int_mode): New type.
19669 (opt_scalar_int_mode): New typedef.
19670 * machmode.h (scalar_int_mode): New class.
19671 (scalar_int_mode::includes_p): New function.
19672 (byte_mode): Change type to scalar_int_mode.
19673 (word_mode): Likewise.
19674 (ptr_mode): Likewise.
19675 * emit-rtl.c (byte_mode): Likewise.
19676 (word_mode): Likewise.
19677 (ptr_mode): Likewise.
19678 (init_derived_machine_modes): Update accordingly.
19679 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19680 and MODE_PARTIAL_INT.
19681 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19682 opt_scalar_int_mode.
19684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19685 Alan Hayward <alan.hayward@arm.com>
19686 David Sherwood <david.sherwood@arm.com>
19688 * target.def (libgcc_floating_mode_supported_p): Take a
19690 * doc/tm.texi: Regenerate.
19691 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19693 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19694 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19698 Alan Hayward <alan.hayward@arm.com>
19699 David Sherwood <david.sherwood@arm.com>
19701 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19702 * doc/tm.texi: Regenerate.
19703 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19704 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19705 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19706 * targhooks.h (default_floatn_mode): Likewise.
19707 * targhooks.c (default_floatn_mode): Likewise.
19708 * tree.c (build_common_tree_nodes): Update accordingly.
19710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19711 Alan Hayward <alan.hayward@arm.com>
19712 David Sherwood <david.sherwood@arm.com>
19714 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19715 (mode_iterator::iterate_p): Likewise.
19716 (mode_iterator::get_wider): Likewise.
19717 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19720 Alan Hayward <alan.hayward@arm.com>
19721 David Sherwood <david.sherwood@arm.com>
19723 * coretypes.h (opt_scalar_float_mode): New typedef.
19724 * machmode.h (float_mode_for_size): New function.
19725 * emit-rtl.c (double_mode): Delete.
19726 (init_emit_once): Use float_mode_for_size.
19727 * stor-layout.c (layout_type): Likewise.
19728 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19731 Alan Hayward <alan.hayward@arm.com>
19732 David Sherwood <david.sherwood@arm.com>
19734 * output.h (assemble_real): Take a scalar_float_mode.
19735 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19736 * config/arm/arm.md (consttable_4): Likewise.
19737 (consttable_8): Likewise.
19738 (consttable_16): Likewise.
19739 * config/mips/mips.md (consttable_float): Likewise.
19740 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19741 * varasm.c (assemble_real): Take a scalar_float_mode.
19742 (output_constant_pool_2): Update accordingly.
19743 (output_constant): Likewise.
19745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19746 Alan Hayward <alan.hayward@arm.com>
19747 David Sherwood <david.sherwood@arm.com>
19749 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19750 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19751 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19752 (native_encode_real): Likewise.
19753 (native_interpret_real): Likewise.
19754 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19755 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19758 Alan Hayward <alan.hayward@arm.com>
19759 David Sherwood <david.sherwood@arm.com>
19761 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19762 <scalar_float_mode>. Simplify.
19763 (gen_extend_conv_libfunc): Likewise.
19765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19766 Alan Hayward <alan.hayward@arm.com>
19767 David Sherwood <david.sherwood@arm.com>
19769 * coretypes.h (scalar_float_mode): New type.
19770 * machmode.h (mode_traits::from_int): Use machine_mode if
19771 USE_ENUM_MODES is defined.
19772 (is_a): New function.
19774 (dyn_cast): Likewise.
19775 (scalar_float_mode): New class.
19776 (scalar_float_mode::includes_p): New function.
19777 (is_float_mode): Likewise.
19778 * gdbhooks.py (MachineModePrinter): New class.
19779 (build_pretty_printer): Use it for scalar_float_mode.
19780 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19781 (format_helper::format_helper): Turn into a template.
19782 * genmodes.c (get_mode_class): New function.
19783 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19784 or machine_mode if none.
19785 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19786 as_a <scalar_float_mode>.
19787 * dwarf2out.c (mem_loc_descriptor): Likewise.
19788 (insert_float): Likewise.
19789 (add_const_value_attribute): Likewise.
19790 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19791 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19792 (expand_unop): Update accordingly.
19793 (expand_abs_nojump): Likewise.
19794 (expand_copysign_absneg): Take a scalar_float_mode.
19795 (expand_copysign_bit): Likewise.
19796 (expand_copysign): Update accordingly.
19798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19799 Alan Hayward <alan.hayward@arm.com>
19800 David Sherwood <david.sherwood@arm.com>
19802 * coretypes.h (opt_mode): New class.
19803 * machmode.h (opt_mode): Likewise.
19804 (opt_mode::else_void): New function.
19805 (opt_mode::require): Likewise.
19806 (opt_mode::exists): Likewise.
19807 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19808 (GET_MODE_2XWIDER_MODE): Likewise.
19809 (mode_iterator::get_wider): Update accordingly.
19810 (mode_iterator::get_2xwider): Likewise.
19811 (mode_iterator::get_known_wider): Likewise, turning into a template.
19812 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19813 forcing a wider mode to exist.
19814 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19815 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19816 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19817 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19818 * lower-subreg.c (init_lower_subreg): Likewise.
19819 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19820 on the final iteration.
19821 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19822 a wider mode exists before asking for a move pattern.
19823 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19824 forcing a wider mode to exist.
19825 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19826 returning false if no such mode exists.
19827 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19828 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19829 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19830 Avoid checking for a MODE_INT if we already know the mode is not a
19832 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19833 forcing a wider mode to exist.
19834 (expmed_mult_highpart_optab): Likewise.
19835 (expmed_mult_highpart): Likewise.
19836 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19838 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19839 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19840 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19841 * internal-fn.c (expand_mul_overflow): Update use of
19842 GET_MODE_2XWIDER_MODE.
19843 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19844 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19845 GET_MODE_WIDER_MODE.
19846 (convert_plusminus_to_widen): Likewise.
19847 * tree-switch-conversion.c (array_value_type): Likewise.
19848 * var-tracking.c (emit_note_insn_var_location): Likewise.
19849 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19850 Return false inside rather than outside the loop if no wider mode
19852 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19853 and GET_MODE_2XWIDER_MODE
19854 (can_compare_p): Use else_void.
19855 * gdbhooks.py (OptMachineModePrinter): New class.
19856 (build_pretty_printer): Use it for opt_mode.
19858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19859 Alan Hayward <alan.hayward@arm.com>
19860 David Sherwood <david.sherwood@arm.com>
19862 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19863 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19866 Alan Hayward <alan.hayward@arm.com>
19867 David Sherwood <david.sherwood@arm.com>
19869 * machmode.h (mode_traits): New structure.
19870 (get_narrowest_mode): New function.
19871 (mode_iterator::start): Likewise.
19872 (mode_iterator::iterate_p): Likewise.
19873 (mode_iterator::get_wider): Likewise.
19874 (mode_iterator::get_known_wider): Likewise.
19875 (mode_iterator::get_2xwider): Likewise.
19876 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19877 (FOR_EACH_MODE): Likewise.
19878 (FOR_EACH_MODE_FROM): Likewise.
19879 (FOR_EACH_MODE_UNTIL): Likewise.
19880 (FOR_EACH_WIDER_MODE): Likewise.
19881 (FOR_EACH_2XWIDER_MODE): Likewise.
19882 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19883 * combine.c (simplify_comparison): Likewise
19884 * config/i386/i386.c (type_natural_mode): Likewise.
19885 * cse.c (cse_insn): Likewise.
19886 * dse.c (find_shift_sequence): Likewise.
19887 * emit-rtl.c (init_derived_machine_modes): Likewise.
19888 (init_emit_once): Likewise.
19889 * explow.c (hard_function_value): Likewise.
19890 * expmed.c (extract_fixed_bit_field_1): Likewise.
19891 (extract_bit_field_1): Likewise.
19892 (expand_divmod): Likewise.
19893 (emit_store_flag_1): Likewise.
19894 * expr.c (init_expr_target): Likewise.
19895 (convert_move): Likewise.
19896 (alignment_for_piecewise_move): Likewise.
19897 (widest_int_mode_for_size): Likewise.
19898 (emit_block_move_via_movmem): Likewise.
19899 (copy_blkmode_to_reg): Likewise.
19900 (set_storage_via_setmem): Likewise.
19901 (compress_float_constant): Likewise.
19902 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19903 * optabs-query.c (get_best_extraction_insn): Likewise.
19904 * optabs.c (expand_binop): Likewise.
19905 (expand_twoval_unop): Likewise.
19906 (expand_twoval_binop): Likewise.
19907 (widen_leading): Likewise.
19908 (widen_bswap): Likewise.
19909 (expand_parity): Likewise.
19910 (expand_unop): Likewise.
19911 (prepare_cmp_insn): Likewise.
19912 (prepare_float_lib_cmp): Likewise.
19913 (expand_float): Likewise.
19914 (expand_fix): Likewise.
19915 (expand_sfix_optab): Likewise.
19916 * postreload.c (move2add_use_add2_insn): Likewise.
19917 * reg-stack.c (reg_to_stack): Likewise.
19918 * reginfo.c (choose_hard_reg_mode): Likewise.
19919 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19920 * stor-layout.c (mode_for_size): Likewise.
19921 (smallest_mode_for_size): Likewise.
19922 (mode_for_vector): Likewise.
19923 (finish_bitfield_representative): Likewise.
19924 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19925 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19926 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19927 * var-tracking.c (prepare_call_arguments): Likewise.
19929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19930 Alan Hayward <alan.hayward@arm.com>
19931 David Sherwood <david.sherwood@arm.com>
19933 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19934 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19935 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19936 * machmode.h (mode_size): Move earlier in file.
19937 (mode_precision): Likewise.
19938 (mode_inner): Likewise.
19939 (mode_nunits): Likewise.
19940 (mode_unit_size): Likewise.
19941 (unit_unit_precision): Likewise.
19942 (mode_wider): Likewise.
19943 (mode_2xwider): Likewise.
19944 (machine_mode): New class.
19945 (mode_to_bytes): New function.
19946 (mode_to_bits): Likewise.
19947 (mode_to_precision): Likewise.
19948 (mode_to_inner): Likewise.
19949 (mode_to_unit_size): Likewise.
19950 (mode_to_unit_precision): Likewise.
19951 (mode_to_nunits): Likewise.
19952 (GET_MODE_SIZE): Use mode_to_bytes.
19953 (GET_MODE_BITSIZE): Use mode_to_bits.
19954 (GET_MODE_PRECISION): Use mode_to_precision.
19955 (GET_MODE_INNER): Use mode_to_inner.
19956 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19957 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19958 (GET_MODE_NUNITS): Use mode_to_nunits.
19959 * system.h (ALWAYS_INLINE): New macro.
19960 * config/powerpcspe/powerpcspe-c.c
19961 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19962 int for arg1_mode and arg2_mode.
19964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19965 Alan Hayward <alan.hayward@arm.com>
19966 David Sherwood <david.sherwood@arm.com>
19968 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19969 Prefix mode names with E_ in case statements.
19970 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19971 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19972 (aarch64_split_simd_move): Likewise.
19973 (aarch64_gen_storewb_pair): Likewise.
19974 (aarch64_gen_loadwb_pair): Likewise.
19975 (aarch64_gen_store_pair): Likewise.
19976 (aarch64_gen_load_pair): Likewise.
19977 (aarch64_get_condition_code_1): Likewise.
19978 (aarch64_constant_pool_reload_icode): Likewise.
19979 (get_rsqrte_type): Likewise.
19980 (get_rsqrts_type): Likewise.
19981 (get_recpe_type): Likewise.
19982 (get_recps_type): Likewise.
19983 (aarch64_gimplify_va_arg_expr): Likewise.
19984 (aarch64_simd_container_mode): Likewise.
19985 (aarch64_emit_load_exclusive): Likewise.
19986 (aarch64_emit_store_exclusive): Likewise.
19987 (aarch64_expand_compare_and_swap): Likewise.
19988 (aarch64_gen_atomic_cas): Likewise.
19989 (aarch64_emit_bic): Likewise.
19990 (aarch64_emit_atomic_swap): Likewise.
19991 (aarch64_emit_atomic_load_op): Likewise.
19992 (aarch64_evpc_trn): Likewise.
19993 (aarch64_evpc_uzp): Likewise.
19994 (aarch64_evpc_zip): Likewise.
19995 (aarch64_evpc_ext): Likewise.
19996 (aarch64_evpc_rev): Likewise.
19997 (aarch64_evpc_dup): Likewise.
19998 (aarch64_gen_ccmp_first): Likewise.
19999 (aarch64_gen_ccmp_next): Likewise.
20000 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
20001 (alpha_emit_xfloating_libcall): Likewise.
20002 (emit_insxl): Likewise.
20003 (alpha_arg_type): Likewise.
20004 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
20005 (arc_preferred_simd_mode): Likewise.
20006 (arc_secondary_reload): Likewise.
20007 (get_arc_condition_code): Likewise.
20008 (arc_print_operand): Likewise.
20009 (arc_legitimate_constant_p): Likewise.
20010 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20011 * config/arc/arc.md (casesi_load): Likewise.
20012 (casesi_compact_jump): Likewise.
20013 * config/arc/predicates.md (proper_comparison_operator): Likewise.
20014 (cc_use_register): Likewise.
20015 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20016 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20017 (arm_init_iwmmxt_builtins): Likewise.
20018 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
20019 (neon_expand_vector_init): Likewise.
20020 (arm_attr_length_move_neon): Likewise.
20021 (maybe_get_arm_condition_code): Likewise.
20022 (arm_emit_vector_const): Likewise.
20023 (arm_preferred_simd_mode): Likewise.
20024 (arm_output_iwmmxt_tinsr): Likewise.
20025 (thumb1_output_casesi): Likewise.
20026 (thumb2_output_casesi): Likewise.
20027 (arm_emit_load_exclusive): Likewise.
20028 (arm_emit_store_exclusive): Likewise.
20029 (arm_expand_compare_and_swap): Likewise.
20030 (arm_evpc_neon_vuzp): Likewise.
20031 (arm_evpc_neon_vzip): Likewise.
20032 (arm_evpc_neon_vrev): Likewise.
20033 (arm_evpc_neon_vtrn): Likewise.
20034 (arm_evpc_neon_vext): Likewise.
20035 (arm_validize_comparison): Likewise.
20036 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
20037 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
20038 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
20039 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
20040 (c6x_preferred_simd_mode): Likewise.
20041 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
20042 (epiphany_rtx_costs): Likewise.
20043 * config/epiphany/predicates.md (proper_comparison_operator):
20045 * config/frv/frv.c (condexec_memory_operand): Likewise.
20046 (frv_emit_move): Likewise.
20047 (output_move_single): Likewise.
20048 (output_condmove_single): Likewise.
20049 (frv_hard_regno_mode_ok): Likewise.
20050 (frv_matching_accg_mode): Likewise.
20051 * config/h8300/h8300.c (split_adds_subs): Likewise.
20052 (h8300_rtx_costs): Likewise.
20053 (h8300_print_operand): Likewise.
20054 (compute_mov_length): Likewise.
20055 (output_logical_op): Likewise.
20056 (compute_logical_op_length): Likewise.
20057 (compute_logical_op_cc): Likewise.
20058 (h8300_shift_needs_scratch_p): Likewise.
20059 (output_a_shift): Likewise.
20060 (compute_a_shift_length): Likewise.
20061 (compute_a_shift_cc): Likewise.
20062 (expand_a_rotate): Likewise.
20063 (output_a_rotate): Likewise.
20064 * config/i386/i386.c (classify_argument): Likewise.
20065 (function_arg_advance_32): Likewise.
20066 (function_arg_32): Likewise.
20067 (function_arg_64): Likewise.
20068 (function_value_64): Likewise.
20069 (ix86_gimplify_va_arg): Likewise.
20070 (ix86_legitimate_constant_p): Likewise.
20071 (put_condition_code): Likewise.
20072 (split_double_mode): Likewise.
20073 (ix86_avx256_split_vector_move_misalign): Likewise.
20074 (ix86_expand_vector_logical_operator): Likewise.
20075 (ix86_split_idivmod): Likewise.
20076 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
20077 (ix86_build_const_vector): Likewise.
20078 (ix86_build_signbit_mask): Likewise.
20079 (ix86_match_ccmode): Likewise.
20080 (ix86_cc_modes_compatible): Likewise.
20081 (ix86_expand_branch): Likewise.
20082 (ix86_expand_sse_cmp): Likewise.
20083 (ix86_expand_sse_movcc): Likewise.
20084 (ix86_expand_int_sse_cmp): Likewise.
20085 (ix86_expand_vec_perm_vpermi2): Likewise.
20086 (ix86_expand_vec_perm): Likewise.
20087 (ix86_expand_sse_unpack): Likewise.
20088 (ix86_expand_int_addcc): Likewise.
20089 (ix86_split_to_parts): Likewise.
20090 (ix86_vectorize_builtin_gather): Likewise.
20091 (ix86_vectorize_builtin_scatter): Likewise.
20092 (avx_vpermilp_parallel): Likewise.
20093 (inline_memory_move_cost): Likewise.
20094 (ix86_tieable_integer_mode_p): Likewise.
20095 (x86_maybe_negate_const_int): Likewise.
20096 (ix86_expand_vector_init_duplicate): Likewise.
20097 (ix86_expand_vector_init_one_nonzero): Likewise.
20098 (ix86_expand_vector_init_one_var): Likewise.
20099 (ix86_expand_vector_init_concat): Likewise.
20100 (ix86_expand_vector_init_interleave): Likewise.
20101 (ix86_expand_vector_init_general): Likewise.
20102 (ix86_expand_vector_set): Likewise.
20103 (ix86_expand_vector_extract): Likewise.
20104 (emit_reduc_half): Likewise.
20105 (ix86_emit_i387_round): Likewise.
20106 (ix86_mangle_type): Likewise.
20107 (ix86_expand_round_sse4): Likewise.
20108 (expand_vec_perm_blend): Likewise.
20109 (canonicalize_vector_int_perm): Likewise.
20110 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
20111 (expand_vec_perm_1): Likewise.
20112 (expand_vec_perm_interleave3): Likewise.
20113 (expand_vec_perm_even_odd_pack): Likewise.
20114 (expand_vec_perm_even_odd_1): Likewise.
20115 (expand_vec_perm_broadcast_1): Likewise.
20116 (ix86_vectorize_vec_perm_const_ok): Likewise.
20117 (ix86_expand_vecop_qihi): Likewise.
20118 (ix86_expand_mul_widen_hilo): Likewise.
20119 (ix86_expand_sse2_abs): Likewise.
20120 (ix86_expand_pextr): Likewise.
20121 (ix86_expand_pinsr): Likewise.
20122 (ix86_preferred_simd_mode): Likewise.
20123 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
20124 * config/i386/sse.md (*andnot<mode>3): Likewise.
20125 (<mask_codefor><code><mode>3<mask_name>): Likewise.
20126 (*<code><mode>3): Likewise.
20127 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
20128 (ia64_expand_atomic_op): Likewise.
20129 (ia64_arg_type): Likewise.
20130 (ia64_mode_to_int): Likewise.
20131 (ia64_scalar_mode_supported_p): Likewise.
20132 (ia64_vector_mode_supported_p): Likewise.
20133 (expand_vec_perm_broadcast): Likewise.
20134 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
20135 (iq2000_function_arg_advance): Likewise.
20136 (iq2000_function_arg): Likewise.
20137 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
20138 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
20139 (m68k_libcall_value): Likewise.
20140 (m68k_function_value): Likewise.
20141 (sched_attr_op_type): Likewise.
20142 * config/mcore/mcore.c (mcore_output_move): Likewise.
20143 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
20145 (microblaze_function_arg): Likewise.
20146 * config/mips/mips.c (mips16_build_call_stub): Likewise.
20147 (mips_print_operand): Likewise.
20148 (mips_mode_ok_for_mov_fmt_p): Likewise.
20149 (mips_vector_mode_supported_p): Likewise.
20150 (mips_preferred_simd_mode): Likewise.
20151 (mips_expand_vpc_loongson_even_odd): Likewise.
20152 (mips_expand_vec_unpack): Likewise.
20153 (mips_expand_vi_broadcast): Likewise.
20154 (mips_expand_vector_init): Likewise.
20155 (mips_expand_vec_reduc): Likewise.
20156 (mips_expand_msa_cmp): Likewise.
20157 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
20158 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
20159 (cc_flags_for_mode): Likewise.
20160 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20161 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20162 (nds32_output_casesi_pc_relative): Likewise.
20163 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20164 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20165 (nvptx_gen_unpack): Likewise.
20166 (nvptx_gen_pack): Likewise.
20167 (nvptx_gen_shuffle): Likewise.
20168 (nvptx_gen_wcast): Likewise.
20169 (nvptx_preferred_simd_mode): Likewise.
20170 * config/pa/pa.c (pa_secondary_reload): Likewise.
20171 * config/pa/predicates.md (base14_operand): Likewise.
20172 * config/powerpcspe/powerpcspe-c.c
20173 (altivec_resolve_overloaded_builtin): Likewise.
20174 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20176 (rs6000_preferred_simd_mode): Likewise.
20177 (output_vec_const_move): Likewise.
20178 (rs6000_expand_vector_extract): Likewise.
20179 (rs6000_split_vec_extract_var): Likewise.
20180 (reg_offset_addressing_ok_p): Likewise.
20181 (rs6000_legitimate_offset_address_p): Likewise.
20182 (rs6000_legitimize_address): Likewise.
20183 (rs6000_emit_set_const): Likewise.
20184 (rs6000_const_vec): Likewise.
20185 (rs6000_emit_move): Likewise.
20186 (spe_build_register_parallel): Likewise.
20187 (rs6000_darwin64_record_arg_recurse): Likewise.
20188 (swap_selector_for_mode): Likewise.
20189 (spe_init_builtins): Likewise.
20190 (paired_init_builtins): Likewise.
20191 (altivec_init_builtins): Likewise.
20192 (do_load_for_compare): Likewise.
20193 (rs6000_generate_compare): Likewise.
20194 (rs6000_expand_float128_convert): Likewise.
20195 (emit_load_locked): Likewise.
20196 (emit_store_conditional): Likewise.
20197 (rs6000_output_function_epilogue): Likewise.
20198 (rs6000_handle_altivec_attribute): Likewise.
20199 (rs6000_function_value): Likewise.
20200 (emit_fusion_gpr_load): Likewise.
20201 (emit_fusion_p9_load): Likewise.
20202 (emit_fusion_p9_store): Likewise.
20203 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20204 (fusion_gpr_mem_load): Likewise.
20205 (fusion_addis_mem_combo_load): Likewise.
20206 (fusion_addis_mem_combo_store): Likewise.
20207 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20208 (fusion_gpr_mem_load): Likewise.
20209 (fusion_addis_mem_combo_load): Likewise.
20210 (fusion_addis_mem_combo_store): Likewise.
20211 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20213 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20214 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20215 (rs6000_preferred_simd_mode): Likewise.
20216 (output_vec_const_move): Likewise.
20217 (rs6000_expand_vector_extract): Likewise.
20218 (rs6000_split_vec_extract_var): Likewise.
20219 (reg_offset_addressing_ok_p): Likewise.
20220 (rs6000_legitimate_offset_address_p): Likewise.
20221 (rs6000_legitimize_address): Likewise.
20222 (rs6000_emit_set_const): Likewise.
20223 (rs6000_const_vec): Likewise.
20224 (rs6000_emit_move): Likewise.
20225 (rs6000_darwin64_record_arg_recurse): Likewise.
20226 (swap_selector_for_mode): Likewise.
20227 (paired_init_builtins): Likewise.
20228 (altivec_init_builtins): Likewise.
20229 (rs6000_expand_float128_convert): Likewise.
20230 (emit_load_locked): Likewise.
20231 (emit_store_conditional): Likewise.
20232 (rs6000_output_function_epilogue): Likewise.
20233 (rs6000_handle_altivec_attribute): Likewise.
20234 (rs6000_function_value): Likewise.
20235 (emit_fusion_gpr_load): Likewise.
20236 (emit_fusion_p9_load): Likewise.
20237 (emit_fusion_p9_store): Likewise.
20238 * config/rx/rx.c (rx_gen_move_template): Likewise.
20239 (flags_from_mode): Likewise.
20240 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20241 (s390_slb_comparison): Likewise.
20242 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20243 (s390_vector_mode_supported_p): Likewise.
20244 (s390_cc_modes_compatible): Likewise.
20245 (s390_match_ccmode_set): Likewise.
20246 (s390_canonicalize_comparison): Likewise.
20247 (s390_emit_compare_and_swap): Likewise.
20248 (s390_branch_condition_mask): Likewise.
20249 (s390_rtx_costs): Likewise.
20250 (s390_secondary_reload): Likewise.
20251 (__SECONDARY_RELOAD_CASE): Likewise.
20252 (s390_expand_cs): Likewise.
20253 (s390_preferred_simd_mode): Likewise.
20254 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20255 * config/sh/sh.c (sh_print_operand): Likewise.
20256 (dump_table): Likewise.
20257 (sh_secondary_reload): Likewise.
20258 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20259 * config/sh/sh.md (casesi_worker_1): Likewise.
20260 (casesi_worker_2): Likewise.
20261 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20262 (fcc_comparison_operator): Likewise.
20263 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20264 (emit_soft_tfmode_cvt): Likewise.
20265 (sparc_preferred_simd_mode): Likewise.
20266 (output_cbranch): Likewise.
20267 (sparc_print_operand): Likewise.
20268 (sparc_expand_vec_perm_bmask): Likewise.
20269 (vector_init_bshuffle): Likewise.
20270 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20271 (spu_vector_mode_supported_p): Likewise.
20272 (spu_expand_insv): Likewise.
20273 (spu_emit_branch_or_set): Likewise.
20274 (spu_handle_vector_attribute): Likewise.
20275 (spu_builtin_splats): Likewise.
20276 (spu_builtin_extract): Likewise.
20277 (spu_builtin_promote): Likewise.
20278 (spu_expand_sign_extend): Likewise.
20279 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20280 (tilegx_simd_int): Likewise.
20281 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20282 (tilepro_simd_int): Likewise.
20283 * config/v850/v850.c (const_double_split): Likewise.
20284 (v850_print_operand): Likewise.
20285 (ep_memory_offset): Likewise.
20286 * config/vax/vax.c (vax_rtx_costs): Likewise.
20287 (vax_output_int_move): Likewise.
20288 (vax_output_int_add): Likewise.
20289 (vax_output_int_subtract): Likewise.
20290 * config/visium/predicates.md (visium_branch_operator): Likewise.
20291 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20292 (visium_print_operand_address): Likewise.
20293 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20294 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20295 (xtensa_expand_conditional_branch): Likewise.
20296 (xtensa_copy_incoming_a7): Likewise.
20297 (xtensa_output_literal): Likewise.
20298 * dfp.c (decimal_real_maxval): Likewise.
20299 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20302 Alan Hayward <alan.hayward@arm.com>
20303 David Sherwood <david.sherwood@arm.com>
20305 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20306 (mode_nunits_inline): Likewise.
20307 (mode_inner_inline): Likewise.
20308 (mode_unit_size_inline): Likewise.
20309 (mode_unit_precision_inline): Likewise.
20310 (emit_insn_modes_h): Likewise. Also emit a #define of the
20312 (emit_mode_wider): Add an E_ prefix to mode names.
20313 (emit_mode_complex): Likewise.
20314 (emit_mode_inner): Likewise.
20315 (emit_mode_adjustments): Likewise.
20316 (emit_mode_int_n): Likewise.
20317 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20318 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20319 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20320 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20321 (CRC32_BUILTIN, ENTRY): Likewise.
20322 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20323 (aarch64_pop_regs): Likewise.
20324 (aarch64_process_components): Likewise.
20325 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20326 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20327 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20328 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20329 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20330 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20332 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20333 (pcmpistr): Likewise.
20334 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20335 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20336 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20338 * config/rl78/rl78.c (mduc_regs): Likewise.
20339 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20340 (htm_expand_builtin): Likewise.
20341 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20342 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20343 * config/xtensa/xtensa.c (print_operand): Likewise.
20344 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20345 (NUM_MODE_VECTOR_INT): Likewise.
20346 * genoutput.c (null_operand): Likewise.
20347 (output_operand_data): Likewise.
20348 * genrecog.c (print_parameter_value): Likewise.
20349 * lra.c (debug_operand_data): Likewise.
20351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20353 * dwarf2out.c (scompare_loc_descriptor_wide)
20354 (scompare_loc_descriptor_narrow): New functions, split out from...
20355 (scompare_loc_descriptor): ...here.
20356 * expmed.c (emit_store_flag_int): New function, split out from...
20357 (emit_store_flag): ...here.
20359 2017-08-30 Richard Biener <rguenther@suse.de>
20361 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20362 (dwarf2out_early_finish): Move setting of AT_pubnames from
20363 early debug output to early finish.
20365 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20367 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20368 and -mdata-region to the assembler.
20370 2017-08-30 Richard Biener <rguenther@suse.de>
20372 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20374 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20375 (dwarf2out_early_global_decl): Only generate a DIE for the
20376 abstract origin if it doesn't already exist or is a declaration DIE.
20377 (resolve_addr): Do not add the linkage name twice when
20378 generating a stub DIE for the DW_TAG_GNU_call_site target.
20380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20382 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20383 Use machine_mode rather than int for arg1_mode.
20385 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20388 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20389 that the second argument of the built-in functions to unpack
20390 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20391 switch statement instead a lot of if statements.
20392 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20393 Allow 64-bit values to be in Altivec registers as well as
20394 traditional floating point registers.
20395 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20397 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20399 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20400 MAX_REGS_PER_ADDRESS == 1.
20402 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20404 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20405 * config/i386/i386.c (ix86_option_override_internal): Simplify
20406 setting of opts->x_flag_entry.
20408 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20409 Jakub Jelinek <jakub@redhat.com>
20410 Richard Biener <rguenther@suse.de>
20412 PR tree-optimization/81503
20413 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20414 folded constant fits in the target type; reorder tests for clarity.
20416 2017-08-29 Martin Liska <mliska@suse.cz>
20418 * passes.def: Include pass_lower_switch.
20419 * stmt.c (dump_case_nodes): Remove and move to
20420 tree-switch-conversion.
20421 (case_values_threshold): Likewise.
20422 (expand_switch_as_decision_tree_p): Likewise.
20423 (emit_case_decision_tree): Likewise.
20424 (expand_case): Likewise.
20425 (balance_case_nodes): Likewise.
20426 (node_has_low_bound): Likewise.
20427 (node_has_high_bound): Likewise.
20428 (node_is_bounded): Likewise.
20429 (emit_case_nodes): Likewise.
20430 (struct simple_case_node): New struct.
20431 (add_case_node): Remove.
20432 (emit_case_dispatch_table): Use vector instead of case_list.
20433 (reset_out_edges_aux): Remove.
20434 (compute_cases_per_edge): Likewise.
20435 (expand_case): Build list of simple_case_node.
20436 (expand_sjlj_dispatch_table): Use it.
20437 * tree-switch-conversion.c (struct case_node): Moved from
20438 stmt.c and adjusted.
20439 (emit_case_nodes): Likewise.
20440 (node_has_low_bound): Likewise.
20441 (node_has_high_bound): Likewise.
20442 (node_is_bounded): Likewise.
20443 (case_values_threshold): Likewise.
20444 (reset_out_edges_aux): Likewise.
20445 (compute_cases_per_edge): Likewise.
20446 (add_case_node): Likewise.
20447 (dump_case_nodes): Likewise.
20448 (balance_case_nodes): Likewise.
20449 (expand_switch_as_decision_tree_p): Likewise.
20450 (emit_jump): Likewise.
20451 (emit_case_decision_tree): Likewise.
20452 (try_switch_expansion): Likewise.
20453 (do_jump_if_equal): Likewise.
20454 (emit_cmp_and_jump_insns): Likewise.
20455 (fix_phi_operands_for_edge): New function.
20456 (record_phi_operand_mapping): Likewise.
20457 (class pass_lower_switch): New pass.
20458 (pass_lower_switch::execute): New function.
20459 (make_pass_lower_switch): Likewise.
20460 (conditional_probability):
20461 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20462 * tree-pass.h: Add make_pass_lower_switch.
20464 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20467 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20470 2017-08-29 Richard Biener <rguenther@suse.de>
20472 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20473 we do not add a DW_AT_inline attribute twice.
20474 (gen_subprogram_die): Remove code setting DW_AT_inline on
20475 DECL_ABSTRACT_P nodes.
20477 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20479 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20480 calls to internal functions.
20481 (gimplify_modify_expr): Likewise.
20482 * tree-call-cdce.c (use_internal_fn): Likewise.
20483 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20484 (convert_to_divmod): Set the nothrow flag.
20485 * tree-if-conv.c (predicate_mem_writes): Likewise.
20486 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20487 (vectorizable_call): Likewise.
20488 (vectorizable_store): Likewise.
20489 (vectorizable_load): Likewise.
20490 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20491 (vect_recog_mask_conversion_pattern): Likewise.
20493 2017-08-29 Martin Liska <mliska@suse.cz>
20496 * gcc.c (driver_handle_option): Add new argument.
20497 * opts-common.c (handle_option): Pass
20498 target_option_override_hook.
20499 * opts-global.c (lang_handle_option): Add new option.
20500 (set_default_handlers): Add new argument.
20501 (decode_options): Likewise.
20502 * opts.c (target_handle_option): Likewise.
20503 (common_handle_option): Call target_option_override_hook.
20504 * opts.h (struct cl_option_handler_func): Add hook for
20505 target option override.
20506 (struct cl_option_handlers): Likewise.
20507 (set_default_handlers): Add new argument.
20508 (decode_options): Likewise.
20509 (common_handle_option): Likewise.
20510 (target_handle_option): Likewise.
20511 * toplev.c (toplev::main): Pass targetm.target_option.override
20514 2017-08-29 Richard Biener <rguenther@suse.de>
20515 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20517 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20518 life to the active subtree.
20520 2017-08-28 Jeff Law <law@redhat.com>
20522 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20523 derive_equivalences.
20524 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20525 Code moved into....
20526 (edge_info::derive_equivalences): New private member function
20528 * tree-ssa-dom.c (class edge_info): Changed from a struct
20529 to a class. Add ctor/dtor, methods and data members.
20530 (edge_info::edge_info): Renamed from allocate_edge_info.
20531 Initialize additional members.
20532 (edge_info::~edge_info): New.
20533 (free_dom_edge_info): Delete the edge info.
20534 (record_edge_info): Use new class & associated member functions.
20535 Tighten forms for testing for edge equivalences.
20536 (record_temporary_equivalences): Iterate over the simple
20537 equivalences rather than assuming there's only one per edge.
20538 (cprop_into_successor_phis): Iterate over the simple
20539 equivalences rather than assuming there's only one per edge.
20540 (optimize_stmt): Use operand_equal_p rather than pointer
20541 equality for mini-DSE code.
20543 2017-08-28 Nathan Sidwell <nathan@acm.org>
20545 * gcc.c (execute): Fold SIGPIPE handling into switch
20546 statement. Adjust internal error message.
20548 2017-08-28 Richard Biener <rguenther@suse.de>
20551 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20552 Do nothing for removed DIEs.
20554 2017-08-28 Richard Biener <rguenther@suse.de>
20556 PR tree-optimization/81977
20557 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20560 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20563 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20564 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20565 using targetm.gen_mem_thread_fence.
20567 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20570 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20571 predicate to register_operand. Reorder operands.
20572 (*btr<mode>): Ditto.
20573 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20574 (*btr<mode>_mask): Ditto.
20576 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20578 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20579 * config/rs6000/xmmintrin.h: New file.
20580 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20582 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20585 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20586 parameter and_insn and return it.
20587 (recombine_lvx_pattern): Insert a copy to ensure availability of
20588 the base register of the copied masking operation at the point of
20589 the instruction replacement.
20590 (recombine_stvx_pattern): Likewise.
20592 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20594 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20595 undocumented switches.
20596 (-mpower9-dform-vector): Likewise.
20597 (-mpower9-dform): Likewise.
20598 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20599 comments to delete references to -mpower9-dform* switches.
20600 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20601 Delete reference to -mpower9-dform* switches, test for
20602 -mpower9-vector instead.
20603 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20604 (OTHER_P9_VECTOR_MASKS): Likewise.
20605 (POWERPC_MASKS): Likewise.
20606 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20607 tests against -mpower9-dform* to -mpower9-vector. Delete code
20608 that checked for -mpower9-dform* consistancy with other options.
20609 Add test for -mpower9-misc to enable other power9 switches.
20610 (rs6000_init_hard_regno_mode_ok): Likewise.
20611 (rs6000_option_override_internal): Likewise.
20612 (rs6000_emit_prologue): Likewise.
20613 (rs6000_emit_epilogue): Likewise.
20614 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20615 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20616 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20618 (emit_fusion_p9_store): Likewise.
20619 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20620 resetting these macros if the assembler does not support ISA 3.0
20622 (TARGET_P9_DFORM_VECTOR): Likewise.
20623 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20624 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20626 2017-08-25 Alan Modra <amodra@gmail.com>
20628 PR rtl-optimization/81747
20629 * cse.c (cse_extended_basic_block): Don't attempt to record
20630 equivalences for degenerate conditional branches that branch
20631 to their fall-through.
20633 2017-08-24 Martin Sebor <msebor@redhat.com>
20635 PR middle-end/81908
20636 * gimple-fold.c (size_must_be_zero_p): New function.
20637 (gimple_fold_builtin_memory_op): Call it.
20639 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20641 * config/rs6000/mm_malloc.h: New file.
20643 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20645 PR tree-optimization/81913
20646 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20647 analysis when either IVs in condition can wrap.
20649 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20651 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20652 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20654 2017-08-24 Richard Biener <rguenther@suse.de>
20657 * targhooks.c (default_target_can_inline_p): Properly
20658 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20659 is present and always compare.
20660 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20661 infer -mfpmath=sse from TARGET_SSE_P.
20662 (ix86_can_inline_p): Properly use target_option_default_node when
20663 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20665 2017-08-24 Richard Biener <rguenther@suse.de>
20668 * dwarf2out.c (output_die): Handle flag_generate_offload like
20670 (output_comp_unit): Likewise.
20671 (gen_array_type_die): Likewise.
20672 (dwarf2out_early_finish): Likewise.
20673 (note_variable_value_in_expr): Likewise.
20674 (dwarf2out_finish): Likewise. Adjust assert.
20675 * cgraphunit.c (symbol_table::compile): Move setting of
20676 flag_generate_offload earlier ...
20677 (symbol_table::finalize_compilation_unit): ... here, before
20678 early debug finalization.
20680 2017-08-24 Richard Biener <rguenther@suse.de>
20682 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20683 and ipa-fnsummary.h.
20684 (ix86_can_inline_p): When ix86_fpmath flags do not match
20685 check whether the callee uses FP math at all.
20687 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20689 PR middle-end/81931
20690 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20691 instead of TYPE_PRECISION.
20693 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20694 Alan Hayward <alan.hayward@arm.com>
20695 David Sherwood <david.sherwood@arm.com>
20697 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20699 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20701 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20703 (ix86_frame::stack_realign_allocate): New field.
20704 (struct machine_frame_state): Modify comments.
20705 (machine_frame_state::sp_realigned_fp_end): New field.
20706 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20707 layout calculation.
20708 (sp_valid_at): Add assertion to assure no attempt to access invalid
20709 offset of a realigned stack.
20710 (fp_valid_at): Likewise.
20711 (choose_baseaddr): Modify comments.
20712 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20713 ix86_expand_prologue.
20714 (ix86_expand_prologue): Modify stack realignment and allocation.
20715 (ix86_expand_epilogue): Modify comments.
20716 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20717 avx2_runtime, avx512f, and avx512f_runtime.
20719 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20721 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20722 (mstackrealign): Do not init to -1.
20723 * config/i386/i386.c (ix86_option_override_internal):
20724 Check opts_set, not opts when setting default value of
20725 opts->x_ix86_force_align_arg_pointer.
20727 2017-08-23 Richard Biener <rguenther@suse.de>
20729 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20730 lang_hooks.decl_printable_name.
20731 * print-rtl-function.c (print_rtx_function): Likewise.
20732 * tree-pretty-print.c (dump_function_header): Likewise.
20734 2017-08-23 Richard Biener <rguenther@suse.de>
20737 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20738 -g0 at compile-time.
20740 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20742 PR middle-end/19706
20743 * doc/sourcebuild.texi (Other hardware attributes):
20746 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20748 PR middle-end/19706
20749 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20750 Fix single-use check.
20752 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20754 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20756 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20758 * doc/install.texi: Modify to add more details on running selected
20761 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20763 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20764 is combined with -mabi=ms.
20765 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20768 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20770 PR tree-optimization/81488
20771 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20772 and cached_basis fields.
20773 (MAX_SPREAD): New constant.
20774 (alloc_cand_and_find_basis): Initialize new fields.
20775 (clear_visited): New function.
20776 (create_phi_basis_1): Rename from create_phi_basis, set visited
20777 and cached_basis fields.
20778 (create_phi_basis): New wrapper function.
20779 (phi_add_costs_1): Rename from phi_add_costs, add spread
20780 parameter, set visited field, short-circuit when limits reached.
20781 (phi_add_costs): New wrapper function.
20782 (record_phi_increments_1): Rename from record_phi_increments, set
20784 (record_phi_increments): New wrapper function.
20785 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20786 (phi_incr_cost): New wrapper function.
20787 (all_phi_incrs_profitable_1): Rename from
20788 all_phi_incrs_profitable, set visited field.
20789 (all_phi_incrs_profitable): New wrapper function.
20791 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20792 Alan Hayward <alan.hayward@arm.com>
20793 David Sherwood <david.sherwood@arm.com>
20795 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20796 that takes the outer and inner modes.
20797 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20798 comparison as the canonical test for a paradoxical subreg.
20799 * combine.c (simplify_set): Use paradoxical_subreg_p.
20800 (make_extraction): Likewise.
20801 (force_to_mode): Likewise.
20802 (rtx_equal_for_field_assignment_p): Likewise.
20803 (gen_lowpart_for_combine): Likewise.
20804 (simplify_comparison): Likewise.
20805 * cse.c (equiv_constant): Likewise.
20806 * expmed.c (store_bit_field_1): Likewise.
20807 * final.c (alter_subreg): Likewise.
20808 * fwprop.c (propagate_rtx): Likewise.
20809 (forward_propagate_subreg): Likewise.
20810 * ira-conflicts.c (ira_build_conflicts): Likewise.
20811 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20812 * lra-constraints.c (curr_insn_transform): Likewise.
20813 (split_reg): Likewise.
20814 * lra-eliminations.c (move_plus_up): Likewise.
20815 (lra_eliminate_regs_1): Likewise.
20816 * recog.c (general_operand): Likewise.
20817 * ree.c (combine_reaching_defs): Likewise.
20818 * reload.c (push_reload): Likewise.
20819 (find_reloads): Likewise.
20820 * reload1.c (elimination_effects): Likewise.
20821 (compute_reload_subreg_offset): Likewise.
20822 (choose_reload_regs): Likewise.
20823 * rtlanal.c (subreg_lsb_1): Likewise.
20824 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20825 (simplify_subreg): Likewise.
20826 * var-tracking.c (track_loc_p): Likewise.
20827 * emit-rtl.c (byte_lowpart_offset): Likewise.
20828 (paradoxical_subreg_p): Delete out-of-line definition.
20830 2017-08-22 Jeff Law <law@redhat.com>
20832 PR tree-optimization/81741
20833 PR tree-optimization/71947
20834 * tree-ssa-dom.c: Include tree-inline.h.
20835 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20836 equivalences if one is more expensive to compute than the other.
20837 * tree-ssa-scopedtables.h (class const_or_copies): Make
20838 record_const_or_copy_raw method private.
20839 (class avail_exprs_stack): New method simplify_binary_operation.
20840 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20841 avail_exprs_stack::simplify_binary_operation as needed.
20842 (avail_exprs_stack::simplify_binary_operation): New function.
20844 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20846 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20847 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20848 (DOT_SYMBOLS): Likewise.
20849 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20850 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20851 (RS6000_ABI_NAME): Likewise.
20852 (TARGET_CMODEL): Likewise.
20853 (TOC_SECTION_ASM_OP): Likewise.
20854 (SET_CMODEL): New macro.
20855 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20857 2017-08-22 Richard Biener <rguenther@suse.de>
20859 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20860 to allow for free-lang-data replacements similar to verify_type_variant.
20862 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20863 Alan Hayward <alan.hayward@arm.com>
20864 David Sherwood <david.sherwood@arm.com>
20866 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20867 VOIDmode for the LABEL_REF.
20869 2017-08-22 Richard Biener <rguenther@suse.de>
20871 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20873 2017-08-22 Richard Biener <rguenther@suse.de>
20875 * common.opt (feliminate-dwarf2-dups): Ignore.
20876 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20877 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20878 same_die_p_wrap, compute_section_prefix,
20879 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20880 (comdat_symbol_id, comdat_symbol_number): Likewise.
20881 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20883 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20884 (output_die): Mark unreachable path unreachable.
20885 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20886 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20887 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20888 (dwarf2out_early_finish): Likewise.
20890 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20892 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20894 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20897 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20898 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20899 (avr_attribute_table) <io, io_low, address>: Initialize
20900 .decl_required with true.
20902 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20904 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20905 undocumented debugging options.
20906 (-mvsx-scalar-double): Likewise.
20907 (-mallow-df-permute): Likewise.
20908 (-mvectorize-builtins): Likewise.
20909 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20910 (rs6000_builtin_vectorized_function): Likewise.
20911 (rs6000_builtin_md_vectorized_function): Likewise.
20912 (rs6000_opt_vars): Likewise.
20914 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20917 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20918 (*btrq_imm): Rename from *btrq.
20919 (*btcq_imm): Rename from *btcq.
20920 (btsc): New code attribute.
20921 (*<btsc><mode>): New insn pattern.
20922 (*btr<mode>): Ditto.
20923 (*<btsc><mode>_mask): New insn_and_split pattern.
20924 (*btr<mode>_mask): Ditto.
20926 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20927 Alan Hayward <alan.hayward@arm.com>
20928 David Sherwood <david.sherwood@arm.com>
20930 * function.c (pad_below): Simplify padding calculation.
20932 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20933 Alan Hayward <alan.hayward@arm.com>
20934 David Sherwood <david.sherwood@arm.com>
20936 * target.def (function_prologue): Remove frame size argument.
20937 (function_epilogue): Likewise.
20938 * doc/tm.texi: Regenerate.
20939 * final.c (final_start_function): Update call to function_prologue.
20940 (final_end_function): Update call to function_epilogue.
20941 (default_function_pro_epilogue): Remove frame size argument.
20942 * output.h (default_function_pro_epilogue): Likewise.
20943 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20944 (arm_output_function_prologue): Likewise.
20945 * config/frv/frv.c (frv_function_prologue): Likewise.
20946 (frv_function_epilogue): Likewise.
20947 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20948 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20949 (ia64_output_function_epilogue): Likewise.
20950 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20951 (m32r_output_function_epilogue): Likewise.
20952 * config/microblaze/microblaze.c (microblaze_function_prologue)
20953 (microblaze_function_epilogue): Likewise.
20954 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20955 (mips_output_function_epilogue): Likewise.
20956 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20957 (mmix_target_asm_function_epilogue): Likewise.
20958 * config/msp430/msp430.c (msp430_start_function): Likewise.
20959 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20960 (nds32_asm_function_epilogue): Likewise.
20961 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20962 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20963 (pa_output_function_epilogue): Likewise.
20964 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20965 (rs6000_output_function_epilogue): Likewise.
20966 * config/rl78/rl78.c (rl78_start_function): Likewise.
20967 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20968 (rs6000_output_function_epilogue): Likewise.
20969 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20970 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20971 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20972 (sparc_asm_function_epilogue): Likewise.
20974 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20976 * tree.h (type_has_mode_precision_p): New function.
20977 * convert.c (convert_to_integer_1): Use it.
20978 * expr.c (expand_expr_real_2): Likewise.
20979 (expand_expr_real_1): Likewise.
20980 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20981 * match.pd: Likewise.
20982 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20983 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20984 * tree-tailcall.c (process_assignment): Likewise.
20985 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20986 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20987 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20988 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20989 (vectorizable_assignment): Likewise.
20990 (vectorizable_shift): Likewise.
20991 (vectorizable_operation): Likewise.
20992 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20994 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20996 * match.pd: Add pow (C, x) simplification.
20998 2017-08-21 Richard Biener <rguenther@suse.de>
21000 PR tree-optimization/81900
21001 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
21002 for blocks with abnormal predecessors.
21003 (compute_antic): Do not set visited flag prematurely.
21005 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
21008 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
21010 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21012 * stor-layout.h (vector_type_mode): Move to...
21013 * tree.h (vector_type_mode): ...here.
21014 * stor-layout.c (vector_type_mode): Move to...
21015 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
21017 2017-08-21 Richard Biener <rguenther@suse.de>
21019 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
21020 register_external_die hooks.
21021 (debug_false_tree_charstarstar_uhwistar): Declare.
21022 (debug_nothing_tree_charstar_uhwi): Likewise.
21023 * debug.c (do_nothing_debug_hooks): Adjust.
21024 (debug_false_tree_charstarstar_uhwistar): New do nothing.
21025 (debug_nothing_tree_charstar_uhwi): Likewise.
21026 * dbxout.c (dbx_debug_hooks): Adjust.
21027 (xcoff_debug_hooks): Likewise.
21028 * sdbout.c (sdb_debug_hooks): Likewise.
21029 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21030 * dwarf2out.c (macinfo_label_base): New global.
21031 (dwarf2out_register_external_die): New function for the
21032 register_external_die hook.
21033 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
21034 (dwarf2_debug_hooks): Use them.
21035 (dwarf2_lineno_debug_hooks): Adjust.
21036 (struct die_struct): Add with_offset flag.
21037 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
21038 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
21039 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
21040 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
21041 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
21042 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
21043 defining section names for the early LTO debug variants.
21044 (reset_indirect_string): New helper.
21045 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
21046 (print_dw_val): Add support for offsetted symbol references.
21047 (get_ultimate_context): Split out from is_cxx.
21048 (is_cxx): Use get_ultimate_context.
21049 (is_fortran): Add decl overload.
21050 (compute_comp_unit_symbol): Split out worker from
21051 compute_section_prefix.
21052 (compute_section_prefix): Call compute_comp_unit_symbol and
21053 set comdat_type_p here.
21054 (output_die): Skip DIE symbol output for the LTO added one.
21055 Handle DIE symbol references with offset.
21056 (output_comp_unit): Guard section name mangling properly.
21057 For LTO debug sections emit a symbol at the section beginning
21058 which we use to refer to its DIEs.
21059 (add_abstract_origin_attribute): For DIEs registered via
21060 dwarf2out_register_external_die directly refer to the early
21061 DIE rather than indirectly through the shadow one we created.
21062 Remove obsolete call to dwarf2out_abstract_function for
21063 non-function/block origins.
21064 (gen_array_type_die): When generating early LTO debug do
21065 not emit DW_AT_string_length.
21066 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
21067 late when in LTO. As suggested place a gcc_unreachable for
21068 the DECL_ABSTRACT_P case.
21069 (gen_subprogram_die): Avoid another specification DIE
21070 for early built declarations/definitions for the late LTO case.
21071 (gen_variable_die): Add type references for late duplicated VLA dies
21073 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
21074 we have the abstract instance already.
21075 (process_scope_var): Adjust decl DIE contexts in LTO which
21076 first puts them in limbo.
21077 (gen_decl_die): Do not generate type DIEs late apart from
21078 types for VLAs or for decls we do not yet have a DIE. Do not
21079 call dwarf2out_abstract_function late.
21080 (dwarf2out_early_global_decl): Make sure to create DIEs
21081 for abstract instances of a decl first.
21082 (dwarf2out_late_global_decl): Adjust comment.
21083 (output_macinfo_op): With multiple macro sections use
21084 macinfo_label_base to distinguish labels.
21085 (output_macinfo): Likewise. Update macinfo_label_base.
21086 Pass in the line info label.
21087 (note_variable_value_in_expr): When generating LTO resolve
21088 all variable values here by generating DIEs as needed.
21089 (init_sections_and_labels): Add early LTO debug flag parameter
21090 and generate different sections and names if set. Add generation
21091 counter for the labels so we can have multiple of them.
21092 (reset_dies): Helper to allow DIEs to be output multiple times.
21093 (dwarf2out_finish): When outputting DIEs to the fat part of an
21094 LTO object first reset DIEs.
21095 (dwarf2out_early_finish): Output early DIEs when generating LTO.
21096 (modified_type_die): Check for decl_ultimate_origin being self
21098 (gen_type_die_with_usage): Likewise.
21099 (gen_typedef_die): Allow decl_ultimate_origin being self.
21100 (set_decl_abstract_flags): Remove.
21101 (set_block_abstract_flags): Likewise.
21102 (dwarf2out_abstract_function): Treat the early generated DIEs
21103 as the abstract copy and only add DW_AT_inline and
21104 DW_AT_artificial here and call set_decl_origin_self.
21105 If the DIE has an abstract origin don't do anything.
21106 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
21107 if we have none yet (Go fails to build one, PR78628).
21108 (variably_modified_type_p): Prevent endless recursion for Ada
21109 cyclic pointer types.
21110 * lto-streamer-in.c: Include debug.h.
21111 (dref_queue): New global.
21112 (lto_read_tree_1): Stream in DIE references.
21113 (lto_input_tree): Register DIE references.
21114 (input_function): Stream DECL_DEBUG_ARGS.
21115 * lto-streamer-out.c: Include debug.h.
21116 (lto_write_tree_1): Output DIE references.
21117 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
21118 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
21119 (output_function): Stream DECL_DEBUG_ARGS.
21120 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
21121 Stream DECL_ABSTRACT_ORIGIN.
21122 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
21123 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
21124 DECL_CONTEXT for file-scope decls.
21125 * lto-streamer.h (struct dref_entry): Declare.
21126 (dref_queue): Likewise.
21127 * cfgexpand.c (pass_expand::execute): Do not call the
21128 outlining_inline_function hook here.
21129 * lto-wrapper.c (debug_obj): New global.
21130 (tool_cleanup): Unlink it if required.
21131 (debug_objcopy): New function.
21132 (run_gcc): Handle early debug sections in the IL files by
21133 extracting them to separate files, partially linkin them and
21134 feeding the result back as result to the linker.
21135 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
21136 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
21137 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
21138 sections into a separate segment.
21139 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
21141 (darwin_asm_dwarf_section): Likewise.
21142 (darwin_asm_output_dwarf_offset): Likewise.
21143 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
21145 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21146 Alan Hayward <alan.hayward@arm.com>
21147 David Sherwood <david.sherwood@arm.com>
21149 * read-md.h (md_reader::record_potential_iterator_use): Replace
21150 pointer argument with an rtx and an index.
21151 * read-rtl.c (iterator_group::apply_iterator): Likewise.
21152 (apply_mode_iterator): Likewise.
21153 (apply_code_iterator): Likewise.
21154 (apply_int_iterator): Likewise.
21155 (apply_subst_iterator): Likewise.
21156 (record_iterator_use): Likewise.
21157 (record_attribute_use): Likewise.
21158 (md_reader::record_potential_iterator_use): Likewise. Update calls
21159 to record_iterator_use and apply_iterator.
21160 (iterator_use): Replace ptr with x and index.
21161 (attribute_use): Likewise.
21162 (apply_attribute_uses): Update calls to apply_iterator.
21163 (apply_iterators): Likewise. Update initialization of iterator_use.
21164 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21165 and record_potential_iterator_use.
21166 (rtx_reader::read_rtx_operand): Likewise.
21168 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21169 Alan Hayward <alan.hayward@arm.com>
21170 David Sherwood <david.sherwood@arm.com>
21172 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21175 2017-08-21 Richard Biener <rguenther@suse.de>
21177 PR middle-end/81884
21178 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21179 at struct end conservatively when comparing common bases.
21181 2017-08-21 Richard Biener <rguenther@suse.de>
21183 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21184 (mem_ref_in_stmt): Remove.
21185 (determine_max_movement): Use ref index to get at the reference.
21186 (invariantness_dom_walker::before_dom_children): Deal with
21187 lim data already initialized.
21188 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21190 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21192 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21193 (TARGET_ISA_ROUND): Ditto.
21194 (TARGET_ROUND): Ditto.
21195 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21196 * config/i386/i386.md: Ditto.
21197 * config/i386/sse.md: Ditto.
21198 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21199 with OPTION_MASK_ISA_SSE4_1.
21201 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21204 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21205 __builtin_ia32_lzcnt_u16.
21207 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21210 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21211 (rs6000_set_current_function): Rewrite function to use it.
21213 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21216 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21217 and TYPE_WARN_IF_NOT_ALIGN.
21218 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21219 (handle_warn_if_not_align): New.
21220 (place_union_field): Call handle_warn_if_not_align.
21221 (place_field): Call handle_warn_if_not_align.
21222 Copy TYPE_WARN_IF_NOT_ALIGN.
21223 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21224 (layout_type): Likewise.
21225 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21227 (tree_decl_common): Add warn_if_not_align.
21228 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21229 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21230 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21231 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21232 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21233 * doc/extend.texi: Document warn_if_not_aligned attribute.
21234 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21236 2017-08-17 Martin Liska <mliska@suse.cz>
21239 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21240 (get_data_dependence): Use it as pointer type.
21241 (distribute_loop): Likewise.
21243 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21245 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21246 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21247 (p8_vmrgow_<mode>_direct): New define_insn.
21248 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21249 handle endianness for vmrgew and vmrgow permute patterns.
21251 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21253 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21254 * config/rs6000/rs6000-cpus.def: Remove comment.
21255 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21256 (POWERPC_MASKS): Likewise.
21257 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21258 use of TARGET_VSX_TIMODE.
21259 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21260 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21261 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21262 (rs6000_option_override_internal): Remove dead code.
21263 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21264 (rs6000_legitimize_reload_address): Likewise.
21265 (rs6000_legitimate_address_p): Likewise.
21266 (rs6000_opt_masks): Delete "vsx-timode".
21267 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21268 from function comment.
21269 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21270 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21271 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21273 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21274 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21275 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21276 (VSX_TI): Likewise.
21278 (define_peephole2): Likewise.
21280 2017-08-17 Martin Sebor <msebor@redhat.com>
21283 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21284 past the end of an array.
21285 (test_pp_format): Add test cases.
21287 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21289 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21290 missing ECF_NOTHROW flags.
21292 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21295 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21296 operands residing in integer registers.
21297 (*vsx_le_perm_load_<mode>): Likewise.
21298 (*vsx_le_perm_store_<mode>): Likewise.
21299 (define_peephole2): Add peepholes to optimize the above.
21301 2017-08-17 Marek Polacek <polacek@redhat.com>
21303 PR middle-end/81814
21304 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21305 to mimic what shorten_compare did. Change the return type to bool.
21306 (fold_cond_expr_with_comparison): Update call to
21307 operand_equal_for_comparison_p.
21308 (fold_ternary_loc): Likewise.
21310 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21312 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21314 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21316 2017-08-17 Richard Biener <rguenther@suse.de>
21318 * tree-ssa-structalias.c (solve_graph): When propagating
21319 to successors update the graphs succ edges and avoid duplicate work.
21321 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21324 * config/i386/i386.c (ix86_option_override_internal): Save target
21325 specific options after ix86_stack_protector_guard_reg was changed.
21327 2017-08-17 Richard Biener <rguenther@suse.de>
21329 PR tree-optimization/81827
21330 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21331 (new_var_info): Initialize it conservatively.
21332 (get_call_vi): Mark register vars.
21333 (new_scalar_tmp_constraint_exp): Likewise.
21334 (handle_rhs_call): Likewise.
21335 (handle_const_call): Likewise.
21336 (create_function_info_for): Likewise.
21337 (solve_constraints): Sort varinfos to separate register from
21338 non-register vars to pack points-to solution bitmaps during
21341 2017-08-17 Marek Polacek <polacek@redhat.com>
21343 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21345 2017-08-17 Richard Biener <rguenther@suse.de>
21347 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21348 to true when overflow is undefined and we saturated the result.
21350 2017-08-17 Alan Modra <amodra@gmail.com>
21353 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21354 Don't use store multiple if only one reg needs saving.
21355 (interesting_frame_related_regno): New function.
21356 (rs6000_frame_related): Don't emit frame info for regs that
21358 (rs6000_emit_epilogue): Likewise.
21360 2017-08-16 Nathan Sidwell <nathan@acm.org>
21362 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21363 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21364 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21365 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21366 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21367 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21368 (verify_type): Adjust for TYPE_BINFO move.
21369 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21370 process TYPE_BINFO directly.
21371 (hash_tree): Likewise.
21372 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21374 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21377 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21379 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21381 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21384 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21385 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21386 Add ix86_binary_operator_ok to insn constraint.
21387 (*iordi_1_bts): Ditto.
21388 (*xordi_1_btc): Ditto.
21389 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21390 Update corresponding peephole2 pattern.
21394 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21396 PR tree-optimization/81832
21397 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21398 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21400 2017-08-16 Marek Polacek <polacek@redhat.com>
21403 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21404 regarding size_zero_node.
21406 2017-08-16 Martin Liska <mliska@suse.cz>
21409 * config.gcc: Respect previously set extra_objs in case
21412 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21414 PR tree-optimization/81835
21415 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21416 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21417 not depend on the phi.
21419 2017-08-16 Alan Modra <amodra@gmail.com>
21421 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21424 2017-08-16 Alan Modra <amodra@gmail.com>
21426 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21427 (save_reg_p): ..into this. Update all callers.
21428 (first_reg_to_save): Simplify.
21430 2017-08-16 Alan Modra <amodra@gmail.com>
21432 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21435 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21439 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21440 constant count if that count is less than 32.
21442 2017-08-15 Nathan Sidwell <nathan@acm.org>
21444 * gcc.c (execute): Emit friendlier message if inferior is killed
21445 by an external cause.
21447 2017-08-15 Richard Biener <rguenther@suse.de>
21449 PR tree-optimization/81790
21450 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21451 CONSTRUCTORs from simplifying and VN.
21453 2017-08-14 Martin Sebor <msebor@redhat.com>
21455 * builtin-attrs.def: Add comments.
21457 2017-08-14 Martin Sebor <msebor@redhat.com>
21460 * doc/extend.texi (attribute nonstring): Document new attribute.
21462 2017-08-14 Martin Sebor <msebor@redhat.com>
21465 * tree-diagnostic.c (default_tree_printer): Handle %G.
21466 * gimple-pretty-print.h (percent_G_format): Declare new function.
21467 * gimple-pretty-print.c (percent_G_format): Define.
21468 * tree-pretty-print.c (percent_K_format): Add argument.
21470 2017-08-14 Martin Sebor <msebor@redhat.com>
21472 PR translation/79998
21473 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21474 Remove a stray space.
21476 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21479 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21480 (*iordi_1_bts): Ditto.
21481 (*xordi_1_btc): Ditto.
21483 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21486 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21487 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21489 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21490 quoted strings, and make more translator-friendly.
21491 (darwin_rs6000_override_options): Likewise.
21492 (rs6000_option_override_internal): Likewise.
21493 (rs6000_return_in_memory): Fix overlong line.
21494 (init_cmulative_args): Use quoted strings, and make more
21495 translator-friendly.
21496 (rs6000_pass_by_reference): Fix overlong line.
21497 (def_builtin): Use quoted strings.
21498 (altivec_expand_predicate_builtin): Use quoted strings, and make
21499 more translator-friendly.
21500 (htm_expand_builtin): Use quoted strings.
21501 (cpu_expand_builtin): Use quoted strings, and make more
21502 translator-friendly.
21503 (altivec_expand_builtin): Likewise.
21504 (paired_expand_predicate_builtin): Likewise.
21505 (rs6000_invalid_builtin): Likewise.
21506 (builtin_function_type): Use quoted strings.
21507 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21508 more translator-friendly.
21509 (rs6000_trampoline_init): Likewise.
21510 (rs6000_handle_altivec_attribute): Likewise.
21511 (rs6000_inner_target_options): Use quoted strings.
21512 (rs6000_disable_incompatible_switches): Likewise.
21513 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21514 strings, and make more translator-friendly.
21515 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21517 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21519 PR tree-optimization/81799
21520 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21521 cond_expr to simple gimple operand.
21523 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21525 PR middle-end/46932
21526 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21528 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21532 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21533 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21534 TARGET_GASISR_PROLOGUES.
21535 * config/avr/avr.c (avr_option_override): Same.
21536 (avr_pass_pre_proep::execute): Same.
21538 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21541 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21542 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21544 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21546 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21547 define_insn_and_split. Split to a memory load from 0 in
21548 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21549 using PTR mode iterator.
21550 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21551 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21552 (*add_tp_<mode>): Redefine as define_insn_and_split.
21553 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21554 address space. Merge with *add_tp_x32 using PTR mode iterator.
21555 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21556 Split to an add with a memory load from 0 in
21557 DEFAULT_TLS_SEG_REG address space.
21559 2017-08-12 Andrew Pinski <apinski@cavium.com>
21561 * config/aarch64/aarch64-option-extensions.def (rdma):
21562 Fix feature string to what Linux prints out in /proc/cpuinfo.
21564 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21567 * dwarf2out.c (modified_type_die): For C typedef types that have
21568 an ultimate origin, process the ultimate origin instead of the
21570 (gen_typedef_die): Assert that input DECLs have no ultimate
21572 (gen_type_die_with_usage): For typedef variants that have an
21573 ultimate origin, just call gen_decl_die on the original DECL.
21574 (process_scope_var): Avoid creating DIEs for local typedefs and
21575 concrete static variables.
21577 2017-08-12 Alan Modra <amodra@gmail.com>
21581 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21582 match gnu-user.h startfile.
21583 (ENDFILE_LINUX_SPEC): Similarly.
21585 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21588 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21590 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21592 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21593 * config/aarch64/aarch64.md (mov<mode>): Change.
21594 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21595 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21596 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21598 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21600 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21601 for storage order barriers.
21603 2017-08-11 Martin Liska <mliska@suse.cz>
21606 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21607 variables of void type.
21609 2017-08-11 Martin Liska <mliska@suse.cz>
21611 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21612 TARGET_SUPPORTS_ALIASES.
21613 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21614 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21615 (optimize_weakref): Likewise.
21616 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21617 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21618 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21620 2017-08-11 Martin Liska <mliska@suse.cz>
21623 * config/i386/i386.c (make_resolver_func): Do complete
21624 refactoring of the function.
21626 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21629 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21630 * config/i386/i386.c (ix86_stack_protect_guard): Use
21631 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21632 * doc/invoke.texi (x86 Options): Document
21633 -mstack-protector-guard-symbol= option.
21635 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21637 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21638 * config/i386/i386.c (ix86_split_stack_guard): New function.
21639 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21640 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21641 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21642 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21643 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21644 (split_stack_space_check): Call ix86_split_stack_guard.
21646 2017-08-10 Martin Sebor <msebor@redhat.com>
21648 * print-tree.c (print_node): Print location using the established
21650 Replace spaces with colons.
21651 (debug_raw, debug): Ditto.
21653 2017-08-10 Martin Sebor <msebor@redhat.com>
21656 * pretty-print.c (pp_format): Correct the handling of %s precision.
21658 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21661 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21663 (ix86_finalize_stack_frame_flags): This. Also clear
21664 frame_pointer_needed if -fno-omit-frame-pointer is used without
21666 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21667 with ix86_finalize_stack_frame_flags.
21668 (ix86_expand_epilogue): Likewise.
21669 (ix86_expand_split_stack_prologue): Likewise.
21670 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21672 2017-08-10 Martin Liska <mliska@suse.cz>
21675 * c-attribs.c (handle_target_attribute):
21676 Report warning for an empty string argument of target attribute.
21678 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21681 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21683 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21684 or DECL_NONLOCAL labels.
21685 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21686 or DECL_NONLOCAL labels here.
21688 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21690 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21691 to indicate when early gimple folding has been disabled.
21692 (rs6000_gimple_fold_builtin): Add debug content.
21693 (rs6000_invalid_builtin): Fix whitespace.
21694 (rs6000_expand_builtin): Fix whitespace.
21695 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21697 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21700 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21701 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21703 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21705 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21706 (qdf24xx): Likewise.
21707 * config/aarch64/aarch64.md: Include falkor.md.
21708 * config/aarch64/falkor.md: New.
21710 2017-08-09 Marek Polacek <polacek@redhat.com>
21713 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21714 * diagnostic.c (emit_diagnostic): Add a comment.
21715 (emit_diagnostic_valist): New function.
21717 2017-08-09 Marek Polacek <polacek@redhat.com>
21720 * input.c (make_location): New overload.
21721 * input.h (make_location): Declare.
21723 2017-08-08 Alan Modra <amodra@gmail.com>
21724 H.J. Lu <hongjiu.lu@intel.com>
21727 * gcc.c (NO_PIE_SPEC): Delete.
21728 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21730 (LINK_PIE_SPEC): ..to here.
21731 (LINK_COMMAND_SPEC): Support -no-pie.
21732 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21733 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21735 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21736 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21737 (ENDFILE_CRTEND_SPEC): Similarly.
21739 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21742 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21743 (mstack-protector-guard-offset=): Ditto.
21744 * config/i386/i386.c (ix86_option_override): Handle
21745 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21747 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21748 ix86_stack_protect_guard_offset variables.
21749 (TARGET_STACK_PROTECT_GUARD): Always define.
21750 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21751 and -mstack-protector-guard-offset= options.
21753 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21755 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21756 boundary case for the last candidate.
21758 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21760 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21761 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21763 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21765 PR middle-end/19706
21766 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21767 * config/aarch64/aarch64-builtins.c
21768 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21769 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21770 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21772 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21773 Andrew Pinski <pinskia@gmail.com>
21775 PR middle-end/19706
21776 * internal-fn.def (XORSIGN): New.
21777 * optabs.def (xorsign_optab): New.
21778 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21779 (convert_expand_mult_copysign): New.
21780 (pass_optimize_widening_mul::execute): Call
21781 convert_expand_mult_copysign.
21783 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21785 PR tree-optimization/81354
21786 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21787 Insert on edges rather than explicitly creating landing pads.
21788 (analyze_candidates_and_replace): Commit edge inserts.
21790 2017-08-08 Richard Biener <rguenther@suse.de>
21792 PR middle-end/81719
21793 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21794 (expand_simple_operations): Also look through ADDR_EXPRs with
21795 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21797 2017-08-08 Richard Biener <rguenther@suse.de>
21799 PR tree-optimization/81723
21800 * tree-vect-slp.c (struct bst_traits): New hash traits.
21801 (bst_fail): New global.
21802 (vect_build_slp_tree_2): New worker, split out from ...
21803 (vect_build_slp_tree): ... this now wrapping it with using
21804 bst_fail set to cache SLP tree build fails. Properly handle
21806 (vect_analyze_slp_instance): Allocate and free bst_fail.
21808 2017-08-08 Martin Liska <mliska@suse.cz>
21811 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21812 LABEL_DECLs that can be from a different function.
21814 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21816 PR tree-optimization/81744
21817 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21818 loop's number of iterations.
21820 2017-08-08 Martin Liska <mliska@suse.cz>
21822 * asan.c: Include header files.
21823 * attribs.c (build_decl_attribute_variant): New function moved
21825 (build_type_attribute_qual_variant): Likewise.
21826 (cmp_attrib_identifiers): Likewise.
21827 (simple_cst_list_equal): Likewise.
21828 (omp_declare_simd_clauses_equal): Likewise.
21829 (attribute_value_equal): Likewise.
21830 (comp_type_attributes): Likewise.
21831 (build_type_attribute_variant): Likewise.
21832 (lookup_ident_attribute): Likewise.
21833 (remove_attribute): Likewise.
21834 (merge_attributes): Likewise.
21835 (merge_type_attributes): Likewise.
21836 (merge_decl_attributes): Likewise.
21837 (merge_dllimport_decl_attributes): Likewise.
21838 (handle_dll_attribute): Likewise.
21839 (attribute_list_equal): Likewise.
21840 (attribute_list_contained): Likewise.
21841 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21842 (lookup_attribute_by_prefix): Likewise.
21843 * bb-reorder.c: Include header files.
21844 * builtins.c: Likewise.
21845 * calls.c: Likewise.
21846 * cfgexpand.c: Likewise.
21847 * cgraph.c: Likewise.
21848 * cgraphunit.c: Likewise.
21849 * convert.c: Likewise.
21850 * dwarf2out.c: Likewise.
21851 * final.c: Likewise.
21852 * fold-const.c: Likewise.
21853 * function.c: Likewise.
21854 * gimple-expr.c: Likewise.
21855 * gimple-fold.c: Likewise.
21856 * gimple-pretty-print.c: Likewise.
21857 * gimple.c: Likewise.
21858 * gimplify.c: Likewise.
21859 * hsa-common.c: Likewise.
21860 * hsa-gen.c: Likewise.
21861 * internal-fn.c: Likewise.
21862 * ipa-chkp.c: Likewise.
21863 * ipa-cp.c: Likewise.
21864 * ipa-devirt.c: Likewise.
21865 * ipa-fnsummary.c: Likewise.
21866 * ipa-inline.c: Likewise.
21867 * ipa-visibility.c: Likewise.
21869 * lto-cgraph.c: Likewise.
21870 * omp-expand.c: Likewise.
21871 * omp-general.c: Likewise.
21872 * omp-low.c: Likewise.
21873 * omp-offload.c: Likewise.
21874 * omp-simd-clone.c: Likewise.
21875 * opts-global.c: Likewise.
21876 * passes.c: Likewise.
21877 * predict.c: Likewise.
21878 * sancov.c: Likewise.
21879 * sanopt.c: Likewise.
21880 * symtab.c: Likewise.
21881 * toplev.c: Likewise.
21882 * trans-mem.c: Likewise.
21883 * tree-chkp.c: Likewise.
21884 * tree-eh.c: Likewise.
21885 * tree-into-ssa.c: Likewise.
21886 * tree-object-size.c: Likewise.
21887 * tree-parloops.c: Likewise.
21888 * tree-profile.c: Likewise.
21889 * tree-ssa-ccp.c: Likewise.
21890 * tree-ssa-live.c: Likewise.
21891 * tree-ssa-loop.c: Likewise.
21892 * tree-ssa-sccvn.c: Likewise.
21893 * tree-ssa-structalias.c: Likewise.
21894 * tree-ssa.c: Likewise.
21895 * tree-streamer-in.c: Likewise.
21896 * tree-vectorizer.c: Likewise.
21897 * tree-vrp.c: Likewise.
21898 * tsan.c: Likewise.
21899 * ubsan.c: Likewise.
21900 * varasm.c: Likewise.
21901 * varpool.c: Likewise.
21902 * tree.c: Remove functions moved to attribs.[ch].
21903 * tree.h: Likewise.
21904 * config/aarch64/aarch64.c: Add attrs.h header file.
21905 * config/alpha/alpha.c: Likewise.
21906 * config/arc/arc.c: Likewise.
21907 * config/arm/arm.c: Likewise.
21908 * config/avr/avr.c: Likewise.
21909 * config/bfin/bfin.c: Likewise.
21910 * config/c6x/c6x.c: Likewise.
21911 * config/cr16/cr16.c: Likewise.
21912 * config/cris/cris.c: Likewise.
21913 * config/darwin.c: Likewise.
21914 * config/epiphany/epiphany.c: Likewise.
21915 * config/fr30/fr30.c: Likewise.
21916 * config/frv/frv.c: Likewise.
21917 * config/ft32/ft32.c: Likewise.
21918 * config/h8300/h8300.c: Likewise.
21919 * config/i386/winnt.c: Likewise.
21920 * config/ia64/ia64.c: Likewise.
21921 * config/iq2000/iq2000.c: Likewise.
21922 * config/lm32/lm32.c: Likewise.
21923 * config/m32c/m32c.c: Likewise.
21924 * config/m32r/m32r.c: Likewise.
21925 * config/m68k/m68k.c: Likewise.
21926 * config/mcore/mcore.c: Likewise.
21927 * config/microblaze/microblaze.c: Likewise.
21928 * config/mips/mips.c: Likewise.
21929 * config/mmix/mmix.c: Likewise.
21930 * config/mn10300/mn10300.c: Likewise.
21931 * config/moxie/moxie.c: Likewise.
21932 * config/msp430/msp430.c: Likewise.
21933 * config/nds32/nds32-isr.c: Likewise.
21934 * config/nds32/nds32.c: Likewise.
21935 * config/nios2/nios2.c: Likewise.
21936 * config/nvptx/nvptx.c: Likewise.
21937 * config/pa/pa.c: Likewise.
21938 * config/pdp11/pdp11.c: Likewise.
21939 * config/powerpcspe/powerpcspe.c: Likewise.
21940 * config/riscv/riscv.c: Likewise.
21941 * config/rl78/rl78.c: Likewise.
21942 * config/rx/rx.c: Likewise.
21943 * config/s390/s390.c: Likewise.
21944 * config/sh/sh.c: Likewise.
21945 * config/sol2.c: Likewise.
21946 * config/sparc/sparc.c: Likewise.
21947 * config/spu/spu.c: Likewise.
21948 * config/stormy16/stormy16.c: Likewise.
21949 * config/tilegx/tilegx.c: Likewise.
21950 * config/tilepro/tilepro.c: Likewise.
21951 * config/v850/v850.c: Likewise.
21952 * config/vax/vax.c: Likewise.
21953 * config/visium/visium.c: Likewise.
21954 * config/xtensa/xtensa.c: Likewise.
21956 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21959 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21960 constraints since the -mupper-regs-* switches have been
21962 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21963 into a vector from a double word element that was extracted from
21964 another vector, and eliminate extra XXPERMDI instructions.
21965 (vsx_concat_<mode>_2): Likewise.
21966 (vsx_concat_<mode>_3): Likewise.
21967 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21968 concat to allow optimizing inserts from previous extracts.
21970 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21972 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21973 memory reference to a SSP offset in TLS address space.
21974 (ix86_print_operand) <case '@'>: Remove.
21975 (ix86_print_operand_punct_valid_p): Remove '@' code.
21976 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21977 UNSPEC_SP_TLS_TEST.
21978 (stack_tls_protect_set_<mode>): Remove.
21979 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21980 (stack_tls_protect_test_<mode>): Remove.
21981 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21983 2017-08-07 Olivier Hainque <hainque@adacore.com>
21986 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21988 2017-08-07 Douglas Rupp <rupp@adacore.com>
21990 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21991 variable was referenced as multidir in command.
21993 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21996 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21999 2017-08-07 Martin Liska <mliska@suse.cz>
22001 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
22002 * config/rl78/rl78.c: Add include of attribs.h.
22003 * config/sh/sh.c: Likewise.
22004 * config/v850/v850.c: Likewise.
22006 2017-08-07 Tom de Vries <tom@codesourcery.com>
22008 PR middle-end/78266
22009 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
22011 2017-08-07 Martin Liska <mliska@suse.cz>
22013 * config/mips/mips.c: Include attribs.h.
22015 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
22018 * doc/invoke.texi: Document change in behvaior for -Ofast for
22021 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
22023 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
22025 (aarch64_pop_regs): Likewise.
22026 (aarch64_gen_load_pair): Likewise.
22027 (aarch64_save_callee_saves): Likewise.
22028 (aarch64_restore_callee_saves): Likewise.
22030 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22032 * config/i386/i386.c: Revert the last change.
22034 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22037 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
22039 (ix86_finalize_stack_frame_flags): This. Also clear
22040 frame_pointer_needed if -fno-omit-frame-pointer is used without
22042 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
22043 with ix86_finalize_stack_frame_flags.
22044 (ix86_expand_epilogue): Likewise.
22045 (ix86_expand_split_stack_prologue): Likewise.
22047 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22050 * config/i386/i386.c (get_builtin_code_for_version): Set priority
22051 to P_AES for Westmere.
22053 2017-08-07 Jonathan Yong <10walls@gmail.com>
22055 * config/i386/mingw.opt (fset-stack-executable): Removed.
22056 * config/i386/cygming.opt (fset-stack-executable): Moved
22058 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
22060 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
22062 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
22064 2017-08-07 Marek Polacek <polacek@redhat.com>
22066 PR middle-end/81737
22067 * fold-const.c (fold_indirect_ref_1): Check type_domain.
22069 2017-08-07 Martin Liska <mliska@suse.cz>
22071 * attribs.h (canonicalize_attr_name): New function.
22072 (cmp_attribs): Move from c-format.c and adjusted.
22073 (is_attribute_p): Moved from tree.h.
22074 * tree-inline.c: Add new includes.
22075 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
22076 (private_is_attribute_p): Remove.
22077 (private_lookup_attribute): Likewise.
22078 (private_lookup_attribute_by_prefix): Simplify.
22079 (remove_attribute): Use is_attribute_p.
22080 * tree.h: Remove removed declarations.
22082 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22084 PR middle-end/81698
22085 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
22086 instead of computing it in the function. Formatting fix.
22087 (expand_case): Don't rely on default_edge being the first edge,
22088 clear it if removing it, pass default_edge to
22089 emit_case_dispatch_table.
22090 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
22093 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
22095 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
22096 insn in the function, emit NOP after the insn.
22098 2017-08-06 Tom de Vries <tom@codesourcery.com>
22100 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
22103 2017-08-06 Tom de Vries <tom@codesourcery.com>
22105 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
22108 2017-08-04 Yury Gribov <tetra2005@gmail.com>
22110 PR tree-optimization/57371
22111 * match.pd: New pattern.
22113 2017-08-04 Marek Polacek <polacek@redhat.com>
22115 PR middle-end/81695
22116 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
22117 perform the computation in offset_int.
22119 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22121 PR tree-optimization/81136
22122 * tree-vectorizer.h: Include tree-hash-traits.h.
22123 (vec_base_alignments): New typedef.
22124 (vec_info): Add a base_alignments field.
22125 (vect_record_base_alignments): Declare.
22126 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
22128 (DR_IS_CONDITIONAL_IN_STMT): New macro.
22129 (create_data_ref): Add an is_conditional_in_stmt argument.
22130 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
22131 the is_conditional_in_stmt field.
22132 (data_ref_loc): Add an is_conditional_in_stmt field.
22133 (get_references_in_stmt): Set the is_conditional_in_stmt field.
22134 (find_data_references_in_stmt): Update call to create_data_ref.
22135 (graphite_find_data_references_in_stmt): Likewise.
22136 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
22137 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
22138 (vect_record_base_alignment): New function.
22139 (vect_record_base_alignments): Likewise.
22140 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
22141 for nested statements even if we fail to compute a misalignment.
22142 Use pooled base alignments for unconditional references.
22143 (vect_find_same_alignment_drs): Compare base addresses instead
22145 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
22146 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
22148 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22150 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
22151 Add an explicit name for the enum. Use auto_vec for slp_instances
22152 and grouped_stores.
22153 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
22155 (_bb_vec_info): Add a constructor and destructor.
22156 (vinfo_for_stmt): Return NULL for uids of -1 as well.
22157 (destroy_loop_vec_info): Delete.
22158 (vect_destroy_datarefs): Likewise.
22159 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
22160 (vec_info::vec_info): New function.
22161 (vec_info::~vec_info): Likewise.
22162 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22163 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22164 destroy_loop_vec_info.
22165 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22166 (_loop_vec_info::_loop_vec_info): ...this.
22167 (destroy_loop_vec_info): Replace with...
22168 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22169 the stmt_vec_infos. Leave handling of vec_info information to its
22170 destructor. Remove explicit vector releases.
22171 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22172 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22173 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22174 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22175 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22176 (destroy_bb_vec_info): Replace with...
22177 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22178 information to its destructor.
22179 (vect_slp_analyze_bb_1): Use new and delete instead of
22180 new_bb_vec_info and destroy_bb_vec_info.
22181 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22184 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22186 * tree-data-ref.h (subscript): Add access_fn field.
22187 (data_dependence_relation): Add could_be_independent_p.
22188 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22189 (same_access_functions): Move to tree-data-ref.c.
22190 * tree-data-ref.c (ref_contains_union_access_p): New function.
22191 (access_fn_component_p): Likewise.
22192 (access_fn_components_comparable_p): Likewise.
22193 (dr_analyze_indices): Add a reference to access_fn_component_p.
22194 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22196 (constant_access_functions): Likewise.
22197 (add_other_self_distances): Likewise.
22198 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22199 (initialize_data_dependence_relation): Use XCNEW and remove
22200 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22201 of access functions that have the same type. Allow the
22202 subsequence to end with different bases in some circumstances.
22203 Record the chosen access functions in SUB_ACCESS_FN.
22204 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22205 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22206 (subscript_dependence_tester_1): Likewise dra and drb.
22207 (build_classic_dist_vector): Update calls accordingly.
22208 (subscript_dependence_tester): Likewise.
22209 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22210 DDR_COULD_BE_INDEPENDENT_P.
22211 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22212 comp_alias_ddrs instead of may_alias_ddrs.
22213 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22215 (vect_analyze_data_ref_dependence): Use it if
22216 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22217 distance vectors if that fails.
22218 (dependence_distance_ge_vf): New function.
22219 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22220 LOOP_VINFO_MAY_ALIAS_DDRS.
22222 2017-08-04 Richard Biener <rguenther@suse.de>
22224 PR middle-end/81705
22225 * fold-const.c (fold_binary_loc): Properly restrict
22226 minus_var0 && minus_var1 case when associating undefined overflow
22229 2017-08-04 Tom de Vries <tom@codesourcery.com>
22231 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22233 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22235 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22236 Don't start diagnostic messages with a capital letter.
22237 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22239 (rs6000_invalid_builtin): Likewise.
22240 (rs6000_trampoline_init): Likewise.
22242 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22245 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22246 after setting changeable df flags.
22248 2017-08-03 Richard Biener <rguenther@suse.de>
22250 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22251 up if the use is in USE - X.
22253 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22255 * toplev.c (dumpfile.h): New include.
22256 (internal_error_reentered): New static function. Use it...
22257 (internal_error_function): ...here to handle reentered internal_error.
22259 2017-08-03 Richard Biener <rguenther@suse.de>
22261 PR middle-end/81148
22262 * fold-const.c (split_tree): Add minus_var and minus_con
22263 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22264 here but always use minus_*.
22265 (associate_trees): Assert we never associate with MINUS_EXPR
22266 and NULL first operand. Do not recurse for PLUS_EXPR operands
22267 when associating as MINUS_EXPR either.
22268 (fold_binary_loc): Track minus_var and minus_con.
22270 2017-08-03 Tom de Vries <tom@codesourcery.com>
22273 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22274 ACCEL_COMPILER, apply finish_options on
22275 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22277 2017-08-03 Tom de Vries <tom@codesourcery.com>
22280 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22281 function_entry_patch_area_size > 0.
22283 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22286 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22287 instead of 10??LU, perform unit multiplication in wide_int,
22288 don't change alloc_object_size_limit if the limit is larger
22291 PR tree-optimization/81655
22292 PR tree-optimization/81588
22293 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22294 the case when ranges[i].low and high are 1 for unsigned type with
22297 PR middle-end/81052
22298 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22299 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22301 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22303 * tree-vrp.h: Add include guard.
22305 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22308 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22309 (ud2): New insn pattern.
22310 * config/i386/i386.c (ix86_expand_epilogue):
22311 For naked functions, generate ud2 instead of trap insn.
22313 2017-08-02 Marek Polacek <polacek@redhat.com>
22316 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22318 2017-08-02 Tom de Vries <tom@codesourcery.com>
22319 Cesar Philippidis <cesar@codesourcery.com>
22321 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22322 Add missing edge probabilities.
22324 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22326 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22327 Correct endianness.
22329 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22331 PR middle-end/79499
22332 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22333 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22334 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22335 prologue_seq sequences - if any.
22337 2017-08-02 Richard Biener <rguenther@suse.de>
22339 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22340 via vectors if supported, integer extracts via punning if supported
22341 or otherwise vector extracts.
22343 2017-08-02 Richard Biener <rguenther@suse.de>
22345 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22347 (bitmap_insert_into_set): ... this.
22349 2017-08-02 Richard Biener <rguenther@suse.de>
22351 PR tree-optimization/81633
22353 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22355 PR tree-optimization/71752
22356 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22358 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22360 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22361 (machine_function::call_ms2sysv_pad_out): Remove field.
22362 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22363 (ix86_compute_frame_layout): Likewise.
22365 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22368 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22369 attribute with interrupt attribute.
22371 2017-08-01 Andrew Pinski <apinski@cavium.com>
22373 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22374 BIT_INSERT_EXPR's operand 1
22375 to see if the types precision matches.
22377 2017-08-01 Martin Liska <mliska@suse.cz>
22379 PR middle-end/70140
22380 * builtins.c (expand_builtin_memcpy_args): Remove.
22381 (expand_builtin_memcpy): Call newly added function
22382 expand_builtin_memory_copy_args.
22383 (expand_builtin_memcpy_with_bounds): Likewise.
22384 (expand_builtin_mempcpy): Remove last argument.
22385 (expand_builtin_mempcpy_with_bounds): Likewise.
22386 (expand_builtin_memory_copy_args): New function created from
22387 expand_builtin_mempcpy_args with small modifications.
22388 (expand_builtin_mempcpy_args): Remove.
22389 (expand_builtin_stpcpy): Remove unused argument.
22390 (expand_builtin): Likewise.
22391 (expand_builtin_with_bounds): Likewise.
22393 2017-08-01 Martin Liska <mliska@suse.cz>
22396 Make mempcpy more optimal (PR middle-end/70140).
22398 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22401 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22402 __builtin_vec_cmpne verify both arguments are compatible vectors
22403 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22404 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22405 move computation of aligned to after checking the argument types.
22409 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22410 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22413 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22414 Doug Rupp <rupp@adacore.com>
22415 Olivier Hainque <hainque@adacore.com>
22417 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22418 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22420 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22421 for TARGET_OS_CPP_BUILTIN.
22422 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22423 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22425 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22426 passing required abi options to the assembler for EABI configurations.
22427 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22428 of .text.hot and .text.unlikely sections for kernel modules when
22429 using ARM style exceptions.
22430 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22431 options. Add EXTRA_CC1_SPEC.
22432 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22434 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22436 (ARM_TARGET2_DWARF_FORMAT): Define.
22437 * config/arm/t-vxworks: Adjust multilib control to removal of the
22438 Diab command line options.
22440 2017-08-01 Martin Liska <mliska@suse.cz>
22442 PR gcov-profile/81561
22443 * gcov.c (unblock): Make unblocking safe as we need to preserve
22444 index correspondence of blocks and block_lists.
22446 2017-08-01 Richard Biener <rguenther@suse.de>
22448 PR tree-optimization/81181
22449 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22450 (compute_antic): ... end of iteration here.
22452 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22454 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22455 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22456 (ftree-slp-vectorize): Likewise.
22457 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22458 can no longer be set independent of flag_tree_loop_vectorize.
22459 * omp-general.c (emp_max_vf): Likewise.
22460 * opts.c (enable_fdo_optimizations): Remove references to
22461 flag_tree_vectorize, these are now implicit.
22462 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22463 and leave it for the options machinery.
22465 2017-08-01 Martin Liska <mliska@suse.cz>
22467 PR middle-end/70140
22468 * builtins.c (expand_builtin_memcpy_args): Remove.
22469 (expand_builtin_memcpy): Call newly added function
22470 expand_builtin_memory_copy_args.
22471 (expand_builtin_memcpy_with_bounds): Likewise.
22472 (expand_builtin_mempcpy): Remove last argument.
22473 (expand_builtin_mempcpy_with_bounds): Likewise.
22474 (expand_builtin_memory_copy_args): New function created from
22475 expand_builtin_mempcpy_args with small modifications.
22476 (expand_builtin_mempcpy_args): Remove.
22477 (expand_builtin_stpcpy): Remove unused argument.
22478 (expand_builtin): Likewise.
22479 (expand_builtin_with_bounds): Likewise.
22481 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22484 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22485 print "ds:" only for immediates in generic address space.
22487 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22490 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22491 (ix86_function_ok_for_sibcall): Return false for naked functions.
22493 2017-08-01 Richard Biener <rguenther@suse.de>
22495 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22496 (compute_antic): Seed worklist with exit block predecessors.
22497 * cfganal.c (dfs_find_deadend): For a cycle return the source
22498 of the edge closing it.
22500 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22502 * config/aarch64/aarch64.c
22503 (aarch64_can_const_movi_rtx_p): Move 0 check.
22505 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22507 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22508 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22510 * match.pd: Ditto in address comparison pattern.
22512 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22514 PR tree-optimization/81627
22515 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22516 closed ssa form for store-store chain.
22518 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22520 PR tree-optimization/81620
22521 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22522 for store-store chain.
22524 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22526 PR tree-optimization/81588
22527 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22528 ranges[i].in_p, invert comparison code ccode. For >/>=,
22529 swap rhs1 and rhs2 and comparison code unconditionally,
22530 for </<= don't do that. Don't swap rhs1/rhs2 again if
22531 ranges[i].in_p, instead invert comparison code ccode if
22532 opcode or oe->rank is BIT_IOR_EXPR.
22535 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22536 a direct optab to conversion optab.
22537 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22538 with GET_MODE_INNER as last argument instead of optab_handler.
22539 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22540 vector extraction if possible and optab is available.
22541 * expr.c (store_constructor): Use convert_optab_handler instead
22542 of optab_handler. Use vector initialization from smaller
22543 vectors if possible and optab is available.
22544 * tree-vect-stmts.c (vectorizable_load): Likewise.
22545 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22546 now have two modes.
22547 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22548 of vec_init from half-sized vectors with the same element mode.
22549 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22550 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22551 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22552 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22553 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22554 after mode in gen_vec_extract* calls.
22555 (vec_extract<mode>): Renamed to ...
22556 (vec_extract<mode><ssescalarmodelower>): ... this.
22557 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22558 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22559 element mode after mode in gen_vec_init* calls.
22560 (VEC_INIT_HALF_MODE): New mode iterator.
22561 (vec_init<mode>): Renamed to ...
22562 (vec_init<mode><ssescalarmodelower>): ... this.
22563 (vec_init<mode><ssehalfvecmodelower>): New expander.
22564 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22565 (vec_extractv2sfsf): ... this.
22566 (vec_initv2sf): Renamed to ...
22567 (vec_initv2sfsf): ... this.
22568 (vec_extractv2si): Renamed to ...
22569 (vec_extractv2sisi): ... this.
22570 (vec_initv2si): Renamed to ...
22571 (vec_initv2sisi): ... this.
22572 (vec_extractv4hi): Renamed to ...
22573 (vec_extractv4hihi): ... this.
22574 (vec_initv4hi): Renamed to ...
22575 (vec_initv4hihi): ... this.
22576 (vec_extractv8qi): Renamed to ...
22577 (vec_extractv8qiqi): ... this.
22578 (vec_initv8qi): Renamed to ...
22579 (vec_initv8qiqi): ... this.
22580 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22581 (vec_init<mode>): Renamed to ...
22582 (vec_init<mode><VEC_base_l>): ... this.
22583 (vec_extract<mode>): Renamed to ...
22584 (vec_extract<mode><VEC_base_l>): ... this.
22585 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22586 (vec_initv2sfsf): ... this.
22587 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22588 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22589 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22590 element mode after mode in gen_vec_init* calls.
22591 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22592 (vec_init<mode><Vel>): ... this.
22593 (vec_extract<mode>): Renamed to ...
22594 (vec_extract<mode><Vel>): ... this.
22595 * config/aarch64/iterators.md (Vel): New mode attribute.
22596 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22597 Add element mode after mode in gen_vec_extract* calls.
22598 * config/s390/vector.md (non_vec_l): New mode attribute.
22599 (vec_extract<mode>): Renamed to ...
22600 (vec_extract<mode><non_vec_l>): ... this.
22601 (vec_init<mode>): Renamed to ...
22602 (vec_init<mode><non_vec_l>): ... this.
22603 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22604 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22606 * config/arm/iterators.md (V_elem_l): New mode attribute.
22607 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22608 (vec_extract<mode><V_elem_l>): ... this.
22609 (vec_extractv2di): Renamed to ...
22610 (vec_extractv2didi): ... this.
22611 (vec_init<mode>): Renamed to ...
22612 (vec_init<mode><V_elem_l>): ... this.
22613 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22614 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22615 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22616 Add element mode after gen_vec_extract* calls.
22617 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22618 (vec_init<mode><unitmode>): ... this.
22619 (vec_extract<mode>): Renamed to ...
22620 (vec_extract<mode><unitmode>): ... this.
22621 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22622 (vec_init<mode><unitmode>): ... this.
22623 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22624 (vec_initv2sfsf): ... this.
22625 (vec_extractv2sf): Renamed to ...
22626 (vec_extractv2sfsf): ... this.
22627 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22628 Add element mode after gen_vec_extract* calls.
22629 * config/mips/mips.md (unitmode): New mode iterator.
22630 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22631 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22632 * config/spu/spu.md (inner_l): New mode attribute.
22633 (vec_init<mode>): Renamed to ...
22634 (vec_init<mode><inner_l>): ... this.
22635 (vec_extract<mode>): Renamed to ...
22636 (vec_extract<mode><inner_l>): ... this.
22637 * config/sparc/sparc.md (veltmode): New mode iterator.
22638 (vec_init<VMALL:mode>): Renamed to ...
22639 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22640 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22641 (vec_initv2sisi): ... this.
22642 (vec_initv2sf): Renamed to ...
22643 (vec_initv2sfsf): ... this.
22644 (vec_extractv2sf): Renamed to ...
22645 (vec_extractv2sfsf): ... this.
22646 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22647 (vec_init<mode>): Renamed to ...
22648 (vec_init<mode><VEC_base_l>): ... this.
22649 (vec_extract<mode>): Renamed to ...
22650 (vec_extract<mode><VEC_base_l>): ... this.
22651 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22652 (vec_initv2sfsf): ... this.
22653 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22654 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22655 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22656 gen_vec_init* calls.
22658 2017-08-01 Richard Biener <rguenther@suse.de>
22660 PR tree-optimization/81297
22661 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22662 TREE_OVERFLOW from INTEGER_CSTs.
22664 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22666 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22668 2017-07-31 Carl Love <cel@us.ibm.com>
22670 * config/rs6000/rs6000-c: Add support for built-in functions
22671 vector signed char vec_xl_be (signed long long, signed char *);
22672 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22673 vector signed int vec_xl_be (signed long long, signed int *);
22674 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22675 vector signed long long vec_xl_be (signed long long, signed long long *);
22676 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22677 vector signed short vec_xl_be (signed long long, signed short *);
22678 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22679 vector double vec_xl_be (signed long long, double *);
22680 vector float vec_xl_be (signed long long, float *);
22681 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22682 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22683 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22685 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22686 (altivec_expand_builtin): Add switch statement to call
22687 altivec_expand_xl_be for each builtin.
22688 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22689 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22690 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22691 __builtin_vsx_le_be_v16qi.
22692 * doc/extend.texi: Update the built-in documentation file for the
22693 new built-in functions.
22695 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22698 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22700 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22702 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22704 * config.gcc: Add z14.
22705 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22706 CPU model numbers for z13s and z14.
22707 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22709 * config/s390/s390-opts.h (enum processor_type): Rename
22710 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22711 * config/s390/s390.c (processor_table): Add field for CPU name to
22712 be passed to Binutils.
22713 (s390_asm_output_machine_for_arch): Use the new field in
22714 processor_table for Binutils.
22715 (s390_expand_builtin): Replace arch12 with z14.
22716 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22717 (s390_get_sched_attrmask): Likewise.
22718 (s390_get_unit_mask): Likewise.
22719 * config/s390/s390.opt: Add z14 to processor_type enum.
22721 2017-07-31 Martin Jambor <mjambor@suse.cz>
22724 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22725 regardless of optimization level.
22727 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22728 Martin Liska <mliska@suse.cz>
22730 * predict.def: Remove old comment and adjust probability.
22731 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22732 PREDICT statements.
22734 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22737 * config/i386/i386.c (ix86_function_naked): New function.
22738 (ix86_can_use_return_insn_p): Return false for naked functions.
22739 (ix86_expand_prologue): Skip prologue for naked functions.
22740 (ix86_expand_epilogue): Skip epilogue for naked functions
22741 and emit trap instruction.
22742 (ix86_warn_func_return): New function.
22743 (ix86_attribute_table): Add "naked" attribute specification.
22744 (TARGET_WARN_FUNC_RETURN): Define.
22745 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22747 2017-07-31 Martin Liska <mliska@suse.cz>
22749 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22750 (dump_gimple_bb_header): Always dump BB info.
22751 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22753 2017-07-31 Martin Liska <mliska@suse.cz>
22756 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22757 also with current_function_decl non-null equality.
22759 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22762 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22763 change type to the element type, instead add eltype variable and
22764 use it where we are interested in the element type.
22766 PR tree-optimization/81603
22767 * ipa-polymorphic-call.c
22768 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22769 offset arithmetic in offset_int, bail out if the resulting bit offset
22770 doesn't fit into shwi.
22772 2017-07-31 Martin Liska <mliska@suse.cz>
22774 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22775 (gimplify_save_expr): Fix comment.
22777 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22780 * config/i386/i386.c (ix86_function_arg): Update arguments for
22782 (ix86_compute_frame_layout): Set the initial stack offset to
22783 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22784 INCOMING_FRAME_SP_OFFSET.
22785 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22786 stack before exception handler returns.
22787 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22788 the 'ERROR_CODE' for exception handler.
22790 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22792 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22793 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22794 (ASM_OUTPUT_REG_POP): Ditto.
22795 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22796 instead of asm_fprintf to output pure string.
22798 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22800 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22801 to imported_module_or_decl hook.
22802 (debug_nothing_tree_tree_tree_bool): Remove.
22803 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22804 * debug.c (do_nothing_debug_hooks): Use
22805 debug_nothing_tree_tree_tree_bool_bool instead of
22806 debug_nothing_tree_tree_tree_bool.
22807 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22808 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22809 * sdbout.c (sdb_debug_hooks): Likewise.
22810 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22811 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22813 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22814 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22815 attribute, don't add anything.
22817 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22819 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22820 (fold_build2_stat_loc): Likewise.
22821 (fold_build3_stat_loc): Likewise.
22822 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22823 (fold_build1_loc): Remove macro.
22824 (fold_build2_loc): Likewise.
22825 (fold_build3_loc): Likewise.
22827 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22829 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22830 (gimple_build_debug_bind_source_stat): Likewise.
22831 * gimple.h (gimple_build_debug_bind): Remove macro.
22832 (gimple_build_debug_bind_source): Likewise.
22834 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22836 * bitmap.c (bitmap_alloc): Adjust.
22837 (bitmap_gc_alloc): Likewise.
22838 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22840 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22842 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22843 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22844 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22845 (bitmap_gc_alloc_stat): Likewise.
22846 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22848 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22850 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22851 * rtl.h (shallow_copy_rtx): Remove macro.
22853 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22855 * emit-rtl.c (gen_raw_REG): Adjust.
22856 * gengenrtl.c (gendef): Likewise.
22857 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22858 * rtl.h (rtx_alloc): Remove macro.
22860 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22862 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22863 (build_tree_list_stat): Likewise.
22864 * tree.h (build_tree_list): Remove macro.
22865 (build_tree_list_vec): Likewise.
22867 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22869 * tree.c (make_vector_stat): Remove _stat from name.
22870 (build_vector_stat): Likewise.
22871 * tree.h (make_vector_stat): Remove macro.
22872 (build_vector_stat): Likewise.
22874 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22876 * tree.h (build_var_debug_value): Remove prototype.
22878 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22880 * tree.c (tree_cons_stat): Remove _stat from name.
22881 * tree.h (tree_cons): Remove macro.
22883 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22885 * tree.c (build_vl_exp_stat): Remove _stat from name.
22886 * tree.h (build_vl_exp): Remove macro.
22888 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22890 * tree.c (build_decl_stat): Remove _stat from name.
22891 * tree.h (build_decl): Remove macro.
22893 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22895 * gimple.c (gimple_build_with_ops_stat): Adjust.
22896 (gimple_alloc_stat): Remove _stat from name.
22897 * gimple.h (gimple_alloc): Remove macro.
22899 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22901 * tree.c (make_tree_vec_stat): Remove _stat from name.
22902 (grow_tree_vec_stat): Likewise.
22903 * tree.h (make_tree_vec_stat): Adjust prototype.
22904 (grow_tree_vec_stat): Likewise.
22905 (make_tree_vec): Remove macro.
22906 (grow_tree_vec): Likewise.
22908 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22910 * fold-const.c (fold_build1_stat_loc): Adjust.
22911 (fold_build2_stat_loc): Likewise.
22912 (fold_build3_stat_loc): Likewise.
22913 * tree.c (build0_stat): Remove _stat from name.
22914 (build1_stat): Likewise.
22915 (build2_stat): Likewise.
22916 (build3_stat): Likewise.
22917 (build4_stat): Likewise.
22918 (build5_stat): Likewise.
22919 * tree.h (build1_loc): Remove macro, and rename _stat function
22921 (build2_loc): Likewise.
22922 (build3_loc): Likewise.
22923 (build4_loc): Likewise.
22924 (build5_loc): Likewise.
22926 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22928 * tree.c (make_int_cst_stat): Remove _stat from name.
22929 * tree.h (make_int_cst_stat): Adjust prototype.
22930 (make_int_cst): Remove macro.
22932 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22934 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22935 * tree.h (make_tree_binfo_stat): Adjust prototype.
22936 (make_tree_binfo): Remove.
22938 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22940 * tree.c (copy_node_stat): Rename to copy_node.
22941 (build_distinct_type_copy): Adjust.
22942 * tree.h (copy_node_stat): Adjust prototype.
22943 (copy_node): Remove macro.
22945 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22947 * tree.c (make_node_stat): rename to make_node.
22948 (build_tree_list_stat): Adjust.
22949 (build0_stat): Likewise.
22950 (build2_stat): Likewise.
22951 (build3_stat): Likewise.
22952 (build4_stat): Likewise.
22953 (build5_stat): Likewise.
22954 (build_decl_stat): Likewise.
22955 * tree.h (make_node_stat): Adjust prototype.
22956 (make_node): remove macro.
22958 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22960 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22961 (PPC_FEATURE2_SCV): Likewise.
22962 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22964 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22966 * config/aarch64/aarch64.c
22967 (aarch64_internal_mov_immediate): Add new special pattern.
22968 * config/aarch64/aarch64.md (*movdi_aarch64):
22969 Add reg/32bit const mov case.
22971 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22972 Richard Sandiford <richard.sandiford@linaro.org>
22974 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22975 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22976 Add integer and movi cases.
22977 (movi-split-hf-df-sf split, fp16): New.
22978 (enabled): Added TARGET_FP_F16INST.
22979 * config/aarch64/iterators.md (GPF_HF): New.
22980 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22982 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22984 * config/aarch64/aarch64.c
22985 (aarch64_simd_container_mode): Add prototype.
22986 (aarch64_expand_mov_immediate): Add HI support.
22987 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22988 (aarch64_can_const_movi_rtx_p): New.
22989 (aarch64_preferred_reload_class):
22990 Remove restrictions of using FP registers for certain SIMD operations.
22991 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22992 (aarch64_valid_floating_const): Add integer move validation.
22993 (aarch64_simd_imm_scalar_p): Remove.
22994 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22995 (aarch64_legitimate_constant_p): Expand list of supported cases.
22996 * config/aarch64/aarch64-protos.h
22997 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22998 (aarch64_reinterpret_float_as_int): New.
22999 (aarch64_simd_imm_scalar_p): Remove.
23000 * config/aarch64/constraints.md (Uvi): New.
23001 (Dd): Split into Ds and new Dd.
23002 * config/aarch64/aarch64.md (*movsi_aarch64):
23004 (*movdi_aarch64): Add SIMD mov case.
23006 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23008 * tree-predcom.c: (struct chain): Handle store-store chain in which
23009 stores for elimination only store loop invariant values.
23010 (execute_pred_commoning_chain): Ditto.
23011 (prepare_initializers_chain_store_elim): Ditto.
23012 (prepare_finalizers): Ditto.
23013 (is_inv_store_elimination_chain): New function.
23014 (initialize_root_vars_store_elim_1): New function.
23016 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23018 * tree-predcom.c: Revise general description of the pass.
23019 (enum chain_type): New enum type for store elimination.
23020 (struct chain): New field supporting store elimination.
23021 (struct component): Ditto.
23022 (dump_chain): Dump store-stores chain.
23023 (release_chain): Release resources.
23024 (split_data_refs_to_components): Compute and create component
23025 contains only stores for elimination.
23026 (get_chain_last_ref_at): New function.
23027 (make_invariant_chain): Initialization.
23028 (make_rooted_chain): Specify chain type in parameter and record it.
23029 (add_looparound_copies): Skip for store-stores chain.
23030 (determine_roots_comp): Compute type of chain and pass it to
23032 (initialize_root_vars_store_elim_2): New function.
23033 (finalize_eliminated_stores): New function.
23034 (remove_stmt): Handle store for elimination.
23035 (execute_pred_commoning_chain): Execute predictive commoning on
23036 store-store chains.
23037 (determine_unroll_factor): Skip unroll for store-stores chain.
23038 (prepare_initializers_chain_store_elim): New function.
23039 (prepare_initializers_chain): Hanlde store-store chain.
23040 (prepare_finalizers_chain, prepare_finalizers): New function.
23041 (tree_predictive_commoning_loop): Return integer value indicating
23042 if loop is unrolled or lcssa form is corrupted.
23043 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
23045 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23047 * tree-predcom.c (initialize_root): Delete.
23048 (execute_pred_commoning_chain): Initialize root vars and replace
23049 reference of non-combined chain directly, rather than call above
23052 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23054 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
23055 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
23057 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23059 * tree-predcom.c (struct chain): New field init_seq.
23060 (release_chain): Release init_seq.
23061 (prepare_initializers_chain): Record intialization stmts in above
23063 (insert_init_seqs): New function.
23064 (tree_predictive_commoning_loop): Call insert_init_seqs.
23066 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23068 * tree-predcom.c (determine_roots_comp): Skip trivial components.
23070 2017-07-28 Richard Biener <rguenther@suse.de>
23072 * match.pd: Remove superfluous :c.
23073 * genmatch.c (simplify::id): Add member.
23074 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
23076 (current_id): New global.
23077 (dt_node::parent): Move from ...
23078 (dt_operand::parent): ... here. Add for_id member.
23079 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
23080 (decision_tree::find_node): Relax order requirement when
23081 merging DT_TRUE nodes to ones inbetween the current simplify
23082 and the one we try to merge with. Add diagnostic whenever
23083 we need to enforce pattern order by not merging.
23084 (decision_tree::insert): Set current_id.
23085 (decision_tree::print_node): Dump parent node and for_id.
23086 (parser::last_id): Add member.
23087 (parser::push_simplify): Assign unique id.
23088 (parser::parser): Initialize last_id.
23090 2017-07-28 Martin Liska <mliska@suse.cz>
23093 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
23094 gimple_build_debug_bind.
23096 2017-07-28 Richard Biener <rguenther@suse.de>
23098 PR tree-optimization/81502
23099 * match.pd: Add pattern combining BIT_INSERT_EXPR with
23101 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
23103 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
23104 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
23105 for BIT_FIELD_REF args.
23106 * fold-const.c (make_bit_field_ref): Likewise.
23107 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
23109 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23112 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
23113 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
23114 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
23115 Or it into SANITIZER_UNDEFINED.
23116 * ubsan.c: Include gimple-fold.h and varasm.h.
23117 (ubsan_expand_ptr_ifn): New function.
23118 (instrument_pointer_overflow): New function.
23119 (maybe_instrument_pointer_overflow): New function.
23120 (instrument_object_size): Formatting fix.
23121 (pass_ubsan::execute): Call instrument_pointer_overflow
23122 and maybe_instrument_pointer_overflow.
23123 * internal-fn.c (expand_UBSAN_PTR): New function.
23124 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
23125 * sanitizer.def (__ubsan_handle_pointer_overflow,
23126 __ubsan_handle_pointer_overflow_abort): New builtins.
23127 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
23128 * internal-fn.def (UBSAN_PTR): New internal function.
23129 * opts.c (sanitizer_opts): Add pointer-overflow.
23130 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
23131 * fold-const.c (build_range_check): Compute pointer range check in
23132 integral type if pointer arithmetics would be needed. Formatting
23135 2017-07-28 Martin Liska <mliska@suse.cz>
23138 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
23139 parameters that are of a variable-length.
23141 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
23143 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
23145 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
23146 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
23147 (CRT_CALL_STATIC_FUNCTION): Likewise.
23148 (ASM_DEFAULT_SPEC): New define.
23149 (ASM_SPEC32): Likewise.
23150 (ASM_SPEC64): Likewise.
23151 (ASM_SPEC_COMMON): Likewise.
23152 (ASM_SPEC): Likewise.
23153 (INVALID_64BIT): Likewise.
23154 (LINK_OS_DEFAULT_SPEC): Likewise.
23155 (LINK_OS_SPEC32): Likewise.
23156 (LINK_OS_SPEC64): Likewise.
23157 (POWERPC_LINUX): Likewise.
23158 (PTRDIFF_TYPE): Likewise.
23159 (RESTORE_FP_PREFIX): Likewise.
23160 (RESTORE_FP_SUFFIX): Likewise.
23161 (SAVE_FP_PREFIX): Likewise.
23162 (SAVE_FP_SUFFIX): Likewise.
23163 (SIZE_TYPE): Likewise.
23164 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23165 (TARGET_64BIT): Likewise.
23166 (TARGET_64BIT): Likewise.
23167 (TARGET_AIX): Likewise.
23168 (WCHAR_TYPE_SIZE): Likewise.
23169 (WCHAR_TYPE): Undefine.
23170 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23171 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23172 (CPP_OS_RTEMS_SPEC): Delete.
23173 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23174 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23176 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23178 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23180 PR tree-optimization/81578
23181 * tree-parloops.c (build_new_reduction): Bail out if
23182 reduction_code isn't one of the standard OpenMP reductions.
23183 Move the details printing after that decision.
23185 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23187 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23188 related to reload_in_progress.
23189 (splat_input_operand): Likewise.
23190 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23192 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23194 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23195 (TARGET_INSTANTIATE_DECLS): Likewise.
23196 (legitimate_indexed_address_p): Delete reload_in_progress code.
23197 (rs6000_debug_legitimate_address_p): Likewise.
23198 (rs6000_eliminate_indexed_memrefs): Likewise.
23199 (rs6000_emit_le_vsx_store): Likewise.
23200 (rs6000_emit_move_si_sf_subreg): Likewise.
23201 (rs6000_emit_move): Likewise.
23202 (register_to_reg_type): Likewise.
23203 (rs6000_pre_atomic_barrier): Likewise.
23204 (rs6000_machopic_legitimize_pic_address): Likewise.
23205 (rs6000_allocate_stack_temp): Likewise.
23206 (rs6000_address_for_fpconvert): Likewise.
23207 (rs6000_address_for_altivec): Likewise.
23208 (rs6000_secondary_memory_needed_rtx): Delete function.
23209 (rs6000_check_sdmode): Likewise.
23210 (rs6000_alloc_sdmode_stack_slot): Likewise.
23211 (rs6000_instantiate_decls): Likewise.
23212 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23213 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23214 Delete reload_in_progress.
23215 (*vec_reload_and_plus_<mptrsize>): Likewise.
23216 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23217 (vsx_div_v2di): Likewise.
23218 (vsx_udiv_v2di): Likewise.
23220 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23222 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23223 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23224 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23225 (rs6000_debug_reg_global): Delete print of LRA status.
23226 (rs6000_option_override_internal): Delete dead LRA related code.
23227 (rs6000_lra_p): Delete function.
23228 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23230 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23232 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23233 * config/riscv/rtems.h: New file.
23235 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23236 Sudakshina Das <sudi.das@arm.com>
23238 * config/aarch64/aarch64.md
23239 (define_split for and<mode>3nr_compare): Move
23240 non aarch64_logical_operand to a register.
23241 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23242 register immediate operand to a register.
23243 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23245 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23247 PR middle-end/81564
23248 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23250 2017-07-27 Richard Biener <rguenther@suse.de>
23252 PR tree-optimization/81573
23253 PR tree-optimization/81494
23254 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23255 multi defuse cycle case.
23257 2017-07-27 Richard Biener <rguenther@suse.de>
23259 PR tree-optimization/81571
23260 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23263 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23265 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23266 earlier and only if MASK_FPU is set. Adjust formatting.
23268 2017-07-27 Martin Liska <mliska@suse.cz>
23270 * opt-functions.awk: Add validation of value of Init.
23271 * optc-gen.awk: Pass new argument.
23273 2017-07-27 Martin Liska <mliska@suse.cz>
23275 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23276 Fix wrong condition.
23278 2017-07-27 Martin Liska <mliska@suse.cz>
23280 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23281 BBs and edges seen by autoFDO.
23283 2017-07-27 Richard Biener <rguenther@suse.de>
23285 PR tree-optimization/81502
23286 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23287 with incompatible but same sized type.
23288 (execute_update_addresses_taken): Likewise.
23290 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23292 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23293 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23295 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23298 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23299 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23300 Change s_operand to memory_operand.
23302 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23304 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23305 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23306 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23307 Emit instructions rather than returning an expression. Handle TFmode
23308 and KFmode by casting to TImode.
23309 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23310 (rs6000_emit_le_vsx_store): Likewise.
23311 * config/rs6000/vsx.md (VSX_TI): New iterator.
23312 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23313 (*vsx_le_undo_permute_<mode>): Likewise.
23314 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23315 emit the split sequence.
23316 (*vsx_le_perm_store_<mode>): Likewise.
23318 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23320 PR tree-optimization/81555
23321 PR tree-optimization/81556
23322 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23323 if true, force CHANGED for the recursive invocation.
23324 (reassociate_bb): Remember original length of ops array, pass
23325 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23327 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23328 attributes for noipa attribute. For naked attribute use
23329 lookup_attribute first before lookup_attribute_spec.
23330 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23332 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23333 for functions with noipa attribute.
23334 (cgraph_externally_visible_p): Return true for functions with noipa
23336 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23337 for functions with noipa attribute.
23338 * doc/extend.texi: Document noipa function attribute.
23339 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23340 also for functions with noipa attribute.
23341 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23343 2017-07-26 Andrew Pinski <apinski@cavium.com>
23345 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23346 vec_unalign_load_cost and vec_unalign_store_cost.
23348 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23350 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23351 -mvsx-small-integer option.
23352 (ISA_3_0_MASKS_IEEE): Likewise.
23353 (OTHER_VSX_VECTOR_MASKS): Likewise.
23354 (POWERPC_MASKS): Likewise.
23355 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23356 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23357 code, only testing for DImode being allowed in non-VSX floating
23359 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23360 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23362 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23363 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23364 TARGET_P8_VECTOR test.
23365 (rs6000_secondary_reload_simple_move): Likewise.
23366 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23367 since TARGET_P9_VECTOR was already tested.
23368 (rs6000_opt_masks): Remove -mvsx-small-integer.
23369 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23370 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23372 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23373 test for TARGET_VEXTRACTUB was used, and that uses
23375 (p9 extract splitter): Likewise.
23376 (vsx_extract_<mode>_di_p9): Likewise.
23377 (vsx_extract_<mode>_store_p9): Likewise.
23378 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23379 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23380 the elimination of TARGET_VSX_SMALL_INTEGER.
23381 (vsx_extract_<mode>_p8): Likewise.
23382 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23383 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23384 (vsx_set_<mode>_p9): Likewise.
23385 (vsx_set_v4sf_p9): Likewise.
23386 (vsx_set_v4sf_p9_zero): Likewise.
23387 (vsx_insert_extract_v4sf_p9): Likewise.
23388 (vsx_insert_extract_v4sf_p9_2): Likewise.
23389 * config/rs6000/rs6000.md (sign extend splitter): Change
23390 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23391 (floatsi<mode>2_lfiwax_mem): Likewise.
23392 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23393 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23394 since a test for TARGET_P9_VECTOR was used.
23395 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23396 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23397 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23398 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23399 TARGET_P8_VECTOR test.
23400 (fix_trunc<mode>si2_stfiwx): Likewise.
23401 (fix_trunc<mode>si2_internal): Likewise.
23402 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23403 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23405 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23406 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23407 TARGET_P8_VECTOR test.
23408 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23409 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23410 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23412 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23413 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23414 since a test for TARGET_P9_VECTOR was used.
23415 (splitter for loading small constants): Likewise.
23417 2017-07-26 Andrew Pinski <apinski@cavium.com>
23419 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23422 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23425 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23426 (fp_valid_at): Likewise.
23428 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23430 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23431 (qdf24xx_addrcost_table): Likewise.
23432 (cortexa57_tunings): Update to use generic_branch_cost.
23433 (cortexa72_tunings): Likewise.
23434 (cortexa73_tunings): Likewise.
23435 (qdf24xx_tunings): Likewise.
23437 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23439 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23440 (thunderx2t99_branch_cost): Likewise.
23441 (cortexa35_tunings): Update to use generic_branch_cost.
23442 (cortexa53_tunings): Likewise.
23443 (cortexa57_tunings): Likewise.
23444 (cortexa72_tunings): Likewise.
23445 (cortexa73_tunings): Likewise.
23446 (thunderx2t99_tunings): Likewise.
23448 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23450 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23451 (sparc_option_override): Honour MASK_FSMULD.
23452 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23453 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23454 * config/sparc/sparc.opt (mfsmuld): New option.
23455 * doc/invoke.texi (mfsmuld): Document option.
23457 2017-07-26 Marek Polacek <polacek@redhat.com>
23459 PR middle-end/70992
23460 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23462 2017-07-26 Richard Biener <rguenther@suse.de>
23464 * gimple-match-head.c (do_valueize): Return OP if valueize
23466 (get_def): New helper to get at the def stmt of a SSA name
23467 if valueize allows.
23468 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23469 do_valueize to get at the def stmt.
23470 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23472 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23474 PR middle-end/46932
23475 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23477 2017-07-26 Martin Liska <mliska@suse.cz>
23480 * function.c (expand_function_start): Make expansion of
23481 nonlocal_goto_save_area after parm_birth_insn.
23483 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23485 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23486 from all CPU target flags enable members.
23488 2017-07-26 Richard Biener <rguenther@suse.de>
23490 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23491 (decision_tree::gen): Make 'type' const.
23492 (write_predicate): Likewise.
23494 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23496 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23497 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23498 (rs6000_option_override_internal): Likewise.
23499 (rs6000_expand_vector_set): Likewise.
23500 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23501 (TARGET_UPPER_REGS_SF): Likewise.
23502 (TARGET_UPPER_REGS_DI): Likewise.
23503 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23504 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23505 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23506 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23507 (Splitters for DI constants in Altivec registers): Likewise.
23508 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23509 (vsx_set_v4sf_p9): Likewise.
23510 (vsx_set_v4sf_p9_zero): Likewise.
23511 (vsx_insert_extract_v4sf_p9): Likewise.
23512 (vsx_insert_extract_v4sf_p9_2): Likewise.
23514 2017-07-25 Carl Love <cel@us.ibm.com>
23516 * doc/extend.texi: Update the built-in documentation file for the
23517 existing built-in functions
23518 vector signed char vec_cnttz (vector signed char);
23519 vector unsigned char vec_cnttz (vector unsigned char);
23520 vector signed short vec_cnttz (vector signed short);
23521 vector unsigned short vec_cnttz (vector unsigned short);
23522 vector signed int vec_cnttz (vector signed int);
23523 vector unsigned int vec_cnttz (vector unsigned int);
23524 vector signed long long vec_cnttz (vector signed long long);
23525 vector unsigned long long vec_cnttz (vector unsigned long long);
23527 2017-07-25 Andrew Pinski <apinski@cavium.com>
23529 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23530 accesses where the use is for the first operand of a BIT_INSERT.
23532 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23535 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23536 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23538 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23540 * config/i386/gstabs.h: Delete.
23541 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23543 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23545 * config/i386/i386.c (ix86_decompose_address): Do not check for
23546 register RTX when looking at index_reg or base_reg.
23547 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23549 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23551 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23552 to update EH info here.
23554 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23556 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23558 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23560 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23562 2017-07-25 Torsten Duwe <duwe@suse.de>
23564 * common.opt: Introduce -fpatchable-function-entry
23565 command line option, and its variables function_entry_patch_area_size
23566 and function_entry_patch_area_start.
23567 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23568 including a two-value parser.
23569 * target.def (print_patchable_function_entry): New target hook.
23570 * targhooks.h (default_print_patchable_function_entry): New function.
23571 * targhooks.c (default_print_patchable_function_entry): Likewise.
23572 * toplev.c (process_options): Switch off IPA-RA if
23573 patchable function entries are being generated.
23574 * varasm.c (assemble_start_function): Look at the
23575 patchable-function-entry command line switch and current
23576 function attributes and maybe generate NOP instructions by
23577 calling the print_patchable_function_entry hook.
23578 * doc/extend.texi: Document patchable_function_entry attribute.
23579 * doc/invoke.texi: Document -fpatchable_function_entry
23580 command line option.
23581 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23583 * doc/tm.texi: Re-generate.
23585 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23588 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23589 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23591 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23593 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23595 2017-07-25 Richard Biener <rguenther@suse.de>
23597 PR tree-optimization/81455
23598 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23599 not walk in cycles when looking for guards.
23601 2017-07-25 Richard Biener <rguenther@suse.de>
23603 PR tree-optimization/81529
23604 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23605 when optimizing backedge uses.
23607 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23609 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23611 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23612 to dl_section_ref. On AIX, append an expression to subtract
23613 the size of the section length to dl_section_ref.
23615 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23617 * configure.ac: If any of the config.* scripts fail, exit 1.
23618 * configure: Regenerate.
23620 2017-07-25 Richard Biener <rguenther@suse.de>
23622 PR middle-end/81546
23623 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23624 of immediate uses, be more verbose on errors.
23626 2017-07-25 Richard Biener <rguenther@suse.de>
23628 PR tree-optimization/81510
23629 * tree-vect-loop.c (vect_is_simple_reduction): When the
23630 reduction stmt is not inside the loop bail out.
23632 2017-07-25 Richard Biener <rguenther@suse.de>
23634 PR tree-optimization/81303
23635 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23636 profitability check against LOOP_VINFO_NITERSM1.
23638 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23640 * domwalk.c (cmp_bb_postorder): Simplify.
23641 (sort_bbs_postorder): New function. Use it...
23642 (dom_walker::walk): ...here to optimize common cases.
23644 2017-07-25 Martin Liska <mliska@suse.cz>
23647 * ipa-visibility.c (function_and_variable_visibility): Make the
23648 redirection just on target that supports aliasing.
23649 Fix GNU coding style.
23651 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23654 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23656 * config/arm/rtems.h: Likewise.
23657 * config/bfin/rtems.h: Likewise.
23658 * config/i386/rtemself.h: Likewise.
23659 * config/lm32/rtems.h: Likewise.
23660 * config/m32c/rtems.h: Likewise.
23661 * config/m68k/rtemself.h: Likewise.
23662 * config/microblaze/rtems.h: Likewise.
23663 * config/mips/rtems.h: Likewise.
23664 * config/moxie/rtems.h: Likewise.
23665 * config/nios2/rtems.h: Likewise.
23666 * config/powerpcspe/rtems.h: Likewise.
23667 * config/rs6000/rtems.h: Likewise.
23668 * config/rtems.h: Likewise.
23669 * config/sh/rtems.h: Likewise.
23670 * config/sh/rtemself.h: Likewise.
23671 * config/sparc/rtemself.h: Likewise.
23673 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23676 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23677 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23678 * tree-ssa-structalias.c (alias_get_name): Same.
23680 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23683 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23684 instructions if no du chain is found.
23686 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23688 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23690 2017-07-25 Richard Biener <rguenther@suse.de>
23692 PR middle-end/81505
23693 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23696 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23698 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23699 upper-regs options.
23700 (ISA_2_7_MASKS_SERVER): Likewise.
23701 (ISA_3_0_MASKS_IEEE): Likewise.
23702 (OTHER_P8_VECTOR_MASKS): Likewise.
23703 (OTHER_VSX_VECTOR_MASKS): Likewise.
23704 (POWERPC_MASKS): Likewise.
23705 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23706 duplicate list of options.
23707 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23708 explicit -mupper-regs options.
23709 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23710 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23711 alias for -mupper-regs-df.
23712 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23713 (rs6000_init_hard_regno_mode_ok): Likewise.
23714 (rs6000_option_override_internal): Likewise.
23715 (rs6000_opt_masks): Likewise.
23716 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23717 options in terms of whether -mvsx or -mpower8-vector was used.
23718 (TARGET_UPPER_REGS_DI): Likewise.
23719 (TARGET_UPPER_REGS_SF): Likewise.
23720 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23721 -mupper-regs-* options.
23723 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23725 * passes.c (emergency_dump_function): Print some empty lines and a
23726 header before the RTL dump.
23728 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23730 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23732 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23735 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23736 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23738 2017-07-24 Carl Love <cel@us.ibm.com>
23740 * config/rs6000/rs6000-c.c: Add support for built-in functions
23741 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23742 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23743 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23744 vec_extract_fp_from_shortl): Add defines for the two builtins.
23745 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23746 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23748 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23749 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23750 * doc/extend.texi: Update the built-in documentation file for the
23751 new built-in function.
23753 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23756 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23758 * doc/generic.texi: Likewise.
23759 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23760 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23762 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23764 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23765 (aarch64_mls_elt_merge<mode>): Likewise.
23767 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23769 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23770 having __cxa_atexit.
23772 2017-07-23 Michael Collison <michael.collison@arm.com>
23774 * config/arm/arm.c (arm_option_override): Deprecate
23775 use of -mstructure-size-boundary.
23776 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23777 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23779 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23782 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23783 Reduce cost estimate for direct moves.
23785 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23788 * config/i386/i386.c (ix86_option_override_internal): Disable
23789 BMI, BMI2 and TBM instructions for -m16.
23791 2017-07-21 Carl Love <cel@us.ibm.com>
23793 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23794 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23795 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23796 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23797 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23798 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23799 VMULOSW): New enum "unspec" values.
23800 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23801 altivec_vmulosw): New patterns.
23802 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23803 VMULOSW): Add definitions.
23805 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23807 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23808 (qdf24xx): Likewise.
23809 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23810 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23811 (AARCH64_FL_V8_1): Renumber.
23812 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23813 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23814 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23815 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23816 rdma to feature modifiers list.
23818 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23820 PR middle-end/56727
23821 * ipa-visibility (function_and_variable_visibility): Convert
23822 recursive PLT call to direct call if appropriate.
23824 2017-07-21 Andrew Pinski <apinski@cavium.com>
23826 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23827 operand 1 to see if the types precision matches.
23828 * fold-const.c (operand_equal_p): Likewise.
23830 2017-07-21 Richard Biener <rguenther@suse.de>
23832 PR tree-optimization/81303
23833 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23834 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23835 (vect_peeling_hash_get_lowest_cost): Adjust.
23836 (vect_enhance_data_refs_alignment): Likewise. Use
23837 vect_get_peeling_costs_all_drs to compute the penalty for no
23838 peeling to match up costs.
23840 2017-07-21 Richard Biener <rguenther@suse.de>
23842 PR tree-optimization/81500
23843 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23844 we didn't identify a reduction path.
23846 2017-07-21 Tom de Vries <tom@codesourcery.com>
23847 Cesar Philippidis <cesar@codesourcery.com>
23849 PR gcov-profile/81442
23850 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23853 2017-07-21 Tom de Vries <tom@codesourcery.com>
23856 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23858 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23859 nvptx_override_options_after_change.
23861 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23863 * dwarf2out.c (output_file_names): Avoid double testing for
23864 dwarf_version >= 5.
23866 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23868 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23870 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23872 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23874 (try_crossjump_to_edge): Do not punt on partitioned functions.
23876 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23878 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23879 Put all BBs reachable only via paths crossing cold region to cold
23881 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23883 2016-07-21 Richard Biener <rguenther@suse.de>
23885 PR tree-optimization/81303
23886 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23887 into account prologue and epilogue iterations when raising
23888 min_profitable_iters to sth at least covering one vector iteration.
23890 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23892 * config/arm/arm.c (arm_test_cpu_arch_dat):
23895 2017-07-20 Nathan Sidwell <nathan@acm.org>
23897 Remove TYPE_METHODS.
23898 * tree.h (TYPE_METHODS): Delete.
23899 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23900 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23901 (dbxout_type_methods): Scan TYPE_FIELDS.
23902 (dbxout_type): Don't check TYPE_METHODS here.
23903 * function.c (use_register_for_decl): Always ignore register for
23904 class types when not optimizing.
23905 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23906 * tree.c (free_lang_data_in_type): Stitch out member functions and
23907 templates from TYPE_FIELDS.
23908 (build_distinct_type_copy, verify_type_variant,
23909 verify_type): Member fns are on TYPE_FIELDS.
23910 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23911 * tree-pretty-print.c (dump_generic_node): Likewise.
23913 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23916 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23917 V2TImode and V4TImode.
23918 (ix86_expand_vector_extract): Likewise.
23919 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23920 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23921 (ssescalarmode): Handle V4TImode and V2TImode.
23922 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23923 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23924 (VEXTRACTI128_MODE): New mode iterator.
23925 (splitter for *vec_extractv?ti first element): New.
23926 (VEC_INIT_MODE): New mode iterator.
23927 (vec_init<mode>): Consolidate 3 expanders into one using
23928 VEC_INIT_MODE mode iterator.
23930 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23932 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23933 non_spilled_static_chain_regno_p.
23935 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23937 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23939 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23941 * bb-reorder.c (connect_traces): Allow copying of blocks within
23944 2017-07-20 Richard Biener <rguenther@suse.de>
23946 * gimple.h (gimple_phi_result): Add gphi * overload.
23947 (gimple_phi_result_ptr): Likewise.
23948 (gimple_phi_arg): Likewise. Adjust index assert to only
23949 allow actual argument accesses rather than all slots available
23951 (gimple_phi_arg_def): Add gphi * overload.
23952 * tree-phinodes.c (make_phi_node): Initialize only actual
23954 (resize_phi_node): Clear memory not covered by old node,
23955 do not initialize excess argument slots.
23956 (reserve_phi_args_for_new_edge): Initialize new argument slot
23959 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23961 PR tree-optimization/81388
23963 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23965 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23966 by removing computation of may_be_zero.
23968 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23969 Tom de Vries <tom@codesourcery.com>
23971 PR middle-end/81030
23972 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23973 when gimple level profile disagrees with what RTL expander did.
23975 2017-07-20 Richard Biener <rguenther@suse.de>
23977 PR tree-optimization/61171
23978 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23979 (vect_analyze_stmt): Add slp instance parameter.
23980 (vectorizable_reduction): Likewise.
23981 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23982 (vect_is_simple_reduction): Deal with chains not detected
23983 as SLP reduction chain, specifically not properly associated
23984 chains containing a mix of plus/minus.
23985 (get_reduction_op): Remove.
23986 (get_initial_defs_for_reduction): Simplify, pass in whether
23987 this is a reduction chain, pass in the SLP node for the PHIs.
23988 (vect_create_epilog_for_reduction): Get the SLP instance as
23990 (vectorizable_reduction): Get the SLP instance as arg.
23991 During analysis remember the SLP node with the PHIs in the
23992 instance. Simplify getting at the vectorized reduction PHIs.
23993 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23994 through SLP instance.
23995 (vect_slp_analyze_operations): Likewise.
23996 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23997 (vect_transform_stmt): Likewise.
23999 2017-07-20 Tom de Vries <tom@codesourcery.com>
24001 PR tree-optimization/81489
24002 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
24003 read of phi arg location to before loop that modifies phi.
24005 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24007 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
24010 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24012 PR middle-end/81331
24013 * except.c (execute): Fix ordering issue.
24015 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24017 PR rtl-optimization/81423
24018 * combine.c (make_compound_operation_int): Don't try to optimize
24019 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
24021 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24023 PR rtl-optimization/81423
24024 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
24025 with a constant that is -1 in the truncated to mode.
24027 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24029 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
24030 (determine_unlikely_bbs): ... here.
24031 * predict.h (propagate_unlikely_bbs_forward): Declare.
24032 * cfgexpand.c (pass_expand::execute): Use it.
24033 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
24035 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
24036 propagate_unlikely_bbs_forward.
24038 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24040 PR middle-end/81331
24041 * except.c (maybe_add_nop_after_section_switch): New function.
24044 2017-07-19 Tom de Vries <tom@codesourcery.com>
24046 * gimple.h (gimple_phi_set_arg): Make assert more strict.
24048 2017-07-19 Tom de Vries <tom@codesourcery.com>
24050 * gimple.h (gimple_phi_arg): Make assert more strict.
24052 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
24054 * config.gcc (powerpc*-*-*): Add mmintrin.h.
24055 * config/rs6000/mmintrin.h: New file.
24056 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
24058 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24060 PR tree-optimization/81346
24061 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
24063 2017-07-19 Tom de Vries <tom@codesourcery.com>
24065 * config/nvptx/nvptx.md (VECIM): Add V2DI.
24067 2017-07-19 Tom de Vries <tom@codesourcery.com>
24069 * config/nvptx/nvptx-modes.def: Add V2DImode.
24070 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
24071 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
24072 (nvptx_output_mov_insn): Handle lack of mov.b128.
24073 (nvptx_print_operand): Handle 'H' and 'L' codes.
24074 (nvptx_vector_mode_supported): Allow V2DImode.
24075 (nvptx_preferred_simd_mode): New function.
24076 (nvptx_data_alignment): New function.
24077 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
24078 nvptx_preferred_simd_mode.
24079 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
24081 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
24083 2017-07-19 Tom de Vries <tom@codesourcery.com>
24085 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
24086 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
24087 (nvptx_vector_mode_supported): New function. Allow V2SImode.
24088 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
24089 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
24090 (mov<VECIM>_insn): New define_insn.
24091 (define_expand "mov<VECIM>): New define_expand.
24093 2017-07-19 Tom de Vries <tom@codesourcery.com>
24095 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
24097 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24099 PR tree-optimization/81346
24100 * fold-const.h (fold_div_compare, range_check_type): Declare.
24101 * fold-const.c (range_check_type): New function.
24102 (build_range_check): Use range_check_type.
24103 (fold_div_compare): No longer static, rewritten into
24104 a match.pd helper function.
24105 (fold_comparison): Don't call fold_div_compare here.
24106 * match.pd (X / C1 op C2): New optimization using fold_div_compare
24107 as helper function.
24109 2017-07-19 Nathan Sidwell <nathan@acm.org>
24111 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
24112 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
24113 * tree.c (find_decls_types_r, verify_type): Use
24114 TYPE_{MIN,MAX}_VALUE_RAW.
24115 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
24116 (hash_tree): Likewise.
24117 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
24119 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
24122 2017-07-18 Tom de Vries <tom@codesourcery.com>
24124 PR middle-end/81464
24125 * omp-expand.c (expand_omp_for_static_chunk): Handle
24126 equal-argument loop exit phi.
24128 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
24131 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
24132 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
24133 operand 2 predicate.
24134 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
24135 operand 2 predicate.
24136 (ror,rol -> rorx splitters): Use const_int_operand as
24137 operand 2 predicate.
24139 2017-06-18 Richard Biener <rguenther@suse.de>
24141 PR tree-optimization/81410
24142 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
24143 the gap in the ! slp_perm SLP case after each group.
24145 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24147 PR middle-end/81463
24148 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
24151 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24153 PR middle-end/81462
24154 * predict.c (set_even_probabilities): Cleanup; do not affect
24155 probabilities that are already known.
24156 (combine_predictions_for_bb): Call even when count is set.
24158 2017-07-18 Nathan Sidwell <nathan@acm.org>
24160 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24163 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24166 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24167 optimization for loop niter analysis.
24169 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24172 * config/avr/avr.c (avr_optimize_casesi): Don't use
24173 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24175 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24177 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24178 body_cost_vec from _vect_peel_extended_info.
24179 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24180 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24183 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24185 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24187 2017-07-18 Richard Biener <rguenther@suse.de>
24189 PR tree-optimization/80620
24190 PR tree-optimization/81403
24191 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24192 info when re-using a VN table entry.
24194 2017-07-18 Richard Biener <rguenther@suse.de>
24196 PR tree-optimization/81418
24197 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24198 vectype_in. Verify that with lane-reducing reduction operations
24199 we have a single def-use cycle.
24201 2017-07-17 Carl Love <cel@us.ibm.com>
24203 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24205 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24206 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24207 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24208 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24209 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24210 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24211 VMULOSW): New enum "unspec" values.
24212 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24213 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24214 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24215 altivec_vmulosw): New patterns.
24216 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24217 VMULOSW): Add definitions.
24219 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24221 * config/alpha/alpha.c: Include predict.h.
24223 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24225 * tree-vrp.c (compare_assert_loc): Fix comparison function
24226 to return predictable results.
24228 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24230 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24232 (subdi3): Likewise.
24233 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24234 * doc/invoke.texi (mexpand-adddi): Update text.
24236 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24238 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24239 that also clobbers the CC register. The old expand code is moved
24241 (*arc_clzsi2): ... here.
24242 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24243 the CC register. The old expand code is moved to ...
24244 (arc_ctzsi2): ... here.
24246 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24248 * config/arc/arc.opt (mindexed-loads): Use initial value
24249 TARGET_INDEXED_LOADS_DEFAULT.
24250 (mauto-modify-reg): Use initial value
24251 TARGET_AUTO_MODIFY_REG_DEFAULT.
24252 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24253 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24254 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24255 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24257 2017-07-17 Martin Liska <mliska@suse.cz>
24260 * opts.c (finish_options): Do not allow -fgnu-tm
24261 w/ -fsanitize={kernel-,}address. Say sorry.
24263 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24266 * tree-loop-distribution.c (classify_partition): Only assert on
24267 numer of iterations.
24268 (merge_dep_scc_partitions): Delete prameter. Update function call.
24269 (distribute_loop): Remove code handling loop with unknown niters.
24270 (pass_loop_distribution::execute): Skip loop with unknown niters.
24272 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24275 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24276 function sort_partitions_by_post_order.
24278 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24280 PR tree-optimization/81374
24281 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24282 the max index of basic blocks, rather than number of basic blocks.
24284 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24286 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24288 (arc_legitimate_pic_operand_p): Likewise.
24289 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24291 (arc_needs_pcl_p): Likewise.
24292 (arc_legitimate_pc_offset_p): Likewise.
24293 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24294 function is also used in constrains.md.
24295 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24296 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24297 PLUS. Only return true/false in known cases, otherwise assert.
24298 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24299 is already called in arc_legitimate_constant_p.
24300 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24302 (LEGITIMATE_PIC_OPERAND_P): Use
24303 arc_raw_symbolic_reference_mentioned_p function.
24304 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24309 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24310 Andrew Burgess <andrew.burgess@embecosm.com>
24312 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24313 (arc_return_address_register): New function.
24314 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24315 (arc_handle_fndecl_attribute): Add naked attribute.
24316 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24317 (TARGET_WARN_FUNC_RETURN): Likewise.
24318 (arc_allocate_stack_slots_for_args): New function.
24319 (arc_warn_func_return): Likewise.
24320 (machine_function): Change type fn_type.
24321 (arc_compute_function_type): Consider new naked function type,
24322 change function return type.
24323 (arc_must_save_register): Adapt to handle new
24324 arc_compute_function_type's return type.
24325 (arc_expand_prologue): Likewise.
24326 (arc_expand_epilogue): Likewise.
24327 (arc_return_address_regs): Delete.
24328 (arc_return_address_register): New function.
24329 (arc_epilogue_uses): Use above function.
24330 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24331 (arc_function_type): Change encoding, add naked type.
24332 (ARC_INTERRUPT_P): Change to handle the new encoding.
24333 (ARC_FAST_INTERRUPT_P): Likewise.
24334 (ARC_NORMAL_P): Define.
24335 (ARC_NAKED_P): Likewise.
24336 (arc_compute_function_type): Delete prototype.
24337 * config/arc/arc.md (in_ret_delay_slot): Use
24338 arc_return_address_register function.
24339 (simple_return): Likewise.
24340 (p_return_i): Likewise.
24342 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24344 PR tree-optimization/81428
24345 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24346 can't be built for those types.
24348 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24350 Remove stuff dead since r239246.
24352 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24353 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24354 (avr_inform_devices): Remove dead stuff.
24356 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24358 * config/arm/arm_neon.h: Fix softp typo.
24360 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24362 PR tree-optimization/81365
24363 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24364 aggregate moves onto bb predecessor edges, make sure there are no
24365 loads that could alias the lhs in between the start of bb and the
24368 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24371 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24372 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24373 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24375 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24377 PR tree-optimization/81396
24378 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24379 (init_symbolic_number): Initialize it to 1.
24380 (perform_symbolic_merge): Add n_ops from both operands into the new
24382 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24383 without base_addr as useless if they need more than one operation.
24384 (bswap_replace): Handle !bswap case for NULL base_addr.
24386 2017-07-17 Tom de Vries <tom@codesourcery.com>
24389 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24392 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24394 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24395 conditional builtin define __FIX_LEON3FT_B2BST.
24397 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24399 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24400 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24403 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24405 PR rtl-optimization/81424
24406 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24407 to remove potential trapping from operands if -fnon-call-exceptions.
24409 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24411 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24412 profile_proability for scalling.
24413 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24415 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24417 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24419 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24421 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24422 fixpoint arithmetics.
24424 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24426 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24427 fixpoint arithmetics.
24429 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24431 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24432 fixpoint arithmetics.
24434 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24436 * profile-count.h (profile_probability::from_reg_br_prob_note,
24437 profile_probability::to_reg_br_prob_note): New functions.
24438 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24439 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24440 * predict.c (probability_reliable_p): Update.
24441 (edge_probability_reliable_p): Update.
24442 (br_prob_note_reliable_p): Update.
24443 (invert_br_probabilities): Update.
24444 (add_reg_br_prob_note): New function.
24445 (combine_predictions_for_insn): Update.
24446 * asan.c (asan_clear_shadow): Update.
24447 * cfgbuild.c (compute_outgoing_frequencies): Update.
24448 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24449 (update_br_prob_note): Update.
24450 (rtl_verify_edges): Update.
24451 (purge_dead_edges): Update.
24452 (fixup_reorder_chain): Update.
24453 * emit-rtl.c (try_split): Update.
24454 * ifcvt.c (cond_exec_process_insns): Update.
24455 (cond_exec_process_if_block): Update.
24456 (dead_or_predicable): Update.
24457 * internal-fn.c (expand_addsub_overflow): Update.
24458 (expand_neg_overflow): Update.
24459 (expand_mul_overflow): Update.
24460 * loop-doloop.c (doloop_modify): Update.
24461 * loop-unroll.c (compare_and_jump_seq): Update.
24462 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24463 * predict.h: Update.
24464 * reorg.c (mostly_true_jump): Update.
24466 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24467 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24468 * config/arc/arc.c: (emit_unlikely_jump): Update.
24469 * config/arm/arm.c: (emit_unlikely_jump): Update.
24470 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24471 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24472 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24473 (ix86_print_operand): Update.
24474 (ix86_split_fp_branch): Update.
24475 (predict_jump): Update.
24476 * config/ia64/ia64.c (ia64_print_operand): Update.
24477 * config/mmix/mmix.c (mmix_print_operand): Update.
24478 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24479 (rs6000_expand_split_stack_prologue): Update.
24480 * config/rs6000/rs6000.c: Update.
24481 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24482 (s390_expand_vec_movstr): Update.
24483 (s390_expand_cs_tdsi): Update.
24484 (s390_expand_split_stack_prologue): Update.
24485 * config/sh/sh.c (sh_print_operand): Update.
24486 (expand_cbranchsi4): Update.
24487 (expand_cbranchdi4): Update.
24488 * config/sparc/sparc.c (output_v9branch): Update.
24489 * config/spu/spu.c (get_branch_target): Update.
24490 (ea_load_store_inline): Update.
24491 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24492 * config/tilepro/tilepro.c: Update.
24494 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24496 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24497 (gimplify_save_expr): Likewise.
24499 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24501 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24503 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24505 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24507 * timevar.def (TV_IPA_FNSUMMARY): Define.
24509 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24511 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24512 to back store errata sensitive sequence from being generated.
24513 (sqrtdf2_fix): Likewise.
24515 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24517 * tree-ssa-threadupdate.c (compute_path_counts,
24518 update_joiner_offpath_counts): Use profile_probability.
24520 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24523 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24525 * config/arm/arm-c.c (arm_cpu_builtins): Define
24526 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24528 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24530 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24531 array entries to represent __ieee128 versions of the
24532 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24533 scalar_extract_sig, and scalar_insert_exp built-in functions.
24534 (altivec_resolve_overloaded_builtin): Add special case handling
24535 for the __builtin_scalar_insert_exp function, as represented by
24536 the P9V_BUILTIN_VEC_VSIEDP constant.
24537 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24538 exponent support for __ieee128 argument.
24539 (VSESQP): Add scalar extract signature support for __ieee128
24541 (VSTDCNQP): Add scalar test negative support for __ieee128
24543 (VSIEQP): Add scalar insert exponent support for __int128 argument
24544 with __ieee128 result.
24545 (VSIEQPF): Add scalar insert exponent support for __ieee128
24546 argument with __ieee128 result.
24547 (VSTDCQP): Add scalar test data class support for __ieee128
24549 (VSTDCNQP): Add overload support for scalar test negative with
24550 __ieee128 argument.
24551 (VSTDCQP): Add overload support for scalar test data class
24552 __ieee128 argument.
24553 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24554 UNSPEC_VSX_SXSIGDP.
24555 (UNSPEC_VSX_SIEXPQP): New constant.
24556 (xsxexpqp): New insn for VSX scalar extract exponent quad
24558 (xsxsigqp): New insn for VSX scalar extract significand quad
24560 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24561 precision with floating point argument.
24562 (xststdcqp): New expand for VSX scalar test data class quad
24564 (xststdcnegqp): New expand for VSX scalar test negative quad
24566 (xststdcqp): New insn to match expansions for VSX scalar test data
24567 class quad precision and VSX scalar test negative quad precision.
24568 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24569 special case operand checking to enforce that second operand of
24570 VSX scalar test data class with quad precision argument is a 7-bit
24572 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24573 prototypes and descriptions of __ieee128 versions of
24574 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24575 scalar_test_data_class, and scalar_test_neg built-in functions.
24577 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24579 PR tree-optimization/81162
24580 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24581 replace a negate with an add.
24583 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24585 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24587 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24589 * config/arm/arm-c.c (arm_cpu_builtins): Define
24590 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24592 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24594 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24595 (armv8-r): Set ARM Cortex-R52 as default CPU.
24596 * config/arm/arm-tables.opt: Regenerate.
24597 * config/arm/arm-tune.md: Regenerate.
24598 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24600 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24601 extension for -mcpu=cortex-r52.
24603 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24605 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24606 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24607 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24608 (fp-armv8): Define it as FP_ARMv8 only.
24609 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24610 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24612 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24613 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24614 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24615 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24616 than TARGET_FPU_ARMV8.
24617 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24618 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24619 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24621 * config/arm/neon.md (neon_vrint): Likewise.
24622 (neon_vcvt): Likewise.
24623 (neon_<fmaxmin_op><mode>): Likewise.
24624 (<fmaxmin><mode>3): Likewise.
24625 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24626 * config/arm/predicates.md (arm_cond_move_operator): Check against
24627 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24629 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24631 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24632 to top of function.
24634 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24636 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24637 loop in comment with memset.
24639 2017-07-14 Martin Liska <mliska@suse.cz>
24641 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24642 * dwarf2out.c (is_java): Remove the function.
24643 (output_pubname): Remove usage of the function.
24644 (lower_bound_default): Remove usage of DW_LANG_Java.
24645 (gen_compile_unit_die): Likewise.
24646 * gcc.c: Remove compiler defaults for .java and .zip files.
24647 * gimple-expr.c (remove_suffix): Change as there's no longer
24648 extension than 4-letter one.
24649 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24650 (gimplify_save_expr): Likewise.
24651 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24652 as it's possible even for other languages than Java.
24653 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24654 * lto-opts.c (lto_write_options): Remove reference to Java.
24655 * opts.c (strip_off_ending): Update file extension handling.
24656 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24657 * tree-eh.c (lower_resx): Likewise.
24658 * tree.c (free_lang_data_in_type): Remove dead code.
24659 (find_decls_types_r): Likewise.
24660 (build_common_builtin_nodes): Remove Java from a comment.
24661 (verify_type): Remove dead code.
24662 * varasm.c (assemble_external): Remove Java from a comment.
24664 2017-07-14 Martin Liska <mliska@suse.cz>
24666 * opts.c (finish_options): Add quotes.
24667 (common_handle_option): Likewise.
24669 2017-07-14 Martin Liska <mliska@suse.cz>
24671 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24672 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24673 Remove N_SO_PASCAL.
24674 * dwarf2out.c (lower_bound_default): Do not handle
24676 (gen_compile_unit_die): Likewise.
24677 * gcc.c: Remove default extension binding for GNU Pascal.
24678 * stmt.c: Remove Pascal language from a comment.
24679 * xcoffout.c: Likewise.
24681 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24684 * diagnostic-show-locus.c (fixit_cmp): New function.
24685 (layout::layout): Sort m_fixit_hints.
24686 (column_range::column_range): Assert that the values are valid.
24687 (struct char_span): New struct.
24688 (correction::overwrite): New method.
24689 (struct source_line): New struct.
24690 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24691 calls in terms of classes source_line and char_span, and
24692 correction::overwrite.
24693 (selftest::test_overlapped_fixit_printing_2): New function.
24694 (selftest::diagnostic_show_locus_c_tests): Call it.
24696 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24698 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24699 early if there is no lhs.
24701 2017-07-13 Martin Liska <mliska@suse.cz>
24703 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24704 (gen_reference_type_die): Likewise.
24705 * stor-layout.c: Remove Pascal-related comment.
24707 2017-07-13 Martin Liska <mliska@suse.cz>
24709 * opts.c (finish_options): Add quotes to error messages.
24710 (parse_sanitizer_options): Likewise.
24712 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24714 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24716 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24718 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24720 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24722 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24724 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24726 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24729 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24730 provides the hardware capability bits, define the macro
24731 __BUILTIN_CPU_SUPPORTS__.
24732 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24733 if GLIBC does not provide the hardware capability bits. Add a
24734 gcc_unreachable call if the built-in cpu function is neither
24735 __builtin_cpu_is nor __builtin_cpu_supports.
24736 (rs6000_get_function_versions_dispatcher): Change the warning
24737 that an old GLIBC is used which does not export the capability
24738 bits to be an error.
24739 * doc/extend.texi (target_clones attribute): Document the
24740 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24741 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24742 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24743 the macros defined by GCC if the newer GLIBC is available.
24745 2017-07-12 Jeff Law <law@redhat.com>
24747 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24748 remaining includes slightly.
24749 * config/riscv/riscv-builtins.c: Include profile-count.h.
24751 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24754 * config/avr/avr.c (avr_set_current_function): In diagnostic
24755 messages: Quote keywords and (parts of) identifiers.
24756 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24759 2017-07-12 Carl Love <cel@us.ibm.com>
24761 * config/rs6000/rs6000-c.c: Add support for built-in functions
24762 vector bool char vec_revb (vector bool char);
24763 vector bool short vec_revb (vector short char);
24764 vector bool int vec_revb (vector bool int);
24765 vector bool long long vec_revb (vector bool long long);
24766 * doc/extend.texi: Update the built-in documentation file for the
24767 new built-in functions.
24769 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24771 * config/s390/s390.md: Remove movcc splitter.
24773 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24775 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24776 load/store on condition.
24778 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24781 * config/avr/avr.c (avr_encode_section_info)
24782 [progmem && !TREE_READONLY]: Error if progmem object needs
24785 2017-07-11 Michael Collison <michael.collison@arm.com>
24787 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24790 2017-07-11 Carl Love <cel@us.ibm.com>
24792 * config/rs6000/rs6000-c.c: Add support for builtins
24793 vector unsigned int vec_parity_lsbb (vector signed int);
24794 vector unsigned int vec_parity_lsbb (vector unsigned int);
24795 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24796 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24797 vector unsigned long long vec_parity_lsbb (vector signed long long);
24798 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24799 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24800 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24801 * doc/extend.texi: Update the built-in documentation file for the
24802 new built-in functions.
24804 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24806 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24807 (layout::m_primary_loc): New field.
24808 (layout::layout): Initialize new field. Move location filtering
24809 logic from here to...
24810 (layout::maybe_add_location_range): ...this new method. Add
24811 support for filtering to just the lines already specified by other
24813 (layout::will_show_line_p): New method.
24814 (gcc_rich_location::add_location_if_nearby): New method.
24815 (selftest::test_add_location_if_nearby): New test function.
24816 (selftest::diagnostic_show_locus_c_tests): Call it.
24817 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24820 2017-07-11 Tom de Vries <tom@codesourcery.com>
24822 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24823 (bb_first_real_insn): New function.
24824 (nvptx_single): Add extra initialization of broadcasted condition
24827 2017-07-11 Nathan Sidwell <nathan@acm.org>
24829 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24831 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24833 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24834 Binutils doc as TEXI will mess them up.
24835 * doc/invoke.texi (AVR Options): Same here.
24837 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24839 * config/sparc/sparc.opt (mfix-ut700): New option.
24840 (mfix-gr712rc): Likewise.
24841 (sparc_fix_b2bst): New variable.
24842 * doc/invoke.texi (SPARC options): Document them.
24843 (ARM options): Fix warnings.
24844 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24845 instructions to prevent sequences that can trigger the store-store
24846 errata for certain LEON3FT processors.
24847 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24848 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24849 * config/sparc/sparc.md (fix_b2bst): New attribute.
24850 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24852 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24855 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24857 (*rsqrtsf2_sse): Ditto.
24859 (div<mode>3): Macroize insn from divdf3 and divsf3
24860 using MODEF mode iterator.
24862 2017-07-10 Martin Sebor <msebor@redhat.com>
24864 PR tree-optimization/80397
24865 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24866 instead of testing for equality to INTEGER_TYPE.
24868 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24870 * config.gcc: Remove uclibc from arc target spec.
24872 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24874 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24876 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24879 * lto-wrapper.c (remove_option): New function.
24880 (merge_and_complain): Merge PIC/PIE options more realistically.
24882 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24884 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24888 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24889 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24890 * config.in: Regenerate.
24891 * configure: Regenerate.
24892 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24893 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24894 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24895 (TARGET_GASISR_PROLOGUES): ...target mask.
24896 * common/config/avr/avr-common.c
24897 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24898 Set -mgas-isr-prologues.
24899 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24900 INSERT_PASS_BEFORE for it.
24901 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24902 * config/avr/avr.c (avr_option_override)
24903 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24904 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24905 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24906 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24907 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24909 (make_avr_pass_pre_proep): New function.
24910 (emit_push_sfr) <treg>: Add argument to function and use it
24911 instead of TMP_REG.
24912 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24913 and set machine->gasisr.yes.
24914 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24915 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24916 __gcc_isr.n_pushed to .L__stack_usage.
24917 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24918 (avr_asm_final_postscan_insn): ...this new static function.
24919 * config/avr/avr.h (machine_function)
24920 <is_no_gccisr, use_L__stack_usage>: New fields.
24921 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24922 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24923 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24924 (gasisr, *gasisr): New expander and insn.
24925 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24926 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24927 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24929 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24931 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24934 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24936 Move jump-tables out of .text again.
24939 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24940 (ASM_OUTPUT_ADDR_VEC): New function.
24941 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24942 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24943 INSN_ADDRESSes as asm comment.
24944 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24945 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24946 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24947 * config/avr/avr.md (*tablejump): Adjust comment.
24948 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24949 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24951 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24952 (avr_output_addr_vec): New proto.
24953 (avr_log_t) <insn_addresses>: New field.
24955 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24958 * config/i386/i386.c (ix86_function_arg_advance): Set
24959 outgoing_args_on_stack to true if there are outgoing arguments
24961 (ix86_function_arg): Likewise.
24962 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24963 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24964 * config/i386/i386.h (machine_function): Add
24965 outgoing_args_on_stack.
24967 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24969 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24970 supporting pthreds.
24971 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24973 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24975 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24976 (REAL_H): Remove $(MACHMODE_H).
24977 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24979 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24980 $(MACHMODE_H) and double-int.h.
24981 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24983 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24986 2017-07-07 Andrew Pinski <apinski@cavium.com>
24988 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24989 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24991 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24993 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24994 Add warning if GCC was not configured to link against a GLIBC that
24995 exports the hardware capability bits.
24996 (make_resolver_func): Make resolver function private and not a
24997 COMDAT function. Create the name with clone_function_name instead
24998 of make_unique_name.
25001 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
25002 correct operand in doing the split.
25004 2017-07-07 Carl Love <cel@us.ibm.com>
25006 * config/rs6000/rs6000-c: Add support for built-in function
25007 vector unsigned short vec_pack_to_short_fp32 (vector float,
25009 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
25010 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
25011 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
25012 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
25013 (convert_4f32_8i16): Add define_expand.
25014 * doc/extend.texi: Update the built-in documentation file for the
25015 new built-in function.
25017 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25019 * config/sparc/m8.md: New file.
25020 * config/sparc/sparc.md: Include m8.md.
25022 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25024 * config/sparc/sparc.opt: New option -mvis4b.
25025 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
25026 (sparc_option_override): Handle VIS4B.
25027 (enum sparc_builtins): Define
25028 SPARC_BUILTIN_DICTUNPACK{8,16,32},
25029 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
25030 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
25031 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
25032 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
25033 (check_constant_argument): New function.
25034 (sparc_vis_init_builtins): Define builtins
25035 __builtin_vis_dictunpack{8,16,32},
25036 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
25037 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
25038 __builtin_vis_fpcmpde{8,16,32}shl and
25039 __builtin_vis_fpcmpur{8,16,32}shl.
25040 (sparc_expand_builtin): Check that the constant operands to
25041 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
25042 constant and in range.
25043 * config/sparc/sparc-c.c (sparc_target_macros): Handle
25045 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
25046 (SPARC_IMM5_P): Likewise.
25047 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
25048 (enabled): Handle vis4b.
25049 (UNSPEC_DICTUNPACK): New unspec.
25050 (UNSPEC_FPCMPSHL): Likewise.
25051 (UNSPEC_FPUCMPSHL): Likewise.
25052 (UNSPEC_FPCMPDESHL): Likewise.
25053 (UNSPEC_FPCMPURSHL): Likewise.
25054 (cpu_feature): New CPU feature `vis4b'.
25055 (dictunpack{8,16,32}): New insns.
25056 (FPCSMODE): New mode iterator.
25057 (fpcscond): New code iterator.
25058 (fpcsucond): Likewise.
25059 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
25060 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
25061 (fpcmpde{8,16,32}{si,di}shl): Likewise.
25062 (fpcmpur{8,16,32}{si,di}shl): Likewise.
25063 * config/sparc/constraints.md: Define constraints `q' for unsigned
25064 2-bit integer constants and `t' for unsigned 5-bit integer
25066 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
25068 (imm5_operand_dictunpack16): Likewise.
25069 (imm5_operand_dictunpack32): Likewise.
25070 (imm2_operand): Likewise.
25071 * doc/invoke.texi (SPARC Options): Document -mvis4b.
25072 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
25073 ditunpack* and fpcmp*shl builtins.
25075 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25077 * config.gcc: Handle m8 in --with-{cpu,tune} options.
25078 * config.in: Add HAVE_AS_SPARC6 define.
25079 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
25081 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
25083 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
25084 (CPP_CPU_SPEC): Handle m8.
25085 (ASM_CPU_SPEC): Likewise.
25086 * config/sparc/sparc-opts.h (enum processor_type): Add
25088 * config/sparc/sparc.c (m8_costs): New struct.
25089 (sparc_option_override): Handle TARGET_CPU_m8.
25090 (sparc32_initialize_trampoline): Likewise.
25091 (sparc64_initialize_trampoline): Likewise.
25092 (sparc_issue_rate): Likewise.
25093 (sparc_register_move_cost): Likewise.
25094 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
25095 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
25096 (ASM_CPU64_DEFAULT_SPEC): Likewise.
25097 (CPP_CPU_SPEC): Handle M8.
25098 (ASM_CPU_SPEC): Likewise.
25099 (AS_M8_FLAG): Define.
25100 * config/sparc/sparc.md: Add m8 to the cpu attribute.
25101 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
25102 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
25104 * configure: Regenerate.
25105 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
25108 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25110 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
25112 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
25113 ("*movdi_insn_sp32"): Do not set v3pipe.
25114 ("*movsi_insn"): Likewise.
25115 ("*movdi_insn_sp64"): Likewise.
25116 ("*movsf_insn"): Likewise.
25117 ("*movdf_insn_sp32"): Likewise.
25118 ("*movdf_insn_sp64"): Likewise.
25119 ("*zero_extendsidi2_insn_sp64"): Likewise.
25120 ("*sign_extendsidi2_insn"): Likewise.
25121 ("*mov<VM32:mode>_insn"): Likewise.
25122 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25123 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25124 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25125 ("<vlop:code><VL:mode>3"): Likewise.
25126 ("*not_<vlop:code><VL:mode>3"): Likewise.
25127 ("*nand<VL:mode>_vis"): Likewise.
25128 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
25129 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
25130 ("one_cmpl<VL:mode>2"): Likewise.
25131 ("faligndata<VM64:mode>_vis"): Likewise.
25132 ("alignaddrsi_vis"): Likewise.
25133 ("alignaddrdi_vis"): Likweise.
25134 ("alignaddrlsi_vis"): Likewise.
25135 ("alignaddrldi_vis"): Likewise.
25136 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25137 ("bmaskdi_vis"): Likewise.
25138 ("bmasksi_vis"): Likewise.
25139 ("bshuffle<VM64:mode>_vis"): Likewise.
25140 ("cmask8<P:mode>_vis"): Likewise.
25141 ("cmask16<P:mode>_vis"): Likewise.
25142 ("cmask32<P:mode>_vis"): Likewise.
25143 ("pdistn<P:mode>_vis"): Likewise.
25144 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25146 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25148 * config/sparc/sparc.md ("subtype"): New insn attribute.
25149 ("*wrgsr_sp64"): Set insn subtype.
25150 ("*rdgsr_sp64"): Likewise.
25151 ("alignaddrsi_vis"): Likewise.
25152 ("alignaddrdi_vis"): Likewise.
25153 ("alignaddrlsi_vis"): Likewise.
25154 ("alignaddrldi_vis"): Likewise.
25155 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25156 ("fexpand_vis"): Likewise.
25157 ("fpmerge_vis"): Likewise.
25158 ("faligndata<VM64:mode>_vis"): Likewise.
25159 ("bshuffle<VM64:mode>_vis"): Likewise.
25160 ("cmask8<P:mode>_vis"): Likewise.
25161 ("cmask16<P:mode>_vis"): Likewise.
25162 ("cmask32<P:mode>_vis"): Likewise.
25163 ("fchksm16_vis"): Likewise.
25164 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25165 ("fmean16_vis"): Likewise.
25166 ("fp<plusminus_insn>64_vis"): Likewise.
25167 ("<plusminus_insn>v8qi3"): Likewise.
25168 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25169 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25170 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25171 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25172 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25173 ("*movqi_insn"): Likewise.
25174 ("*movhi_insn"): Likewise.
25175 ("*movsi_insn"): Likewise.
25176 ("movsi_pic_gotdata_op"): Likewise.
25177 ("*movdi_insn_sp32"): Likewise.
25178 ("*movdi_insn_sp64"): Likewise.
25179 ("movdi_pic_gotdata_op"): Likewise.
25180 ("*movsf_insn"): Likewise.
25181 ("*movdf_insn_sp32"): Likewise.
25182 ("*movdf_insn_sp64"): Likewise.
25183 ("*zero_extendhisi2_insn"): Likewise.
25184 ("*zero_extendqihi2_insn"): Likewise.
25185 ("*zero_extendqisi2_insn"): Likewise.
25186 ("*zero_extendqidi2_insn"): Likewise.
25187 ("*zero_extendhidi2_insn"): Likewise.
25188 ("*zero_extendsidi2_insn_sp64"): Likewise.
25189 ("ldfsr"): Likewise.
25190 ("prefetch_64"): Likewise.
25191 ("prefetch_32"): Likewise.
25192 ("tie_ld32"): Likewise.
25193 ("tie_ld64"): Likewise.
25194 ("*tldo_ldub_sp32"): Likewise.
25195 ("*tldo_ldub1_sp32"): Likewise.
25196 ("*tldo_ldub2_sp32"): Likewise.
25197 ("*tldo_ldub_sp64"): Likewise.
25198 ("*tldo_ldub1_sp64"): Likewise.
25199 ("*tldo_ldub2_sp64"): Likewise.
25200 ("*tldo_ldub3_sp64"): Likewise.
25201 ("*tldo_lduh_sp32"): Likewise.
25202 ("*tldo_lduh1_sp32"): Likewise.
25203 ("*tldo_lduh_sp64"): Likewise.
25204 ("*tldo_lduh1_sp64"): Likewise.
25205 ("*tldo_lduh2_sp64"): Likewise.
25206 ("*tldo_lduw_sp32"): Likewise.
25207 ("*tldo_lduw_sp64"): Likewise.
25208 ("*tldo_lduw1_sp64"): Likewise.
25209 ("*tldo_ldx_sp64"): Likewise.
25210 ("*mov<VM32:mode>_insn"): Likewise.
25211 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25212 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25214 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25216 * config/sparc/sparc.md ("type"): New insn type viscmp.
25217 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25219 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25220 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25221 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25222 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25224 ("n7_vis_logical_11cycle"): Likewise.
25225 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25226 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25227 * config/sparc/niagara.md ("niag_vis"): Likewise.
25228 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25229 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25231 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25233 * config/sparc/sparc.md: New instruction type `bmask'.
25234 (bmaskdi_vis): Use the `bmask' type.
25235 (bmasksi_vis): Likewise.
25236 * config/sparc/ultra3.md (us3_array): Likewise.
25237 * config/sparc/niagara7.md (n7_array): Likewise.
25238 * config/sparc/niagara4.md (n4_array): Likewise.
25239 * config/sparc/niagara2.md (niag2_vis): Likewise.
25240 (niag3_vis): Likewise.
25241 * config/sparc/niagara.md (niag_vis): Likewise.
25243 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25245 * ipa-comdats.c: Remove optimize check from gate.
25246 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25247 for functions not optimized.
25248 (ipa_fn_summary_read): Skip optimize check.
25249 (ipa_fn_summary_write): Likewise.
25250 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25252 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25254 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25255 (check_callers): Give up on caller which is not optimized.
25256 (inline_small_functions): Likewise.
25257 (ipa_inline): Do not give up when not optimizing.
25258 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25259 away unoptimizes cdtors.
25260 (whole_program_function_and_variable_visibility): Do
25261 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25262 * ipa.c (process_references): Do not check optimize.
25263 (symbol_table::remove_unreachable_nodes): Update optimize check.
25264 (set_writeonly_bit): Update optimize check.
25265 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25266 (pass_ipa_single_use::gate): Remove.
25268 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25270 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25271 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25272 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25273 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25274 permute_load, permute_store, adjust_extract, adjust_splat,
25275 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25276 replace_swap_with_copy, dump_swap_insn_table,
25277 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25278 recombine_lvx_pattern, recombine_stvx_pattern,
25279 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25280 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25281 to file rs6000-p8swap.c.
25282 * config/rs6000/rs6000-p8swap.c: New file.
25283 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25284 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25285 and rs6000*-*-* targets.
25287 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25289 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25291 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25293 * lto-wrapper.c (merge_and_complain): Do not merge
25294 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25295 fsigned_zeros, ftrapping_math, fwrapv.
25296 (append_compiler_options): Do not track these options.
25297 (append_linker_options): Likewie
25299 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25301 * cgraphunit.c (cgraph_node::finalize_function): When
25302 !flag_toplevel_reorde set no_reorder flag.
25303 (varpool_node::finalize_decl): Likewise.
25304 (symbol_table::compile): Drop no toplevel reorder path.
25306 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25308 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25309 edges; zero probability is not better than uninitialized.
25311 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25313 * asan.h (asan_sanitize_allocas_p): Declare.
25314 * asan.c (asan_sanitize_allocas_p): New function.
25315 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25316 (handle_builtin_alloca): Likewise.
25317 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25318 if !asan_sanitize_allocas_p.
25319 * params.def (asan-instrument-allocas): Add new option.
25320 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25321 * opts.c (common_handle_option): Disable allocas sanitization for
25324 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25326 * asan.c: Include gimple-fold.h.
25327 (get_last_alloca_addr): New function.
25328 (handle_builtin_stackrestore): Likewise.
25329 (handle_builtin_alloca): Likewise.
25330 (asan_emit_allocas_unpoison): Likewise.
25331 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25332 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25333 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25334 (instrument_builtin_call): Pass gimple iterator to
25335 get_mem_refs_of_builtin_call.
25336 (last_alloca_addr): New global.
25337 * asan.h (asan_emit_allocas_unpoison): Declare.
25338 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25339 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25340 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25341 if function calls alloca.
25342 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25343 * gimple-fold.h (replace_call_with_value): Declare.
25344 * internal-fn.c: Include asan.h.
25345 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25346 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25348 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25350 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25351 (C_SELFTEST_FLAGS): New.
25352 (CPP_SELFTEST_FLAGS): New.
25353 (SELFTEST_DEPS): New, from deps of s-selftest.
25354 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25355 (CPP_SELFTEST_DEPS): New.
25356 (selftest): Add dependency on s-selftest-c++.
25357 (s-selftest): Rename to...
25358 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25359 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25360 than SELFTEST_FLAGS.
25361 (selftest-gdb): Rename to...
25362 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25364 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25365 (selftest-valgrind): Rename to...
25366 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25368 (selftest-valgrind): Reintroduce as an alias for
25369 selftest-c-valgrind.
25370 (s-selftest-c++): New.
25371 (selftest-c++-gdb): New.
25372 (selftest-c++-valgrind): New.
25374 2017-07-06 Olivier Hainque <hainque@adacore.com>
25376 * gcc.c (process_command): When deciding if undefined variables
25377 should be ignored when processing specs, accept "gcc -v" as well.
25379 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25381 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25382 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25384 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25386 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25387 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25388 * config/arm/arm-tables.opt: Regenerate.
25389 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25391 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25393 2017-07-06 Carl Love <cel@us.ibm.com>
25395 * ChangeLog: Clean up from mid air collision
25397 2017-07-06 Carl Love <cel@us.ibm.com>
25399 * config/rs6000/rs6000-c.c: Add support for built-in functions
25400 vector signed int vec_subc (vector signed int, vector signed int);
25401 vector signed __int128 vec_subc (vector signed __int128,
25402 vector signed __int128);
25403 vector unsigned __int128 vec_subc (vector unsigned __int128,
25404 vector unsigned __int128);
25405 vector signed int vec_sube (vector signed int, vector signed int,
25406 vector signed int);
25407 vector unsigned int vec_sube (vector unsigned int,
25408 vector unsigned int,
25409 vector unsigned int);
25410 vector signed __int128 vec_sube (vector signed __int128,
25411 vector signed __int128,
25412 vector signed__int128);
25413 vector unsigned __int128 vec_sube (vector unsigned __int128,
25414 vector unsigned __int128,
25415 vector unsigned __int128);
25416 vector signed int vec_subec (vector signed int, vector signed int,
25417 vector signed int);
25418 vector unsigned int vec_subec (vector unsigned int,
25419 vector unsigned int,
25420 vector unsigned int);
25421 vector signed __int128 vec_subec (vector signed __int128,
25422 vector signed __int128,
25423 vector signed__int128);
25424 vector unsigned __int128 vec_subec (vector unsigned __int128,
25425 vector unsigned __int128,
25426 vector unsigned __int128);
25427 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25428 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25429 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25430 BU_ALTIVEC_OVERLOAD_X definitions.
25431 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25432 * doc/extend.texi: Update the built-in documentation file for the new
25433 built-in functions.
25435 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25438 * diagnostic-show-locus.c (layout::layout): Use start and finish
25439 spelling location for the start and finish of each range.
25440 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25441 Add unused aspect param.
25442 * input.c (expand_location_1): Add "aspect" param, and use it
25443 to access the correct part of the location.
25444 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25446 (expand_location_to_spelling_point): Likewise.
25447 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25448 param, and pass it to expand_location_1.
25450 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25452 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25453 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25454 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25455 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25456 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25457 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25458 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25459 _mm_maskz_getmant_ss): New intrinsics.
25460 (__builtin_ia32_getexpss128_mask): Changed to ...
25461 __builtin_ia32_getexpss128_round ... this.
25462 (__builtin_ia32_getexpsd128_mask): Changed to ...
25463 __builtin_ia32_getexpsd128_round ... this.
25464 * config/i386/i386-builtin-types.def
25465 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25466 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25467 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25468 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25469 __builtin_ia32_getmantss_mask_round): New builtins.
25470 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25471 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25472 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25473 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25474 * config/i386/sse.md
25475 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25476 avx512f_sgetexp<mode><mask_scalar_name>
25477 <round_saeonly_scalar_name> ... this.
25478 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25479 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25480 vgetexp<ssescalarmodesuffix>
25481 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25482 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25483 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25484 avx512f_vgetmant<mode><mask_scalar_name>
25485 <round_saeonly_scalar_name> ... this.
25486 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25487 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25488 vgetmant<ssescalarmodesuffix>
25489 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25490 %0<mask_scalar_operand4>, %1, %2
25491 <round_saeonly_scalar_mask_op4>, %3} ... this.
25492 * config/i386/subst.md (mask_scalar_operand4,
25493 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25494 round_saeonly_scalar_nimm_predicate): New subst attributes.
25496 2017-07-06 Julia Koval <julia.koval@intel.com>
25498 * config/i386/i386.c (ix86_erase_embedded_rounding):
25499 Remove code for old rounding pattern.
25501 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25503 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25505 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25507 * doc/sourcebuild.texi (Test Directives, Variants of
25508 dg-require-support): Add documentation for dg-require-stack-check.
25510 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25512 * config/i386/subst.md (mask_scalar, round_scalar,
25513 round_saeonly_scalar): New meta-templates.
25514 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25515 round_scalar_mask_operand3, round_scalar_mask_op3,
25516 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25517 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25518 round_saeonly_scalar_constraint,
25519 round_saeonly_scalar_prefix): New subst attribute.
25520 * config/i386/sse.md
25521 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25522 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25523 <round_scalar_name> ... this.
25524 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25525 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25526 <round_scalar_name> ... this.
25527 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25528 <sse>_vm<code><mode>3<mask_scalar_name>
25529 <round_saeonly_scalar_name> ... this.
25530 (v<plusminus_mnemonic><ssescalarmodesuffix>
25531 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25532 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25533 v<plusminus_mnemonic><ssescalarmodesuffix>
25534 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25535 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25536 (v<multdiv_mnemonic><ssescalarmodesuffix>
25537 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25538 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25539 v<multdiv_mnemonic><ssescalarmodesuffix>
25540 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25541 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25542 (v<maxmin_float><ssescalarmodesuffix>
25543 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25544 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25545 v<maxmin_float><ssescalarmodesuffix>
25546 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25547 %0<mask_scalar_operand3>, %1, %<iptr>2
25548 <round_saeonly_scalar_mask_op3>} ... this.
25550 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25552 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25553 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25555 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25556 Alan Hayward <alan.hayward@arm.com>
25557 David Sherwood <david.sherwood@arm.com>
25559 * combine.c (simplify_if_then_else): Remove "enum" before
25561 * compare-elim.c (can_eliminate_compare): Likewise.
25562 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25564 (aarch64_lookup_simd_builtin_type): Likewise.
25565 (aarch64_simd_builtin_type): Likewise.
25566 (aarch64_init_simd_builtin_types): Likewise.
25567 (aarch64_simd_expand_args): Likewise.
25568 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25570 (aarch64_reverse_mask): Likewise.
25571 (aarch64_simd_emit_reg_reg_move): Likewise.
25572 (aarch64_gen_adjusted_ldpstp): Likewise.
25573 (aarch64_ccmp_mode_to_code): Likewise.
25574 (aarch64_operands_ok_for_ldpstp): Likewise.
25575 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25576 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25578 (aarch64_min_divisions_for_recip_mul): Likewise.
25579 (aarch64_reassociation_width): Likewise.
25580 (aarch64_get_condition_code_1): Likewise.
25581 (aarch64_simd_emit_reg_reg_move): Likewise.
25582 (aarch64_simd_attr_length_rglist): Likewise.
25583 (aarch64_reverse_mask): Likewise.
25584 (aarch64_operands_ok_for_ldpstp): Likewise.
25585 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25586 (aarch64_gen_adjusted_ldpstp): Likewise.
25587 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25589 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25590 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25591 (arm_lookup_simd_builtin_type): Likewise.
25592 (arm_simd_builtin_type): Likewise.
25593 (arm_init_simd_builtin_types): Likewise.
25594 (arm_expand_builtin_args): Likewise.
25595 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25596 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25597 (ft32_setup_incoming_varargs): Likewise.
25598 (ft32_function_arg): Likewise.
25599 (ft32_function_arg_advance): Likewise.
25600 (ft32_pass_by_reference): Likewise.
25601 (ft32_arg_partial_bytes): Likewise.
25602 (ft32_valid_pointer_mode): Likewise.
25603 (ft32_addr_space_pointer_mode): Likewise.
25604 (ft32_addr_space_legitimate_address_p): Likewise.
25605 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25607 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25608 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25609 (iamcu_alignment): Likewise.
25610 (canonicalize_vector_int_perm): Likewise.
25611 (ix86_noce_conversion_profitable_p): Likewise.
25612 (ix86_mpx_bound_mode): Likewise.
25613 (ix86_operands_ok_for_move_multiple): Likewise.
25614 * config/microblaze/microblaze-protos.h
25615 (microblaze_expand_conditional_branch_reg): Likewise.
25616 * config/microblaze/microblaze.c
25617 (microblaze_expand_conditional_branch_reg): Likewise.
25618 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25620 (rs6000_reassociation_width): Likewise.
25621 (rs6000_invalid_binary_op): Likewise.
25622 (fusion_p9_p): Likewise.
25623 (emit_fusion_p9_load): Likewise.
25624 (emit_fusion_p9_store): Likewise.
25625 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25627 (riscv_hard_regno_mode_ok_p): Likewise.
25628 (riscv_address_insns): Likewise.
25629 (riscv_split_symbol): Likewise.
25630 (riscv_legitimize_move): Likewise.
25631 (riscv_function_value): Likewise.
25632 (riscv_hard_regno_nregs): Likewise.
25633 (riscv_expand_builtin): Likewise.
25634 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25635 (riscv_build_integer): Likewise.
25636 (riscv_split_integer): Likewise.
25637 (riscv_legitimate_constant_p): Likewise.
25638 (riscv_cannot_force_const_mem): Likewise.
25639 (riscv_regno_mode_ok_for_base_p): Likewise.
25640 (riscv_valid_base_register_p): Likewise.
25641 (riscv_valid_offset_p): Likewise.
25642 (riscv_valid_lo_sum_p): Likewise.
25643 (riscv_classify_address): Likewise.
25644 (riscv_legitimate_address_p): Likewise.
25645 (riscv_address_insns): Likewise.
25646 (riscv_load_store_insns): Likewise.
25647 (riscv_force_binary): Likewise.
25648 (riscv_split_symbol): Likewise.
25649 (riscv_force_address): Likewise.
25650 (riscv_legitimize_address): Likewise.
25651 (riscv_move_integer): Likewise.
25652 (riscv_legitimize_const_move): Likewise.
25653 (riscv_legitimize_move): Likewise.
25654 (riscv_address_cost): Likewise.
25655 (riscv_subword): Likewise.
25656 (riscv_output_move): Likewise.
25657 (riscv_canonicalize_int_order_test): Likewise.
25658 (riscv_emit_int_order_test): Likewise.
25659 (riscv_function_arg_boundary): Likewise.
25660 (riscv_pass_mode_in_fpr_p): Likewise.
25661 (riscv_pass_fpr_single): Likewise.
25662 (riscv_pass_fpr_pair): Likewise.
25663 (riscv_get_arg_info): Likewise.
25664 (riscv_function_arg): Likewise.
25665 (riscv_function_arg_advance): Likewise.
25666 (riscv_arg_partial_bytes): Likewise.
25667 (riscv_function_value): Likewise.
25668 (riscv_pass_by_reference): Likewise.
25669 (riscv_setup_incoming_varargs): Likewise.
25670 (riscv_print_operand): Likewise.
25671 (riscv_elf_select_rtx_section): Likewise.
25672 (riscv_save_restore_reg): Likewise.
25673 (riscv_for_each_saved_reg): Likewise.
25674 (riscv_register_move_cost): Likewise.
25675 (riscv_hard_regno_mode_ok_p): Likewise.
25676 (riscv_hard_regno_nregs): Likewise.
25677 (riscv_class_max_nregs): Likewise.
25678 (riscv_memory_move_cost): Likewise.
25679 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25680 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25681 (rl78_addr_space_address_mode): Likewise.
25682 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25684 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25685 (rs6000_reassociation_width): Likewise.
25686 (rs6000_invalid_binary_op): Likewise.
25687 (fusion_p9_p): Likewise.
25688 (emit_fusion_p9_load): Likewise.
25689 (emit_fusion_p9_store): Likewise.
25690 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25691 (ok_for_simple_move_operands): Likewise.
25692 (ok_for_simple_move_strict_operands): Likewise.
25693 (ok_for_simple_arith_logic_operands): Likewise.
25694 (visium_legitimize_reload_address): Likewise.
25695 (visium_select_cc_mode): Likewise.
25696 (output_cbranch): Likewise.
25697 (visium_split_double_move): Likewise.
25698 (visium_expand_copysign): Likewise.
25699 (visium_expand_int_cstore): Likewise.
25700 (visium_expand_fp_cstore): Likewise.
25701 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25702 (visium_function_arg): Likewise.
25703 (visium_function_arg_advance): Likewise.
25704 (visium_libcall_value): Likewise.
25705 (visium_setup_incoming_varargs): Likewise.
25706 (visium_legitimate_constant_p): Likewise.
25707 (visium_legitimate_address_p): Likewise.
25708 (visium_legitimize_address): Likewise.
25709 (visium_secondary_reload): Likewise.
25710 (visium_register_move_cost): Likewise.
25711 (visium_memory_move_cost): Likewise.
25712 (prepare_move_operands): Likewise.
25713 (ok_for_simple_move_operands): Likewise.
25714 (ok_for_simple_move_strict_operands): Likewise.
25715 (ok_for_simple_arith_logic_operands): Likewise.
25716 (visium_function_value_1): Likewise.
25717 (rtx_ok_for_offset_p): Likewise.
25718 (visium_legitimize_reload_address): Likewise.
25719 (visium_split_double_move): Likewise.
25720 (visium_expand_copysign): Likewise.
25721 (visium_expand_int_cstore): Likewise.
25722 (visium_expand_fp_cstore): Likewise.
25723 (visium_split_cstore): Likewise.
25724 (visium_select_cc_mode): Likewise.
25725 (visium_split_cbranch): Likewise.
25726 (output_cbranch): Likewise.
25727 (visium_print_operand_address): Likewise.
25728 * expmed.c (flip_storage_order): Likewise.
25729 * expmed.h (emit_cstore): Likewise.
25730 (flip_storage_order): Likewise.
25731 * genrecog.c (validate_pattern): Likewise.
25732 * hsa-gen.c (gen_hsa_addr): Likewise.
25733 * internal-fn.c (expand_arith_overflow): Likewise.
25734 * ira-color.c (allocno_copy_cost_saving): Likewise.
25735 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25736 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25737 (process_invariant_for_inheritance): Likewise.
25738 * lra-eliminations.c (move_plus_up): Likewise.
25739 * omp-low.c (lower_oacc_reductions): Likewise.
25740 * simplify-rtx.c (simplify_subreg): Likewise.
25741 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25742 (TARGET_CHKP_BOUND_MODE): Likewise..
25743 * targhooks.c (default_chkp_bound_mode): Likewise.
25744 (default_setup_incoming_vararg_bounds): Likewise.
25745 * targhooks.h (default_chkp_bound_mode): Likewise.
25746 (default_setup_incoming_vararg_bounds): Likewise.
25747 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25748 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25749 (have_whole_vector_shift): Likewise.
25750 * tree-vect-stmts.c (vectorizable_load): Likewise.
25751 * doc/tm.texi: Regenerate.
25753 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25755 Graceful degrade if Binutils PR21472 is not available.
25758 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25759 .rodata in flash test fails.
25760 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25761 * confgure: Regenerate.
25762 * config.in: Regenerate.
25763 * config/avr/avr.c (avr_asm_named_section)
25764 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25765 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25766 (avr_asm_init_sections): Same.
25768 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25770 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25771 (fma<VH:mode>4_intrinsic): Likewise.
25772 (*fmsub<VCVTF:mode>4): Likewise.
25773 (*fmsub<VH:mode>4_intrinsic): Likewise.
25775 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25778 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25779 Don't depend on "optimize > 0".
25780 (out_movhi_r_mr, out_movqi_mr_r): Same.
25781 (out_movhi_mr_r, out_movqi_r_mr): Same.
25782 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25783 io_address_operand on "optimize > 0".
25785 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25787 * tree-loop-distribution.c: Add general explanantion on the pass.
25788 (generate_loops_for_partition): Mark distributed loop.
25789 (pg_add_dependence_edges): New parameter. Handle alias data
25790 dependence specially and record it in the parameter if asked.
25791 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25792 (init_partition_graph_vertices, add_partition_graph_edge): New.
25793 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25794 (free_partition_graph_vdata, build_partition_graph): New.
25795 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25796 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25797 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25798 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25799 (version_for_distribution_p, finalize_partitions): New.
25800 (distribute_loop): Handle alias data dependence specially. Factor
25801 out loop fusion code as functions and call these functions.
25803 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25805 * tree-loop-distribution.c (classify_partition): New parameter and
25806 better handle reduction statement.
25807 (rdg_build_partitions): Revise comment.
25808 (distribute_loop): Compute statements in all partitions and pass it
25809 to classify_partition.
25811 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25813 * tree-loop-distribution.c (enum partition_type): New.
25814 (struct partition): New field type.
25815 (partition_merge_into): Add parameter. Update partition type.
25816 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25817 (build_rdg_partition_for_vertex): Compute partition type.
25818 (rdg_build_partitions): Dump partition type.
25819 (distribute_loop): Update calls to partition_merge_into.
25821 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25823 * tree-loop-distribution.c (struct ddr_hasher): New.
25824 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25826 (classify_partition): Call get_data_dependence.
25827 (pg_add_dependence_edges): Ditto.
25828 (distribute_loop): Release data dependence hash table.
25830 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25832 * tree-loop-distribution.c (ref_base_address): Delete.
25833 (similar_memory_accesses): Rename ...
25834 (share_memory_accesses): ... to this. Check if partitions access
25835 the same memory reference.
25836 (distribute_loop): Call share_memory_accesses.
25838 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25840 * tree-loop-distribution.c (struct partition): New field recording
25841 its data reference.
25842 (partition_alloc, partition_free): Init and release data refs.
25843 (partition_merge_into): Merge data refs.
25844 (build_rdg_partition_for_vertex): Collect data refs for partition.
25845 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25847 (distribute_loop): Remve data refs from vertice data of partition
25850 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25852 * tree-loop-distribution.c (params.h): Include header file.
25853 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25854 (datarefs_vec): New global var.
25855 (create_rdg_vertices): Use datarefs_vec directly.
25856 (free_rdg): Don't free data references.
25857 (build_rdg): Update use. Don't free data references.
25858 (distribute_loop): Compute global variable for data references.
25859 Bail out if there are too many data references.
25861 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25863 * tree-loop-distribution.c (loop_nest): New global var.
25864 (build_rdg): Use loop directly, rather than loop nest.
25865 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25867 (distribute_loop): Compute global variable loop nest. Update use.
25869 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25871 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25872 (partition_merge_into): New parameter. Dump reason for fusion.
25873 (distribute_loop): Update use of partition_merge_into.
25875 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25877 * tree-loop-distribution.c (bb_top_order_index): New.
25878 (bb_top_order_index_size, bb_top_order_cmp): New.
25879 (stmts_from_loop): Use topological order.
25880 (pass_loop_distribution::execute): Compute and release topological
25881 order for basic blocks.
25883 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25885 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25888 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25890 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25891 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25892 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25893 * internal-fn.def (LOOP_DIST_ALIAS): New.
25894 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25895 (fold_loop_internal_call): ... this.
25896 (vect_loop_dist_alias_call): New function.
25897 (set_uid_loop_bbs): Call fold_loop_internal_call.
25898 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25901 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25904 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25905 Require dead FLAGS_REG at the beginning of a peephole.
25907 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25910 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25911 arguments in the call to __builtin_ia32_sbb_u32.
25912 (_subborrow_u64): Swap _X and _Y arguments in the call to
25913 __builtin_ia32_sbb_u64.
25915 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25918 * tree-vrp.c (compare_assert_loc): Turn into a function template
25919 with stable template parameter. Only test if a->e is NULL,
25920 !a->e == !b->e has been verified already. Use e == NULL or
25921 e != NULL instead of e or ! e tests. If stable is true, don't use
25922 iterative_hash_expr, on the other side allow a or b or both NULL
25923 and sort the NULLs last.
25924 (process_assert_insertions): Sort using compare_assert_loc<false>
25925 instead of compare_assert_loc, later sort using
25926 compare_assert_loc<true> before calling process_assert_insertions_for
25927 in a loop. Use break instead of continue once seen NULL pointer.
25929 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25931 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25932 Cortex-R7 and Cortex-R8 processors.
25934 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25936 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25937 uninitialized while src is not.
25939 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25941 * common/config/arm/arm-common.c: Adjust include path for
25943 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25944 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25945 (arm-cpu-data.h): Likewise.
25946 (arm-cpu-cdata.h): Likewise.
25947 * config/arm/arm-cpu.h: Delete.
25948 * config/arm/arm-cpu-cdata.h: Delete.
25949 * config/arm/arm-cpu-data.h: Delete.
25951 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25953 * config/arm/arm-cpus.in (cortex-a55): New.
25954 (cortex-a75): Likewise.
25955 (cortex-a75.cortex-a55): Likewise.
25956 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25958 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25959 * config/arm/arm-cpu-cdata.h: Regenerate.
25960 * config/arm/arm-cpu-data.h: Regenerate.
25961 * config/arm/arm-cpu.h: Regenerate.
25962 * config/arm/arm-tables.opt: Regenerate.
25963 * config/arm/arm-tune.md: Regenerate.
25965 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25967 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25969 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25971 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25974 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25976 PR tree-optimization/81292
25977 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25978 full_string_p, also call adjust_related_strinfos if the adjustment
25979 is simple, otherwise invalidate related strinfos.
25981 2017-07-04 Martin Liska <mliska@suse.cz>
25984 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25985 newly created variable as DECL_IGNORED_P.
25987 2017-07-04 Martin Liska <mliska@suse.cz>
25990 * ipa-inline.c (inline_small_functions):
25991 Use xstrdup_for_dump.
25993 2017-07-04 Tom de Vries <tom@codesourcery.com>
25995 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25997 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
26000 * config/darwin.c (darwin_function_switched_text_sections):
26003 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
26005 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
26007 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26009 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
26011 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26013 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
26014 min_profitable_iters, and th as inclusive lower bounds.
26015 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
26016 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
26017 for min_profitable_iters and min_profitable_estimate.
26018 (vect_transform_loop): Treat th as an inclusive lower bound.
26019 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26021 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
26024 * config/darwin.c (darwin_function_switched_text_sections):
26025 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
26026 in two pieces, and suppress the use of buf.
26028 2017-07-03 Nathan Sidwell <nathan@acm.org>
26030 * hash-table.h (hash_table_mod1): Fix indentation.
26032 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26034 PR middle-end/81290
26035 * predict.c (force_edge_cold): Be more careful about propagation
26037 * profile-count.h (profile_probability::guessed,
26038 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
26040 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
26042 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
26044 * doc/invoke.texi (rcpc architecture extension): Document it.
26046 2017-07-03 Richard Biener <rguenther@suse.de>
26048 PR tree-optimization/60510
26049 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
26050 the scalar reduction PHI and use it.
26051 (vectorizable_reduction): Properly guard the single_defuse_cycle
26052 path for non-SLP reduction chains where we cannot use it.
26053 Rework reduc_def/index and vector type deduction. Rework
26054 vector operand gathering during reduction op code-gen.
26055 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
26056 chains dissolve the chain and leave it to non-SLP reduction
26059 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26061 * tree-data-ref.h (dr_alignment): Declare.
26062 * tree-data-ref.c (dr_alignment): New function.
26063 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
26064 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
26066 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
26068 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26070 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
26071 and base_misalignment fields.
26072 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
26073 * tree-data-ref.c: Include builtins.h.
26074 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
26075 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
26076 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
26077 * tree-vect-data-refs.c: Include tree-cfg.h.
26078 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
26079 fields instead of calculating an alignment here.
26080 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
26081 innermost_loop_behavior fields.
26083 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26085 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
26087 (DR_STEP_ALIGNMENT): New macro.
26088 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
26089 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
26090 (create_data_ref): Print it.
26091 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
26092 to tell whether the step preserves vector (mis)alignment.
26093 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26094 Move the check for an integer step and generalise to all INTEGER_CST.
26095 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
26096 Print the outer step alignment.
26098 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26100 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
26101 with offset_alignment.
26102 (DR_ALIGNED_TO): Delete.
26103 (DR_OFFSET_ALIGNMENT): New macro.
26104 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
26105 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
26106 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
26107 (create_data_ref): Likewise.
26108 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26109 (vect_analyze_data_refs): Likewise.
26110 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
26111 creating dummy innermost behavior.
26113 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26115 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
26116 with a "innermost_loop_behavior *" and refeence tree.
26117 * tree-data-ref.c (dr_analyze_innermost): Likewise.
26118 (create_data_ref): Update call accordingly.
26119 * tree-predcom.c (find_looparound_phi): Likewise.
26121 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26123 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
26124 fields with dr_wrt_vec_loop.
26125 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
26126 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
26127 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
26128 (vect_dr_behavior): New function.
26129 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26130 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
26131 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
26132 track whether the step preserves the misalignment.
26133 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26134 Use vect_dr_behavior.
26135 (vect_setup_realignment): Update call accordingly.
26136 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
26137 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
26138 call to vect_create_addr_base_for_vector_ref.
26139 (vect_create_cond_for_align_checks): Likewise.
26140 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
26141 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
26142 (vect_recog_mask_conversion_pattern): Likewise.
26143 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
26144 (new_stmt_vec_info): Remove redundant zeroing.
26146 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26148 * common/config/arm/arm-common.c (arm_be8_option): New function.
26149 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
26150 (ISA_ARMv6): Add isa_bit_be8.
26151 * config/arm/arm.h (arm_be8_option): Add prototype.
26152 (BE8_SPEC_FUNCTION): New define.
26153 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
26154 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
26155 (mlittle-endian): Similarly.
26156 (mbe8, mbe32): New options.
26157 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
26158 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
26160 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26162 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26164 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26166 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26167 (cleanup_tree_cfg_bb): Use it.
26168 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26170 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26172 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26175 * loop-doloop.c (add_test): Update profile.
26177 2017-07-03 Martin Liska <mliska@suse.cz>
26180 * sanopt.c (rewrite_usage_of_param): New function.
26181 (sanitize_rewrite_addressable_params): Likewise.
26182 (pass_sanopt::execute): Call rewrite_usage_of_param.
26184 2017-07-03 Richard Biener <rguenther@suse.de>
26186 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26187 back to using VIEW_CONVERT_EXPR.
26189 2017-07-03 Martin Liska <mliska@suse.cz>
26192 * doc/extend.texi: Document when a resolver function is
26193 generated for target_clones.
26195 2017-07-03 Martin Liska <mliska@suse.cz>
26197 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26198 and shadow memory of auto variables which are subject of
26199 use-after-scope sanitization.
26200 (asan_expand_mark_ifn): Add do set only when is_poison.
26202 2016-07-03 Richard Biener <rguenther@suse.de>
26204 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26206 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26207 (vectorizable_reduction): Perform reduction PHI creation when
26208 visiting a reduction PHI and adjust and simplify code generation
26209 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26210 (vect_transform_loop): Visit reduction PHIs.
26211 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26212 defs into the SLP tree.
26213 (vect_build_slp_tree): Reduction defs terminate the recursion.
26214 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26216 (vect_get_vec_defs_for_stmt_copy): Export.
26217 (vect_get_vec_defs): Likewise.
26218 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26220 (vect_get_vec_defs_for_stmt_copy): Declare.
26221 (vect_get_vec_defs): Likewise.
26223 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26225 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26226 parameter with a "loop" parameter and use it instead of the
26227 loop containing DR_STMT. Don't check simple_iv when doing
26228 BB analysis. Describe the two analysis modes in the comment.
26230 2017-07-03 Tom de Vries <tom@codesourcery.com>
26232 PR tree-optimization/69468
26233 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26234 (find_same_succ_bb): Handle ignore_edge_flags.
26236 2017-07-03 Tom de Vries <tom@codesourcery.com>
26238 PR tree-optimization/81192
26239 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26241 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26243 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26245 2017-07-03 Tom de Vries <tom@codesourcery.com>
26247 PR tree-optimization/81192
26248 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26249 BB_SAME_SUCC (bb) == NULL.
26251 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26253 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26256 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26258 * dumpfile.c: Include profile-count.h
26259 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26261 (insert_cond_bb): Update profile.
26262 * tree-cfg.h (insert_cond_bb): Update prototype.
26263 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26264 * tree-dump.c: Do not include tree-cfg.
26266 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26268 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26270 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26272 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26275 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26277 * tree-complex.c (expand_complex_div_wide): update profile.
26279 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26280 Alan Hayward <alan.hayward@arm.com>
26281 David Sherwood <david.sherwood@arm.com>
26283 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26284 (CORETYPES_H): New define.
26285 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26286 (insn-modes-inline.h, s-modes-inline-h): New rules.
26287 (generated_files): Add insn-modes-inline.h.
26288 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26289 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26290 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26291 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26292 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26293 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26294 (build/gencodes.o, build/genconditions.o): Likewise.
26295 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26296 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26297 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26298 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26299 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26300 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26301 * coretypes.h: Include everything up to real.h for generators.
26302 Include insn-modes.h first. Include wide-int-print.h after
26303 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26304 * machmode.h: Don't include insn-modes.h here.
26305 * function-tests.c: Remove includes of signop.h, machmode.h,
26306 double-int.h and wide-int.h.
26308 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26310 * optc-save-gen.awk: Likewise.
26311 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26312 * godump.c: Remove include of wide-int-print.h.
26313 * pretty-print.h: Likewise.
26314 * wide-int-print.cc: Likewise.
26315 * wide-int.cc: Likewise.
26316 * hash-map-tests.c: Remove include of signop.h.
26317 * hash-set-tests.c: Likewise.
26318 * rtl-tests.c: Likewise.
26319 * mkconfig.sh: Remove include of machmode.h.
26320 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26322 (emit_insn_modes_inline_h): ...this new function. Emit the code
26323 into an insn-modes-inline.h header file, adding appropriate
26324 include guards and end comments.
26325 (emit_insn_modes_c_header): Remove include of machmode.h.
26326 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26328 (main): Handle -i flag and call emit_insn_modes_inline_h when
26331 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26333 * tree-ssa-strlen.c (strinfo): Rename the length field to
26334 nonzero_chars. Add a full_string_p field.
26335 (compare_nonzero_chars, zero_length_string_p): New functions.
26336 (get_addr_stridx): Add an offset_out parameter.
26337 Use compare_nonzero_chars.
26338 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26339 (new_strinfo): Update after above changes to strinfo.
26340 (set_endptr_and_length): Set full_string_p.
26341 (get_string_length): Update after above changes to strinfo.
26342 (unshare_strinfo): Update call to new_strinfo.
26343 (maybe_invalidate): Likewise.
26344 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26345 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26346 as a uhwi instead of an shwi. Update after above changes to
26347 strinfo and new_strinfo.
26348 (zero_length_string): Assert that chainsi contains full strings.
26349 Use zero_length_string_p. Update call to new_strinfo.
26350 (adjust_related_strinfos): Update after above changes to strinfo.
26351 Copy full_string_p from origsi.
26352 (adjust_last_stmt): Use zero_length_string_p.
26353 (handle_builtin_strlen): Update after above changes to strinfo and
26354 new_strinfo. Install the lhs as the string length if the previous
26355 entry didn't describe a full string.
26356 (handle_builtin_strchr): Update after above changes to strinfo
26358 (handle_builtin_strcpy): Likewise.
26359 (handle_builtin_strcat): Likewise.
26360 (handle_builtin_malloc): Likewise.
26361 (handle_pointer_plus): Likewise.
26362 (handle_builtin_memcpy): Likewise. Track nonzero characters
26363 that aren't necessarily followed by a nul terminator.
26364 (handle_char_store): Likewise.
26366 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26368 PR tree-optimization/80769
26369 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26370 for malloc and calloc. Document the new invariant that all related
26371 strinfos have delayed lengths or none do.
26372 (verify_related_strinfos): Move earlier in file.
26373 (set_endptr_and_length): New function, split out from...
26374 (get_string_length): ...here. Also set the lengths of related
26376 (zero_length_string): Assert that chainsi has known (rather than
26378 (adjust_related_strinfos): Likewise.
26380 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26382 PR tree-optimization/81136
26383 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26384 assert that two references with the same misalignment have the same
26385 compile-time misalignment if those compile-time misalignments
26388 2017-07-01 Andi Kleen <ak@linux.intel.com>
26390 * print-tree.c (print_node): Print all attributes.
26392 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26394 * cfg.c (scale_bbs_frequencies): New function.
26395 * cfg.h (scale_bbs_frequencies): Declare it.
26396 * cfgloopanal.c (single_likely_exit): Cleanup.
26397 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26399 (scale_loop_profile): Likewise.
26400 (loop_version): Likewise.
26401 (create_empty_loop_on_edge): Update.
26402 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26403 scale_loop_frequencies, scale_loop_profile, loopify,
26404 loop_version): Update prototypes.
26405 * modulo-sched.c (sms_schedule): Update.
26406 * predict.c (unlikely_executed_edge_p): Also check probability.
26407 (probably_never_executed_edge_p): Fix typo.
26408 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26409 * tree-parloops.c (gen_parallel_loop): Update.
26410 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26411 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26412 * tree-ssa-loop-split.c (split_loop): Update.
26413 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26414 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26415 (vect_loop_versioning): Update.
26416 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26418 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26420 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26422 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26424 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26425 to keep profile consistent.
26427 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26429 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26430 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26431 * profile-count.h (max_safe_multiplier): Make unsigned.
26432 (profile_count::guessed_zero): New.
26434 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26436 * bb-reorder.c (fix_up_crossing_landing_pad,
26437 fix_crossing_conditional_branches): Use make_single_succ_edge
26438 to keep profile consistent.
26440 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26442 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26445 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26448 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26449 the right scopes, make sure cond_jump isn't preserved between multiple
26450 iterations. Search for fallthru edge whenever there are 3+ edges and
26451 use find_fallthru_edge for it.
26453 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26455 Patch by Alexander Monakov <amonakov@ispras.ru>
26456 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26457 probabilities consistently.
26459 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26461 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26462 emit_cmp_and_jump_insns.
26464 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26467 * tree-inline.c (expand_call_inline): Combine profile statuses.
26469 2017-06-30 Andrew Pinski <apinski@cavium.com>
26471 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26472 fold_stmt returned true.
26474 2017-06-30 Nathan Sidwell <nathan@acm.org>
26476 * ggc.h (empty_string): Delete.
26477 * cfgexpand.c (expand_asm_stmt): Use plain "".
26478 * optabs.c (expand_asm_memory_barrier): Likewise.
26479 * stringpool.c (empty_string): Delete.
26480 (digit_vector, digit_string): Delete.
26481 (ggc_alloc_string): Use plain "", don't optimize single digit
26482 strings. Use ggc_alloc_atomic.
26484 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26486 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26487 comparison set and one other set, use the cost of the non-comparison
26490 2017-06-30 Nathan Sidwell <nathan@acm.org>
26492 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26495 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26497 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26498 loops. Remove now unneeded calls to gimple_switch_set_label() that
26499 just set removed labels to NULL_TREE.
26501 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26503 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26504 (set_range_info): ...here. Only call set_range_info_raw if domain
26506 (set_nonzero_bits): Call set_range_info_raw.
26507 * tree-ssanames.h (set_range_info_raw): New.
26509 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26512 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26513 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26514 of nonimmediate_operand and <store_mask_constraint> instead of m
26515 for the input operand. For V8FI iterator, always split if input
26516 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26517 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26518 <store_mask_predicate> instead of register_operand and
26519 <store_mask_constraint> instead of v for the input operand. Make
26520 sure both operands aren't MEMs for if not <mask_applied>.
26522 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26524 * lto-wrapper.c (copy_file) Close both file descriptors before
26527 2017-06-30 Martin Liska <mliska@suse.cz>
26530 * multiple_target.c (create_dispatcher_calls): Make ifunc
26531 also for function that don't have calls or are not referenced.
26533 2017-06-30 Richard Biener <rguenther@suse.de>
26535 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26536 analyze the first scalar stmt. Move vector type computation
26537 for the BB case here from ...
26538 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26539 live operation processing in the SLP case properly.
26541 2017-06-30 Richard Biener <rguenther@suse.de>
26543 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26545 2017-06-30 Martin Liska <mliska@suse.cz>
26548 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26549 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26551 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26553 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26556 2017-06-30 Richard Biener <rguenther@suse.de>
26558 PR tree-optimization/81249
26559 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26560 condition reduction result to original scalar type.
26562 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26564 * profile-count.h (enum profile_quality): Fix typos and whitespace
26567 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26569 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26570 type for branch probabilities.
26572 2017-06-29 Julian Brown <julian@codesourcery.com>
26573 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26575 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26576 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26577 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26578 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26580 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26582 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26583 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26584 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26585 CC usage from generic code to here.
26586 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26587 CC usage into the target macros.
26589 2017-06-29 Maya Rashish <coypu@sdf.org>
26591 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26594 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26596 * arm/arm-builtins.c: Include profile-count.h
26597 * except.c (sjlj_emit_function_enter): Use
26598 profile_probability::unlikely.
26600 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26602 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26603 and tocrel_offset be pointer args rather than implicitly using
26605 (legitimate_constant_pool_address_p, rs6000_emit_move,
26606 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26607 tocrel_offset and use in toc_relative_expr_p call.
26608 (print_operand, print_operand_address): Use static tocrel_base_oac
26609 and tocrel_offset_oac.
26610 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26613 2017-06-29 Maya Rashish <coypu@sdf.org>
26615 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26617 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26619 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26620 objects, take into account only the alignment of 'op0' and 'mode1' if
26623 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26625 * ccmp.c (ccmp_tree_comparison_p): New function.
26626 (ccmp_candidate_p): Update to use above function.
26627 (get_compare_parts): New function.
26628 (expand_ccmp_next): Update to use new functions.
26629 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26631 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26632 take mode as argument.
26633 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26634 * expr.c (expand_expr_real_1): Pass mode as argument.
26636 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26638 * combine.c (combine_instructions): Print insns to dump_file, together
26641 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26643 * asan.c (asan_emit_stack_protection): Update.
26644 (create_cond_insert_point): Update.
26645 * auto-profile.c (afdo_propagate_circuit): Update.
26646 * basic-block.h (struct edge_def): Turn probability to
26647 profile_probability.
26648 (EDGE_FREQUENCY): Update.
26649 * bb-reorder.c (find_traces_1_round): Update.
26650 (better_edge_p): Update.
26651 (sanitize_hot_paths): Update.
26652 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26653 (make_single_succ_edge): Update.
26654 (check_bb_profile): Update.
26655 (dump_edge_info): Update.
26656 (update_bb_profile_for_threading): Update.
26657 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26658 probabilitycount to 0.
26659 * cfgbuild.c (compute_outgoing_frequencies): Update.
26660 * cfgcleanup.c (try_forward_edges): Update.
26661 (outgoing_edges_match): Update.
26662 (try_crossjump_to_edge): Update.
26663 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26664 (expand_gimple_tailcall): Update.
26665 (construct_init_block): Use make_single_succ_edge.
26666 (construct_exit_block): Use make_single_succ_edge.
26667 * cfghooks.c (verify_flow_info): Update.
26668 (redirect_edge_succ_nodup): Update.
26669 (split_edge): Update.
26670 (account_profile_record): Update.
26671 * cfgloopanal.c (single_likely_exit): Update.
26672 * cfgloopmanip.c (scale_loop_profile): Update.
26673 (set_zero_probability): Remove.
26674 (duplicate_loop_to_header_edge): Update.
26675 * cfgloopmanip.h (loop_version): Update prototype.
26676 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26677 (force_nonfallthru_and_redirect): Update.
26678 (update_br_prob_note): Update.
26679 (rtl_verify_edges): Update.
26680 (purge_dead_edges): Update.
26681 (rtl_lv_add_condition_to_bb): Update.
26682 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26683 * cgraphunit.c (init_lowered_empty_function): Update.
26684 (cgraph_node::expand_thunk): Update.
26685 * cilk-common.c: Include profile-count.h
26686 * dojump.c (inv): Remove.
26687 (jumpifnot): Update.
26688 (jumpifnot_1): Update.
26689 (do_jump_1): Update.
26691 (do_jump_by_parts_greater_rtx): Update.
26692 (do_compare_rtx_and_jump): Update.
26693 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26694 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26695 * dwarf2cfi.c: Include profile-count.h
26696 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26697 (sjlj_emit_dispatch_table): Likewise.
26698 * explow.c: Include profile-count.h
26699 * expmed.c (emit_store_flag_force): Update.
26700 (do_cmp_and_jump): Update.
26701 * expr.c (compare_by_pieces_d::generate): Update.
26702 (compare_by_pieces_d::finish_mode): Update.
26703 (emit_block_move_via_loop): Update.
26704 (store_expr_with_bounds): Update.
26705 (store_constructor): Update.
26706 (expand_expr_real_2): Update.
26707 (expand_expr_real_1): Update.
26708 * expr.h (try_casesi, try_tablejump): Update prototypes.
26709 * gimple-pretty-print.c (dump_probability): Update.
26710 (dump_profile): New.
26711 (dump_gimple_label): Update.
26712 (dump_gimple_bb_header): Update.
26713 * graph.c (draw_cfg_node_succ_edges): Update.
26714 * hsa-gen.c (convert_switch_statements): Update.
26715 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26716 (find_if_case_1): Update.
26717 (find_if_case_2): Update.
26718 * internal-fn.c (expand_arith_overflow_result_store): Update.
26719 (expand_addsub_overflow): Update.
26720 (expand_neg_overflow): Update.
26721 (expand_mul_overflow): Update.
26722 (expand_vector_ubsan_overflow): Update.
26723 * ipa-cp.c (good_cloning_opportunity_p): Update.
26724 * ipa-split.c (split_function): Use make_single_succ_edge.
26725 * ipa-utils.c (ipa_merge_profiles): Update.
26726 * loop-doloop.c (add_test): Update.
26727 (doloop_modify): Update.
26728 * loop-unroll.c (compare_and_jump_seq): Update.
26729 (unroll_loop_runtime_iterations): Update.
26730 * lra-constraints.c (lra_inheritance): Update.
26731 * lto-streamer-in.c (input_cfg): Update.
26732 * lto-streamer-out.c (output_cfg): Update.
26733 * mcf.c (adjust_cfg_counts): Update.
26734 * modulo-sched.c (sms_schedule): Update.
26735 * omp-expand.c (expand_omp_for_init_counts): Update.
26736 (extract_omp_for_update_vars): Update.
26737 (expand_omp_ordered_sink): Update.
26738 (expand_omp_for_ordered_loops): Update.
26739 (expand_omp_for_generic): Update.
26740 (expand_omp_for_static_nochunk): Update.
26741 (expand_omp_for_static_chunk): Update.
26742 (expand_cilk_for): Update.
26743 (expand_omp_simd): Update.
26744 (expand_omp_taskloop_for_outer): Update.
26745 (expand_omp_taskloop_for_inner): Update.
26746 * omp-simd-clone.c (simd_clone_adjust): Update.
26747 * optabs.c (expand_doubleword_shift): Update.
26748 (expand_abs): Update.
26749 (emit_cmp_and_jump_insn_1): Update.
26750 (expand_compare_and_swap_loop): Update.
26751 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26752 * predict.c (predictable_edge_p): Update.
26753 (edge_probability_reliable_p): Update.
26754 (set_even_probabilities): Update.
26755 (combine_predictions_for_insn): Update.
26756 (combine_predictions_for_bb): Update.
26757 (propagate_freq): Update.
26758 (estimate_bb_frequencies): Update.
26759 (force_edge_cold): Update.
26760 * profile-count.c (profile_count::dump): Add missing space into dump.
26761 (profile_count::debug): Add newline.
26762 (profile_count::differs_from_p): Explicitly convert to unsigned.
26763 (profile_count::stream_in): Update.
26764 (profile_probability::dump): New member function.
26765 (profile_probability::debug): New member function.
26766 (profile_probability::differs_from_p): New member function.
26767 (profile_probability::differs_lot_from_p): New member function.
26768 (profile_probability::stream_in): New member function.
26769 (profile_probability::stream_out): New member function.
26770 * profile-count.h (profile_count_quality): Rename to ...
26771 (profile_quality): ... this one.
26772 (profile_probability): New.
26773 (profile_count): Update.
26774 * profile.c (compute_branch_probabilities): Update.
26775 * recog.c (peep2_attempt): Update.
26776 * sched-ebb.c (schedule_ebbs): Update.
26777 * sched-rgn.c (find_single_block_region): Update.
26778 (compute_dom_prob_ps): Update.
26779 (schedule_region): Update.
26780 * sel-sched-ir.c (compute_succs_info): Update.
26781 * stmt.c (struct case_node): Update.
26782 (do_jump_if_equal): Update.
26783 (get_outgoing_edge_probs): Update.
26784 (conditional_probability): Update.
26785 (emit_case_dispatch_table): Update.
26786 (expand_case): Update.
26787 (expand_sjlj_dispatch_table): Update.
26788 (emit_case_nodes): Update.
26789 * targhooks.c: Update.
26790 * tracer.c (better_p): Update.
26791 (find_best_successor): Update.
26792 * trans-mem.c (expand_transaction): Update.
26793 * tree-call-cdce.c: Update.
26794 * tree-cfg.c (gimple_split_edge): Upate.
26795 (move_sese_region_to_fn): Upate.
26796 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26797 * tree-eh.c (lower_resx): Upate.
26798 (cleanup_empty_eh_move_lp): Upate.
26799 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26800 * tree-inline.c (copy_edges_for_bb): Update.
26801 (copy_cfg_body): Update.
26802 * tree-parloops.c (gen_parallel_loop): Update.
26803 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26804 (gimple_gen_time_profiler): Update.
26805 * tree-ssa-dce.c (remove_dead_stmt): Update.
26806 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26807 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26808 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26809 (unloop_loops): Update.
26810 (try_peel_loop): Update.
26811 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26812 * tree-ssa-loop-split.c (connect_loops): Update.
26813 (split_loop): Update.
26814 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26815 (hoist_guard): Update.
26816 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26817 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26818 (value_replacement): Update.
26819 * tree-ssa-reassoc.c (branch_fixup): Update.
26820 * tree-ssa-tail-merge.c (replace_block_by): Update.
26821 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26822 (create_edge_and_update_destination_phis): Update.
26823 (compute_path_counts): Update.
26824 (recompute_probabilities): Update.
26825 (update_joiner_offpath_counts): Update.
26826 (freqs_to_counts_path): Update.
26827 (duplicate_thread_path): Update.
26828 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26829 (struct switch_conv_info): Update.
26830 (gen_inbound_check): Update.
26831 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26832 (vect_do_peeling): Update.
26833 (vect_loop_versioning): Update.
26834 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26835 (optimize_mask_stores): Update.
26836 * ubsan.c (ubsan_expand_null_ifn): Update.
26837 * value-prof.c (gimple_divmod_fixed_value): Update.
26838 (gimple_divmod_fixed_value_transform): Update.
26839 (gimple_mod_pow2): Update.
26840 (gimple_mod_pow2_value_transform): Update.
26841 (gimple_mod_subtract): Update.
26842 (gimple_mod_subtract_transform): Update.
26843 (gimple_ic): Update.
26844 (gimple_stringop_fixed_value): Update.
26845 (gimple_stringops_transform): Update.
26846 * value-prof.h: Update.
26848 2017-06-29 Carl Love <cel@us.ibm.com>
26850 * config/rs6000/rs6000-c.c: Add support for built-in functions
26851 vector signed int vec_signed (vector float);
26852 vector signed long long vec_signed (vector double);
26853 vector signed int vec_signed2 (vector double, vector double);
26854 vector signed int vec_signede (vector double);
26855 vector signed int vec_signedo (vector double);
26856 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26857 instruction generator.
26858 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26859 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26860 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26862 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26863 vunsignede_v2df): Add define_expands.
26864 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26865 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26866 VEC_UNSIGNEDO): Add definitions.
26867 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26868 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26869 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26870 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26871 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26872 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26873 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26874 vec_unsignede, vec_unsignedo): Add builtin defines.
26875 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26877 * doc/extend.texi: Update the built-in documentation file for the
26878 new built-in functions.
26880 2017-06-29 Richard Biener <rguenther@suse.de>
26882 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26883 reduction chains to LOOP_VINFO_REDUCTIONS.
26884 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26885 SLP reductions after processing reduction chains.
26887 2017-06-29 Nathan Sidwell <nathan@acm.org>
26889 * builtins.c (fold_builtin_FUNCTION): Use
26890 lang_hooks.decl_printable_name.
26892 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26894 PR middle-end/81194
26895 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26896 with only one label.
26897 * stmt.c (expand_case): Assert NCASES is greater than one.
26899 2017-06-29 Richard Biener <rguenther@suse.de>
26901 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26903 (group_case_labels): Likewise.
26904 (find_taken_edge): Push sanity checking on val to workers...
26905 (find_taken_edge_cond_expr): ... here
26906 (find_taken_edge_switch_expr): ... and here, handle cases
26907 with just a default label.
26908 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26909 (group_case_labels): Likewise.
26910 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26911 group_case_labels does anything cleanup the CFG again.
26913 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26915 PR tree-optimization/81196
26916 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26917 exit condition comparing two IVs.
26919 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26921 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26922 profile to the dummy entry at the end of the list of architectures.
26923 * config/arm/arm-cpu-cdata.h: Regenerated.
26925 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26926 Michael Collison <michael.collison@arm.com>
26929 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26931 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26932 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26933 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26934 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26935 of shift when the shift amount is masked with constant equal to
26936 the size of the mode.
26937 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26940 2017-06-29 Martin Liska <mliska@suse.cz>
26942 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26944 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26947 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26950 (edge_growth_cache_entry::edge_growth_cache_entry): New
26952 (reset_edge_growth_cache): Update to use constructor.
26954 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26956 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26957 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26958 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26960 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26962 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26963 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26965 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26967 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26968 (*-linux-uclibc*): Add t-uclibc tmake_file.
26969 * config/t-musl: New.
26970 * config/t-uclibc: New.
26972 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26974 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26976 (gen_comm_data): Emit architectural setting of arch_prof.
26977 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26979 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26980 (armv8-m.base, armv8-m.main): Likewise.
26981 * arm-protos.h (arm_build_target): Add profile field.
26982 (arch_option): Likewise.
26983 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26985 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26986 arm_active_target.profile.
26988 2017-06-28 Richard Biener <rguenther@suse.de>
26990 PR middle-end/81227
26991 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26992 TYPE_OVERFLOW_WRAPS.
26993 * match.pd (negate_expr_p): Likewise.
26994 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26995 fold_build2, not fold_binary.
26997 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26999 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
27000 Convert memory address to Pmode.
27001 (aarch64_print_operand): Assert MEM operands are always Pmode.
27003 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
27006 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
27007 Remove redundant if.
27008 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
27009 * config/arm/aarch-common-protos.h
27010 (aarch_forward_to_shift_is_not_shifted_re): Remove.
27011 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
27013 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
27016 * multiple_target.c (create_dispatcher_calls): Set the default
27017 clone to be static, not public.
27019 2017-06-28 Richard Biener <rguenther@suse.de>
27021 * tree-vect-loop.c (vectorizable_reduction): Move special
27022 cond reduction IV var creation ...
27023 (vect_create_epilog_for_reduction): ... here. Remove induction_index
27024 parameter. Use STMT_VINFO_VECTYPE.
27025 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
27028 2017-06-28 Martin Liska <mliska@suse.cz>
27031 * ipa-visibility.c (non_local_p): Handle visibility.
27033 2017-06-28 Martin Liska <mliska@suse.cz>
27036 * common.opt: Add IntegerRange to various options.
27037 * opt-functions.awk (integer_range_info): New function.
27038 * optc-gen.awk: Add integer_range_info to cl_options struct.
27039 * opts-common.c (decode_cmdline_option): Handle
27040 CL_ERR_INT_RANGE_ARG.
27041 (cmdline_handle_error): Likewise.
27042 * opts.c (print_filtered_help): Show valid interval in
27043 when --help is provided.
27044 * opts.h (struct cl_option): Add range_min and range_max fields.
27045 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
27047 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
27049 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
27050 (x * C EQ/NE y * C): New transformation.
27052 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
27054 * genmultilib (combination_space): Accept '+' in option names.
27056 2017-06-28 Martin Liska <mliska@suse.cz>
27059 * asan.c (instrument_derefs): Bail out inner references
27060 that are hard register variables.
27062 2017-06-28 Jakub Jelinek <jakub@redhat.com>
27065 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
27066 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
27068 2017-06-28 Richard Biener <rguenther@suse.de>
27070 * tree-vectorizer.h (vect_get_vec_defs): Remove.
27071 (vect_get_slp_defs): Adjust.
27072 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
27074 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
27076 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
27077 get_initial_defs_for_reduction instead of vect_get_vec_defs.
27078 (vectorizable_reduction): Adjust.
27079 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
27081 (vect_get_slp_defs): Likewise.
27082 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
27083 (vectorizable_bswap): Adjust.
27084 (vectorizable_call): Likewise.
27085 (vectorizable_conversion): Likewise.
27086 (vectorizable_assignment): Likewise.
27087 (vectorizable_shift): Likewise.
27088 (vectorizable_operation): Likewise.
27089 (vectorizable_store): Likewise.
27090 (vectorizable_condition): Likewise.
27091 (vectorizable_comparison): Likewise.
27093 2017-06-28 Michael Collison <michael.collison@arm.com>
27096 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
27098 (arm_gen_movmemqi): Removed unused variable 'i'.
27099 Convert 'for' loop into 'while' loop.
27100 (arm_expand_prologue): Remove last unnecessary set of insn.
27101 (thumb_pop): Remove unused variable 'pushed_words'.
27102 (thumb_exit): Remove last unnecessary set of regs_to_pop.
27104 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27106 * config/s390/predicates.md: Use s390_rel_address_ok_p.
27107 * config/s390/s390-protos.h: Add prototype of
27108 s390_rel_address_ok_p.
27109 * config/s390/s390.c (s390_got_symbol): New function.
27110 (s390_rel_address_ok_p): New function.
27111 (legitimize_pic_address): Use s390_rel_address_ok_p.
27112 (s390_load_got): Use s390_got_symbol.
27113 (s390_option_override): Issue error if
27114 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
27115 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
27117 * config/s390/s390.opt: New option mpic-data-is-text-relative.
27119 2017-06-27 Andrew Pinski <apinski@cavium.com>
27121 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
27122 (X * copysign (1.0, X)): New pattern.
27123 (X * copysign (1.0, -X)): New pattern.
27124 (copysign (-1.0, CST)): New pattern.
27126 2017-06-27 Joseph Myers <joseph@codesourcery.com>
27128 * genmultilib (combination_space): Remove variable.
27129 Validate reuse rules against regular expression for any sequence
27130 of multilib options in any order.
27132 2017-06-27 Michael Collison <michael.collison@arm.com>
27134 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
27135 call aarch64_split_simd_combine.
27136 * (aarch64_combine_internal<mode>): Delete pattern.
27137 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
27138 Allow register and subreg operands.
27140 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27142 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
27143 specific need, just fallback on defaults.
27144 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
27146 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27147 Olivier Hainque <hainque@adacore.com>
27149 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
27151 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
27152 targets. Pick a default if no particular attempt applied.
27153 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
27156 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27158 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
27159 (x86_64-wrs-vxworks7): Likewise.
27161 2017-06-27 Marek Polacek <polacek@redhat.com>
27164 * ubsan.c (instrument_null): Check get_base_address's result for null.
27166 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27168 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27170 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27172 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27173 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27174 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27175 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27176 New function types.
27177 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27178 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27179 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27180 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27181 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27182 BUILT_IN_FEUPDATEENV): New builtins.
27183 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27184 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27185 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27186 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27188 (builtin_structptr_types): Adjust size.
27189 * tree.c (builtin_structptr_types): Add four entries.
27191 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27192 Olivier Hainque <hainque@adacore.com>
27194 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27195 (TLS_SYM): New local macro, forcing reference to __tls__ on
27196 link command lines for VxWorks 7 RTPs, triggering initialization
27198 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27199 OS features TLS support, true for RTPs on VxWorks 7.
27200 * config/vxworks.c (vxworks_override_options): Setup emutls
27203 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27205 * predict.c (test_prediction_value_range): Use -1U instead of -1
27206 to avoid narrowing conversion warning.
27207 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27208 to avoid narrowing conversion warning.
27209 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27211 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27213 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27215 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27216 64bit configurations.
27217 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27218 (SIZE_TYPE): Likewise.
27219 * config/vxworks.c (vxworks_emutls_var_fields): Use
27220 long_unsigned_type_node instead of unsigned_type_node as the offset
27221 field type, which is "pointer" mode in emutls.c.
27223 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27226 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27228 PR middle-end/81207
27229 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27230 gimple_vuse copying separately from gimple_vdef copying.
27232 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27234 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27237 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27238 Olivier Hainque <hainque@adacore.com>
27240 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27241 for all vxworks7 targets.
27242 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27243 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27244 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27245 variations for VX6/VX7 and 32/64bits later on in ...
27246 (VXWORKS_LIB_SPEC): Leverage new macros.
27247 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27248 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27250 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27252 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27253 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27255 2017-06-26 Carl Love <cel@us.ibm.com>
27257 * config/rs6000/rs6000-c.c: Add support for built-in functions
27258 vector bool char vec_reve (vector bool char);
27259 vector signed char vec_reve (vector signed char);
27260 vector unsigned char vec_reve (vector unsigned char);
27261 vector bool int vec_reve (vector bool int);
27262 vector signed int vec_reve (vector signed int);
27263 vector unsigned int vec_reve (vector unsigned int);
27264 vector bool long long vec_reve (vector bool long long);
27265 vector signed long long vec_reve (vector signed long long);
27266 vector unsigned long long vec_reve (vector unsigned long long);
27267 vector bool short vec_reve (vector bool short);
27268 vector signed short vec_reve (vector signed short);
27269 vector double vec_reve (vector double);
27270 vector float vec_reve (vector float);
27271 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27272 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27273 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27274 (altivec_vreve): New pattern.
27275 * config/rs6000/altivec.h (vec_reve): New define.
27276 * doc/extend.texi (vec_rev): Update the built-in documentation file
27277 for the new built-in functions.
27279 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27281 PR tree-optimization/71815
27282 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27284 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27286 (slsr_process_phi): Likewise.
27287 (replace_uncond_cands_and_profitable_phis): Don't replace a
27288 multiply candidate with a stride of 1 (copy or cast).
27289 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27291 (lowest_cost_path): Likewise.
27292 (total_savings): Likewise.
27294 2017-06-26 Richard Biener <rguenther@suse.de>
27297 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27298 Use def_builtin_pure for all gather builtins.
27300 2017-06-26 Richard Biener <rguenther@suse.de>
27302 PR tree-optimization/81203
27303 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27304 non-dominating BBs.
27306 2017-06-26 Marek Polacek <polacek@redhat.com>
27309 * doc/invoke.texi: Document -Wmultistatement-macros.
27311 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27313 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27314 arm_neon_ok_no_float_abi effective target.
27316 2017-06-26 Richard Biener <rguenther@suse.de>
27318 PR tree-optimization/80928
27319 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27320 (copy_bbs): Set BB_DUPLICATED flag early.
27321 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27323 (execute_on_shrinking_pred): Likewise.
27324 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27325 BB_DUPLICATED blocks.
27326 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27327 iterate over all PHIs considering removal of *gsi.
27329 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27331 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27334 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27336 * config/rs6000/rs6000-string.c: (expand_block_clear,
27337 do_load_for_compare, select_block_compare_mode,
27338 compute_current_alignment, expand_block_compare,
27339 expand_strncmp_align_check, expand_strn_compare,
27340 expand_block_move, rs6000_output_load_multiple)
27341 Move functions related to string/block move/compare
27342 to a separate file.
27343 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27344 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27345 for this function which is now used in two files.
27346 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27347 * config.gcc: Add rs6000-string.o to extra_objs for
27348 targets powerpc*-*-* and rs6000*-*-*.
27350 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27353 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27354 32-bit, since indexed is not valid for DImode.
27355 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27356 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27357 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27358 (define_peephole2 for Altivec d-form store): Likewise.
27361 * multiple_target.c (create_dispatcher_calls): Only create the
27362 dispatcher call if the function is the default clone of a
27363 versioned function.
27365 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27367 PR middle-end/80902
27368 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27369 a call, force the call to not be a tail call.
27371 2017-06-23 Jeff Law <law@redhat.com>
27373 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27376 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27378 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27379 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27380 handling for early expansion of vector shifts (sl,sr,sra,rl).
27381 (builtin_function_type): Add vector shift right instructions
27382 to the unsigned argument list.
27384 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27386 rtl-optimizatoin/79286
27387 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27388 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27389 trap. PIC register plus a const unspec without offset can never trap.
27391 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27393 * tree.h (builtin_structptr_type): New type.
27394 (builtin_structptr_types): Declare new array.
27395 * tree.c (builtin_structptr_types): New array.
27396 (free_lang_data, build_common_tree_nodes): Use it.
27398 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27401 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27404 2017-06-22 Matt Turner <mattst88@gmail.com>
27406 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27407 Lake models to skylake case. Assume skylake for unknown
27408 models with clflushopt.
27410 2017-06-22 Jeff Law <law@redhat.com>
27412 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27413 frame sizes that do not satisfy aarch64_uimm12_shift.
27415 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27417 * profile-count.h (apply_probability,
27418 apply_scale, probability_in): Fix checks for zero.
27420 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27422 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27423 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27425 2016-06-22 Richard Biener <rguenther@suse.de>
27427 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27428 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27429 REDUC_MAX_EXPR support.
27430 (vectorizable_reduction): Likewise.
27431 (vect_create_epilog_for_reduction): Likewise.
27433 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27435 * match.pd (A / (1 << B) -> A >> B): New.
27436 * generic-match-head.c: Include optabs-tree.h.
27437 * gimple-match-head.c: Likewise.
27438 * optabs-tree.h (target_supports_op_p): New.
27439 * optabs-tree.c (target_supports_op_p): New.
27441 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27443 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27444 $gcc_cv_ld --help output.
27445 (gcc_cv_ld_demangle): Likewise.
27446 (gcc_cv_ld_eh_frame_hdr): Likewise.
27447 (gcc_cv_ld_pie): Likewise.
27448 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27449 (gcc_cv_ld_buildid): Likewise.
27450 (gcc_cv_ld_sysroot): Likewise.
27451 (ld_bndplt_support): Likewise.
27452 (ld_pushpopstate_support): Likewise.
27453 * configure: Regenerate.
27454 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27456 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27459 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27460 operands indexes to avoid gap between operands and match_dups.
27462 2017-06-21 Andrew Pinski <apinski@cavium.com>
27464 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27465 Increment Arith_shift and Arith_shift_reg by 1.
27466 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27468 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27469 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27470 (aarch64_strip_extend): Add new argument and test for it.
27471 (aarch64_cheap_mult_shift_p): New function.
27472 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27473 add a cost if it is true.
27474 Update calls to aarch64_strip_extend.
27475 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27477 2017-06-21 Andrew Pinski <apinski@cavium.com>
27479 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27481 (thunderxt88): Likewise.
27482 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27483 (thunderx_prefetch_tune): New variable.
27484 (thunderx2t99_prefetch_tune): Update for the correct values.
27485 (thunderxt88_tunings): New variable.
27486 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27487 generic_prefetch_tune.
27488 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27490 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27492 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27493 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27494 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27495 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27496 (aarch64_atomic_cas<mode>, GPI): Likewise.
27498 2017-06-21 Martin Liska <mliska@suse.cz>
27500 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27501 statements on cold and hot labels.
27502 * predict.c (tree_estimate_probability_bb): Remove the
27503 prediction from this place.
27505 2017-06-21 Martin Liska <mliska@suse.cz>
27507 PR tree-optimization/79489
27508 * gimplify.c (maybe_add_early_return_predict_stmt): New
27510 (gimplify_return_expr): Call the function.
27511 * predict.c (tree_estimate_probability_bb): Remove handling
27513 * predict.def: Update comment about early return predictor.
27514 * gimple-predict.h (is_gimple_predict): New function.
27515 * predict.def: Change default value of early return to 66.
27516 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27518 * passes.def: Put pass_strip_predict_hints to the beginning of
27521 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27523 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27524 FUNCTION_DECL declarations.
27525 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27527 (dwaf2out_decl): Likewise.
27528 * godump.c (go_early_global_decl): Skip call to the real debug hook
27529 for FUNCTION_DECL declarations.
27530 * passes.c (rest_of_decl_compilation): Skip call to the
27531 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27532 -fdump-go-spec is passed.
27534 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27536 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27537 Reorder for compactness.
27538 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27539 (def_builtin_pure, def_builtin_pure2): New functions.
27540 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27542 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27544 * match.pd (nop_convert): New predicate.
27545 ((A +- CST1) +- CST2): Allow some NOP conversions.
27547 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27550 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27551 with ctors/dtors if GOVD_SHARED is set.
27553 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27555 * config/aarch64/aarch64.md (movti_aarch64):
27556 Emit mov rather than orr.
27557 (movtf_aarch64): Likewise.
27558 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27559 Emit mov rather than orr.
27561 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27563 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27564 Swap alternatives, make integer dup more expensive.
27566 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27568 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27569 Return true for non-tls symbols.
27571 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27573 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27574 (cortex-a75): Likewise.
27575 (cortex-a75.cortex-a55): Likewise.
27576 * config/aarch64/aarch64-tune.md: Regenerate.
27577 * doc/invoke.texi (-mtune): Document new values for -mtune.
27579 2017-06-21 Tom de Vries <tom@codesourcery.com>
27581 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27582 stack_size feature.
27583 (Effective-Target Keywords, Other attributes): Suggest using
27584 dg-add-options stack_size feature to get stack limit in stack_size
27585 effective target documentation.
27587 2017-06-21 Julian Brown <julian@codesourcery.com>
27588 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27590 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27591 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27592 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27594 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27595 attribute type list for neon_multiply.
27596 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27597 attribute type list for neon_multiply.
27598 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27599 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27600 attribute type list for neon_multiply.
27601 * config/arm/types.md (crypto_pmull): Add.
27602 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27603 attribute type list.
27605 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27607 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27610 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27612 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27613 to make sure not to dereference a NULL cost_classes_ptr pointer.
27615 2017-06-20 Carl Love <cel@us.ibm.com>
27617 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27618 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27619 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27620 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27621 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27622 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27623 VMULOSW): New enum "unspec" values.
27624 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27625 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27626 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27627 altivec_vmulosw): New patterns.
27628 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27629 VMULOSW): Add definitions.
27631 2017-06-20 Julia Koval <julia.koval@intel.com>
27633 * config/i386/i386.c: Fix rounding expand for new pattern.
27634 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27636 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27638 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27639 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27641 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27643 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27646 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27648 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27649 architecture, then by implementer ID.
27650 * config/aarch64/aarch64-tune.md: Regenerate.
27652 2017-06-20 Richard Biener <rguenther@suse.de>
27654 PR middle-end/81097
27655 * fold-const.c (split_tree): Fold to type before negating.
27657 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27659 * diagnostic-show-locus.c
27660 (selftest::test_fixit_deletion_affecting_newline): New function.
27661 (selftest::diagnostic_show_locus_c_tests): Call it.
27663 2017-06-20 Andreas Schwab <schwab@suse.de>
27666 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27669 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27671 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27672 __ARM_FEATURE_COPROC according to support.
27674 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27676 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27677 Rewritten to avoid overflow for > 32-bit pointers.
27680 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27681 by removing enum keyword.
27682 (ubsan_type_descriptor): Likewise. Formatting fix.
27685 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27686 splitter): Require TARGET_SSE2 in the condition.
27688 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27691 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27692 for doing vector set of SFmode on ISA 3.0.
27693 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27694 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27696 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27697 SFmode value into a V4SF variable that was extracted from another
27698 V4SF variable without converting the element to double precision
27699 and back to single precision vector format.
27700 (vsx_insert_extract_v4sf_p9_2): Likewise.
27702 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27704 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27705 in UWHI to avoid undefined overflow.
27708 * ubsan.h (enum ubsan_encode_value_phase): New.
27709 (ubsan_encode_value): Change second argument to
27710 enum ubsan_encode_value_phase with default value of
27711 UBSAN_ENCODE_VALUE_GENERIC.
27712 * ubsan.c (ubsan_encode_value): Change second argument to
27713 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27714 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27715 create_tmp_var_raw instead of create_tmp_var and use a
27717 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27718 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27719 ubsan_encode_value callers.
27722 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27723 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27724 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27726 2017-06-19 Richard Biener <rguenther@suse.de>
27728 PR middle-end/81118
27729 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27730 estimates if we changed anything.
27732 2017-06-19 Richard Biener <rguenther@suse.de>
27734 PR tree-optimization/80887
27735 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27736 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27737 simplified lookups, then reset mprts_hook.
27738 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27740 (try_to_simplify): Likewise.
27742 2017-06-19 Martin Liska <mliska@suse.cz>
27745 * gimplify.c (gimplify_switch_expr):
27746 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27748 2017-06-19 Martin Liska <mliska@suse.cz>
27750 * doc/install.texi: Document that PGO runs in 4 stages.
27752 2017-06-19 Martin Liska <mliska@suse.cz>
27755 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27756 to dispatcher function name.
27757 * multiple_target.c (replace_function_decl): New function.
27758 (create_dispatcher_calls): Redirect both edges and references.
27760 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27762 * profile-count.c (profile_count::dump): Dump quality.
27763 (profile_count::differs_from_p): Update for unsigned val.
27764 * profile-count.h (profile_count_quality): New enum.
27765 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27767 2017-06-19 Richard Biener <rguenther@suse.de>
27769 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27770 struct function as arg.
27771 (estimate_numbers_of_iterations): Export overload with loop arg.
27772 (free_numbers_of_iterations_estimates_loop): Use an overload of
27773 free_numbers_of_iterations_estimates instead.
27774 * tree-cfg.c (remove_bb): Adjust.
27775 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27776 * tree-parloops.c (gen_parallel_loop): Likewise.
27777 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27779 (tree_unroll_loops_completely): Likewise.
27780 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27781 Use an overload instead and export.
27782 (estimated_loop_iterations): Adjust.
27783 (max_loop_iterations): Likewise.
27784 (likely_max_loop_iterations): Likewise.
27785 (estimate_numbers_of_iterations): Take struct function as arg
27787 (loop_exits_before_overflow): Adjust.
27788 (free_numbers_of_iterations_estimates_loop): Use an overload.
27789 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27790 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27792 2017-06-19 Richard Biener <rguenther@suse.de>
27795 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27796 RANGE_EXPR conservatively.
27798 2017-06-16 Carl Love <cel@us.ibm.com>
27800 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27801 definitions for vec_float, vec_float2, vec_floato,
27802 vec_floate built-ins.
27803 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27804 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27806 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27807 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27808 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27809 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27810 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27811 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27812 vec_floato): Add builtin defines.
27813 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27814 Update the built-in documentation file for the new built-in
27817 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27819 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27820 (mthumb): Mark as the negative of -marm.
27822 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27824 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27826 (ARM Options, -mtune): Document that this accepts the same
27827 extension options as -mcpu.
27828 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27830 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27832 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27833 permitted extensions.
27835 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27837 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27838 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27839 (armv8-m.main): Add option +nodsp.
27840 * config/arm/arm-cpu-cdata.h: Regenerated.
27842 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27844 * config/arm/t-fuchsia: New file.
27845 * config.gcc (arm*-*-fuchsia*): Use it.
27847 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27849 * config/arm/t-symbian: Rewrite for new option infrastructure.
27851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27853 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27854 (MULTILIB_REQUIRED): Likewise.
27856 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27858 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27859 (MULTILIB_RESUE): Likewise.
27860 (MULTILIB_MATCHES): Likewise.
27861 (MULTLIB_REQUIRED): Likewise.
27863 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27865 * config/arm/t-rtems: Rewrite for new option framework.
27867 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27869 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27870 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27871 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27872 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27873 * config/arm/t-multilib: ... here.
27874 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27875 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27876 armv7-a and armv8*-a when A-profile libraries have not been built.
27877 * config/arm/t-rmprofile: Rewrite.
27879 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27881 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27882 with a backslash. Remove the backslash after substituting unescaped
27884 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27886 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27888 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27889 the driver to pass through the default setting of -mfloat-abi.
27890 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27892 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27893 (all_feat_combs): New rule.
27894 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27896 * config/arm/t-aprofile: Rewrite.
27898 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27900 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27901 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27902 fpu is not specified by the user/command-line.
27903 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27904 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27905 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27906 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27907 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27908 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27910 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27912 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27913 * config/arm/t-arm-elf: Rewritten.
27915 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27917 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27918 have some floating-point instructions.
27919 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27920 (TARGET_MAYBE_HARD_FLOAT): New macro.
27921 * config/arm/arm-builtins.c (arm_init_builtins): Use
27922 TARGET_MAYBE_HARD_FLOAT.
27923 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27925 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27927 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27928 (configargs.h): Include it.
27929 (arm_print_hint_for_fpu_option): New function.
27930 (arm_parse_fpu_option): New function.
27931 (candidate_extension): New class.
27932 (arm_canon_for_multilib): New function.
27933 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27934 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27935 (ARCH_CANONICAL_SPECS): New macro.
27936 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27938 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27940 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27941 are set after handling multilib fragments. Set target_cpu_default2
27944 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27946 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27948 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27949 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27951 (check_arch): Likewise.
27952 * config/arm/arm.c (arm_configure_build_target): Handle
27953 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27954 options in the default.
27956 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27958 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27959 when an option is an alias of another.
27960 * config/arm/parsecpu.awk (optalias): New parser token.
27961 (gen_comm_data): Mark non-alias options as such. Emit entries
27962 for extension aliases.
27963 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27964 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27965 (armv6kz, armv6zk, armv6t2): Likewise.
27966 (armv7): Make vfpv3-d16 an alias.
27967 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27969 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27970 Sort in canonical order.
27971 (armv8-a): Sort in canonical order.
27972 (armv8.1-a, armv8.2-a): Likewise.
27973 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27975 (cortex-a9): Sort in canonical order.
27976 * config/arm/arm.c (selftests.h): Include it.
27977 (arm_test_cpu_arch_data): New function.
27978 (arm_run_self_tests): New function.
27979 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27980 (targetm): Move declaration to the end of the file.
27981 * arm-cpu-cdata.h: Regenerated.
27983 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27985 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27986 call to target_mode_check describing the type of option passed.
27987 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27988 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27989 arm_parse_cpu_option_name to match parameters against list of
27991 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27992 arm_arch_core_flags data structure.
27993 * config/arm/arm-cpu_cdata.h: Regenerated.
27995 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27997 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27999 (arm_print_hint_for_cpu_option): Likewise.
28000 (arm_print_hint_for_arch_option): Likewise.
28001 (arm_parse_cpu_option_name): Likewise.
28002 (arm_parse_arch_option_name): Likewise.
28003 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
28004 of entries in the all_fpus list.
28005 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
28006 (arm_parse_cpu_option_name): Declare.
28007 (arm_parse_arch_option_name): Declare.
28008 (arm_parse_option_features): Declare.
28009 (arm_intialize_isa): Declare.
28010 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
28012 (gen_comm_data): ... here. Make definitions non-static.
28013 * config/arm/arm-cpu-data.h: Regenerated.
28014 * config/arm/arm-cpu-cdata.h: Regenerated.
28016 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28018 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
28019 (cpu_arch_extension): New structure.
28020 (cpu_arch_option, arch_option, cpu_option): New structures.
28021 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
28022 architecture types.
28023 (gen_data): Generate new format data tables.
28024 * config/arm/arm.c (cpu_tune): New structure.
28025 (cpu_option, processors): Delete.
28026 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
28027 (arm_print_hint_for_cpu_option): ... this and ...
28028 (arm_print_hint_for_arch_option): ... this.
28029 (arm_parse_arch_cpu_name): Delete. Replace with ...
28030 (arm_parse_cpu_option_name): ... this and ...
28031 (arm_parse_arch_option_name): ... this.
28032 (arm_unrecognized_feature): Change type of target parameter to
28034 (arm_parse_arch_cpu_features): Delete. Replace with ...
28035 (arm_parse_option_features): ... this.
28036 (arm_configure_build_target): Rework to use new configuration data
28038 (arm_print_tune_info): Rework for new configuration data tables.
28039 * config/arm/arm-cpu-data.h: Regenerated.
28040 * config/arm/arm-cpu.h: Regenerated.
28042 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28044 * Makefile.in (OBJS): Move sbitmap.o from here ...
28045 (OBJS-libcommon): ... to here.
28047 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28049 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
28050 (ISA_ALL_CRYPTO): New macro.
28051 (ISA_ALL_SIMD): New macro
28052 (ISA_ALL_FP): New macro.
28053 * config/arm/arm.c (fpu_bitlist): Update initializer.
28054 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
28056 (arm9e): Add fpu. Add option for nofp
28057 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
28058 (arm926ej-s, arm1026ej-s): Likewise.
28059 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
28060 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
28061 neon-fp16, neon-vfpv4, nofp and nosimd.
28062 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
28063 (cortex-a8): Add fpu. Add option for nofp.
28064 (cortex-a9): Add fpu. Add options for nosimd and nofp.
28065 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
28066 (cortex-r4f): Add fpu.
28067 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
28068 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
28070 (cortex-r8): Likewise.
28071 (cortex-m4): Add fpu. Add option for nofp.
28072 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
28073 (cortex-a17.cortex-a7): Likewise.
28074 (cortex-a32): Add fpu. Add options for crypto and nofp.
28075 (cortex-a35, cortex-a53): Likewise.
28076 (cortex-a57): Add fpu. Add option for crypto.
28077 (cortex-a72, cortex-a73): Likewise.
28078 (exynos-m1): Likewise.
28079 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
28080 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
28081 (cortex-m33): Add fpu. Add option for nofp.
28082 * config/arm/arm-cpu-cdata.h: Regenerated
28083 * config/arm/arm-cpu-data.h: Regenerated.
28085 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28087 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
28088 (armv5te, armv5tej): Likewise.
28089 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
28090 (armv7): Add options fp and vfpv3-d16.
28091 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
28092 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
28094 (armv7ve): Likewise.
28095 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
28096 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
28097 (armv8-a): Add nocrypto option.
28098 (armv8.1-a, armv8.2-a): Likewise.
28099 (armv8-m.main): add options fp, fp.dp and nofp.
28101 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28103 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
28105 (armv8-a+crc): Delete.
28106 (armv8.1-a): Add options simd, crypto and nofp.
28107 (armv8.2-a): Add options fp16, simd, crypto and nofp.
28108 (armv8.2-a+fp16): Delete.
28109 (armv8-m.main): Add option dsp.
28110 (armv8-m.main+dsp): Delete.
28111 (cortex-a8): Add fpu. Add nofp option.
28112 (cortex-a9): Add fpu. Add nofp and nosimd options.
28113 * config/arm/parsecpu.awk (gen_data): Generate option tables and
28114 link to main cpu and architecture data structures.
28115 (gen_comm_data): Only put isa attributes from the main architecture
28117 (option): New statement for architecture and CPU entries.
28118 * arm.c (struct cpu_option): New structure.
28119 (struct processors): Add entry for options.
28120 (arm_unrecognized_feature): New function.
28121 (arm_parse_arch_cpu_name): Ignore any characters after the first
28123 (arm_parse_arch_cpu_feature): New function.
28124 (arm_configure_build_target): Separate out any CPU and architecture
28125 features and parse separately. Don't error out if -mfpu=auto is
28126 used with only an architecture string.
28127 (arm_print_asm_arch_directives): New function.
28128 (arm_file_start): Call it.
28129 * config/arm/arm-cpu-cdata.h: Regenerated.
28130 * config/arm/arm-cpu-data.h: Likewise.
28131 * config/arm/arm-tables.opt: Likewise.
28133 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28135 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
28136 assembler when it is not -mfpu=auto.
28138 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28140 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
28141 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
28142 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
28143 (ASM_CPU_SPEC): Rewrite.
28144 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
28145 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
28146 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
28147 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
28148 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
28149 copied string is NUL-terminated. Also strip any characters prefixed
28151 (arm_rewrite_selected_arch): New function.
28152 (arm_rewrite_march): New function.
28154 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28156 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
28157 (x_arm_cpu_string, x_arm_tune_string): Likewise.
28158 (march, mcpu, mtune): Convert to string-based options.
28159 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
28160 (arm_parse_arch_cpu_name): New function.
28161 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28162 identify selected architecture or CPU.
28163 (arm_option_save): New function.
28164 (TARGET_OPTION_SAVE): Redefine.
28165 (arm_option_restore): Restore string options.
28166 (arm_option_print): Print string options.
28168 2017-06-16 Martin Sebor <msebor@redhat.com>
28170 PR tree-optimization/80933
28171 PR tree-optimization/80934
28172 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28173 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28174 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28175 (gimple_fold_builtin): Call them.
28177 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28179 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28180 as unlikely; update profile.
28182 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28184 * predict.c (force_edge_cold): Handle declaring edges impossible
28187 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28189 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28191 (try_unroll_loop_completely): Fix reporting.
28193 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28195 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28197 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28200 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28201 if given a non-constant argument for an intrinsic which requires a
28204 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28206 * profile.c (compare_freqs): New function.
28207 (branch_prob): Sort edge list.
28208 (find_spanning_tree): Assume that the list is priority sorted.
28210 2017-06-16 Richard Biener <rguenther@suse.de>
28212 PR tree-optimization/81090
28213 * passes.def (pass_record_bounds): Remove.
28214 * tree-pass.h (make_pass_record_bounds): Likewise.
28215 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28216 make_pass_record_bounds): Likewise.
28217 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28218 not free niter estimates at the beginning but at the end.
28219 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28221 2017-06-16 Richard Biener <rguenther@suse.de>
28223 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28224 initializer to workaround ICE in host GCC 4.8.
28226 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28228 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28230 (clone_inlined_nodes): Update.
28232 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28234 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28235 prefetch settings, and enable prefetching by default at -O3.
28237 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28239 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28240 Set flag_prefetch_loop_arrays according to tuning data.
28242 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28244 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28245 New tune structure.
28246 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28247 [Unrelated to main purpose of the patch] Place the pointer field last
28248 to enable type checking errors when tune structure are wrongly merged.
28249 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28250 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28251 (thunderx2t99_prefetch_tune): New tune constants.
28252 (tune_params *_tunings): Update all tunings (no functional change).
28253 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28254 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28255 from tunings structures.
28257 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28260 * ubsan.c (instrument_null): Add T argument, use it instead
28261 of computing it based on IS_LHS.
28262 (instrument_object_size): Likewise.
28263 (pass_ubsan::execute): Adjust instrument_null and
28264 instrument_object_size callers to pass gimple_get_lhs or
28265 gimple_assign_rhs1 result to it. Use instrument_null instead of
28266 calling get_base_address and instrument_mem_ref. Handle
28267 aggregate call arguments for object-size sanitization.
28269 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28271 PR tree-optimization/81089
28272 * tree-vrp.c (is_masked_range_test): Validate operands of
28275 2017-06-15 Martin Sebor <msebor@redhat.com>
28278 * dumpfile.c (dump_register): Avoid calling memset to initialize
28279 a class with a default ctor.
28280 * gcc.c (struct compiler): Remove const qualification.
28281 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28282 * hash-table.h: Ditto.
28283 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28285 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28286 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28288 * params.h (struct param_info): Make struct members non-const.
28289 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28290 with default initialization.
28291 * vec.h (vec_copy_construct, vec_default_construct): New helper
28293 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28294 with vec_copy_construct.
28295 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28296 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28297 * doc/invoke.texi (-Wclass-memaccess): Document.
28299 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28301 * emit-rtl.h (is_leaf): Update comment about local
28302 register allocator.
28304 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28307 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28308 for a variable to have a section before checking if the section has a
28310 Set section to.persistent if persistent attribute is set.
28311 Warn if .persistent attribute is used on an automatic variable.
28313 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28315 PR rtl-optimization/80474
28316 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28318 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28320 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28323 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28325 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28326 (casesi): Emit bounds checking as RTL.
28327 (casesi_internal_mips16_<mode>): Remove bounds checking.
28329 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28331 * config/xtensa/xtensa.c (xtensa_option_override): Append
28332 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28333 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28334 xtensa_doloop_hooks): Define unconditionally.
28335 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28337 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28338 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28339 for it in xtensa_option_override.
28340 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28341 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28343 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28345 * doc/cppopts.texi: Document '-' special value to -MF.
28347 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28349 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28350 (cortex_a53_fconst): Likewise.
28351 (cortex_a53_fpmul): Likewise.
28352 (cortex_a53_f_load_64): Likewise.
28353 (cortex_a53_f_load_many): Likewise.
28354 (cortex_a53_advsimd_alu): Likewise.
28355 (cortex_a53_advsimd_alu_q): Likewise.
28356 (cortex_a53_advsimd_mul): Likewise.
28357 (cortex_a53_advsimd_mul_q): Likewise.
28358 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28359 Add missing fmul, r2f_cvt and fconst cases.
28361 2017-06-14 Richard Biener <rguenther@suse.de>
28363 PR middle-end/81088
28364 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28366 (fold_binary_loc): When associating do not treat pre-existing
28367 TREE_OVERFLOW on literal constants as a reason to allow
28368 TREE_OVERFLOW on associated literal constants.
28370 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28372 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28373 (MASK_FEATURES): New macro.
28374 * config/sparc/sparc.c (sparc_option_override): Remove the special
28375 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28377 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28379 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28380 a division of 0 if non-call exceptions are enabled.
28382 2017-06-14 Andrew Pinski <apinski@cavium.com>
28383 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28386 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28387 Improve vector initialization code gen for only variable case.
28389 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28391 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28393 2017-06-14 Richard Biener <rguenther@suse.de>
28395 PR tree-optimization/81083
28396 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28399 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28401 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28402 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28403 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28404 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28405 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28406 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28408 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28410 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28411 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28413 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28415 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28417 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28419 * config/rs6000/t-rtems: Don't handle SPE.
28421 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28423 * config/rs6000/t-linux: Don't handle SPE.
28425 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28427 * config/rs6000/eabispe.h: Delete file.
28429 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28431 * config/rs6000/t-spe: Delete file.
28433 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28435 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28436 (rs6000_legitimate_offset_address_p): Return false for anything in
28437 V2SImode or V2SFmode.
28439 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28441 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28442 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28443 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28445 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28446 (rs6000_legitimate_offset_address_p): Ditto.
28447 (rs6000_emit_move): Ditto.
28448 (rs6000_init_builtins): Remove V4HI_type_node.
28450 2017-06-13 Martin Liska <mliska@suse.cz>
28453 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28454 (gate_asan): Likewise.
28455 * asan.h (asan_no_sanitize_address_p): Remove the function.
28456 (sanitize_flags_p): New function.
28457 * builtins.def: Fix coding style.
28458 * common.opt: Use renamed enum value.
28459 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28460 * doc/extend.texi: Document no_sanitize attribute.
28461 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28462 to SANITIZE_UNDEFINED_NONDEFAULT.
28463 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28464 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28465 Use sanitize_flags_p.
28466 * gimplify.c (gimplify_function_tree): Likewise.
28467 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28468 * opts.c (parse_no_sanitize_attribute): New function.
28469 (common_handle_option): Use renamed enum value.
28470 * opts.h (parse_no_sanitize_attribute): Declare.
28471 * tree.c (sanitize_flags_p): New function.
28472 * tree.h: Declared here.
28473 * tsan.c: Use sanitize_flags_p.
28474 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28475 (instrument_mem_ref): Likewise.
28476 (instrument_bool_enum_load): Likewise.
28477 (do_ubsan_in_current_function): Remove the function.
28478 (pass_ubsan::execute): Use sanitize_flags_p.
28479 * ubsan.h: Remove do_ubsan_in_current_function
28480 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28481 (dump_function_to_file): Use it here.
28483 2017-06-13 Martin Jambor <mjambor@suse.cz>
28485 PR tree-optimization/80803
28486 PR tree-optimization/81063
28487 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28488 (propagate_subaccesses_across_link): Enqueue subtree whenever
28489 necessary instead of relying on the caller.
28491 2017-06-13 Martin Jambor <mjambor@suse.cz>
28493 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28494 that have a first_link.
28495 (sort_and_splice_var_accesses): Do not check first_link before
28497 (subtree_mark_written_and_enqueue): Likewise.
28498 (propagate_all_subaccesses): Likewise and do not stop at first
28499 parent with a first_link.
28501 2017-06-13 Martin Jambor <mjambor@suse.cz>
28503 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28506 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28508 * match.pd: New pattern.
28510 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28512 * tree-vrp.c (is_masked_range_test): New function.
28513 (register_edge_assert_for): Determine ranges for
28516 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28518 PR tree-optimization/67328
28519 * fold-const.c (maskable_range_p): New function.
28520 (build_range_check): Generate bittests if possible.
28522 2017-06-13 Martin Liska <mliska@suse.cz>
28524 * gimple-pretty-print.c (dump_probability): Add new argument.
28525 (dump_edge_probability): Dump both probability and count.
28526 (dump_gimple_label): Likewise.
28527 (dump_gimple_bb_header): Likewise.
28529 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28532 * config/avr/avr-devices.c: Fix indentation.
28533 * config/avr/gen-avr-mmcu-specs.c: Dito.
28535 2017-06-13 Richard Biener <rguenther@suse.de>
28537 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28538 instead get vector type from stmt_info.
28539 (vectorizable_reduction): Adjust. Remove dead code.
28541 2017-06-13 Richard Biener <rguenther@suse.de>
28543 PR middle-end/81065
28544 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28545 case of C * (x * C2 + C3).
28546 (fold_addr_of_array_ref_difference): Properly fold index difference.
28548 2017-06-12 David S. Miller <davem@davemloft.net>
28551 * config/sparc/sparc.md (return expander): Emit frame blockage if
28552 function uses alloca.
28554 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28556 * combine.c (make_field_assignment): Check len rather than the mode
28557 precision when calling force_to_mode.
28559 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28561 Support multilibs and devices that see flash in RAM address range.
28564 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28565 (avr_mcu_t) <flash_pm_offset>: New field.
28566 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28567 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28568 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28569 (AVR_TINY_PM_OFFSET): Remove macro.
28570 * config/avr/avr.opt (-mshort-calls): New option.
28571 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28572 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28573 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28574 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28575 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28576 instead of avr_arch->have_jmp_call.
28577 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28578 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28579 avr_arch->flash_pm_offset to define.
28580 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28581 new field flash_pm_offset. Add entry for avrxmega3.
28582 (avr_texinfo): Add entry for avrxmega3.
28583 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28584 attiny212, attiny214,
28585 attiny412, attiny414, attiny416, attiny417,
28586 attiny814, attiny816, attiny817,
28587 attiny1614, attiny1616, attiny1617,
28588 attiny3214, attiny3216, attiny3217.
28589 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28590 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28591 (avr_print_operand_address) [AVR_TINY]: Same.
28592 (avr_asm_init_sections) <readonly_data_section>: Only patch
28593 callback if avr_arch->flash_pm_offset = 0.
28594 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28595 for rodata if avr_arch->flash_pm_offset != 0.
28596 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28597 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28598 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28599 (m_options): Append opt_rcall.
28600 (m_dirnames): Append dir_rcall.
28601 * config/avr/t-multilib: Regenerate.
28603 * configure.ac [target=avr]: Check whether avrxmega3 default
28604 linker description file works as needed.
28605 * configure: Regenerate.
28606 * doc/avr-mmcu.texi: Regenerate.
28607 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28608 <__AVR_ARCH__>: Document avrxmega3 and 103.
28609 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28610 <__AVR_SHORT_CALLS__>: Document it.
28611 <__AVR_PM_BASE_ADDRESS__>: Document it.
28612 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28613 (AVR Variable Attributes) <progmem>: Document this is
28614 not needed for avrxmega3.
28615 (AVR Named Address Spaces) <__flash>: Dito.
28617 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28619 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28621 2017-06-12 Doug Rupp <rupp@adacore.com>
28623 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28624 Append vxworks-stdint.h to the tm_file list.
28625 * config/vxworks-stdint.h: New file.
28627 2017-06-12 Martin Liska <mliska@suse.cz>
28629 PR tree-optimization/81041
28630 * tree-profile.c (gimple_gen_ic_func_profiler):
28631 Create an extra BB in profile-generate
28632 (gimple_gen_time_profiler): Likewise.
28634 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28636 PR tree-optimization/81003
28637 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28638 (update_range_test): Use it instead of force_gimple_operand_gsi.
28640 2017-06-12 Richard Biener <rguenther@suse.de>
28642 PR tree-optimization/81053
28643 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28644 with backedge value not defined in loop. Simplify def stmt
28647 2017-06-11 Tom de Vries <tom@codesourcery.com>
28650 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28652 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28653 nvptx_cannot_force_const_mem.
28655 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28657 * opts.c (finish_options): Move test for flag_split_stack after
28658 it has been initialized.
28660 2017-06-11 Jason Merrill <jason@redhat.com>
28662 * tree.h (id_equal): New.
28663 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28664 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28665 instead of strcmp of IDENTIFIER_POINTER.
28667 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28669 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28670 (mark_all_inlined_calls_cdtor): Fix formating.
28671 (inline_transform): Rescale profile before inlining.
28673 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28675 * cgraph.h (cgraph_edge::clone): Update prototype.
28676 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28677 (cgraph_node::create_clone): Update.
28678 (cgraph_node::create_version_clone): Update.
28679 * tree-inline.c (copy_bb): Update.
28680 (expand_call_inline): Update.
28682 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28684 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28685 factored out from ...
28686 (rs6000_emit_prologue): ... here.
28688 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28690 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28691 factored out from ...
28692 (rs6000_emit_prologue): ... here.
28694 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28696 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28698 (handle_missing_profiles): Fix computation of tp_first_run.
28699 (counts_to_freqs): Do not touch freqs when count is 0.
28701 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28703 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28706 2017-06-10 Tom de Vries <tom@codesourcery.com>
28708 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28709 attributes): Document signal effective target.
28711 2017-06-10 Tom de Vries <tom@codesourcery.com>
28713 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28714 Document effective target stack_size.
28716 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28718 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28719 to the edit_context if they can be auto-applied.
28721 2017-06-9 Ian Lance Taylor <iant@golang.org>
28723 * opts.c (finish_options): If -fsplit-stack, disable implicit
28724 -forder-blocks-and-partition.
28725 * doc/invoke.texi (Optimize Options): Document that when using
28726 -fsplit-stack -forder-blocks-and-partition is not implicitly
28729 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28731 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28732 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28733 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28734 * builtins.def (abort, trap, unreachable): Declare cold.
28735 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28736 * tree-core.h (ECF_COLD): New.
28737 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28738 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28740 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28742 * predict.c (unlikely_executed_stmt_p): Cleanup.
28744 2017-06-09 Richard Biener <rguenther@suse.de>
28746 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28747 model if the ref is always written to.
28749 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28751 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28753 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28755 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28758 2017-06-09 Tom de Vries <tom@codesourcery.com>
28761 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28762 "target cannot support label values" when encountering LABEL_REF.
28764 2017-06-09 Martin Liska <mliska@suse.cz>
28766 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28767 (gimple_gen_ic_func_profiler): Emit direct comparison
28768 of __gcov_indirect_call_callee with NULL.
28769 (gimple_gen_time_profiler): Change probability from
28770 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28772 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28774 * profile.c (edge_gcov_counts): Turn to pointer.
28775 (compute_branch_probabilities, compute_branch_probabilities): Update.
28776 (branch_prob): Do not clear edge_gcov_count.
28777 * profile.h (edge_gcov_counts): Turn to pointer.
28778 (edge_gcov_count): Update.
28780 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28782 * gimple.h (gimple_check_failed): Mark cold.
28784 2017-06-09 Richard Biener <rguenther@suse.de>
28786 PR tree-optimization/66623
28787 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28788 refactor check_reduction into two parts, properly computing
28789 whether we have to check reduction validity for outer loop
28792 2017-06-09 Richard Biener <rguenther@suse.de>
28794 PR tree-optimization/79483
28795 * graphite-scop-detection.c (order): New global.
28796 (get_order): Compute bb to order mapping that satisfies code
28797 generation constraints.
28798 (cmp_pbbs): New helper.
28799 (build_scops): Start domwalk at entry block, sort generated
28802 2017-06-09 Richard Biener <rguenther@suse.de>
28804 PR middle-end/81007
28805 * ipa-polymorphic-call.c
28806 (ipa_polymorphic_call_context::restrict_to_inner_class):
28807 Skip FIELD_DECLs with error_mark_node type.
28808 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28811 2017-06-09 Martin Liska <mliska@suse.cz>
28813 * predict.c (struct branch_predictor): New struct.
28814 (test_prediction_value_range): New test.
28815 (predict_c_tests): New function.
28816 * selftest-run-tests.c (selftest::run_tests): Run the function.
28817 * selftest.h: Declare new tests.
28819 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28822 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28823 gen_add3_insn did not fail.
28824 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28825 r0, construct that number in a temporary reg and add that reg to r0.
28826 If asked to put the result in r0 as well, fail.
28828 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28830 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28831 for early expansion of vec_eqv.
28833 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28835 PR middle-end/81005
28836 * ubsan.c (instrument_null): Avoid pointless code temporary.
28837 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28839 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28843 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28846 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28849 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28851 * predict.c (unlikely_executed_edge_p): Move ahead.
28852 (probably_never_executed_edge_p): Use it.
28854 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28856 PR middle-end/79988
28857 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28858 gimple_call_builtin_p call.
28860 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28862 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28863 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28864 rtl_check_failed_type2, rtl_check_failed_code1,
28865 rtl_check_failed_code2, rtl_check_failed_code_mode,
28866 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28867 rtvec_check_failed_bounds, rtl_check_failed_flag,
28868 _fatal_insn_not_found, _fatal_insn): Likewise.
28869 * tree.h (tree_contains_struct_check_failed,
28870 tree_check_failed, tree_not_check_failed,
28871 tree_class_check_failed, tree_range_check_failed,
28872 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28873 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28874 tree_operand_check_failed, omp_clause_check_failed,
28875 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28878 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28880 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28881 flag_branch_probabilities.
28882 * ipa-inline.c (edge_badness): Likewise.
28883 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28884 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28885 * predict.c (maybe_hot_frequency_p): Likewise.
28886 (probably_never_executed): Likewise.
28887 * sched-ebb.c (schedule_ebbs): Likewise.
28888 * sched-rgn.c (find_single_block_region): Likewise.
28889 * tracer.c (tail_duplicate): Likewise.
28891 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28893 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28894 longer requires x_flag_profile_use.
28896 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28898 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28899 instead of flag_reorder_blocks_and_partition.
28900 * dbxout.c (dbxout_function_end): Likewise.
28901 * dwarf2out.c (gen_subprogram_die): Likewise.
28902 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28903 * hw-doloop.c (reorg_loops): Likewise.
28904 * varasm.c (assemble_start_function,
28905 assemble_end_function): Likewise.
28906 (decide_function_section): Do not check for
28907 flag_reorder_blocks_and_partition.
28909 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28911 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28913 (chkp_get_hard_register_fake_addr_expr): Ditto.
28914 (chkp_build_addr_expr): Add check for hard reg case.
28915 (chkp_parse_array_and_component_ref): Ditto.
28916 (chkp_find_bounds_1): Ditto.
28917 (chkp_process_stmt): Don't generate bounds store for
28920 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28922 * predict.c (maybe_hot_bb_p): Do not check profile status.
28923 (maybe_hot_edge_p): Likewise.
28924 (probably_never_executed): Check for zero counts even if profile
28926 (unlikely_executed_edge_p): New function.
28927 (unlikely_executed_stmt_p): New function.
28928 (unlikely_executed_bb_p): New function.
28929 (set_even_probabilities): Use unlikely predicates.
28930 (combine_predictions_for_bb): Likewise.
28931 (predict_paths_for_bb): Likewise.
28932 (predict_paths_leading_to_edge): Likewise.
28933 (determine_unlikely_bbs): New function.
28934 (estimate_bb_frequencies): Use it.
28935 (compute_function_frequency): Use zero counts even if profile is
28937 * profile-count.h: Fix typo.
28939 2017-08-08 Julia Koval <julia.koval@intel.com>
28941 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28942 _mm512_mask_cvtsepi16_storeu_epi8,
28943 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28944 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28945 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28946 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28947 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28948 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28949 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28950 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28951 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28952 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28953 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28954 __builtin_ia32_pmovuswb256mem_mask,
28955 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28956 __builtin_ia32_pmovwb512mem_mask): New builtins.
28958 2017-08-08 Julia Koval <julia.koval@intel.com>
28960 PR target/73350,80862
28961 * config/i386/subst.md (round): Fix round pattern.
28962 * config/i386/i386.c (ix86_erase_embedded_rounding):
28963 Fix erasing rounding for the fixed pattern.
28965 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28967 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28969 2017-06-08 Martin Liska <mliska@suse.cz>
28971 PR gcov-profile/80911
28972 * gcov.c (block_info::block_info): New constructor.
28974 2017-06-07 Carl Love <cel@us.ibm.com>
28976 * config/rs6000/rs6000-c: The return type of the following
28977 built-in functions was implemented as int not long long. Fix sign
28978 of return value for the unsigned version of vec_mulo and vec_mule.
28979 vector unsigned long long vec_bperm (vector unsigned long long,
28980 vector unsigned char)
28981 vector signed long long vec_mule (vector signed int,
28983 vector unsigned long long vec_mule (vector unsigned int,
28984 vector unsigned int)
28985 vector signed long long vec_mulo (vector signed int,
28987 vector unsigned long long vec_mulo (vector unsigned int,
28988 vector unsigned int)
28989 * doc/extend.texi: Fix the documentation for the built-in
28992 2017-06-07 Carl Love <cel@us.ibm.com>
28995 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28998 2017-06-07 Carl Love <cel@us.ibm.com>
29000 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
29001 support, Generate doublehv for signed int/float for BE case only.
29003 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
29005 * doc/invoke.texi (mcx16): Rewrite.
29007 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29009 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
29010 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
29011 *mov<mode>_softfloat, and an anonymous splitter): Use
29012 nonimmediate_operand instead of rs6000_nonimmediate_operand.
29014 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29016 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
29018 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29019 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
29020 (rs6000_debug_reg_global): Adjust.
29021 (rs6000_init_hard_regno_mode_ok): Adjust.
29022 (rs6000_dbx_register_number): Adjust.
29023 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
29024 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29025 Remove SPE_ACC and SPEFSCR.
29026 (REG_ALLOC_ORDER): Ditto.
29027 (FRAME_POINTER_REGNUM): Change to 111.
29028 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
29029 (REG_CLASS_NAMES): Ditto.
29030 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
29031 (REGISTER_NAMES): Ditto.
29032 (ADDITIONAL_REG_NAMES): Ditto.
29033 (rs6000_reg_names): Ditto.
29034 * config/rs6000/rs6000.md: Renumber some register number
29037 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29039 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
29041 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29042 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
29044 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
29045 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29046 Delete the SPE high registers.
29047 (REG_ALLOC_ORDER): Ditto.
29048 (enum reg_class): Remove SPE_HIGH_REGS.
29049 (REG_CLASS_NAMES): Ditto.
29050 (REG_CLASS_CONTENTS): Delete the SPE high registers.
29051 (REGISTER_NAMES): Ditto.
29052 (rs6000_reg_names): Ditto.
29053 * doc/tm.texi.in: Remove SPE as example.
29054 * doc/tm.texi: Regenerate.
29056 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29058 * config/rs6000/8540.md (ppc8540_brinc): Delete.
29059 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
29060 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
29061 * config/rs6000/rs6000.md (type): Remove "brinc".
29063 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29065 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
29066 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
29067 * config/rs6000/linuxspe.h: Delete file.
29068 * config/rs6000/rs6000.md: Don't include spe.md.
29069 * config/rs6000/spe.h: Delete file.
29070 * config/rs6000/spe.md: Delete file.
29071 * config/rs6000/t-rs6000: Remove spe.md.
29073 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29075 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
29076 (reg_or_none500mem_operand): Delete.
29077 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
29078 instead of reg_or_none500mem_operand.
29080 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29082 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29083 handling of SPE flags.
29084 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
29086 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29088 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
29090 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
29091 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
29092 paired_divv2sf3): Similar.
29093 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
29094 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
29095 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
29097 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
29098 Rename the paired_* instruction patterns.
29099 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
29101 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
29102 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
29103 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
29104 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
29105 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
29106 PAIRED_VECTOR_MODE.
29107 (struct machine_function): Delete field spe_insn_chain_scanned_p.
29108 (spe_func_has_64bit_regs_p): Delete.
29109 (spe_expand_predicate_builtin): Delete.
29110 (spe_expand_evsel_builtin): Delete.
29111 (TARGET_DWARF_REGISTER_SPAN): Do not define.
29112 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
29113 (invalid_e500_subreg): Delete.
29114 (rs6000_legitimize_address): Always force_reg op2 as well, for
29115 paired single memory accesses.
29116 (rs6000_member_type_forces_blk): Delete.
29117 (rs6000_spe_function_arg): Delete.
29118 (rs6000_expand_unop_builtin): Delete SPE handling.
29119 (rs6000_expand_binop_builtin): Ditto.
29120 (spe_expand_stv_builtin): Delete.
29121 (bdesc_2arg_spe): Delete.
29122 (spe_expand_builtin): Delete.
29123 (spe_expand_predicate_builtin): Delete.
29124 (spe_expand_evsel_builtin): Delete.
29125 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
29126 (spe_init_builtins): Delete.
29127 (spe_func_has_64bit_regs_p): Delete.
29128 (savres_routine_name): Delete "info" parameter. Adjust callers.
29129 (rs6000_emit_stack_reset): Ditto.
29130 (rs6000_dwarf_register_span): Delete.
29131 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
29132 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
29133 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
29135 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
29137 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
29138 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
29139 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
29140 mulv2sf3, divv2sf3): Delete expanders.
29142 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29144 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
29146 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29148 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
29149 * config/rs6000/rs6000.c: Ditto.
29151 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29153 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
29154 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
29155 comparison_operator.
29157 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29159 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
29160 * config/rs6000/rs6000.opt: Ditto.
29161 * config/rs6000/t-rtems: Ditto.
29163 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29165 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29166 TARGET_E500_SINGLE by 0, simplify.
29167 * config/rs6000/rs6000.c: Ditto.
29168 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29169 (spe_build_register_parallel): Delete.
29170 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29171 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29172 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29173 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29174 (E500_CONVERT): Delete.
29175 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29177 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29179 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29180 * config/rs6000/dfp.md: Ditto.
29181 (negdd2, *negdd2_fpr): Merge.
29182 (absdd2, *absdd2_fpr): Merge.
29183 (negtd2, *negtd2_fpr): Merge.
29184 (abstd2, *abstd2_fpr): Merge.
29185 * config/rs6000/e500.h: Delete file.
29186 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29187 TARGET_FPRS by 1 and simplify.
29188 * config/rs6000/rs6000-c.c: Ditto.
29189 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29190 TARGET_DF_SPE by 0.
29191 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29193 * config/rs6000/rs6000.md: Ditto.
29194 (floatdidf2, *floatdidf2_fpr): Merge.
29195 (move_from_CR_gt_bit): Delete.
29196 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29197 (E500_CR_IOR_COMPARE): Delete.
29198 (All patterns that require !TARGET_FPRS): Delete.
29199 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29201 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29203 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29205 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29207 * graphds.c (add_edge): Intitialize edge's attached data.
29208 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29209 pointer parameter. Call pointed function on each edge during
29210 graph traversing. Skip traversing the edge when the function
29212 (graphds_dfs, graphds_scc): Ditto.
29213 (for_each_edge): New parameter. Pass the new parameter to callback
29215 * graphds.h (skip_edge_callback): New function pointer type.
29216 (graphds_dfs, graphds_scc): New function pointer parameter.
29217 (graphds_edge_callback, for_each_edge): New parameter.
29219 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29221 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29222 out code checking if runtime alias check is possible to below ...
29223 Call the new function.
29224 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29225 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29227 2017-06-07 Marek Polacek <polacek@redhat.com>
29230 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29231 TYPE_OVERFLOW_WRAPS check.
29233 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29235 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29236 if versioning is required.
29237 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29238 peeling with the check for versioning.
29240 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29242 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29243 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29244 Set true to new parameter if new ssa variable is defined.
29245 (vect_gen_vector_loop_niters): Refactor. Set range information
29246 for the new vector loop bound variable.
29247 (vect_do_peeling): Ditto.
29249 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29251 * tree-affine.c (ssa.h): Include header file.
29252 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29253 has wrapping overflow behavior.
29255 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29257 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29259 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29261 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29262 (tree_to_aff_combination): ... here.
29264 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29266 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29267 reg_pressure model function.
29268 (ivopts_global_cost_for_size): Delete.
29269 (determine_set_costs, iv_ca_recount_cost): Call new model function
29270 ivopts_estimate_reg_pressure.
29272 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29274 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29275 expensive than udiv. Remove floating point cases from mod.
29277 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29279 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29280 Increase idiv cost.
29282 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29284 * config/aarch64/aarch64.md
29285 (copysignsf3): Fix mask generation.
29287 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29289 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29291 (class dump_manager): Add register_dumps method.
29292 * dumpfile.c: Include langhooks.h.
29293 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29294 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29295 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29296 (dump_manager::dump_register): Start with 512 entries instead of 32.
29297 (dump_manager::register_dumps): New method.
29298 * toplev.c (general_init): Instead of invoking register_dumps
29299 langhook, invoke register_dumps method on the dump manager.
29300 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29303 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29305 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29306 Say that # requires an associated define_split to exist, and that
29307 the define_split must be suitable for use after register allocation.
29309 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29311 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29312 (compute_outgoing_frequencies): Also initialize zero counts.
29313 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29314 around loops; preserve more of profile when nothing changes.
29316 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29318 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29320 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29321 * config/arm/arm-cpu-cdata.h: Regenerate.
29322 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29323 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29324 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29325 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29326 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29328 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29330 * config/arm/t-rmprofile: Likewise.
29331 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29333 2017-06-06 David S. Miller <davem@davemloft.net>
29336 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29337 blockage if function uses alloca.
29339 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29341 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29342 New "uid" fields to hold pretty-print IDs of group and ref.
29343 Memory references are now identified as <group_id>:<ref_id>
29344 instead of using [random] addresses.
29345 (dump_mem_details): Simplify, no functional change.
29346 (dump_mem_ref): Simplify and make output more concise.
29347 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29348 (find_or_create_group): Initialize group uid.
29349 (record_ref): Initialize ref uid. Improve debug output.
29350 (prune_group_by_reuse, should_issue_prefetch_p,)
29351 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29352 (mark_nontemporal_store, determine_loop_nest_reuse):
29353 Improve debug output.
29355 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29357 * dbgcnt.def (prefetch): New debug counter.
29358 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29359 (schedule_prefetches): Stop issueing prefetches if debug counter
29362 2017-06-06 Tom de Vries <tom@codesourcery.com>
29364 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29365 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29367 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29369 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29370 Use aarch64_reg_or_zero predicate for operand 4.
29371 (aarch64_compare_and_swap<mode> define_insn_and_split):
29372 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29373 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29375 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29377 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29378 (arm_compute_save_core_reg_mask): This.
29379 (thumb1_compute_save_reg_mask): Rename into ...
29380 (thumb1_compute_save_core_reg_mask): This.
29381 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29382 (arm_compute_frame_layout): Likewise.
29384 2017-06-06 Richard Biener <rguenther@suse.de>
29386 PR tree-optimization/80974
29387 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29388 keep or clear leaders SSA info.
29390 2017-06-06 Tom de Vries <tom@codesourcery.com>
29392 * config/nvptx/nvptx.c (split_mode_p): New function.
29393 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29395 2017-06-06 Tom de Vries <tom@codesourcery.com>
29397 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29399 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29402 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29405 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29407 * shrink-wrap.c (handle_simple_exit): Update profile.
29408 (try_shrink_wrapping): Upate profile.
29410 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29412 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29413 (tree_guess_outgoing_edge_probabilities): New.
29414 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29415 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29417 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29419 * ipa-split.c (split_function): Initialize return bb profile.
29421 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29423 * profile.c (compute_branch_probabilities): Also initialize
29424 EXIT_BLOCK profile.
29426 2017-06-06 Richard Biener <rguenther@suse.de>
29428 PR tree-optimization/80928
29429 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29430 (vect_analyze_loop_operations): Properly guard analysis for
29432 (vect_transform_loop): Likewise.
29433 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29434 (vect_model_induction_cost): Do not cost for pure SLP.
29435 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29436 of induction in inner loop vectorization.
29437 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29438 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29439 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29441 (vect_analyze_slp_cost_1): Cost induction.
29442 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29443 (vect_get_slp_vect_defs): Likewise.
29444 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29445 (vect_transform_stmt): Handle SLP reductions.
29446 * tree-vectorizer.h (vectorizable_induction): Adjust.
29448 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29450 * config/rs6000/rs6000.c (make_resolver_func): Update
29451 init_lowered_empty_function call.
29453 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29455 * doc/invoke.texi: Document the -fprofile-abs-path option.
29456 * common.opt (fprofile-abs-path): New option.
29457 * gcov-io.h (gcov_write_filename): Declare.
29458 * gcov-io.c (gcov_write_filename): New function.
29459 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29460 * profile.c (output_location): Likewise.
29462 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29464 * shring-wrap.c: Revert accidental commit.
29466 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29468 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29470 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29472 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29474 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29475 profile in callgraph edge.
29476 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29477 (apply_scale): Likewise.
29478 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29479 Also scale profile when inlining function with zero profile.
29480 (initialize_cfun): Update exit block profile even when it is zero.
29481 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29482 when profile is read.
29484 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29486 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29487 (CLONE_*): New constants to define the processors we can generate
29488 code for with the target_clone attribute.
29489 (rs6000_clone_map): New array to identify which clone processors
29490 the current program is running on.
29491 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29492 target_clone attribute.
29493 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29494 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29495 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29496 (cpu_expand_builtin): Add support for target_clone attribute.
29497 (rs6000_valid_attribute_p): Allow "default" attribute.
29498 (get_decl_name): New debug function to simplify printing the
29499 current function name in debugging statements.
29500 (rs6000_clone_priority): New functions to support the target_clone
29501 attribute, and be able to generate code to switch between ISA 2.05
29502 through ISA 3.0 (power6 through power9).
29503 (rs6000_compare_version_priority): Likewise.
29504 (rs6000_get_function_versions_dispatcher): Likewise.
29505 (make_resolver_func): Likewise.
29506 (add_condition_to_bb): Likewise.
29507 (dispatch_function_versions): Likewise.
29508 (rs6000_generate_version_dispatcher_body): Likewise.
29509 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29510 (fusion_gpr_load_p): Fix a spacing issue.
29511 * doc/extend.texi (Common Function Attributes): Document that the
29512 PowerPC supports the target_clone attribute.
29514 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29516 * config/arm/arm.h: explain F symbol found in description of ARM
29517 register allocation in its legend.
29519 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29521 * config/mips/frame-header-opt.c: Include profile-count.h.
29522 * config/riscv/riscv.c: Include profile-count.h
29524 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29526 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29528 (sm_set_flag_if_changed): Add bbs field.
29529 (execute_sm_if_changed_flag_set): Pass BBS.
29530 (execute_sm): Update.
29532 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29534 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29537 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29539 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29540 (peephole2): New peephole2 to emit the above.
29541 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29543 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29545 * config/aarch64/aarch64.c (define_peephole2 above
29546 *sub_<shift>_<mode>): New peephole.
29548 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29550 * config/i386/i386.c (make_resolver_func): Update.
29551 * Makefile.in: Add profile-count.h and profile-count.o
29552 * auto-profile.c (afdo_indirect_call): Update to new API.
29553 (afdo_set_bb_count): Update.
29554 (afdo_propagate_edge): Update.
29555 (afdo_propagate_circuit): Update.
29556 (afdo_calculate_branch_prob): Update.
29557 (afdo_annotate_cfg): Update.
29558 * basic-block.h: Include profile-count.h
29559 (struct edge_def): Turn count to profile_count.
29560 (struct basic_block_def): Likewie.
29561 (REG_BR_PROB_BASE): Move to profile-count.h
29562 (RDIV): Move to profile-count.h
29563 * bb-reorder.c (max_entry_count): Turn to profile_count.
29564 (find_traces): Update.
29565 (rotate_loop):Update.
29566 (connect_traces):Update.
29567 (sanitize_hot_paths):Update.
29568 * bt-load.c (migrate_btr_defs): Update.
29569 * cfg.c (RDIV): Remove.
29570 (init_flow): Use alloc_block.
29571 (alloc_block): Uninitialize count.
29572 (unchecked_make_edge): Uninitialize count.
29573 (check_bb_profile): Update.
29574 (dump_edge_info): Update.
29575 (dump_bb_info): Update.
29576 (update_bb_profile_for_threading): Update.
29577 (scale_bbs_frequencies_int): Update.
29578 (scale_bbs_frequencies_gcov_type): Update.
29579 (scale_bbs_frequencies_profile_count): New.
29580 * cfg.h (update_bb_profile_for_threading): Update.
29581 (scale_bbs_frequencies_profile_count): Declare.
29582 * cfgbuild.c (compute_outgoing_frequencies): Update.
29583 (find_many_sub_basic_blocks): Update.
29584 * cfgcleanup.c (try_forward_edges): Update.
29585 (try_crossjump_to_edge): Update.
29586 * cfgexpand.c (expand_gimple_tailcall): Update.
29587 (construct_exit_block): Update.
29588 * cfghooks.c (verify_flow_info): Update.
29589 (dump_bb_for_graph): Update.
29590 (split_edge): Update.
29591 (make_forwarder_block): Update.
29592 (duplicate_block): Update.
29593 (account_profile_record): Update.
29594 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29595 (get_estimated_loop_iterations): Update.
29596 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29597 (single_likely_exit): Update.
29598 * cfgloopmanip.c (scale_loop_profile): Update.
29600 (set_zero_probability): Update.
29601 (lv_adjust_loop_entry_edge): Update.
29602 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29603 (purge_dead_edges): Update.
29604 (rtl_account_profile_record): Update.
29605 * cgraph.c (cgraph_node::create): Uninitialize count.
29606 (symbol_table::create_edge): Uninitialize count.
29607 (cgraph_update_edges_for_call_stmt_node): Update.
29608 (cgraph_edge::dump_edge_flags): Update.
29609 (cgraph_node::dump): Update.
29610 (cgraph_edge::maybe_hot_p): Update.
29611 * cgraph.h: Include profile-count.h
29612 (create_clone), create_edge, create_indirect_edge): Update.
29613 (cgraph_node): Turn count to profile_count.
29614 (cgraph_edge0: Likewise.
29615 (make_speculative, clone): Update.
29616 (create_edge): Update.
29617 (init_lowered_empty_function): Update.
29618 * cgraphclones.c (cgraph_edge::clone): Update.
29619 (duplicate_thunk_for_node): Update.
29620 (cgraph_node::create_clone): Update.
29621 * cgraphunit.c (cgraph_node::analyze): Update.
29622 (cgraph_node::expand_thunk): Update.
29623 * final.c (dump_basic_block_info): Update.
29624 * gimple-streamer-in.c (input_bb): Update.
29625 * gimple-streamer-out.c (output_bb): Update.
29626 * graphite.c (print_global_statistics): Update.
29627 (print_graphite_scop_statistics): Update.
29628 * hsa-brig.c: Include basic-block.h.
29629 * hsa-dump.c: Include basic-block.h.
29630 * hsa-gen.c (T sum_slice): Update.
29631 (convert_switch_statements):Update.
29632 * hsa-regalloc.c: Include basic-block.h.
29633 * ipa-chkp.c (chkp_produce_thunks): Update.
29634 * ipa-cp.c (struct caller_statistics): Update.
29635 (init_caller_stats): Update.
29636 (gather_caller_stats): Update.
29637 (ipcp_cloning_candidate_p): Update.
29638 (good_cloning_opportunity_p): Update.
29639 (get_info_about_necessary_edges): Update.
29640 (dump_profile_updates): Update.
29641 (update_profiling_info): Update.
29642 (update_specialized_profile): Update.
29643 (perhaps_add_new_callers): Update.
29644 (decide_about_value): Update.
29645 (ipa_cp_c_finalize): Update.
29646 * ipa-devirt.c (struct odr_type_warn_count): Update.
29647 (struct decl_warn_count): Update.
29648 (struct final_warning_record): Update.
29649 (possible_polymorphic_call_targets): Update.
29650 (ipa_devirt): Update.
29651 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29652 * ipa-icf.c (sem_function::merge): Update.
29653 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29654 * ipa-inline.c (compute_uninlined_call_time): Update.
29655 (compute_inlined_call_time): Update.
29656 (want_inline_small_function_p): Update.
29657 (want_inline_self_recursive_call_p): Update.
29658 (edge_badness): Update.
29659 (lookup_recursive_calls): Update.
29660 (recursive_inlining): Update.
29661 (inline_small_functions): Update.
29662 (dump_overall_stats): Update.
29663 (dump_inline_stats): Update.
29664 * ipa-profile.c (ipa_profile_generate_summary): Update.
29665 (ipa_propagate_frequency): Update.
29666 (ipa_profile): Update.
29667 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29668 * ipa-utils.c (ipa_merge_profiles): Update.
29669 * loop-doloop.c (doloop_modify): Update.
29670 * loop-unroll.c (report_unroll): Update.
29671 (unroll_loop_runtime_iterations): Update.
29672 * lto-cgraph.c (lto_output_edge): Update.
29673 (lto_output_node): Update.
29674 (input_node): Update.
29675 (input_edge): Update.
29676 (merge_profile_summaries): Update.
29677 * lto-streamer-in.c (input_cfg): Update.
29678 * lto-streamer-out.c (output_cfg): Update.
29679 * mcf.c (create_fixup_graph): Update.
29680 (adjust_cfg_counts): Update.
29681 (sum_edge_counts): Update.
29682 * modulo-sched.c (sms_schedule): Update.
29683 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29684 * predict.c (maybe_hot_count_p): Update.
29685 (probably_never_executed): Update.
29686 (dump_prediction): Update.
29687 (combine_predictions_for_bb): Update.
29688 (propagate_freq): Update.
29689 (handle_missing_profiles): Update.
29690 (counts_to_freqs): Update.
29691 (rebuild_frequencies): Update.
29692 (force_edge_cold): Update.
29693 * predict.h: Include profile-count.h
29694 (maybe_hot_count_p, counts_to_freqs): UPdate.
29695 * print-rtl-function.c: Do not include cfg.h
29696 * print-rtl.c: Include basic-block.h
29697 * profile-count.c: New file.
29698 * profile-count.h: New file.
29699 * profile.c (is_edge_inconsistent): Update.
29700 (correct_negative_edge_counts): Update.
29701 (is_inconsistent): Update.
29702 (set_bb_counts): Update.
29703 (read_profile_edge_counts): Update.
29704 (compute_frequency_overlap): Update.
29705 (compute_branch_probabilities): Update; Initialize and deinitialize
29707 (branch_prob): Update.
29708 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29709 (edge_gcov_count): New.
29710 (bb_gcov_count): New.
29711 * shrink-wrap.c (try_shrink_wrapping): Update.
29712 * tracer.c (better_p): Update.
29713 * trans-mem.c (expand_transaction): Update.
29714 (ipa_tm_insert_irr_call): Update.
29715 (ipa_tm_insert_gettmclone_call): Update.
29716 * tree-call-cdce.c: Update.
29717 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29718 (gimple_duplicate_sese_tail): Update.
29719 (gimple_account_profile_record): Update.
29720 (execute_fixup_cfg): Update.
29721 * tree-inline.c (copy_bb): Update.
29722 (copy_edges_for_bb): Update.
29723 (initialize_cfun): Update.
29724 (freqs_to_counts): Update.
29725 (copy_cfg_body): Update.
29726 (expand_call_inline): Update.
29727 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29728 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29729 (try_unroll_loop_completely): Update.
29730 (try_peel_loop): Update.
29731 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29732 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29733 * tree-ssa-loop-split.c (connect_loops): Update.
29734 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29735 * tree-ssa-reassoc.c (branch_fixup): Update.
29736 * tree-ssa-tail-merge.c (replace_block_by): Update.
29737 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29738 (compute_path_counts): Update.
29739 (update_profile): Update.
29740 (recompute_probabilities): Update.
29741 (update_joiner_offpath_counts): Update.
29742 (estimated_freqs_path): Update.
29743 (freqs_to_counts_path): Update.
29744 (clear_counts_path): Update.
29745 (ssa_fix_duplicate_block_edges): Update.
29746 (duplicate_thread_path): Update.
29747 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29748 (struct switch_conv_info): Update.
29749 * tree-tailcall.c (decrease_profile): Update.
29750 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29751 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29752 * value-prof.c (check_counter): Update.
29753 (gimple_divmod_fixed_value): Update.
29754 (gimple_mod_pow2): Update.
29755 (gimple_mod_subtract): Update.
29756 (gimple_ic_transform): Update.
29757 (gimple_stringop_fixed_value): Update.
29758 * value-prof.h (gimple_ic): Update.
29760 2017-06-02 Carl Love <cel@us.ibm.com>
29762 * config/rs6000/rs6000-c: Add support for built-in functions
29763 vector double vec_doublee (vector signed int);
29764 vector double vec_doublee (vector unsigned int);
29765 vector double vec_doublee (vector float);
29766 vector double vec_doubleh (vector signed int);
29767 vector double vec_doubleh (vector unsigned int);
29768 vector double vec_doubleh (vector float);
29769 vector double vec_doublel (vector signed int);
29770 vector double vec_doublel (vector unsigned int);
29771 vector double vec_doublel (vector float);
29772 vector double vec_doubleo (vector signed int);
29773 vector double vec_doubleo (vector unsigned int);
29774 vector double vec_doubleo (vector float);.
29775 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29776 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29778 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29779 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29780 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29782 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29783 vec_doublel, vec_doubleh.
29784 * doc/extend.texi: Update the built-in documentation file for the
29785 new built-in functions.
29787 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29790 * ipa-inline-analysis.c (free_growth_caches): Set
29791 edge_removal_hook_holder to NULL after removing it.
29793 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29795 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29796 comparision with zero.
29798 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29799 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29800 for early expansion of vec_min and vec_max builtins.
29801 (builtin_function_type): Add min/max unsigned variants to those
29802 identified as having unsigned arguments.
29804 2017-06-02 Olivier Hainque <hainque@adacore.com>
29806 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29808 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29810 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29811 Use VALL_F16 iterator rather than VALL.
29813 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29815 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29816 Emit CBNZ inside loop when doing a strong exchange and comparing
29817 against zero. Generate the CC flags after the loop.
29819 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29821 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29822 (dl_section_ref): New.
29823 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29824 On AIX, append an expression to subtract the size of the
29825 section length to dl_section_ref.
29827 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29829 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29830 for early expansion of vector absolute builtins.
29832 2017-06-02 Richard Biener <rguenther@suse.de>
29834 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29835 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29837 2017-06-02 Richard Biener <rguenther@suse.de>
29839 PR tree-optimization/80948
29840 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29841 stmt order as well.
29843 2017-06-02 Richard Biener <rguenther@suse.de>
29845 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29847 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29848 for inductions as relevant.
29850 2017-06-02 Richard Biener <rguenther@suse.de>
29852 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29853 (vectorizable_induction): ... this. Remove dead code.
29855 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29857 * builtins. (expand_builtin_alloca): Remove second parameter and
29858 infer its value from the first parameter instead.
29859 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29861 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29863 PR rtl-optimization/80903
29864 * loop-doloop.c (add_test): Unshare sequence.
29866 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29868 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29870 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29872 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29874 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29875 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29876 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29877 (xlogue_layout::get_stub_rtx): Make static.
29878 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29879 (xlogue_layout::compute_stub_managed_regs): Rename to...
29880 (xlogue_layout::count_stub_managed_regs): ...this.
29881 (xlogue_layout::is_stub_managed_reg): New function.
29882 (xlogue_layout::m_stub_names): Rename to...
29883 (xlogue_layout::s_stub_names): ...this, make static.
29884 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29885 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29886 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29887 xlogue_layout::s_stub_names): Instantiate statics.
29888 (stub_managed_regs): Remove.
29889 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29890 (disable_call_ms2sysv_xlogues): Rename to...
29891 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29892 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29894 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29895 change after reload_completed.
29896 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29898 (ix86_expand_prologue): Likewise.
29899 (ix86_expand_epilogue): Likewise.
29900 (ix86_expand_split_stack_prologue): Likewise.
29901 (ix86_compute_frame_layout): Remove frame parameter ...
29902 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29903 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29905 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29906 (ix86_frame): Move from here ...
29907 * config/i386/i386.h (ix86_frame): ... to here.
29908 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29909 complete ix86_frame data structure instead. Remove some_ld_name.
29911 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29913 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29914 symbols that hold a DECL_VALUE_EXPR.
29916 2017-06-01 Martin Jambor <mjambor@suse.cz>
29918 PR tree-optimization/80898
29919 * tree-sra.c (process_subtree_disqualification): Removed.
29920 (disqualify_candidate): Do not acll
29921 process_subtree_disqualification.
29922 (subtree_mark_written_and_enqueue): New function.
29923 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29924 RHS has been disqualified and re-queue LHS if necessary. Apart
29925 from that, ignore disqualified RHS.
29927 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29929 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29930 address fetch for z10 or later.
29932 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29934 * config/arc/arc.md (tst_movb): Add guard when splitting.
29936 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29938 * config/arc/arc.c (arc_can_eliminate): Test against
29939 arc_frame_pointer_needed.
29941 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29943 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29944 to prevent store reordering.
29945 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29946 (type): Add block type.
29947 (stack_tie): Define special instruction to be used in
29950 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29952 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29953 constraint. It is not valid for the pattern.
29954 (noncommutative_binary_comparison): Likewise.
29956 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29958 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29961 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29963 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29964 be used by the reg-alloc.
29966 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29968 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29969 reg-alloc when having mul64 or mul32x16 instructions.
29970 (mulsidi3): Likewise.
29971 (umulsidi3): Likewise.
29972 (mulsi32x16): New pattern.
29973 (mulsi64): Likewise.
29974 (mulsidi64): Likewise.
29975 (umulsidi64): Likewise.
29976 (MUL32x16_REG): Define.
29977 (mul64_600): Use MUL32x16_REG.
29978 (mac64_600): Likewise.
29979 (umul64_600): Likewise.
29980 (umac64_600): Likewise.
29982 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29984 * config/arc/arc.md (mulsi3_700): Make it commutative.
29986 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29988 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29990 (*sign_extendsidi2_insn): Likewise for movstosw.
29992 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29994 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29995 the type of the input discriminant value. Convert the
29996 discriminant value of signedness vary.
29998 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
30000 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
30001 Add to -Wall section.
30003 2017-06-01 Richard Biener <rguenther@suse.de>
30005 PR middle-end/66313
30006 * fold-const.c (fold_plusminus_mult_expr): If the factored
30007 factor may be zero use a wrapping type for the inner operation.
30008 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
30009 and handle moved defs.
30010 (process_assignment): Properly guard the unary op case. Return a
30011 tri-state indicating that moving the stmt before the call may allow
30012 to continue. Pass through to_move.
30013 (find_tail_calls): Handle moving unrelated defs before
30016 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
30019 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
30020 splitter result in the canonical way.
30022 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
30024 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
30025 also for 32bit target. Update insn attributes.
30026 (zero-extendsidi2 splitter): Allow all registers for operand 1.
30028 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
30030 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
30031 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
30032 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
30033 (_mm_maskz_min_ss): New intrinsics.
30035 2017-05-31 Martin Liska <mliska@suse.cz>
30037 * tree-vect-loop.c (vect_create_epilog_for_reduction):
30038 Change comment style to one we normally use.
30039 (vectorizable_reduction): Likewise.
30040 (vectorizable_induction): Likewise.
30041 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
30042 (vectorizable_call): Likewise.
30043 (vectorizable_simd_clone_call): Likewise.
30044 (vectorizable_conversion): Likewise.
30045 (vectorizable_assignment): Likewise.
30046 (vectorizable_shift): Likewise.
30047 (vectorizable_operation): Likewise.
30048 (vectorizable_store): Likewise.
30049 (vectorizable_load): Likewise.
30050 * tree-vectorizer.h: Likewise.
30052 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
30054 * passes.c (emergency_dump_function): New.
30055 * tree-pass.h (emergency_dump_function): Declare.
30056 * plugin.c (plugins_internal_error_function): Remove.
30057 * plugin.h (plugins_internal_error_function): Remove declaration.
30058 * toplev.c (internal_error_function): New static function. Use it...
30059 (general_init): ...here.
30061 2017-05-31 Graham Markall <graham.markall@embecosm.com>
30063 * config/arc/arc.c (arc_print_operand): Handle constant operands.
30064 (arc_rtx_costs): Add costs for new patterns.
30065 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
30066 * config/arc/predicates.md: Add _1_2_3_operand predicate.
30068 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30070 * tree-ssa-strlen.c (get_next_strinfo): New function.
30071 (get_stridx_plus_constant): Use it.
30072 (zero_length_string): Likewise.
30073 (adjust_related_strinfos): Likewise.
30074 (adjust_last_stmt): Likewise.
30076 2017-05-31 Richard Biener <rguenther@suse.de>
30079 * config/i386/i386.c (ix86_expand_builtin): Remove assert
30080 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
30082 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30084 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
30085 loop_vinfo argument and use of dependence distance vectors.
30086 Check instead whether the two references differ only in their
30087 initial value and assume that they have the same alignment if the
30088 difference is a multiple of the vector alignment.
30089 (vect_analyze_data_refs_alignment): Update call accordingly.
30091 2017-05-31 Martin Liska <mliska@suse.cz>
30094 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
30096 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30098 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
30099 (create_intersect_range_checks): Move from ...
30100 * tree-data-ref.c (create_intersect_range_checks_index)
30101 (create_intersect_range_checks): ... to here.
30102 (create_runtime_alias_checks): New function factored from ...
30103 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
30104 here. Call above function.
30105 * tree-data-ref.h (create_runtime_alias_checks): New function.
30107 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30109 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
30110 segment length for dr_b and compute it in wide_int.
30112 2017-05-31 Richard Biener <rguenther@suse.de>
30114 PR tree-optimization/80906
30115 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
30116 and pass through iv_map.
30117 (copy_bb_and_scalar_dependences): Adjust.
30118 (translate_pending_phi_nodes): Likewise.
30119 (copy_loop_close_phi_args): Handle code-generating IVs instead
30122 2017-05-30 David Malcolm <dmalcolm@redhat.com>
30124 * diagnostic-color.c (color_dict): Add "type-diff".
30125 (parse_gcc_colors): Update comment.
30126 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
30127 -fdiagnostics-show-template-tree and -fno-elide-type.
30128 (GCC_COLORS): Add type-diff to example.
30130 (-fdiagnostics-show-template-tree): New.
30131 (-fno-elide-type): New.
30132 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
30133 the pp_format_decoder callback. Call any m_format_postprocessor's
30135 (pretty_printer::pretty_printer): Initialize
30136 m_format_postprocessor.
30137 (pretty_printer::~pretty_printer): Delete any
30138 m_format_postprocessor.
30139 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
30140 (class format_postprocessor): New class.
30141 (struct pretty_printer::format_decoder): Document the new parameters.
30142 (struct pretty_printer::m_format_postprocessor): New field.
30143 * tree-diagnostic.c (default_tree_printer): Update for new
30144 bool and const char ** params.
30145 * tree-diagnostic.h (default_tree_printer): Likewise.
30147 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
30149 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
30150 (lwa_operand): Delete rs6000_gen_cell_microcode test.
30151 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30152 rs6000_gen_cell_microcode code.
30153 (rs6000_final_prescan_insn): Delete.
30154 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
30155 "warn-cell-microcode" entries.
30156 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
30157 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
30158 throughout. Change cc_reg_not_micro_cr0_operand to
30159 cc_reg_not_cr0_operand throughout.
30160 (*extendhi<mode>2_noload): Delete.
30161 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30162 (mwarn-cell-microcode): Delete.
30163 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30164 -mgen-cell-microcode and -mwarn-cell-microcode.
30166 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30169 * config/i386/constraints.md (Yd): New constraint.
30171 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30172 and (?Yd, r) alternatives. Update insn attributes.
30173 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30174 and (?*Yd, r) alternatives. Update insn attributes.
30175 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30177 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30179 * gimplify.c (gimplify_modify_expr): Don't create a
30180 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30183 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30185 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30187 2017-05-30 Richard Biener <rguenther@suse.de>
30189 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30190 and reduc_def fields.
30191 (STMT_VINFO_REDUC_TYPE): New define.
30192 (STMT_VINFO_REDUC_DEF): Likewise.
30193 (vect_force_simple_reduction): Adjust prototype.
30194 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30195 (vect_is_simple_reduction): Remove check_reduction argument.
30196 (vect_force_simple_reduction): Adjust and set
30197 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30198 (vectorizable_reduction): Do not re-do reduction analysis
30199 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30200 * tree-parloops.c (gather_scalar_reductions): Adjust.
30202 2017-05-30 Richard Biener <rguenther@suse.de>
30204 PR middle-end/80901
30205 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30208 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30210 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30211 Introduce unknown_misalignment parameter and remove vf.
30212 (vect_peeling_hash_get_lowest_cost):
30213 Pass unknown_misalignment parameter.
30214 (vect_enhance_data_refs_alignment):
30215 Fix unsupportable data ref treatment.
30217 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30219 * tree-vect-data-refs.c (vect_get_data_access_cost):
30220 Workaround for SLP handling.
30221 (vect_enhance_data_refs_alignment):
30222 Compute costs for doing no peeling at all, compare to the best
30223 peeling costs so far and avoid peeling if cheaper.
30225 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30227 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30228 Return peeling info and set costs to zero for unlimited cost
30230 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30231 with unknown misalignment. Compute and costs for unknown
30232 misalignment, compare them to the costs for known misalignment
30233 and choose the cheapest for peeling.
30235 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30237 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30238 (vect_get_peeling_costs_all_drs): Create function.
30239 (vect_peeling_hash_get_lowest_cost):
30240 Use vect_get_peeling_costs_all_drs.
30241 (vect_peeling_supportable): Create function.
30242 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30244 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30246 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30247 DR_HAS_NEGATIVE_STEP.
30248 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30249 (vect_enhance_data_refs_alignment): Use.
30250 (vect_duplicate_ssa_name_ptr_info): Use.
30251 * tree-vectorizer.h (dr_misalignment): Use.
30252 (known_alignment_for_access_p): Use.
30254 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30257 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30259 (has_section_name): New function.
30261 2017-05-30 Martin Liska <mliska@suse.cz>
30264 * auto-profile.c (get_function_decl_from_block): Fix
30267 2017-05-30 Richard Biener <rguenther@suse.de>
30269 PR middle-end/80876
30270 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30272 2017-05-30 Martin Liska <mliska@suse.cz>
30274 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30275 * dumpfile.h (struct dump_file_info): Remove ctors.
30277 2017-05-30 Martin Liska <mliska@suse.cz>
30279 * predict.def: Fix GNU coding style.
30281 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30283 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30284 Mark 'to' argument with ATTRIBUTE_UNUSED.
30286 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30288 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30289 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30290 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30293 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30295 * doc/install.texi (Options specification): Restore entry of
30296 --enable-sjlj-exceptions.
30298 2017-05-27 Michael Eager <eager@eagercon.com>
30301 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30303 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30305 * config/microblaze/microblaze.h
30306 (FIXED_REGISTERS): Update in macro.
30307 (CALL_USED_REGISTERS): Update in macro.
30309 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30311 * doc/install.texi: Add links to macOS binary distributions.
30313 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30317 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30319 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30321 2017-05-26 Martin Liska <mliska@suse.cz>
30323 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30325 2017-05-26 Martin Liska <mliska@suse.cz>
30327 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30328 always leading ';; '.
30329 (dump_bb_info): Likewise.
30330 (brief_dump_cfg): Likewise.
30331 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30332 * dumpfile.c: Remove usage of TDF_VERBOSE.
30333 * dumpfile.h (enum dump_kind): Likewise.
30334 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30335 * print-tree.c (debug_verbose): Remove.
30336 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30337 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30338 * tree-diagnostic.c (default_tree_printer): Replace
30339 TDF_DIAGNOSTIC with TDF_SLIM.
30341 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30343 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30344 in parameter loop, rather than loop_vinfo.
30345 (create_intersect_range_checks): Ditto.
30346 (vect_create_cond_for_alias_checks): Update call to above functions.
30348 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30350 PR tree-optimization/80815
30351 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30352 for merging runtime alias checks. Handle negative DR_STEPs.
30354 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30356 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30358 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30359 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30360 out code pruning runtime alias checks.
30361 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30362 factored out from above.
30363 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30365 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30367 (prune_runtime_alias_test_list): New decalaration.
30369 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30371 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30372 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30373 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30374 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30375 (operator==, comp_dr_with_seg_len_pair): Ditto.
30376 (vect_prune_runtime_alias_test_list): Ditto.
30378 2017-05-26 Martin Liska <mliska@suse.cz>
30381 * params.def: Bound partial-inlining-entry-probability param.
30383 2017-05-26 Marek Polacek <polacek@redhat.com>
30386 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30389 2017-05-26 Richard Biener <rguenther@suse.de>
30391 PR tree-optimization/80842
30392 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30395 2017-05-26 Richard Biener <rguenther@suse.de>
30397 PR tree-optimization/80844
30398 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30400 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30402 * doc/md.texi (Machine Constraints): Update x86 family
30403 machine constraints section to match 'config/i386/constraints.md'.
30405 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30407 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30409 2017-05-25 Nathan Sidwell <nathan@acm.org>
30411 * doc/invoke.texi (--enable-languages): Update documentation.
30413 2017-05-25 Martin Liska <mliska@suse.cz>
30415 * dumpfile.c: Add TDF_FOLDING.
30416 * dumpfile.h (enum dump_kind): Likewise.
30417 * genmatch.c (dt_simplify::gen_1): Use it.
30419 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30421 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30423 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30425 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30426 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30428 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30430 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30431 * match.pd (X == C): Rewrite it here.
30432 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30433 with_certain_nonzero_bits2): New predicates.
30434 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30436 2017-05-24 Nathan Sidwell <nathan@acm.org>
30438 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30441 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30444 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30446 * config/powerpcspe: New port. Files are copied from the rs6000
30447 port, with "rs6000" in filenames replaced by "powerpcspe".
30449 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30451 PR rtl-optimization/80754
30452 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30454 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30456 * config/sparc/sparc.md (length): Return the correct value for -mflat
30457 sibcalls to match output_sibcall.
30459 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30463 * config/rs6000/rs6000.c (struct machine_function): Add new field
30465 (rs6000_get_separate_components): Init that field, use it.
30466 (rs6000_components_for_bb): Use the field.
30468 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30470 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30472 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30474 PR middle-end/80823
30475 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30477 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30480 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30481 address against address_operand predicate.
30482 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30484 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30486 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30487 parameters passed indirectly.
30489 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30491 * config/i386/i386.md (*movdi_internal): Remove SSE4
30492 alternative 18 (?r, *v). Update insn attributes.
30493 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30494 Update insn attributes.
30495 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30496 Update insn attributes.
30497 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30498 alternative 1 (r, v). Remove isa attribute.
30499 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30500 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30501 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30503 2017-05-23 Tom de Vries <tom@codesourcery.com>
30505 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30508 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30510 * cgraphunit.c (symbol_table::process_new_functions): Update.
30511 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30512 (inline_generate_summary): Rename to ...
30513 (ipa_fn_summary_generate): ... this one.
30514 (inline_read_summary): Rename to ...
30515 (ipa_fn_summary_read): ... this one.
30516 (inline_write_summary): Rename to ...
30517 (ipa_fn_summary_write): ... this one.
30518 (inline_free_summary): Rename to ...
30519 (ipa_free_fn_summary): ... this one.
30520 (pass_data_local_fn_summary, pass_local_fn_summary,
30521 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30522 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30523 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30524 make_pass_ipa_fn_summary): New.
30525 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30526 inline_write_summary, inline_free_summary): Remove.
30527 (ipa_free_fn_summary) : New.
30528 * ipa-inline.c (ipa_inline): Update.
30529 (pass_ipa_inline): Do not generate summaries.
30530 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30532 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30533 and add pass_ipa_fn_summary.
30534 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30536 (make_pass_inline_parameters): Remove.
30538 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30540 * omp-low.c (struct omp_context): Remove "default_kind" member.
30543 * omp-offload.c (execute_oacc_device_lower): Remove the
30544 parallelism dimensions function attributes for unparallelized
30545 OpenACC kernels constructs.
30547 2017-05-23 Martin Liska <mliska@suse.cz>
30549 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30551 (cgraph_edge::make_speculative): Likewise.
30552 (cgraph_edge::resolve_speculation): Likewise.
30553 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30554 (cgraph_node::dump): Likewise.
30555 * cgraph.h: Likewise.
30556 * cgraphunit.c (analyze_functions): Likewise.
30557 (symbol_table::compile): Likewise.
30558 * ipa-cp.c (print_all_lattices): Likewise.
30559 (determine_versionability): Likewise.
30560 (initialize_node_lattices): Likewise.
30561 (ipcp_verify_propagated_values): Likewise.
30562 (estimate_local_effects): Likewise.
30563 (update_profiling_info): Likewise.
30564 (create_specialized_node): Likewise.
30565 (perhaps_add_new_callers): Likewise.
30566 (decide_about_value): Likewise.
30567 (decide_whether_version_node): Likewise.
30568 (identify_dead_nodes): Likewise.
30569 (ipcp_store_bits_results): Likewise.
30570 * ipa-devirt.c (dump_targets): Likewise.
30571 (ipa_devirt): Likewise.
30572 * ipa-icf.c (sem_item::dump): Likewise.
30573 (sem_function::equals): Likewise.
30574 (sem_variable::equals): Likewise.
30575 (sem_item_optimizer::read_section): Likewise.
30576 (sem_item_optimizer::execute): Likewise.
30577 (congruence_class::dump): Likewise.
30578 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30579 (dump_inline_summary): Likewise.
30580 (estimate_node_size_and_time): Likewise.
30581 (inline_analyze_function): Likewise.
30582 * ipa-inline-transform.c (inline_call): Likewise.
30583 * ipa-inline.c (report_inline_failed_reason): Likewise.
30584 (want_early_inline_function_p): Likewise.
30585 (edge_badness): Likewise.
30586 (update_edge_key): Likewise.
30587 (inline_small_functions): Likewise.
30588 * ipa-profile.c (ipa_profile): Likewise.
30589 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30590 (ipa_make_edge_direct_to_target): Likewise.
30591 (remove_described_reference): Likewise.
30592 (ipa_impossible_devirt_target): Likewise.
30593 (propagate_controlled_uses): Likewise.
30594 (ipa_print_node_params): Likewise.
30595 (ipcp_transform_function): Likewise.
30596 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30597 (propagate_pure_const): Likewise.
30598 * ipa-reference.c (generate_summary): Likewise.
30599 (read_write_all_from_decl): Likewise.
30600 (propagate): Likewise.
30601 (ipa_reference_read_optimization_summary): Likewise.
30602 * ipa-utils.c (ipa_merge_profiles): Likewise.
30603 * ipa.c (walk_polymorphic_call_targets): Likewise.
30604 (symbol_table::remove_unreachable_nodes): Likewise.
30605 (ipa_single_use): Likewise.
30606 * passes.c (execute_todo): Likewise.
30607 * predict.c (drop_profile): Likewise.
30608 * symtab.c (symtab_node::get_dump_name): New function.
30609 (symtab_node::dump_name): Likewise.
30610 (symtab_node::dump_asm_name): Likewise.
30611 (symtab_node::dump_references): Likewise.
30612 (symtab_node::dump_referring): Likewise.
30613 (symtab_node::dump_base): Likewise.
30614 (symtab_node::debug_symtab): Likewise.
30615 * tree-sra.c (convert_callers_for_node): Likewise.
30616 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30617 * value-prof.c (init_node_map): Likewise.
30619 2017-05-23 Martin Liska <mliska@suse.cz>
30621 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30622 and symtab_node::debug_symtab to symbol_table::debug.
30623 * cgraphunit.c (analyze_functions): Use the renamed function.
30624 (symbol_table::compile): Likewise.
30625 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30626 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30627 * passes.c (execute_todo): Likewise.
30628 * symtab.c (symbol_table::dump): New function.
30629 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30631 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30633 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30634 that nonconst implies exec.
30636 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30638 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30639 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30640 (inline_edge_summary_vec): Turn into ...
30641 (ipa_call_summaries): ... this one.
30642 (redirect_to_unreachable, edge_set_predicate,
30643 evaluate_properties_for_edge, inline_summary_alloc,
30644 reset_ipa_call_summary, reset_inline_summary,
30645 inline_summary_t::duplicate): Update.
30646 (inline_edge_duplication_hook): Turn to ...
30647 (ipa_call_summary_t::duplicate): ... this one.
30648 (inline_edge_removal_hook): Turn to ...
30649 (ipa_call_summary_t::remove): ... this one.
30650 (dump_inline_edge_summary): Turn to ...
30651 (dump_ipa_call_summary): ... this one.
30652 (estimate_function_body_sizes): Update.
30653 (inline_update_callee_summaries): Update.
30654 (remap_edge_change_prob): Update.
30655 (remap_edge_summaries): Update.
30656 (inline_merge_summary): Update.
30657 (do_estimate_edge_time): Update.
30658 (inline_generate_summary): Update.
30659 (inline_read_section): Update.
30660 (inline_read_summary): Update.
30661 (inline_free_summary): Update.
30662 * ipa-inline.c (can_inline_edge_p): Update.
30663 (compute_inlined_call_time): Update.
30664 (want_inline_small_function_p): Update.
30665 (edge_badness): Update.
30666 (early_inliner): Update.
30667 * ipa-inline.h (inline_edge_summary): Turn to ...
30668 (ipa_call_summary): ... this one.
30669 (ipa_call_summary_t): New class.
30670 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30671 (ipa_call_summaries): New.
30672 (inline_edge_summary): Remove.
30673 (estimate_edge_growth): Update.
30674 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30675 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30676 * ipa-split.c (execute_split_functions): Update.
30677 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30679 2017-05-23 Tom de Vries <tom@codesourcery.com>
30681 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30682 attributes): Document rdrand effective target.
30684 2017-05-23 Tom de Vries <tom@codesourcery.com>
30686 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30687 attributes): Sort alphabetically.
30689 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30691 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30693 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30696 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30697 V2DF/V2DI splat into two separate patterns, one that handles
30698 registers, and the other that only handles memory. Drop support
30699 for splatting from a GPR on ISA 2.07 and then splitting the
30700 splat into direct move and splat.
30701 (vsx_splat_<mode>_reg): Likewise.
30702 (vsx_splat_<mode>_mem): Likewise.
30704 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30706 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30708 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30710 PR middle-end/80809
30711 * omp-low.c (finish_taskreg_remap): New function.
30712 (finish_taskreg_scan): If unit size of ctx->record_type
30713 is non-constant, unshare the size expression and replace
30714 decls in it with possible outer var refs.
30716 PR middle-end/80809
30717 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30718 GOVD_SHARED rather than GOVD_PRIVATE with it.
30719 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30720 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30722 PR middle-end/80853
30723 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30724 as last argument to build_outer_var_ref for pointer bases of array
30725 section reductions.
30727 2017-05-19 Martin Sebor <msebor@redhat.com>
30729 * print-tree.c (print_node): Print DECL_READ_P flag.
30731 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30733 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30734 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30735 * cgraph.c: Likewise.
30736 * cgraphunit.c: Likewise.
30737 * gengtype.c: Likewise.
30738 * ipa-cp.c: Likewise.
30739 * ipa-devirt.c: Likewise.
30740 * ipa-icf.c: Likewise.
30741 * ipa-predicate.c: Likewise.
30742 * ipa-profile.c: Likewise.
30743 * ipa-prop.c: Likewise.
30744 * ipa-split.c: Likewise.
30746 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30747 edge_predicate_pool, dump_inline_hints,
30748 inline_summary::account_size_time, redirect_to_unreachable,
30749 edge_set_predicate, set_hint_predicate,
30750 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30751 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30752 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30753 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30754 ipa_call_summary_t::remove, initialize_growth_caches,
30755 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30756 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30757 mark_modified, unmodified_parm_1, unmodified_parm,
30758 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30759 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30760 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30761 will_be_nonconstant_predicate, record_modified_bb_info,
30762 get_minimal_bb, record_modified, param_change_prob,
30763 phi_result_unknown_predicate, predicate_for_phi_result,
30764 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30765 estimate_function_body_sizes, compute_inline_parameters,
30766 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30767 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30768 inline_update_callee_summaries, remap_edge_change_prob,
30769 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30770 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30771 inline_analyze_function, inline_summary_t::insert,
30772 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30773 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30774 inline_free_summary): Move to ipa-fnsummary.h
30775 (predicate_t): Remove.
30776 * ipa-fnsummary.c: New file.
30777 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30778 (enum inline_hints_vals, inline_hints, agg_position_info,
30779 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30780 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30781 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30782 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30783 inline_read_summary, inline_write_summary, inline_free_summary,
30784 inline_analyze_function, initialize_inline_failed,
30785 inline_merge_summary, inline_update_overall_summary,
30786 compute_inline_parameters): Move to ipa-fnsummary.h
30787 * ipa-fnsummary.h: New file.
30788 * ipa-inline-transform.h: Include ipa-inline.h.
30789 * ipa-inline.c: LIkewise.
30791 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30793 * ipa-inline.c (edge_badness): Use inlined_time instead of
30794 inline_summaries->get.
30796 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30798 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30800 2017-05-22 Nathan Sidwell <nathan@acm.org>
30802 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30803 (fdump-lang): Document 'raw' option.
30804 * dumpfile.h (TDI_tu): Delete.
30805 * dumpfile.c (dump_files): Remove translation-unit.
30806 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30808 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30810 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30811 command option from $(AWK) call.
30812 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30813 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30814 [FORMAT]: Remove handling of variable.
30815 * config/avr/t-multilib: Regenerate.
30817 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30819 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30821 (dump_inline_summary): Do not print self_time.
30822 (estimate_function_body_sizes): Do not set self_time.
30823 (compute_inline_parameters): Likewise.
30824 (inline_read_section, inline_write_summary): Do not stream self_time.
30825 * ipa-inline.h (inline_summary): Drop self_time.
30827 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30829 * ipa-inline-analysis.c (account_size_time): Rename to ...
30830 (inline_summary::account_size_time): ... this one.
30831 (reset_ipa_call_summary): Turn to ...
30832 (ipa_call_summary::reset): ... this one.
30833 (reset_inline_summary): Turn to ...
30834 (inline_summary::reset): ... this one.
30835 (inline_summary_t::remove): Update.
30836 (inline_summary_t::duplicate): Update.
30837 (ipa_call_summary_t::remove): Update.
30838 (dump_inline_summary): Update.
30839 (estimate_function_body_sizes): Update.
30840 (compute_inline_parameters): Update.
30841 (estimate_node_size_and_time): Update.
30842 (inline_merge_summary): Update.
30843 (inline_update_overall_summary): Update.
30844 (inline_read_section): Update.
30845 (inline_write_summary): Update.
30846 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30847 add account_size_time and reset member functions.
30848 (ipa_call_summary): Add reset function.
30849 * ipa-predicate.h (predicate::operator &): Constify.
30851 2017-05-22 Richard Biener <rguenther@suse.de>
30853 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30855 2017-05-19 Jason Merrill <jason@redhat.com>
30857 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30859 2017-05-19 Marek Polacek <polacek@redhat.com>
30862 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30863 TYPE_OVERFLOW_WRAPS checks.
30865 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30867 * tree-core.h (enum omp_clause_default_kind): Add
30868 "OMP_CLAUSE_DEFAULT_PRESENT".
30869 * tree-pretty-print.c (dump_omp_clause): Handle it.
30870 * gimplify.c (enum gimplify_omp_var_data): Add
30871 "GOVD_MAP_FORCE_PRESENT".
30872 (gimplify_adjust_omp_clauses_1): Map it to
30873 "GOMP_MAP_FORCE_PRESENT".
30874 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30876 * gimplify.c (oacc_default_clause): Clarify.
30878 2017-05-19 Nathan Sidwell <nathan@acm.org>
30880 LANG_HOOK_REGISTER_DUMPS
30881 * toplev.c (general_init): Call register dump lang hook.
30882 * doc/invoke.texi: Document -fdump-lang option family.
30883 * dumpfile.c (dump_files): Remove class dump here.
30884 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30885 * dumpfile.h (tree_dump_index): Remove TDI_class.
30886 * langhooks-def.h (lhd_register_dumps): Declare.
30887 (LANG_HOOKS_REGISTER_DUMPS): Define.
30888 (LANG_HOOKS_INITIALIZER): Add it.
30889 * langhooks.c (lhd_register_dumps): Define.
30890 * langhooks.h (struct lang_hooks): Add register_dumps.
30892 2017-05-19 Nathan Sidwell <nathan@acm.org>
30894 * context.h (context::set_passes): New.
30895 * context.c (context::context): Do not create pass manager.
30896 * toplev.c (general_init): Create pass manager here.
30898 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30900 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30901 use this splitter if two add or or instructions would also work for
30902 the constant we want to generate.
30904 2017-05-19 Richard Biener <rguenther@suse.de>
30907 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30908 predicate evaluation.
30910 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30912 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30914 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30917 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30919 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30920 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30921 (inline_edge_summary_vec): Turn into ...
30922 (ipa_call_summaries): ... this one.
30923 (redirect_to_unreachable, edge_set_predicate,
30924 evaluate_properties_for_edge, inline_summary_alloc,
30925 reset_ipa_call_summary, reset_inline_summary,
30926 inline_summary_t::duplicate): Update.
30927 (inline_edge_duplication_hook): Turn to ...
30928 (ipa_call_summary_t::duplicate): ... this one.
30929 (inline_edge_removal_hook): Turn to ...
30930 (ipa_call_summary_t::remove): ... this one.
30931 (dump_inline_edge_summary): Turn to ...
30932 (dump_ipa_call_summary): ... this one.
30933 (estimate_function_body_sizes): Update.
30934 (inline_update_callee_summaries): Update.
30935 (remap_edge_change_prob): Update.
30936 (remap_edge_summaries): Update.
30937 (inline_merge_summary): Update.
30938 (do_estimate_edge_time): Update.
30939 (inline_generate_summary): Update.
30940 (inline_read_section): Update.
30941 (inline_read_summary): Update.
30942 (inline_free_summary): Update.
30943 * ipa-inline.c (can_inline_edge_p): Update.
30944 (compute_inlined_call_time): Update.
30945 (want_inline_small_function_p): Update.
30946 (edge_badness): Update.
30947 (early_inliner): Update.
30948 * ipa-inline.h (inline_edge_summary): Turn to ...
30949 (ipa_call_summary): ... this one.
30950 (ipa_call_summary_t): New class.
30951 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30952 (ipa_call_summaries): New.
30953 (inline_edge_summary): Remove.
30954 (estimate_edge_growth): Update.
30955 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30956 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30957 * ipa-split.c (execute_split_functions): Update.
30958 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30960 2017-05-19 Richard Biener <rguenther@suse.de>
30962 PR middle-end/80764
30963 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30965 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30967 * config/rs6000/rs6000.c (struct machine_function): Add field
30968 fpr_is_wrapped_separately.
30969 (rs6000_get_separate_components): Use 64 components. Handle the
30970 new FPR components.
30971 (rs6000_components_for_bb): Handle the FPR components.
30972 (rs6000_emit_prologue_components): Handle the FPR components.
30973 (rs6000_emit_epilogue_components): Handle the FPR components.
30974 (rs6000_set_handled_components): Handle the FPR components.
30975 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30976 that are already separately shrink-wrapped.
30977 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30978 that are already separately shrink-wrapped.
30980 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30983 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30986 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30987 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30988 cases where the register allocator uses a move and an offsettable
30989 memory operation to/from a FPR register on ISA 2.06/2.07.
30990 (define_peephole2 for Altivec d-form store): Likewise.
30992 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30995 * config/i386/mmx.md (*mov<mode>_internal): Enable
30996 alternatives 11, 12, 13 and 14 also for 32bit targets.
30997 Remove alternatives 15, 16, 17 and 18.
30998 * config/i386/sse.md (vec_concatv2di): Change
30999 alternative (!x, *y) to (x, ?!*Yn).
31001 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
31003 * dumpfile.h (enum dump_kind): Remove stray comma.
31005 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31007 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
31008 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
31009 predicate::num_conditions
31010 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
31011 (CHANGED): turn into predicate::changed.
31012 (agg_position_info): Move to ipa-predicate.h
31013 (add_condition, predicate::add_clause, predicate::operator &=,
31014 predicate::or_with, predicate::evaluate, predicate::probability,
31015 dump_condition, dump_clause, predicate::dump,
31016 predicate::remap_after_duplication, predicate::remap_after_inlining,
31017 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
31018 (evaluate_conditions_for_known_args): Update.
31019 (set_cond_stmt_execution_predicate): Update.
31020 * ipa-inline.h: Include ipa-predicate.h
31021 (condition, inline_param_summary, conditions, agg_position_info,
31022 predicate): Move to ipa-predicate.h
31023 * ipa-predicate.c: New file.
31024 * ipa-predicate.h: New file.
31026 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
31028 * final.c (leaf_function_p): Check we are not in a sequence.
31030 2017-05-18 Martin Liska <mliska@suse.cz>
31032 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
31033 * dumpfile.c (dump_register): Use new enum dump_kind.
31034 (get_dump_file_name): Likewise.
31035 (dump_enable_all): Likewise.
31036 (dump_switch_p_1): Likewise.
31037 (enable_rtl_dump_file): Remove usage of TDF_RTL.
31038 * dumpfile.h (enum dump_kind): New enum type.
31039 (struct dump_file_info): Create constructor and
31040 format fields and comments.
31041 * passes.c (pass_manager::register_one_dump_file):
31043 * statistics.c (statistics_early_init): Likewise.
31044 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
31045 TDF_TREE with TDF_SLIM.
31046 (gather_memory_references_ref): Likewise.
31048 2017-05-18 Martin Liska <mliska@suse.cz>
31050 * vec.h (struct vnull): Use it.
31052 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31054 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
31055 (true_predicate, false_predicate, true_predicate_p,
31056 false_predicate_p): Remove.
31057 (single_cond_predicate, not_inlined_predicate): Turn to member function
31059 (add_condition): Update.
31060 (add_clause): Turn to...
31061 (predicate::add_clause): ... this one; update; allow passing NULL
31063 (and_predicates): Turn to ...
31064 (predicate::operator &=): ... this one.
31065 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
31066 (or_predicates): Turn to ...
31067 (predicate::or_with): ... this one.
31068 (evaluate_predicate): Turn to ...
31069 (predicate::evaluate): ... this one.
31070 (predicate_probability): Turn to ...
31071 (predicate::probability): ... this one.
31072 (dump_condition): Update.
31073 (dump_predicate): Turn to ...
31074 (predicate::dump): ... this one.
31075 (account_size_time): Update.
31076 (edge_set_predicate): Update.
31077 (set_hint_predicate): UPdate.
31078 (evaluate_conditions_for_known_args): Update.
31079 (evaluate_properties_for_edge): Update.
31080 (remap_predicate_after_duplication): Turn to...
31081 (predicate::remap_after_duplication): ... this one.
31082 (remap_hint_predicate_after_duplication): Update.
31083 (inline_summary_t::duplicate): UPdate.
31084 (dump_inline_edge_summary): Update.
31085 (dump_inline_summary): Update.
31086 (set_cond_stmt_execution_predicate): Update.
31087 (set_switch_stmt_execution_predicate): Update.
31088 (compute_bb_predicates): Update.
31089 (will_be_nonconstant_expr_predicate): Update.
31090 (will_be_nonconstant_predicate): Update.
31091 (phi_result_unknown_predicate): Update.
31092 (predicate_for_phi_result): Update.
31093 (array_index_predicate): Update.
31094 (estimate_function_body_sizes): Update.
31095 (estimate_node_size_and_time): Update.
31096 (estimate_ipcp_clone_size_and_time): Update.
31097 (remap_predicate): Rename to ...
31098 (predicate::remap_after_inlining): ... this one.
31099 (remap_hint_predicate): Update.
31100 (inline_merge_summary): Update.
31101 (inline_update_overall_summary): Update.
31102 (estimate_size_after_inlining): Update.
31103 (read_predicate): Rename to ...
31104 (predicate::stream_in): ... this one.
31105 (read_inline_edge_summary): Update.
31106 (write_predicate): Rename to ...
31107 (predicate::stream_out): ... this one.
31108 (write_inline_edge_summary): Update.
31109 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
31110 (clause_t): Turn to uint32_t
31111 (predicate): Turn to class; implement constructor and operators
31113 (size_time_entry): Update.
31114 (inline_summary): Update.
31115 (inline_edge_summary): Update.
31117 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
31119 * fold-const.c (fold_binary_loc): Move transformation...
31120 * match.pd (C - X CMP X): ... here.
31122 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
31124 * config/sparc/sparc.c (sparc_option_override): Set function
31125 alignment for -mcpu=niagara7 to 64 to match the I$ line.
31126 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
31128 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
31130 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
31132 2017-05-18 Marek Polacek <polacek@redhat.com>
31135 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
31136 (pass_ubsan::execute): Call gimple_assign_single_p instead of
31137 gimple_assign_load_p.
31139 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
31141 PR middle-end/80692
31142 * real.c (do_compare): Give decimal_do_compare preference over
31143 comparing just the signs.
31145 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
31147 * doc/md.texi (Canonicalization of Instructions): Describe the
31148 canonical form of instructions that inherently set a condition
31151 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
31153 PR middle-end/80775
31154 * tree-cfg.c: Move deletion of unreachable case statements to after
31155 the merging of consecutive case labels.
31157 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
31159 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
31160 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31161 restoring of callee-saved registers.
31163 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31165 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31166 * config/visium/visium.c (single_set_and_flags): Likewise.
31167 * config/visium/visium.md (Substitutions): Likewise.
31169 2017-05-17 Martin Liska <mliska@suse.cz>
31171 * cfg.c: Introduce dump_flags_t type and
31172 use it instead of int type.
31174 * cfghooks.c: Likewise.
31175 * cfghooks.h (struct cfg_hooks): Likewise.
31176 * cfgrtl.c: Likewise.
31177 * cfgrtl.h: Likewise.
31178 * cgraph.c (cgraph_node::get_body): Likewise.
31179 * coretypes.h: Likewise.
31180 * domwalk.c: Likewise.
31181 * domwalk.h: Likewise.
31182 * dumpfile.c (struct dump_option_value_info): Likewise.
31183 (dump_enable_all): Likewise.
31184 (dump_switch_p_1): Likewise.
31185 (opt_info_switch_p): Likewise.
31186 * dumpfile.h (enum tree_dump_index): Likewise.
31187 (struct dump_file_info): Likewise.
31188 * genemit.c: Likewise.
31189 * generic-match-head.c: Likewise.
31190 * gengtype.c (open_base_files): Likewise.
31191 * gimple-pretty-print.c: Likewise.
31192 * gimple-pretty-print.h: Likewise.
31193 * graph.c (print_graph_cfg): Likewise.
31194 * graphite-scop-detection.c (dot_all_sese): Likewise.
31195 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31196 * loop-unroll.c (report_unroll): Likewise.
31197 * passes.c (pass_manager::register_one_dump_file): Likewise.
31198 * print-tree.c: Likewise.
31199 * statistics.c: Likewise.
31200 * tree-cfg.c: Likewise.
31201 * tree-cfg.h: Likewise.
31202 * tree-dfa.c: Likewise.
31203 * tree-dfa.h: Likewise.
31204 * tree-dump.c (dump_function): Likewise.
31205 * tree-dump.h (struct dump_info): Likewise.
31206 * tree-pretty-print.c: Likewise.
31207 * tree-pretty-print.h: Likewise.
31208 * tree-ssa-live.c: Likewise.
31209 * tree-ssa-live.h: Likewise.
31210 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31211 * tree-vect-loop.c: Likewise.
31212 * tree-vect-slp.c: Likewise.
31214 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31215 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31217 PR tree-optimization/80457
31218 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31219 of all arguments to a statement as scalar_to_vec operations.
31220 (vectorizable_call): Adjust call to vect_model_simple_cost for
31222 (vectorizable_conversion): Likewise.
31223 (vectorizable_assignment): Likewise.
31224 (vectorizable_shift): Likewise.
31225 (vectorizable_operation): Likewise.
31226 (vectorizable_comparison): Likewise.
31227 (vect_is_simple_cond): Record the def types for operands.
31228 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31229 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31230 for statement argument count.
31232 2017-05-16 Carl Love <cel@us.ibm.com>
31234 * config/rs6000/rs6000-c: Add support for built-in functions
31235 vector unsigned long long vec_bperm (vector unsigned long long,
31236 vector unsigned char)
31237 vector signed long long vec_mule (vector signed int,
31239 vector unsigned long long vec_mule (vector unsigned int,
31240 vector unsigned int)
31241 vector signed long long vec_mulo (vector signed int,
31243 vector unsigned long long vec_mulo (vector unsigned int,
31244 vector unsigned int)
31245 vector signed char vec_sldw (vector signed char,
31246 vector signed char,
31248 vector unsigned char vec_sldw (vector unsigned char,
31249 vector unsigned char,
31251 vector signed short vec_sldw (vector signed short,
31252 vector signed short,
31254 vector unsigned short vec_sldw (vector unsigned short,
31255 vector unsigned short,
31257 vector signed int vec_sldw (vector signed int,
31260 vector unsigned int vec_sldw (vector unsigned int,
31261 vector unsigned int,
31263 vector signed long long vec_sldw (vector signed long long,
31264 vector signed long long,
31266 vector unsigned long long vec_sldw (vector unsigned long long,
31267 vector unsigned long long,
31269 * config/rs6000/rs6000-c: Add support for built-in functions
31270 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31271 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31272 * doc/extend.texi: Update the built-in documentation for the
31273 new built-in functions.
31275 2017-05-16 Marek Polacek <polacek@redhat.com>
31279 * tree.c (save_expr): Don't fold the expression.
31281 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31283 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31284 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31285 and (?*y,m). Update insn attributes.
31287 2017-05-16 Martin Liska <mliska@suse.cz>
31289 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31290 flags argument of print_gimple_stmt, print_gimple_expr,
31291 print_generic_stmt and print_generic_expr.
31292 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31293 * coretypes.h: Likewise.
31294 * except.c (dump_eh_tree): Likewise.
31295 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31296 * gimple-pretty-print.h: Likewise.
31297 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31298 (backprop::push_to_worklist): Likewise.
31299 (backprop::pop_from_worklist): Likewise.
31300 (backprop::process_use): Likewise.
31301 (backprop::intersect_uses): Likewise.
31302 (note_replacement): Likewise.
31303 * gimple-ssa-store-merging.c
31304 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31305 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31306 (pass_store_merging::execute): Likewise.
31307 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31308 (ssa_base_cand_dump_callback): Likewise.
31309 (dump_incr_vec): Likewise.
31310 (replace_refs): Likewise.
31311 (replace_mult_candidate): Likewise.
31312 (create_add_on_incoming_edge): Likewise.
31313 (create_phi_basis): Likewise.
31314 (insert_initializers): Likewise.
31315 (all_phi_incrs_profitable): Likewise.
31316 (introduce_cast_before_cand): Likewise.
31317 (replace_one_candidate): Likewise.
31318 * gimplify.c (gimplify_expr): Likewise.
31319 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31320 (set_rename): Likewise.
31321 (rename_uses): Likewise.
31322 (copy_loop_phi_nodes): Likewise.
31323 (add_close_phis_to_merge_points): Likewise.
31324 (copy_loop_close_phi_args): Likewise.
31325 (copy_cond_phi_args): Likewise.
31326 (graphite_copy_stmts_from_block): Likewise.
31327 (translate_pending_phi_nodes): Likewise.
31328 * graphite-poly.c (print_pdr): Likewise.
31329 (dump_gbb_cases): Likewise.
31330 (dump_gbb_conditions): Likewise.
31331 (print_scop_params): Likewise.
31332 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31333 (build_cross_bb_scalars_use): Likewise.
31334 (gather_bbs::before_dom_children): Likewise.
31335 * hsa-dump.c (dump_hsa_immed): Likewise.
31336 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31337 (get_replacement_map): Likewise.
31338 * ipa-inline-analysis.c (dump_condition): Likewise.
31339 (estimate_function_body_sizes): Likewise.
31340 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31341 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31342 * ipa-prop.c (ipa_dump_param): Likewise.
31343 (ipa_print_node_jump_functions_for_edge): Likewise.
31344 (ipa_modify_call_arguments): Likewise.
31345 (ipa_modify_expr): Likewise.
31346 (ipa_dump_param_adjustments): Likewise.
31347 (ipa_dump_agg_replacement_values): Likewise.
31348 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31349 * ipa-pure-const.c (check_stmt): Likewise.
31350 (pass_nothrow::execute): Likewise.
31351 * ipa-split.c (execute_split_functions): Likewise.
31352 * omp-offload.c (dump_oacc_loop_part): Likewise.
31353 (dump_oacc_loop): Likewise.
31354 * trans-mem.c (tm_log_emit): Likewise.
31355 (tm_memopt_accumulate_memops): Likewise.
31356 (dump_tm_memopt_set): Likewise.
31357 (dump_tm_memopt_transform): Likewise.
31358 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31359 (print_loop): Likewise.
31360 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31361 (chkp_gather_checks_info): Likewise.
31362 (chkp_get_check_result): Likewise.
31363 (chkp_remove_check_if_pass): Likewise.
31364 (chkp_use_outer_bounds_if_possible): Likewise.
31365 (chkp_reduce_bounds_lifetime): Likewise.
31366 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31367 (chkp_mark_completed_bounds): Likewise.
31368 (chkp_register_incomplete_bounds): Likewise.
31369 (chkp_mark_invalid_bounds): Likewise.
31370 (chkp_maybe_copy_and_register_bounds): Likewise.
31371 (chkp_build_returned_bound): Likewise.
31372 (chkp_get_bound_for_parm): Likewise.
31373 (chkp_build_bndldx): Likewise.
31374 (chkp_get_bounds_by_definition): Likewise.
31375 (chkp_generate_extern_var_bounds): Likewise.
31376 (chkp_get_bounds_for_decl_addr): Likewise.
31377 * tree-chrec.c (chrec_apply): Likewise.
31378 * tree-data-ref.c (dump_data_reference): Likewise.
31379 (dump_subscript): Likewise.
31380 (dump_data_dependence_relation): Likewise.
31381 (analyze_overlapping_iterations): Likewise.
31382 * tree-inline.c (expand_call_inline): Likewise.
31383 (tree_function_versioning): Likewise.
31384 * tree-into-ssa.c (dump_defs_stack): Likewise.
31385 (dump_currdefs): Likewise.
31386 (dump_names_replaced_by): Likewise.
31387 (dump_update_ssa): Likewise.
31388 (update_ssa): Likewise.
31389 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31390 * tree-parloops.c (build_new_reduction): Likewise.
31391 (try_create_reduction_list): Likewise.
31392 (ref_conflicts_with_region): Likewise.
31393 (oacc_entry_exit_ok_1): Likewise.
31394 (oacc_entry_exit_single_gang): Likewise.
31395 * tree-pretty-print.h: Likewise.
31396 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31397 (get_scalar_evolution): Likewise.
31398 (add_to_evolution): Likewise.
31399 (get_loop_exit_condition): Likewise.
31400 (analyze_evolution_in_loop): Likewise.
31401 (analyze_initial_condition): Likewise.
31402 (analyze_scalar_evolution): Likewise.
31403 (instantiate_scev): Likewise.
31404 (number_of_latch_executions): Likewise.
31405 (gather_chrec_stats): Likewise.
31406 (final_value_replacement_loop): Likewise.
31407 (scev_const_prop): Likewise.
31408 * tree-sra.c (dump_access): Likewise.
31409 (disqualify_candidate): Likewise.
31410 (create_access): Likewise.
31411 (reject): Likewise.
31412 (maybe_add_sra_candidate): Likewise.
31413 (create_access_replacement): Likewise.
31414 (analyze_access_subtree): Likewise.
31415 (analyze_all_variable_accesses): Likewise.
31416 (sra_modify_assign): Likewise.
31417 (initialize_constant_pool_replacements): Likewise.
31418 (find_param_candidates): Likewise.
31419 (decide_one_param_reduction): Likewise.
31420 (replace_removed_params_ssa_names): Likewise.
31421 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31422 * tree-ssa-copy.c (dump_copy_of): Likewise.
31423 (copy_prop_visit_cond_stmt): Likewise.
31424 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31425 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31426 (record_equivalences_from_stmt): Likewise.
31427 * tree-ssa-dse.c (compute_trims): Likewise.
31428 (delete_dead_call): Likewise.
31429 (delete_dead_assignment): Likewise.
31430 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31431 (forward_propagate_into_cond): Likewise.
31432 (pass_forwprop::execute): Likewise.
31433 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31434 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31436 (move_computations_worker): Likewise.
31437 (execute_sm): Likewise.
31438 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31439 (remove_exits_and_undefined_stmts): Likewise.
31440 (remove_redundant_iv_tests): Likewise.
31441 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31442 (adjust_iv_update_pos): Likewise.
31443 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31444 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31445 (value_replacement): Likewise.
31446 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31447 * tree-ssa-pre.c (print_pre_expr): Likewise.
31448 (get_representative_for): Likewise.
31449 (create_expression_by_pieces): Likewise.
31450 (insert_into_preds_of_block): Likewise.
31451 (eliminate_insert): Likewise.
31452 (eliminate_dom_walker::before_dom_children): Likewise.
31453 (eliminate): Likewise.
31454 (remove_dead_inserted_code): Likewise.
31455 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31456 * tree-ssa-reassoc.c (get_rank): Likewise.
31457 (eliminate_duplicate_pair): Likewise.
31458 (eliminate_plus_minus_pair): Likewise.
31459 (eliminate_not_pairs): Likewise.
31460 (undistribute_ops_list): Likewise.
31461 (eliminate_redundant_comparison): Likewise.
31462 (update_range_test): Likewise.
31463 (optimize_range_tests_var_bound): Likewise.
31464 (optimize_vec_cond_expr): Likewise.
31465 (rewrite_expr_tree): Likewise.
31466 (rewrite_expr_tree_parallel): Likewise.
31467 (linearize_expr): Likewise.
31468 (break_up_subtract): Likewise.
31469 (linearize_expr_tree): Likewise.
31470 (attempt_builtin_powi): Likewise.
31471 (attempt_builtin_copysign): Likewise.
31472 (transform_stmt_to_copy): Likewise.
31473 (transform_stmt_to_multiply): Likewise.
31474 (dump_ops_vector): Likewise.
31475 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31476 (print_scc): Likewise.
31477 (set_ssa_val_to): Likewise.
31478 (visit_reference_op_store): Likewise.
31479 (visit_use): Likewise.
31480 (sccvn_dom_walker::before_dom_children): Likewise.
31481 (run_scc_vn): Likewise.
31482 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31484 (expr_hash_elt::print): Likewise.
31485 (const_and_copies::pop_to_marker): Likewise.
31486 (const_and_copies::record_const_or_copy_raw): Likewise.
31487 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31488 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31489 (dump_predicates): Likewise.
31490 (find_uninit_use): Likewise.
31491 (warn_uninitialized_phi): Likewise.
31492 (pass_late_warn_uninitialized::execute): Likewise.
31493 * tree-ssa.c (verify_vssa): Likewise.
31494 (verify_ssa): Likewise.
31495 (maybe_optimize_var): Likewise.
31496 * tree-vrp.c (dump_value_range): Likewise.
31497 (dump_all_value_ranges): Likewise.
31498 (dump_asserts_for): Likewise.
31499 (register_edge_assert_for_2): Likewise.
31500 (vrp_visit_cond_stmt): Likewise.
31501 (vrp_visit_switch_stmt): Likewise.
31502 (vrp_visit_stmt): Likewise.
31503 (vrp_visit_phi_node): Likewise.
31504 (simplify_cond_using_ranges_1): Likewise.
31505 (fold_predicate_in): Likewise.
31506 (evrp_dom_walker::before_dom_children): Likewise.
31507 (evrp_dom_walker::push_value_range): Likewise.
31508 (evrp_dom_walker::pop_value_range): Likewise.
31509 (execute_early_vrp): Likewise.
31511 2017-05-16 Richard Biener <rguenther@suse.de>
31513 * dwarf2out.c (loc_list_from_tree_1): Do not create
31514 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31516 2017-05-16 Richard Biener <rguenther@suse.de>
31518 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31520 (note_variable_value_in_expr): If we resolved the decl ref
31521 do not push to the stack.
31523 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31525 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31526 operations in fast-math mode.
31527 (vaddq_f16): Likewise.
31528 (vmul_f16): Likewise.
31529 (vmulq_f16): Likewise.
31530 (vsub_f16): Likewise.
31531 (vsubq_f16): Likewise.
31532 * config/arm/neon.md (add<mode>3): New.
31534 (fma:<VH:mode>3): New. Also remove outdated comment.
31537 2017-05-16 Martin Liska <mliska@suse.cz>
31541 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31542 (odr_types_equivalent_p): Likewise.
31544 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31546 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31548 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31551 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31552 non-interunit SSE move alternatives with '?'.
31553 (zero-extendsidi peephole2): New peephole to skip intermediate
31554 general register in SSE zero-extend sequence.
31556 2017-05-15 Jeff Law <law@redhat.com>
31558 * reorg.c (relax_delay_slots): Create a new variable to hold
31559 the temporary target rather than clobbering TARGET_LABEL.
31561 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31562 missing argument to extract_bit_field call.
31563 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31565 2017-05-15 Martin Liska <mliska@suse.cz>
31568 * gcc.c (process_command): Do not allow empty argument of -o option.
31570 2017-05-15 Renlin Li <renlin.li@arm.com>
31572 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31573 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31574 * config/aarch64/constraints.md (Usf): Add long call check.
31575 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31576 (call_value): Likewise.
31577 (sibcall): Likewise.
31578 (sibcall_value): Likewise.
31580 (call_value_insn): New.
31581 (sibcall_insn): Update rtx pattern.
31582 (sibcall_value_insn): Likewise.
31583 (call_internal): Remove.
31584 (call_value_internal): Likewise.
31585 (sibcall_internal): Likewise.
31586 (sibcall_value_internal): Likewise.
31587 (call_reg): Likewise.
31588 (call_symbol): Likewise.
31589 (call_value_reg): Likewise.
31590 (call_value_symbol): Likewise.
31592 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31595 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31597 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31599 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31600 compatible with CCGOCmode and with CCZmode.
31602 2017-05-14 Martin Sebor <msebor@redhat.com>
31604 PR middle-end/77671
31605 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31606 (gimple_fold_builtin_snprintf): Same.
31607 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31608 (gimple_fold_builtin_snprintf): Same.
31609 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31610 of character types.
31611 (is_call_safe): New function.
31612 (try_substitute_return_value): Call it.
31613 (try_simplify_call): New function.
31614 (pass_sprintf_length::handle_gimple_call): Call it.
31616 2017-05-14 Martin Sebor <msebor@redhat.com>
31618 PR middle-end/80669
31619 * builtins.c (expand_builtin_stpncpy): Simplify.
31621 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31623 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31624 * config/i386/i386.h
31625 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31626 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31627 (struct machine_function): Add new members call_ms2sysv,
31628 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31629 (struct machine_frame_state): New fields sp_realigned and
31630 sp_realigned_offset.
31631 * config/i386/i386.c
31632 (enum xlogue_stub): New enum.
31633 (enum xlogue_stub_sets): New enum.
31634 (class xlogue_layout): New class.
31635 (struct ix86_frame): New fields stack_realign_allocate_offset,
31636 stack_realign_offset and outlined_save_offset. Modify comments to
31637 detail stack layout when using out-of-line stubs.
31638 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31639 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31640 -mcall-ms2sysv-xlogues.
31641 (stub_managed_regs): New static variable.
31642 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31643 registers managed by out-of-line stub.
31644 (disable_call_ms2sysv_xlogues): New function.
31645 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31646 m->call_ms2sysv when appropriate and compute frame layout for
31648 (sp_valid_at, fp_valid_at): New inline functions.
31649 (choose_basereg): New function.
31650 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31652 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31653 Use align parameter of choose_baseaddr to generated aligned SSE movs
31655 (pro_epilogue_adjust_stack): Modify to track
31656 machine_frame_state::sp_realigned.
31657 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31658 (ix86_nsaved_sseregs): Likewise.
31659 (ix86_emit_save_regs): Likewise.
31660 (ix86_emit_save_regs_using_mov): Likewise.
31661 (ix86_emit_save_sse_regs_using_mov): Likewise.
31662 (get_scratch_register_on_entry): Likewise.
31663 (gen_frame_set): New function.
31664 (gen_frame_load): Likewise.
31665 (gen_frame_store): Likewise.
31666 (emit_outlined_ms2sysv_save): Likewise.
31667 (emit_outlined_ms2sysv_restore): Likewise.
31668 (ix86_expand_prologue): Modify stack re-alignment code and call
31669 emit_outlined_ms2sysv_save when appropriate.
31670 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31671 parameter rtx_insn *insn, which allows the function to be used to only
31672 generate the notes.
31673 (ix86_expand_epilogue): Modify validity checks of frame and stack
31674 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31675 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31676 * config/i386/predicates.md
31677 (save_multiple): New predicate.
31678 (restore_multiple): Likewise.
31679 * config/i386/sse.md
31680 (save_multiple<mode>): New pattern.
31681 (save_multiple_realign<mode>): Likewise.
31682 (restore_multiple<mode>): Likewise.
31683 (restore_multiple_and_return<mode>): Likewise.
31684 (restore_multiple_leave_return<mode>): Likewise.
31685 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31687 2017-05-14 Julia Koval <julia.koval@intel.com>
31689 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31690 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31691 (__builtin_ia32_xsetbv): New builtins.
31692 * config/i386/i386.c (ix86_expand_special_args_builtin):
31694 (ix86_expand_builtin): Special expand for new intrinsics.
31695 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31696 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31697 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31699 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31701 * cfganal.c (inverted_post_order_compute): Change argument type
31703 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31704 * df-core.c (rest_of_handle_df_initialize): Adjust.
31705 (rest_of_handle_df_finish): Likewise.
31706 (df_analyze_1): Likewise.
31707 (df_analyze): Likewise.
31708 (loop_inverted_post_order_compute): Change argument to be a vec *.
31709 (df_analyze_loop): Adjust.
31710 (df_get_n_blocks): Likewise.
31711 (df_get_postorder): Likewise.
31712 * df.h (struct df_d): Change field to be a vec.
31713 * lcm.c (compute_laterin): Adjust.
31714 (compute_available): Likewise.
31715 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31716 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31717 * tree-ssa-pre.c (compute_antic): Likewise.
31719 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31721 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31722 (depth_first_search::depth_first_search): Change structure init
31723 function to this constructor.
31724 (depth_first_search::add_bb): Rename function to this member.
31725 (depth_first_search::execute): Likewise.
31726 (flow_dfs_compute_reverse_finish): Adjust.
31728 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31730 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31731 (longest_simple_path): Likewise.
31732 * shrink-wrap.c (spread_components): Likewise.
31733 (disqualify_problematic_components): Likewise.
31734 (emit_common_heads_for_components): Likewise.
31735 (emit_common_tails_for_components): Likewise.
31736 (insert_prologue_epilogue_for_components): Likewise.
31738 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31740 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31743 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31745 * df-core.c (df_set_blocks): Start using auto_bitmap.
31746 (df_compact_blocks): Likewise.
31747 * df-problems.c (df_rd_confluence_n): Likewise.
31748 * df-scan.c (df_insn_rescan_all): Likewise.
31749 (df_process_deferred_rescans): Likewise.
31750 (df_update_entry_block_defs): Likewise.
31751 (df_update_exit_block_uses): Likewise.
31752 (df_entry_block_bitmap_verify): Likewise.
31753 (df_exit_block_bitmap_verify): Likewise.
31754 (df_scan_verify): Likewise.
31755 * lra-constraints.c (lra_constraints): Likewise.
31756 (undo_optional_reloads): Likewise.
31757 (lra_undo_inheritance): Likewise.
31758 * lra-remat.c (calculate_gen_cands): Likewise.
31759 (do_remat): Likewise.
31760 * lra-spills.c (assign_spill_hard_regs): Likewise.
31761 (spill_pseudos): Likewise.
31762 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31763 (bitmap_set_subtract_values): Likewise.
31765 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31767 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31768 management with auto_bitmap.
31769 (fix_inter_tick): Likewise.
31770 (fix_recovery_deps): Likewise.
31771 * ira.c (add_store_equivs): Likewise.
31772 (find_moveable_pseudos): Likewise.
31773 (split_live_ranges_for_shrink_wrap): Likewise.
31774 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31775 (rtx_reuse_manager::seen_def_p): Likewise.
31776 (rtx_reuse_manager::set_seen_def): Likewise.
31777 * print-rtl.h (class rtx_reuse_manager): Likewise.
31779 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31781 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31783 (migrate_btr_def): Likewise.
31784 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31785 * df-core.c (loop_post_order_compute): Likewise.
31786 (loop_inverted_post_order_compute): Likewise.
31787 * hsa-common.h: Likewise.
31788 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31789 * init-regs.c (initialize_uninitialized_regs): Likewise.
31790 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31791 (inline_small_functions): Likewise.
31792 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31793 * ira.c (combine_and_move_insns): Likewise.
31794 (build_insn_chain): Likewise.
31795 * loop-invariant.c (find_invariants): Likewise.
31796 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31797 * predict.c (tree_predict_by_opcode): Likewise.
31798 (predict_paths_leading_to): Likewise.
31799 (predict_paths_leading_to_edge): Likewise.
31800 (estimate_loops_at_level): Likewise.
31801 (estimate_loops): Likewise.
31802 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31803 (spread_components): Likewise.
31804 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31805 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31806 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31807 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31808 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31809 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31810 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31811 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31812 (mark_threaded_blocks): Likewise.
31813 (thread_through_all_blocks): Likewise.
31814 * tree-ssa.c (verify_ssa): Likewise.
31815 (execute_update_addresses_taken): Likewise.
31816 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31818 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31820 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31822 (post_order_compute): Likewise.
31823 (inverted_post_order_compute): Likewise.
31824 (pre_and_rev_post_order_compute_fn): Likewise.
31826 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31828 * genrecog.c (int_set::int_set): Explicitly construct our
31829 auto_vec base class.
31830 * vec.h (auto_vec::auto_vec): New constructor.
31832 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31834 * bitmap.h (class auto_bitmap): New constructor taking
31835 bitmap_obstack * argument.
31837 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31839 * bitmap.h (class auto_bitmap): Change type of m_bits to
31840 bitmap_head, and adjust ctor / dtor and member operators.
31842 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31844 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31845 when returned register mode doesn't match original mode.
31847 2017-05-12 Jeff Law <law@redhat.com>
31848 Jakub Jelinek <jakub@redhat.com>
31850 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31851 we look for cc setter after the compare-elim changes.
31852 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31853 within the vector to match what compare-elim now expects.
31854 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31855 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31857 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31858 after the compare-elim changes.
31859 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31860 the vector to match what compare-elim now expects.
31861 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31862 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31863 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31864 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31865 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31867 * config/visium/visium.c (single_set_and_flags): Fix where
31868 we look for cc setter after the compare-elim changes.
31869 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31870 with the vector to match what compare-elim now expects.
31871 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31872 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31873 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31874 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31875 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31876 (neg<mode>2_insn_set_overflow): Likewise.
31878 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31880 PR middle-end/79794
31881 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31882 maybe_expand_insn call, set ops[0].target. If still set after call,
31883 set alt_rtl. Add extra arg to recursive calls.
31884 (extract_bit_field): Add alt_rtl argument. Pass to
31886 * expmed.h (extract_bit_field): Fix prototype.
31887 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31888 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31889 to extract_bit_field_calls.
31890 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31891 Pass alt_rtl to extract_bit_field calls.
31892 * calls.c (store_unaligned_arguments_into_psuedos)
31893 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31894 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31896 * optabs.h (struct expand_operand): Add target bitfield.
31898 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31900 * compare-elim.c (try_eliminate_compare): Canonicalize
31901 operation with embedded compare to
31902 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31903 (set (reg) (operation)].
31905 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31907 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31910 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31911 cost of adding a carry flag for ADC instruction.
31912 [case MINUS]: Ignore the cost of subtracting a carry flag
31913 for SBB instruction.
31915 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31917 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31919 * config/rs6000/bmiintrin.h: New file.
31920 * config/rs6000/bmi2intrin.h: New file.
31921 * config/rs6000/x86intrin.h: New file.
31923 2017-05-12 Jeff Law <law@redhat.com>
31925 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31928 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31930 PR middle-end/80707
31931 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31933 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31935 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31936 early expansion of vector divide builtins.
31937 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31938 builtins identified as having unsigned arguments.
31940 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31942 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31943 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31944 expansion of vector logical operations (and, andc, or, xor,
31947 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31949 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31950 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31952 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31954 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31955 early GIMPLE expansion of vector multiplies.
31957 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31959 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31960 TARGET_HAVE_MOVT conditional.
31961 (movt splitter): Likewise.
31963 2017-05-12 Richard Biener <rguenther@suse.de>
31965 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31966 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31967 Fold all stmts not inplace.
31969 2017-05-12 Richard Biener <rguenther@suse.de>
31971 PR tree-optimization/80713
31972 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31973 inserted_exprs bit for not removed stmts.
31975 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31977 PR middle-end/69921
31978 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31979 parallelized" attribute for parallelized OpenACC kernels.
31980 * omp-offload.c (execute_oacc_device_lower): Use it.
31982 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31983 Set "oacc kernels" attribute.
31984 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31985 parameter. Adjust all users.
31986 (oacc_fn_attrib_kernels_p): Remove function.
31987 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31988 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31989 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31990 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31991 assert "oacc kernels" attribute is set.
31993 2017-05-11 Carl Love <cel@us.ibm.com>
31995 * config/rs6000/rs6000-c: Add support for built-in functions
31996 vector unsigned char vec_popcnt (vector signed char)
31997 vector unsigned char vec_popcnt (vector unsigned char)
31998 vector unsigned short vec_popcnt (vector signed short)
31999 vector unsigned short vec_popcnt (vector unsigned short)
32000 vector unsigned int vec_popcnt (vector signed int)
32001 vector unsigned int vec_popcnt (vector unsigned int)
32002 vector unsigned long long vec_popcnt (vector signed long long)
32003 vector unsigned long long vec_popcnt (vector unsigned long long)
32004 vector signed long long vec_slo (vector signed long long,
32005 vector signed char)
32006 vector signed long long vec_slo (vector signed long long,
32007 vector unsigned char)
32008 vector unsigned long long vec_slo (vector unsigned long long,
32009 vector signed char)
32010 vector unsigned long long vec_slo (vector unsigned long long,
32011 vector unsigned char)
32012 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
32013 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
32014 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
32015 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
32016 * doc/extend.texi: Update the built-in documentation file for the
32017 new built-in functions.
32019 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
32021 * attribs.h (sorted_attr_string): Move machine independent
32022 functions for target clone support from the i386 port to common
32023 code. Rename ix86_function_versions to common_function_versions.
32024 Rename make_name to make_unique_name.
32025 (common_function_versions): Likewise.
32026 (make_unique_name): Likewise.
32027 (make_dispatcher_decl): Likewise.
32028 (is_function_default_version): Likewise.
32029 * attribs.c (attr_strcmp): Likewise.
32030 (sorted_attr_string): Likewise.
32031 (common_function_versions): Likewise.
32032 (make_unique_name): Likewise.
32033 (make_dispatcher_decl): Likewise.
32034 (is_function_default_version): Likewise.
32035 * config/i386/i386.c (attr_strcmp): Likewise.
32036 (sorted_attr_string): Likewise.
32037 (ix86_function_versions): Likewise.
32038 (make_name): Likewise.
32039 (make_dispatcher_decl): Likewise.
32040 (is_function_default_version): Likewise.
32041 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
32043 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32046 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
32047 Account for direct move costs for vec_construct of integer
32050 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
32053 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
32054 (UNSPEC_STX_ATOMIC): Ditto.
32055 (loaddi_via_sse): New insn.
32056 (storedi_via_sse): Ditto.
32057 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
32058 Update corresponding peephole2 patterns.
32059 (atomic_storedi_fpu): Ditto.
32061 2017-05-11 Julia Koval <julia.koval@intel.com>
32063 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
32064 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
32066 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
32067 (__builtin_ia32_rsqrt14ss_mask): New builtins.
32068 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
32070 2017-05-11 Nathan Sidwell <nathan@acm.org>
32072 * graphite-poly.c: Include dumpfile.h.
32074 * dumpfle.h (dump_function): Declare here ...
32075 * tree-dump.h (dump_function): ... not here.
32076 * dumpfile.c: #include tree-cfg.h.
32077 (dump_function): Move here from ...
32078 * tree-dump.c (dump_function): ... here.
32079 * gimplify.c: #include splay-tree.h, not tree-dump.h.
32080 * graphite-poly.c: Don't include tree-dump.h.
32081 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
32082 * print-tree.c: Likewise.
32083 * stor-layout.c: Likewise.
32084 * tree-nested.c: Likewise.
32086 * dumpfile.c (dump_start): Use TDF_FLAGS.
32087 (dump_enable_all): Fix TDF_KIND check thinko.
32089 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
32091 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32092 array entries to represent two legal parameterizations of the
32093 overloaded __builtin_cmpb function, as represented by the
32094 P6_OV_BUILTIN_CMPB constant.
32095 (altivec_resolve_overloaded_builtin): Add special case handling
32096 for the __builtin_cmpb function, as represented by the
32097 P6_OV_BUILTIN_CMPB constant.
32098 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
32099 (BU_P6_64BIT_2): New macro.
32100 (BU_P6_OVERLOAD_2): New macro
32101 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
32102 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
32103 (CMPB): Add overload support to represent both 32-bit and 64-bit
32104 compare-bytes function.
32105 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
32106 support for TARGET_CMPB.
32107 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
32108 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32109 documentation of the __builtin_cmpb overloaded built-in function.
32111 2017-05-11 Richard Biener <rguenther@suse.de>
32113 PR tree-optimization/80705
32114 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
32115 bases are not vectorizable.
32117 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32119 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
32120 when counting register pressure.
32122 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32124 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
32125 (iv_ca_more_deps): Renamed to ...
32126 (iv_ca_compare_deps): ... this.
32127 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
32129 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32131 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
32133 (determine_group_iv_costs): ... here.
32134 (find_inv_vars_cb): Record inv var if it's not recorded before.
32136 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32138 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
32139 (get_shiftadd_cost): Ditto.
32141 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32143 * tree-ssa-address.c: Include header file.
32144 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
32146 (add_to_parts): Refactor.
32147 (addr_to_parts): New parameter. Update use of move_hint_to_base.
32148 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
32151 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32153 PR tree-optimization/53090
32154 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
32156 (extract_cond_operands): Detect condition with IV on both sides
32157 and return COMP_IV_EXPR_2.
32158 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
32159 (rewrite_use_compare): Simplify by removing call to function
32160 extract_cond_operands.
32162 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32164 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32165 (extract_cond_operands): Detect condition comparing against non-
32166 invariant bound and return appropriate enum value.
32167 (find_interesting_uses_cond): Update use of extract_cond_operands.
32168 Handle its return value accordingly.
32169 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32171 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32173 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32174 nonlinear iv_use computation in loop invariant sensitive way.
32176 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32178 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32179 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32181 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32183 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32184 (dump_cand): Support iv_cand.inv_exprs.
32185 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32187 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32190 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32192 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32194 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32195 as local function. Include necessary header files.
32196 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32198 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32200 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32202 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32204 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32205 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32206 RSHIFT_EXPR and BIT_NOT_EXPR.
32208 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32210 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32211 (adjust_setup_cost): New parameter supporting round up adjustment.
32212 (struct address_cost_data): Delete.
32213 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32214 (split_address_cost, ptr_difference_cost): Delete.
32215 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32216 (struct ainc_cost_data): New struct.
32217 (get_address_cost_ainc): New function.
32218 (get_address_cost, get_computation_cost): Reimplement.
32219 (determine_group_iv_cost_address): Record inv_expr for all uses of
32221 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32222 (iv_ca_has_deps): Reimplemented to ...
32223 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32225 (iv_ca_extend): Call iv_ca_more_deps.
32227 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32229 * tree-ssa-address.c (struct mem_address): Move to header file.
32230 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32231 * tree-ssa-address.h (struct mem_address): Move from C file.
32232 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32234 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32236 * tree-affine.h (aff_combination_type): New interface.
32237 (aff_combination_zero_p): Remove static.
32238 (aff_combination_const_p): New interface.
32239 (aff_combination_singleton_var_p): New interfaces.
32241 2017-05-11 Richard Biener <rguenther@suse.de>
32243 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32244 Skip unreachable blocks and destinations.
32245 (eliminate): Move stmt removal and fixup ...
32246 (fini_eliminate): ... here. Skip inserted exprs.
32247 (pass_pre::execute): Move fini_pre after fini_eliminate.
32248 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32249 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32250 PRE to get rid of dead code that has invalid SSA form and
32251 split critical edges again.
32253 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32255 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32257 2017-05-11 Richard Biener <rguenther@suse.de>
32259 * passes.c (execute_function_todo): Verify loops if they are
32260 said to be up-to-date.
32261 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32262 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32264 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32267 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32268 handle calling assemble_external ourself.
32271 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32272 modes with zero size. Enhance comment.
32274 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32276 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32277 built-ins for vec_xl and vec_xst with short and char pointer
32280 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32282 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32283 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32284 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32285 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32286 (_mm_maskz_min_round_ss): New intrinsics.
32287 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32288 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32289 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32290 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32291 (__builtin_ia32_minss_mask_round): New builtins.
32292 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32293 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32294 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32296 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32297 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32299 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32302 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32304 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32305 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32306 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32307 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32308 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32309 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32310 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32311 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32312 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32313 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32314 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32315 (__builtin_ia32_mulss_mask_round): New builtins.
32316 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32317 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32318 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32320 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32321 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32323 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32326 2017-05-10 Julia Koval <julia.koval@intel.com>
32328 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32329 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32330 (_mm256_setr_m128i): New intrinsics.
32332 2017-05-10 Julia Koval <julia.koval@intel.com>
32334 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32335 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32336 (_mm_maskz_rcp14_ss): New intrinsics.
32337 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32338 (__builtin_ia32_rcp14ss_mask): New builtins.
32339 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32341 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32343 PR tree-optimization/51513
32344 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32345 (assert_unreachable_fallthru_edge_p): Use it.
32346 (group_case_labels_stmt): Likewise.
32347 * tree-cfg.h: Prototype it.
32348 * stmt.c: Include cfghooks.h and tree-cfg.h.
32349 (emit_case_dispatch_table) <gap_label>: New local variable.
32350 Use it to fill dispatch table gaps.
32351 Test for default_label before updating probabilities.
32352 (expand_case) <default_label>: Remove unneeded initialization.
32353 Test for unreachable default case statement and remove its edge.
32354 Set default_label accordingly.
32355 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32357 2017-05-10 Carl Love <cel@us.ibm.com>
32359 * config/rs6000/rs6000-c: Add support for built-in functions
32360 vector signed char vec_neg (vector signed char)
32361 vector signed short int vec_neg (vector short int)
32362 vector signed int vec_neg (vector signed int)
32363 vector signed long long vec_neg (vector signed long long)
32364 vector float vec_neg (vector float)
32365 vector double vec_neg (vector double)
32366 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32368 * config/rs6000/altivec.h: Add define for vec_neg
32369 * doc/extend.texi: Update the built-in documentation for the
32370 new built-in functions.
32372 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32374 PR tree-optimization/77644
32375 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32377 2017-05-10 Nathan Sidwell <nathan@acm.org>
32379 * dumpfile.h (TDI_lang_all): New.
32380 (TDF_KIND): New. Renumber others
32381 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32383 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32385 (get_dump_file_name): Adjust suffix generation.
32386 (dump_enable_all): Use TDF_KIND.
32387 * doc/invoke.texi (-fdump-lang-all): Document.
32389 * dumpfile.h: Tabify.
32391 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32394 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32395 Move member access before delete.
32397 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32399 * tree-inline.c (expand_call_inline): Split block at stmt
32402 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32405 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32406 are now unused after splitting mov{sf,sd}_hardfloat.
32407 (f32_lr2): Likewise.
32408 (f32_lm): Likewise.
32409 (f32_lm2): Likewise.
32410 (f32_li): Likewise.
32411 (f32_li2): Likewise.
32412 (f32_lv): Likewise.
32413 (f32_sr): Likewise.
32414 (f32_sr2): Likewise.
32415 (f32_sm): Likewise.
32416 (f32_sm2): Likewise.
32417 (f32_si): Likewise.
32418 (f32_si2): Likewise.
32419 (f32_sv): Likewise.
32420 (f32_dm): Likewise.
32421 (f32_vsx): Likewise.
32422 (f32_av): Likewise.
32423 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32424 For movsf, order stores so the VSX stores occur before the GPR
32425 store which encourages the register allocator to use a traditional
32426 FPR instead of a GPR. For movsd, order the stores so that the GPR
32427 store comes before the VSX stores to allow the power6 to work.
32428 This is due to the power6 not having a 32-bit integer store
32429 instruction from a FPR.
32430 (movsf_hardfloat): Likewise.
32431 (movsd_hardfloat): Likewise.
32433 2017-05-09 Martin Sebor <msebor@redhat.com>
32435 PR translation/80280
32436 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32439 PR translation/80280
32440 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32441 data member added in r247778.
32442 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32444 2017-05-09 Nathan Sidwell <nathan@acm.org>
32446 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32448 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32451 2017-05-09 Marek Polacek <polacek@redhat.com>
32453 * doc/invoke.texi: Fix typo.
32455 2017-05-09 Richard Biener <rguenther@suse.de>
32457 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32458 (vrp_val_is_min): Likewise.
32459 (set_value_range_to_value): Likewise.
32460 (set_value_range_to_nonnegative): Likewise.
32461 (gimple_assign_nonzero_p): Likewise.
32462 (gimple_stmt_nonzero_p): Likewise.
32463 (vrp_int_const_binop): Likewise. Remove unreachable case.
32464 (adjust_range_with_scev): Adjust comments.
32465 (compare_range_with_value): Likewise.
32466 (extract_range_from_phi_node): Likewise.
32467 (test_for_singularity): Likewise.
32469 2017-05-09 Richard Biener <rguenther@suse.de>
32471 * tree-vrp.c (get_single_symbol): Add assert that we don't
32472 get overflowed constants as invariant part.
32473 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32474 checks. Use wi::cmp instead of recursing for integer constants.
32475 (compare_values): Just ignore whether we assumed undefined
32476 overflow instead of failing the compare.
32477 (extract_range_for_var_from_comparison_expr): Add comment before the
32478 TREE_NO_WARNING sets.
32479 (test_for_singularity): Likewise.
32480 (extract_range_from_comparison): Do not disable optimization
32481 when we assumed undefined overflow.
32482 (extract_range_basic): Remove init of unused var.
32484 2017-05-09 Richard Biener <rguenther@suse.de>
32486 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32487 (extract_range_from_multiplicative_op_1): Adjust.
32488 (extract_range_from_binary_expr_1): Use int_const_binop.
32490 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32493 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32494 rs6000_store_data_bypass_p in seven define_bypass directives and
32495 in several comments.
32496 * config/rs6000/rs6000-protos.h: Add prototype for
32497 rs6000_store_data_bypass_p function.
32498 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32499 function implements slightly different (rs6000-specific) semantics
32500 than store_data_bypass_p, returning false rather than aborting
32501 with assertion error when arguments do not satisfy the
32502 requirements of store data bypass.
32503 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32504 rs6000_store_data_bypass_p.
32506 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32508 * config/xtensa/xtensa-protos.h
32509 (xtensa_initial_elimination_offset): New declaration.
32510 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32511 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32512 macro definition, add case for FRAME_POINTER_REGNUM when
32513 FRAME_GROWS_DOWNWARD.
32514 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32515 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32516 xtensa_initial_elimination_offset.
32518 2017-05-08 Nathan Sidwell <nathan@acm.org>
32520 * doc/invoke.texi: Alphabetize -fdump options.
32522 2017-05-08 Martin Sebor <msebor@redhat.com>
32524 PR translation/80280
32525 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32527 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32529 * target.def (compute_frame_layout): New optional target hook.
32530 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32531 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32532 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32534 * reload1.c (verify_initial_elim_offsets): Likewise.
32535 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32536 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32537 (arm_get_frame_offsets): Split up into this ...
32538 (arm_compute_frame_layout): ... and this function.
32540 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32542 * config/aarch64/constraints.md (Usa): New constraint.
32543 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32545 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32547 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32548 with_multilib_list after it has been checked.
32550 2017-05-08 Richard Biener <rguenther@suse.de>
32552 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32553 (bitmap_set_subtract_values): Likewise.
32555 2017-05-08 Richard Biener <rguenther@suse.de>
32557 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32558 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32560 (gimple_stmt_nonzero_warnv_p): Rename to ...
32561 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32563 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32564 (extract_range_basic): Adjust, do not disable propagation on
32565 strict overflow sensitive simplification.
32566 (vrp_visit_cond_stmt): Likewise.
32568 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32570 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32571 body size unconditionally.
32573 2017-05-07 Jeff Law <law@redhat.com>
32576 2017-05-06 Jeff Law <law@redhat.com>
32577 PR tree-optimization/78496
32578 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32581 PR tree-optimization/78496
32582 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32583 (simplify_stmt_using_ranges): Call it.
32584 (vrp_dom_walker::before_dom_children): Extract equivalences
32585 from an ASSERT_EXPR with an equality comparison against a
32588 2017-05-06 Jeff Law <law@redhat.com>
32590 PR tree-optimization/78496
32591 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32594 PR tree-optimization/78496
32595 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32596 (simplify_stmt_using_ranges): Call it.
32597 (vrp_dom_walker::before_dom_children): Extract equivalences
32598 from an ASSERT_EXPR with an equality comparison against a
32601 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32603 * lra-constraints.c (lra_copy_reg_equiv): New function.
32604 (split_reg): Use it to copy equivalence information from the
32605 original register to the spill register.
32607 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32609 PR rtl-optimization/75964
32610 * simplify-rtx.c (simplify_const_relational_operation): Remove
32611 invalid handling of comparisons of integer ABS.
32613 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32615 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32616 initialize to zero.
32617 (init_regs): Remove declaration.
32618 (function_arg_advance_32): Initialize error_p as boolean variable.
32620 2017-05-05 Nathan Sidwell <nathan@acm.org>
32622 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32623 lines. Use for (;;).
32625 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32627 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32628 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32629 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32630 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32631 VF=2 that require versioning.
32633 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32635 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32638 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32640 * diagnostic.h (diagnostic_override_option_index): Convert from
32641 macro to inline function.
32643 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32645 * diagnostic.c (last_module_changed_p): New function.
32646 (set_last_module): New function.
32647 (diagnostic_report_current_module): Convert macro usage to
32648 the above functions.
32649 * diagnostic.h (diagnostic_context::last_module): Strengthen
32650 from const line_map * to const line_map_ordinary *.
32651 (diagnostic_last_module_changed): Delete macro.
32652 (diagnostic_set_last_module): Delete macro.
32654 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32656 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32657 with diagnostic_report_diagnostic.
32658 (diagnostic_n_impl_richloc): Likewise.
32659 * diagnostic.h (report_diagnostic): Delete macro.
32660 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32661 with diagnostic_report_diagnostic.
32662 * substring-locations.c (format_warning_va): Likewise.
32664 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32666 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32667 save/restor of format_spec. Move option-printing code to...
32668 (print_option_information): ...this new function, and
32669 reimplement by simply printing to the pretty_printer,
32670 rather than appending to the format string.
32672 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32674 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32675 handling logic into...
32676 (update_effective_level_from_pragmas): ...this new function.
32678 2017-05-04 Andrew Waterman <andrew@sifive.com>
32680 * config/riscv/riscv.opt (mstrict-align): New option.
32681 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32682 (SLOW_UNALIGNED_ACCESS): Define.
32683 (riscv_slow_unaligned_access): Declare.
32684 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32686 (riscv_slow_unaligned_access): New variable.
32687 (rocket_tune_info): Set slow_unaligned_access to true.
32688 (optimize_size_tune_info): Set slow_unaligned_access to false.
32689 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32690 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32691 (riscv_option_override): Set riscv_slow_unaligned_access.
32692 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32694 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32696 * config/riscv/riscv.md: Unify indentation.
32698 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32703 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32705 (extendsi<mode>2): Add support for doing sign extension via
32706 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32707 don't have ISA 3.0 instructions.
32708 (extendsi<mode>2 splitter): Likewise.
32709 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32710 generate the normal insns since SImode can now go in vector
32711 registers. Disallow the special UNSPECs needed for previous
32712 machines to hide SImode being used. Add new insns
32713 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32714 (fix_trunc<mode>si2_stfiwx): Likewise.
32715 (fix_trunc<mode>si2_internal): Likewise.
32716 (fixuns_trunc<mode>si2): Likewise.
32717 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32718 (fctiw<u>z_<mode>_smallint): Likewise.
32719 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32720 of floating point to 32-bit integer from doing a direct move to
32721 the GPR registers to do a store.
32722 (fctiwz_<mode>): Break long line.
32724 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32726 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32727 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32728 (addr_list, addr_offset_valid_p): New.
32729 (split_address_groups): Check offset validity with above function.
32730 (gt-tree-ssa-loop-ivopts.h): Include header file.
32732 2017-05-05 Nathan Sidwell <nathan@acm.org>
32734 * config.gcc (arm*-*-*): Add missing 'fi'.
32736 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32738 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32739 included in -fopt-info does not matter.
32740 * doc/optinfo.texi (-fopt-info): Fix description of default
32741 behavour. Explicitly say order of options included in -fopt-info
32744 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32746 * config.gcc: Allow combinations of aprofile and rmprofile values for
32747 --with-multilib-list.
32748 * config/arm/t-multilib: New file.
32749 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32750 variables. Remove setting of ISA and floating-point ABI in
32751 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32752 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32753 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32754 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32756 * config/arm/t-rmprofile: Likewise except for the matches changes.
32757 * doc/install.texi (--with-multilib-list): Document the combination of
32758 aprofile and rmprofile values and warn about pitfalls in doing that.
32760 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32762 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32763 (movdi_aarch64): Likewise.
32765 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32767 PR tree-optimization/80632
32768 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32770 (build_arrays): Initialize it for virtual phis.
32771 (fix_phi_nodes): Use it for virtual phis.
32773 PR tree-optimization/80558
32774 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32775 [x, y] op z into [x op, y op z] for op & or | if conditions
32778 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32779 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32782 * config/arm/arm.md (use_literal_pool): Remove.
32783 (64-bit immediate split): No longer takes cost into consideration
32784 if arm_disable_literal_pool is enabled.
32785 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32786 used when arm_disable_literal_pool is enabled.
32787 (arm_max_const_double_inline_cost): Remove use of
32788 arm_disable_literal_pool.
32789 (push_minipool_fix): Add assert.
32790 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32791 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32792 (no_literal_pool_sf_immediate): New.
32794 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32796 PR tree-optimization/80613
32797 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32798 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32800 2017-05-05 Richard Biener <rguenther@suse.de>
32802 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32804 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32806 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32807 of this flag from insn conditions due to removal from r247495.
32809 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32811 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32813 (arm_early_store_addr_dep_ptr): Likewise.
32814 * config/arm/aarch-common-protos.h
32815 (arm_early_load_addr_dep_ptr): Add prototype.
32816 (arm_early_store_addr_dep_ptr): Likewise.
32817 * config/arm/cortex-a53.md: Add new bypasses.
32819 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32821 * tree.c (next_type_uid): Change type to unsigned.
32822 (type_hash_canon): Decrement back next_type_uid if
32823 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32824 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32827 2017-05-04 Martin Sebor <msebor@redhat.com>
32829 * builtins.c: Fix a trivial typo in a comment.
32831 PR middle-end/79234
32832 * builtins.c (check_sizes): Adjust to handle reading past the end.
32833 Avoid printing excessive upper bound of ranges. Use %E to print
32834 tree nodes instead of converting them to %wu.
32835 (expand_builtin_memchr): New function.
32836 (compute_dest_size): Rename...
32837 (compute_objsize): ...to this.
32838 (expand_builtin_memcpy): Adjust.
32839 (expand_builtin_mempcpy): Adjust.
32840 (expand_builtin_strcat): Adjust.
32841 (expand_builtin_strcpy): Adjust.
32842 (check_strncat_sizes): Adjust.
32843 (expand_builtin_strncat): Adjust.
32844 (expand_builtin_strncpy): Adjust and simplify.
32845 (expand_builtin_memset): Adjust.
32846 (expand_builtin_bzero): Adjust.
32847 (expand_builtin_memcmp): Adjust.
32848 (expand_builtin): Handle memcmp.
32849 (maybe_emit_chk_warning): Check strncat just once.
32851 2017-05-04 Martin Sebor <msebor@redhat.com>
32853 PR preprocessor/79214
32854 PR middle-end/79222
32855 PR middle-end/79223
32856 * builtins.c (check_sizes): Add inlining context and issue
32857 warnings even when -Wno-system-headers is set.
32858 (check_strncat_sizes): Same.
32859 (expand_builtin_strncat): Same.
32860 (expand_builtin_memmove): New function.
32861 (expand_builtin_stpncpy): Same.
32862 (expand_builtin): Handle memmove and stpncpy.
32864 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32866 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32867 which is not used any more.
32869 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32871 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32873 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32875 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32876 (cortexa53_tunings): Likewise.
32877 (cortexa57_tunings): Likewise.
32878 (cortexa72_tunings): Likewise.
32879 (cortexa73_tunings): Likewise.
32881 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32883 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32884 Set loop alignment to 8.
32886 2017-05-04 Martin Sebor <msebor@redhat.com>
32888 PR translation/80280
32889 * builtins.c (expand_builtin_object_size): Add missing quoting to
32890 %D and like directives.
32891 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32892 (hsa_type_for_tree_type): Same.
32893 (verify_function_arguments): Same.
32894 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32895 * varasm.c (get_section): Same.
32898 2017-05-04 Martin Sebor <msebor@redhat.com>
32900 PR translation/80280
32901 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32903 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32905 * config/aarch64/aarch64.c (generic_addrcost_table):
32906 Change HI/TI mode setting.
32908 2017-05-04 Martin Jambor <mjambor@suse.cz>
32910 PR tree-optimization/80622
32911 * tree-sra.c (comes_initialized_p): New function.
32912 (build_accesses_from_assign): Only set write lazily when
32913 comes_initialized_p is false.
32914 (analyze_access_subtree): Use comes_initialized_p.
32915 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32916 instead of testing for PARM_DECL.
32918 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32920 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32921 constraint on operand 0 to allow more general addressing modes.
32922 Adjust output template.
32923 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32925 * config/aarch64/aarch64-protos.h
32926 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32927 * config/aarch64/constraints.md (Dp): New address constraint.
32928 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32931 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32933 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32934 update use of estimate_ipcp_clone_size_and_time.
32935 (estimate_local_effects): Update use of
32936 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32937 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32938 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32939 Return nonspecialized time.
32941 2017-05-04 Richard Biener <rguenther@suse.de>
32943 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32944 for the last VUSE which def dominates the PHI. Directly call
32946 (get_continuation_for_phi_1): Remove.
32948 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32950 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32951 to explain the use of truncating division. Cap the number of
32952 iterations to the maximum given by nb_iterations_upper_bound,
32955 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32957 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32958 * configure: Regenerate.
32959 * config.in: Regenerate.
32960 * config/i386/driver-mingw32.c: new file.
32961 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32962 * config.host: Link driver-mingw32.o on MinGW host.
32963 * doc/install.texi: Document new --enable-mingw-wildcard configure
32966 2017-05-04 Marek Polacek <polacek@redhat.com>
32968 PR tree-optimization/80612
32969 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32971 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32972 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32974 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32975 (movt splitter): Likewise.
32976 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32977 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32978 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32979 block for Thumb-1 with MOVT.
32980 (thumb2_legitimate_address_p): Move code block ...
32981 (can_avoid_literal_pool_for_label_p): ... into this new function.
32982 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32984 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32985 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32986 "M-profile targets with the MOVT instruction".
32988 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32990 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32991 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32992 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32994 2017-05-04 Martin Liska <mliska@suse.cz>
32996 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32997 variable cond_code.
32999 2017-05-04 Richard Biener <rguenther@suse.de>
33001 * tree.c (array_at_struct_end_p): Handle arrays at struct
33002 end with flexarrays more conservatively. Refactor and treat
33003 arrays of arrays or aggregates more strict. Fix
33004 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
33005 * tree.c (array_at_struct_end_p): Adjust prototype.
33006 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
33007 * gimple-fold.c (get_range_strlen): Likewise.
33008 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
33010 2017-05-04 Richard Biener <rguenther@suse.de>
33012 PR tree-optimization/31130
33013 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
33015 (supports_overflow_infinity): Likewise.
33016 (is_negative_overflow_infinity): Likewise.
33017 (is_positive_overflow_infinity): Likewise.
33018 (is_overflow_infinity): Likewise.
33019 (stmt_overflow_infinity): Likewise.
33020 (overflow_infinity_range_p): Likewise.
33021 (usable_range_p): Remove as always returning true.
33022 (make_overflow_infinity): Remove.
33023 (negative_overflow_infinity): Likewise.
33024 (positive_overflow_infinity): Likewise.
33025 (avoid_overflow_infinity): Likewise.
33026 (set_value_range): Adjust accordingly.
33027 (set_value_range_to_nonnegative): Likewise, remove now unused
33028 overflow_infinity arg.
33029 (vrp_operand_equal_p): Adjust.
33030 (update_value_range): Likewise.
33031 (range_int_cst_singleton_p): Likewise.
33032 (operand_less_p): Likewise.
33033 (compare_values_warnv): Likewise.
33034 (extract_range_for_var_from_comparison_expr): Likewise.
33035 (vrp_int_const_binop): Likewise.
33036 (zero_nonzero_bits_from_vr): Likewise.
33037 (extract_range_from_multiplicative_op_1): Likewise.
33038 (extract_range_from_binary_expr_1): Likewise.
33039 (extract_range_from_unary_expr): Likewise.
33040 (extract_range_from_comparison): Likewise.
33041 (extract_range_basic): Likewise.
33042 (adjust_range_with_scev): Likewise.
33043 (compare_ranges): Likewise.
33044 (compare_range_with_value): Likewise.
33045 (dump_value_range): Likewise.
33046 (test_for_singularity): Likewise, remove strict_overflow_p parameter
33048 (simplify_cond_using_ranges): Adjust.
33050 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33052 * brig-builtins.def: Added a builtin for class_f64.
33053 * builtin-types.def: Added a builtin type needed by class_f64.
33055 2017-05-03 Jason Merrill <jason@redhat.com>
33057 * timevar.def: Add TV_CONSTEXPR.
33059 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33061 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
33063 2017-05-03 Martin Jambor <mjambor@suse.cz>
33065 * ipa-prop.c (ipa_update_after_lto_read): Removed.
33066 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
33067 * ipa-cp.c (ipcp_propagate_stage): Do not call
33068 ipa_update_after_lto_read.
33069 * ipa-inline.c (ipa_inline): Likewise.
33071 2017-05-03 Martin Jambor <mjambor@suse.cz>
33073 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
33074 tag. Added a default constructor and a destructor.
33075 (ipa_edge_args_sum_t): New class;
33076 (ipa_edge_args_sum): Declare.
33077 (ipa_edge_args_vector): Remove declaration.
33078 (IPA_EDGE_REF): Use ipa_edge_args_sum.
33079 (ipa_free_edge_args_substructures): Remove declaration.
33080 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
33081 (ipa_edge_args_info_available_for_edge_p): Likewise.
33082 * ipa-prop.c (ipa_edge_args_vector): Removed.
33083 (edge_removal_hook_holder): Likewise.
33084 (edge_duplication_hook_holder): Likewise.
33085 (ipa_edge_args_sum): New variable.
33086 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
33087 ipa_edge_args_vector.
33088 (ipa_free_edge_args_substructures): Likewise.
33089 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
33090 ipa_edge_args_vector.
33091 (ipa_edge_removal_hook): Turned into method
33092 ipa_edge_args_sum_t::remove.
33093 (ipa_edge_duplication_hook): Turned into method
33094 ipa_edge_args_sum_t::duplicate.
33095 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
33096 registering edge hooks.
33097 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
33098 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
33099 ipa_edge_args_sum instead of ipa_edge_args_vector.
33100 * ipa-profile.c (ipa_profile): Likewise.
33102 2017-05-03 Martin Jambor <mjambor@suse.cz>
33104 * symbol-summary.h (function_summary): New method exists.
33105 (function_summary::symtab_removal): Deallocate through release.
33106 (call_summary): New class.
33107 (gt_ggc_mx): New overload.
33108 (gt_pch_nx): Likewise.
33109 (gt_pch_nx): Likewise.
33111 2017-05-03 Jeff Law <law@redhat.com>
33113 PR tree-optimization/78496
33114 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
33115 from simplify_cond_using_ranges. Split off code to walk
33116 backwards through casts into ...
33117 (simplify_cond_using_ranges_2): New function.
33118 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
33119 (execute_vrp): After identifying jump threads, call
33120 simplify_cond_using_ranges_2.
33122 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
33125 * ipa-inline.h (inline_summary): Add ctor.
33126 (create_ggc): Do not use ggc_cleared_alloc.
33128 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
33129 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33131 * gcc.c (handle_braces): Support escaping in switch matching
33133 * doc/invoke.texi (Spec Files): Document it.
33134 Remove superfluous @code markup in items.
33136 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33138 * diagnostic-show-locus.c (struct column_range): New struct.
33139 (get_affected_columns): New function.
33140 (get_printed_columns): New function.
33141 (struct correction): New struct.
33142 (correction::ensure_capacity): New function.
33143 (correction::ensure_terminated): New function.
33144 (struct line_corrections): New struct.
33145 (line_corrections::~line_corrections): New dtor.
33146 (line_corrections::add_hint): New function.
33147 (layout::print_trailing_fixits): Reimplement in terms of the new
33149 (selftest::test_overlapped_fixit_printing): New function.
33150 (selftest::diagnostic_show_locus_c_tests): Call it.
33152 2017-05-03 Nathan Sidwell <nathan@acm.org>
33154 Canonicalize canonical type hashing
33155 * tree.h (type_hash_canon_hash): Declare.
33156 * tree.c (type_hash_list, attribute_hash_list): Move into
33157 type_hash_canon_hash.
33158 (build_type_attribute_qual_variant): Break out hash code calc into
33159 type_hash_canon_hash.
33160 (type_hash_canon_hash): New. Generic type hash computation.
33161 (build_range_type_1, build_array_type_1, build_function_type,
33162 build_method_type_directly, build_offset_type, build_complex_type,
33163 make_vector_type): Call it.
33165 2017-05-03 Richard Biener <rguenther@suse.de>
33167 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33168 When all DRs have unknown misaligned do not always peel
33169 when there is a store but apply the same costing model as if
33170 there were only loads.
33172 2017-05-03 Richard Biener <rguenther@suse.de>
33175 PR tree-optimization/80492
33176 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33177 compare_base_decls returning dont-know properly.
33179 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33181 * config/arm/iterators.md (CCSI): New mode iterator.
33182 (arch): New mode attribute.
33183 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33184 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33185 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33186 code iterator for success result mode.
33187 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33188 the corresponding new insn generators.
33190 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33193 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33194 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33196 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33198 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33199 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33200 (DDR_A): Wrap DDR argument in brackets.
33201 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33202 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33203 (DDR_REVERSED_P): Likewise.
33205 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33207 PR tree-optimization/79472
33208 * tree-switch-conversion.c (struct switch_conv_info): Add
33209 contiguous_range and default_case_nonstandard fields.
33210 (collect_switch_conv_info): Compute contiguous_range and
33211 default_case_nonstandard fields, don't clear final_bb if
33212 contiguous_range and only the default case doesn't have the required
33214 (check_all_empty_except_final): Set default_case_nonstandard instead
33215 of failing if contiguous_range and the default case doesn't have empty
33217 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33218 and only the default case doesn't have the required constants. Skip
33220 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33221 if default_case_nonstandard.
33222 (build_constructors): Build constant 1 just once. Assert that default
33223 values aren't inserted in between cases if contiguous_range. Skip
33225 (build_arrays): Skip virtual phis.
33226 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33227 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33228 Handle virtual phis.
33229 (gen_inbound_check): Handle default_case_nonstandard case.
33230 (process_switch): Adjust check_final_bb caller. Call
33231 gather_default_values with the first non-default case instead of
33232 default case if default_case_nonstandard.
33234 2017-05-02 Nathan Sidwell <nathan@acm.org>
33236 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33237 check. Fix formatting.
33239 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33241 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33242 errors when comparing specialized and unspecialized times.
33244 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33246 * diagnostic-show-locus.c
33247 (layout::should_print_annotation_line_p): Make private.
33248 (layout::print_annotation_line): Make private.
33249 (layout::annotation_line_showed_range_p): Make private.
33250 (layout::show_ruler): Make private.
33251 (layout::print_source_line): Make private. Pass in line and
33252 line_width, rather than calling location_get_source_line. Drop
33254 (layout::print_leading_fixits): New method.
33255 (layout::print_any_fixits): Rename to...
33256 (layout::print_trailing_fixits): ...this, and make private.
33257 Don't print newline fixits.
33258 (diagnostic_show_locus): Move logic for printing one row into...
33259 (layout::print_line): ...this new function. Move the
33260 location_get_source_line call and error-handling from
33261 print_source_line to here. Call print_leading_fixits, and rename
33262 print_any_fixits to print_trailing_fixits.
33263 (selftest::test_fixit_insert_containing_newline): Update now that
33264 newlines are partially supported.
33265 (selftest::test_fixit_insert_containing_newline_2): New test.
33266 (selftest::test_fixit_replace_containing_newline): Update comments.
33267 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33268 * edit-context.c (class added_line): New class.
33269 (class edited_line): Describe newline handling in comment.
33270 (edited_line::actually_edited_p): New method.
33271 (edited_line::print_content): Delete redundant decl.
33272 (edited_line::m_predecessors): New field.
33273 (edited_file::print_content): Call edited_line::print_content.
33274 (edited_file::print_diff): Update to support newlines.
33275 (edited_file::print_diff_hunk): Likewise.
33276 (edited_file::print_run_of_changed_lines): New function.
33277 (edited_file::print_diff_line): Convert to...
33278 (print_diff_line): ...this.
33279 (edited_file::get_effective_line_count): New function.
33280 (edited_line::edited_line): Initialize new field m_predecessors.
33281 (edited_line::~edited_line): Clean up m_predecessors.
33282 (edited_line::apply_fixit): Handle newlines.
33283 (edited_line::get_effective_line_count): New function.
33284 (edited_line::print_content): New function.
33285 (edited_line::print_diff_lines): New function.
33286 (selftest::test_applying_fixits_insert_containing_newline): New
33288 (selftest::test_applying_fixits_replace_containing_newline): New
33290 (selftest::insert_line): New function.
33291 (selftest::test_applying_fixits_multiple_lines): Add example of
33293 (selftest::edit_context_c_tests): Call the new tests.
33295 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33297 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33298 parameter cand. Update dump information.
33299 (get_computation_cost): Update uses.
33301 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33303 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33304 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33305 (get_computation_at, rewrite_use_address): Update use of
33306 get_computation_aff.
33308 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33310 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33311 (get_computation): Delete.
33312 (get_computation_cost): Implement like get_computation_cost_at.
33313 Use get_computation_at.
33314 (get_computation_cost_at): Delete.
33315 (rewrite_use_nonlinear_expr): Use get_computation_at.
33316 (rewrite_use_compare, remove_unused_ivs): Ditto.
33318 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33320 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33322 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33324 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33325 (ivopts_global_cost_for_size): Rename parameter and update uses.
33326 (iv_ca_recount_cost): Update uses.
33327 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33328 candidates seperately in n_invs and n_cands.
33329 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33331 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33333 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33334 (find_inv_vars_cb): New.
33335 (find_depends): Renamed to ...
33336 (find_inv_vars): ... this.
33337 (add_candidate_1, force_var_cost): Call find_inv_vars.
33338 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33340 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33342 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33343 inv_vars. Add inv_exprs.
33344 (struct iv_cand): Rename depends_on to inv_vars.
33345 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33346 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33347 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33348 (dump_cand): Dump inv_vars.
33349 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33350 (record_invariant, find_depends, add_candidate_1): Ditto.
33351 (set_group_iv_cost, force_var_cost): Ditto.
33352 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33353 (get_computation_cost_at, get_computation_cost): Ditto.
33354 (determine_group_iv_cost_generic): Ditto.
33355 (determine_group_iv_cost_address): Ditto.
33356 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33357 (determine_group_iv_costs): Ditto.
33358 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33359 (iv_ca_set_remove_invariants): Renamed to ...
33360 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33361 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33362 (iv_ca_set_add_invariants): Renamed to ...
33363 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33364 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33365 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33366 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33367 (create_new_ivs): Remove useless dump.
33369 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33371 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33373 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33374 (iv_ca_set_no_cp, create_new_iv): Ditto.
33376 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33378 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33380 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33382 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33383 function tree_check2.
33385 2017-05-02 Martin Liska <mliska@suse.cz>
33387 * doc/gcov.texi: Add missing preposition.
33388 * gcov.c (function_info::function_info): Properly fill up
33389 all member variables.
33391 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33393 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33395 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33397 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33399 2017-05-02 Martin Liska <mliska@suse.cz>
33402 * lto-streamer-in.c (lto_read_tree_1): Remove
33403 LTO_STREAMER_DEBUG.
33404 * lto-streamer.c (struct tree_hash_entry): Likewise.
33405 (struct tree_entry_hasher): Likewise.
33406 (tree_entry_hasher::hash): Likewise.
33407 (tree_entry_hasher::equal): Likewise.
33408 (lto_streamer_init): Likewise.
33409 (lto_orig_address_map): Likewise.
33410 (lto_orig_address_get): Likewise.
33411 (lto_orig_address_remove): Likewise.
33412 * lto-streamer.h: Likewise.
33413 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33414 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33416 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33418 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33419 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33420 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33421 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33422 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33423 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33424 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33425 (mm_maskz_sub_ss): New intrinsics.
33426 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33427 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33428 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33429 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33430 (__builtin_ia32_subss_mask_round): New builtins.
33431 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33432 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33433 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33435 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33436 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33438 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33441 2017-05-02 Martin Jambor <mjambor@suse.cz>
33443 PR tree-optimization/78687
33444 * tree-sra.c (access): New field parent.
33445 (process_subtree_disqualification): New function.
33446 (disqualify_candidate): Call it.
33447 (build_accesses_from_assign): Reset write flag if creating an
33449 (build_access_subtree): Fill in parent field and also prpagate
33450 down grp_write flag.
33451 (create_artificial_child_access): New parameter set_grp_write, set
33452 grp_write to its value.
33453 (propagate_subaccesses_across_link): Also propagate grp_write flag
33455 (propagate_all_subaccesses): Push the closest parent back to work
33456 queue if add_access_to_work_queue returned true.
33458 2017-05-02 Richard Biener <rguenther@suse.de>
33460 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33461 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33462 -fstrict-overflow documentation.
33463 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33464 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33465 flag_strict_overflow.
33466 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33467 * lto-opts.c (lto_write_options): Do not stream it.
33468 * lto-wrapper.c (merge_and_complain): Do not handle it.
33469 * opts.c (default_options_table): Do not set -fstrict-overflow.
33470 (finish_options): Likewise do not clear it when sanitizing.
33471 * simplify-rtx.c (simplify_const_relational_operation): Do not
33472 test flag_strict_overflow.
33474 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33476 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33477 using enabled attribute.
33478 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33479 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33480 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33481 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33482 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33483 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33484 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33485 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33486 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33487 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33489 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33491 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33493 2017-05-02 Richard Biener <rguenther@suse.de>
33495 PR tree-optimization/80591
33497 2017-04-10 Richard Biener <rguenther@suse.de>
33499 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33502 2017-05-02 Richard Biener <rguenther@suse.de>
33504 PR tree-optimization/80549
33505 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33506 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33507 headers if they do not have a preheader.
33509 2017-05-02 Martin Liska <mliska@suse.cz>
33512 * common.opt: Fix typo.
33513 * doc/invoke.texi: Likewise.
33515 2017-05-01 Jan Beulich <jbeulich@suse.com>
33517 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33518 swapping, add (x,x,m,x,n) alternative.
33520 2017-05-01 Nathan Sidwell <nathan@acm.org>
33522 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33523 unnecessary unadjusted_alignment check.
33525 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33528 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33530 * gimplify.c (gimplify_cilk_detach): New function.
33531 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33532 * tree-core.h: Document EXPR_CILK_SPAWN.
33533 * tree.h (EXPR_CILK_SPAWN): Define.
33535 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33537 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33538 to use new fixit_hint representation, using the "replace" logic.
33539 (get_line_span_for_fixit_hint): Likewise.
33540 (layout::print_any_fixits): Likewise.
33541 (selftest::test_one_liner_many_fixits): Rename to...
33542 (selftest::test_one_liner_many_fixits_1): ...this, and update
33543 comment and expected output to reflect that the multiple fix-it
33544 hints are now consolidated into one insertion.
33545 (selftest::test_one_liner_many_fixits_2): New test.
33546 (selftest::test_diagnostic_show_locus_one_liner): Update for
33548 (selftest::test_fixit_consolidation): Update for fix-it API
33550 * diagnostic.c (print_parseable_fixits): Likewise.
33551 * edit-context.c (edited_line::m_line_events): Convert from
33552 auto_vec <line_event *> to auto_vec <line_event>.
33553 (class line_event): Convert from abstract base class to a concrete
33554 class, taking over the role of replace_event.
33555 (class insert_event): Delete.
33556 (class replace_event): Rename to class line_event. Convert to
33558 (edit_context::add_fixits): Reimplement.
33559 (edit_context::apply_insert): Delete.
33560 (edit_context::apply_replace): Rename to...
33561 (edit_context::apply_fixit): ...this. Convert to half-open range.
33562 (edited_file::apply_insert): Delete.
33563 (edited_file::apply_replace): Rename to...
33564 (edited_file::apply_fixit): ...this.
33565 (edited_line::~edited_line): Drop deletion of events.
33566 (edited_line::apply_insert): Delete.
33567 (edited_line::apply_replace): Rename to...
33568 (edited_line::apply_fixit): ...this. Convert to half-open range.
33569 Update for change to type of m_line_events.
33570 * edit-context.h (edit_context::apply_insert): Delete.
33571 (edit_context::apply_replace): Rename to...
33572 (edit_context::apply_fixit): ...this.
33574 2017-05-01 Martin Sebor <msebor@redhat.com>
33576 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33579 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33582 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33583 __get_cpuid_max returns 0.
33584 (__get_cpuid_count): Ditto.
33586 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33588 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33589 replacement expression is another instance of one of its arguments.
33591 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33594 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33595 check for stack push/pop autoinc.
33596 * config/i386/i386.c (ix86_agi_dependent): Return false
33597 if the only reason why modified_in_p returned true is that
33598 addr is SP based and set_insn is a push or pop.
33600 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33602 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33605 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33608 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33609 (account_size_time): Use two predicates - exec_pred and
33611 (evaluate_conditions_for_known_args): Compute both clause and
33613 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33614 (inline_summary_t::duplicate): Update.
33615 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33617 (compute_inline_parameters): Likewise.
33618 (estimate_edge_size_and_time): Update caluclation of time.
33619 (estimate_node_size_and_time): Compute both time and nonspecialized
33621 (estimate_ipcp_clone_size_and_time): Update.
33622 (inline_merge_summary): Update.
33623 (do_estimate_edge_time): Update.
33624 (do_estimate_edge_size): Update.
33625 (do_estimate_edge_hints): Update.
33626 (inline_read_section, inline_write_summary): Stream both new predicates.
33627 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33629 (compute_inlined_call_time): Cleanup.
33630 (big_speedup_p): Update.
33631 (edge_badness): Update.
33632 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33633 (size_time_entry): Replace predicate by exec_predicate and
33634 nonconst_predicate.
33635 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33636 (estimate_edge_time): Return also nonspec_time.
33637 (reset_edge_growth_cache): Update.
33639 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33641 PR rtl-optimization/80491
33642 * ifcvt.c (noce_process_if_block): When looking for x setter
33643 with missing else_bb, don't check only the insn right before
33644 cond_earliest, but look for the last insn that x is modified in
33645 within the same bb.
33647 PR rtl-optimization/80491
33648 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33650 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33652 PR tree-optimization/80487
33653 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33655 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33657 PR tree-optimization/79697
33658 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33659 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33660 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33662 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33663 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33665 2017-04-28 Martin Sebor <msebor@redhat.com>
33667 PR tree-optimization/80523
33668 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33669 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33671 (maybe_warn, format_directive, parse_directive): Use new functions.
33672 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33674 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33676 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33678 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33680 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33681 target_header_dir): Set correctly.
33682 * configure: Regenerated.
33683 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33684 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33685 instead of SYSTEM_HEADER_DIR.
33687 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33689 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33690 (estimate_local_effects): Likewise.
33691 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33692 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33693 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33694 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33695 do_estimate_edge_time, estimate_edge_time): Likewise.
33696 * ipa-inline-analysis.c (estimate_node_size_and_time,
33697 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33698 (estimate_time_after_inlining): Remove.
33700 2017-04-28 Martin Liska <mliska@suse.cz>
33702 * doc/gcov.texi: Enhance documentation of gcov.
33704 2017-04-28 Martin Liska <mliska@suse.cz>
33706 * doc/gcov.texi: Sort options in alphabetic order.
33707 * doc/gcov-dump.texi: Likewise.
33708 * doc/gcov-tool.texi: Likewise.
33709 * gcov.c (print_usage): Likewise.
33710 * gcov-dump.c (print_usage): Likewise.
33711 * gcov-tool.c (print_merge_usage_message): Likewise.
33712 (print_rewrite_usage_message): Likewise.
33713 (print_overlap_usage_message): Likewise.
33715 2017-04-28 Martin Liska <mliska@suse.cz>
33717 PR gcov-profile/53915
33718 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33720 2017-04-28 Martin Liska <mliska@suse.cz>
33722 PR gcov-profile/79891
33723 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33724 is marked by compiler as living on a line.
33725 (get_cycles_count): Remove usage of the union.
33726 (output_intermediate_file): Likewise.
33727 (find_source): Fix GNU coding style.
33728 (accumulate_line_counts): Remove old non-all block mode.
33729 (output_lines): Remove usage of the union.
33730 * profile.c (output_location): Include all BBs, even if
33731 belonging to a same line (and file) as a previous BB.
33733 2017-04-28 Martin Liska <mliska@suse.cz>
33735 * gcov.c (process_args): Handle new argument 'w'.
33736 (read_graph_file): Assign ID to BBs.
33737 (output_branch_count): Display BB # if verbose flag is set.
33738 (output_lines): Likewise for arcs.
33739 (print_usage): Add '--verbose' option help.
33740 * doc/gcov.texi: Document --verbose (-w) option.
33742 2017-04-28 Martin Liska <mliska@suse.cz>
33744 * gcov.c (struct block_location_info): New struct.
33745 (process_file): Fill up the new structure.
33746 (read_graph_file): Replace usage of encoding by the newly added
33748 (add_line_counts): Likewise.
33749 (accumulate_line_counts): Remove usage of the union.
33750 (function_info::function_info): New function.
33751 (function_info::~function_info): Likewise.
33752 (process_file): Call delete instead of release_function.
33753 (release_function): Release the function.
33754 (release_structures): Call delete instead of release_function.
33755 (solve_flow_graph): Replace usage of num_blocks.
33756 (find_exception_blocks): Likewise.
33757 (output_lines): Fix GNU coding style.
33759 2017-04-28 Martin Liska <mliska@suse.cz>
33762 * coverage.c (coverage_remove_note_file): New function.
33763 * coverage.h: Declare the function.
33764 * toplev.c (finalize): Clean if an error has been seen.
33766 2017-04-28 Martin Liska <mliska@suse.cz>
33768 PR gcov-profile/80031
33769 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33770 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33771 * gcov.c (read_graph_file): Read just number of blocks.
33772 * profile.c (branch_prob): Do not stream 0 flags per a basic
33775 2017-04-28 Martin Liska <mliska@suse.cz>
33777 * gcov-dump.c (tag_*): Add new argument to declarations.
33778 (dump_gcov_file): Likewise.
33779 (tag_blocks): Add and use new argument depth.
33780 (tag_arcs): Likewise.
33781 (tag_lines): Likewise.
33782 (tag_counters): Likewise.
33783 (tag_summary): Likewise.
33784 (dump_working_sets): Use depth to do a proper indentation.
33786 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33789 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33790 * symtab.c (symtab_node::debug_symtab): Move definition here.
33792 2017-04-28 Richard Biener <rguenther@suse.de>
33794 * lto-streamer.h (LTO_major_version): Bump to 7.
33796 2017-04-28 Richard Biener <rguenther@suse.de>
33798 * tree-vrp.c (assert_info): New struct.
33799 (add_assert_info): New helper.
33800 (register_edge_assert_for_2): Refactor to add asserts to a vector
33802 (register_edge_assert_for_1): Likewise.
33803 (register_edge_assert_for): Likewise.
33804 (finish_register_edge_assert_for): New helper actually registering
33805 asserts where live on edge.
33806 (find_conditional_asserts): Adjust.
33807 (find_switch_asserts): Likewise.
33808 (evrp_dom_walker::try_find_new_range): Generalize.
33809 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33811 2017-04-27 Marek Polacek <polacek@redhat.com>
33814 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33815 arg10 and arg11 to itype.
33817 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33819 * doc/extend.texi (Object Size Checking): Improve grammar.
33821 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33824 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33825 that the logic for permitting reciprocal estimates matches that
33828 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33831 * tree.c (type_cache_hasher::equal): Only compare
33832 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33833 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33834 non-aggregate element types.
33835 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33836 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33838 2017-04-27 Richard Biener <rguenther@suse.de>
33840 PR middle-end/80533
33841 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33842 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33843 keeping a reference to a trailing array.
33845 2017-04-27 Richard Biener <rguenther@suse.de>
33847 PR middle-end/80539
33848 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33849 being in loop-closed SSA form conservatively.
33850 (chrec_fold_multiply_poly_poly): Likewise.
33852 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33854 PR middle-end/79665
33855 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33856 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33858 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33861 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33862 (aarch64_function_arg_alignment): Return unsigned int again, but still
33863 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33864 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33865 Don't emit -Wpsabi note.
33866 (aarch64_function_arg_boundary): Likewise.
33867 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33870 2017-04-26 Nathan Sidwell <nathan@acm.org>
33872 * tree.h (crc32_unsigned_n): Declare.
33873 (crc32_unsigned, crc32_unsigned): Make inline.
33874 * tree.c (crc32_unsigned_bits): Replace with ...
33875 (crc32_unsigned_n): ... this.
33876 (crc32_unsigned, crc32_byte): Remove.
33877 (crc32_string): Remove unnecessary braces.
33879 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33881 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33882 * ipa-inline-analysis.c (MAX_TIME): Remove.
33883 (account_size_time): Use sreal for time.
33884 (dump_inline_summary): Update.
33885 (estimate_function_body_sizes): Update.
33886 (estimate_edge_size_and_time): Update.
33887 (estimate_calls_size_and_time): Update.
33888 (estimate_node_size_and_time): Update.
33889 (inline_merge_summary): Update.
33890 (inline_update_overall_summary): Update.
33891 (estimate_time_after_inlining): Update.
33892 (inline_read_section): Update.
33893 (inline_write_summary): Update.
33894 * ipa-inline.c (compute_uninlined_call_time): Update.
33895 (compute_inlined_call_time): Update.
33896 (recursive_inlining): Update.
33897 (inline_small_functions): Update.
33898 (dump_overall_stats): Update.
33899 * ipa-inline.h: Include sreal.h.
33900 (size_time_entry): Turn time to sreal.
33901 (inline_summary): Turn self_time nad time to sreal.
33903 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33905 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33907 (sreal::stream_out, sreal::stream_in): New.
33908 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33910 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33912 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33915 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33918 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33920 (dimode_scalar_chain::compute_convert_gain): Ditto.
33921 (dimode_scalar_chain::make_vector_copies): Ditto.
33922 (dimode_scalar_chain::convert_reg): Ditto.
33923 (dimode_scalar_chain::convert_insn): Ditto.
33924 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33925 (VI248_AVX512BW_1): New mode iterator.
33926 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33927 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33930 2017-04-25 Martin Sebor <msebor@redhat.com>
33932 PR tree-optimization/80497
33933 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33934 constants are representable in HOST_WIDE_INT.
33935 (parse_directive): Ditto.
33937 2017-04-25 Martin Sebor <msebor@redhat.com>
33940 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33941 (new_zero_array): Adjust signature.
33942 (dom_info::dom_init): Used unsigned rather that size_t.
33943 (dom_info::dom_info): Same.
33945 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33946 Jakub Jelinek <jakub@redhat.com>
33949 * config/arm/arm.c: Include gimple.h.
33950 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33951 returns negative, increment ncrn only if it returned positive.
33952 (arm_needs_doubleword_align): Return int instead of bool,
33953 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33954 members, but if there is any such non-FIELD_DECL
33955 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33956 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33957 returns negative, increment nregs only if it returned positive.
33958 (arm_setup_incoming_varargs): Likewise.
33959 (arm_function_arg_boundary): Emit -Wpsabi note if
33960 arm_needs_doubleword_align returns negative, return
33961 DOUBLEWORD_ALIGNMENT only if it returned positive.
33963 2017-04-25 Marek Polacek <polacek@redhat.com>
33966 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33967 first argument to type.
33969 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33972 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33973 type checks to test for compatibility instead of equality.
33975 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33976 Jakub Jelinek <jakub@redhat.com>
33979 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33981 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33982 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33983 the alignment computation, but return their maximum in warn_alignment.
33984 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33985 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33987 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33988 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33991 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33993 * config/arc/simdext.md (dmpyh): Fix typo.
33995 2017-04-25 Richard Biener <rguenther@suse.de>
33997 PR tree-optimization/80492
33998 * alias.c (compare_base_decls): Handle registers with asm
33999 specification conservatively.
34000 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
34001 compare_base_decls returning dont-know properly.
34003 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34005 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
34006 (legitimate_offset_address_p): New function.
34007 (arc_legitimate_address_p): Use above function.
34009 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34011 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
34013 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34015 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
34016 ACCH registers whenever they are available.
34018 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34020 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
34021 double regs fix when not used.
34023 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34025 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
34027 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
34028 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
34030 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34032 * config/arc/arc.c (arc_output_addsi): Check for h-register class
34033 when emitting short ADD instructions.
34035 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34037 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
34039 (cmpsi_cc_c_insn): Likewise.
34040 (cbranchsi4_scratch): Compute proper instruction length using
34041 compact_hreg_operand.
34042 * config/arc/predicates.md (compact_hreg_operand): New predicate.
34044 2017-04-25 Richard Biener <rguenther@suse.de>
34046 PR middle-end/80509
34047 * passes.c (pass_manager::pass_manager): Initialize
34048 m_name_to_pass_map.
34050 2017-04-25 Richard Biener <rguenther@suse.de>
34052 PR tree-optimization/79201
34053 * tree-ssa-sink.c (statement_sink_location): Handle calls.
34055 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34058 * config/s390/vector.md: Split MEM->GPR vector moves for
34059 non-s_operand addresses.
34061 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34064 * config/s390/predicates.md (reload_const_wide_int_operand): New
34066 * config/s390/s390.md ("movti"): Remove d/P alternative.
34067 ("movti_bigconst"): New pattern definition.
34069 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34072 * s390-protos.h (s390_expand_cs_hqi): Removed.
34073 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
34074 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
34075 modes as well as CCZ1mode and CCZmode.
34076 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
34077 signature of s390_emit_compare_and_swap.
34078 (s390_expand_cs_hqi): Likewise, make static.
34079 (s390_expand_cs_tdsi): Generate an explicit compare before trying
34080 compare-and-swap, in some cases.
34081 (s390_expand_cs): Wrapper function.
34082 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
34084 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
34085 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
34086 patterns for small and large integers. Forbid symref memory operands.
34087 Move expander to s390.c. Require cc register.
34088 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
34089 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
34090 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
34091 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
34092 symref memory operands. Remove CC mode and call s390_match_ccmode
34094 ("atomic_exchange<mode>"): Allow and implement all integer modes.
34096 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34098 * config/s390/s390.md (define_peephole2): New peephole to help
34099 combining the load-and-test pattern with volatile memory.
34101 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34103 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
34104 with CCZmode for TARGET_Z196.
34106 2017-04-25 Jakub Jelinek <jakub@redhat.com>
34108 PR rtl-optimization/80501
34109 * combine.c (make_compound_operation_int): Set subreg_code to SET
34110 even for AND with mask of the sign bit of mode.
34112 PR rtl-optimization/80500
34113 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
34114 sum's initial value.
34116 2017-04-25 Julian Brown <julian@codesourcery.com>
34117 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34119 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
34121 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
34123 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
34125 2017-04-25 Julian Brown <julian@codesourcery.com>
34126 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34128 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
34129 (thunderx2t99_sha): New Reservation.
34131 2017-04-25 Julian Brown <julian@codesourcery.com>
34132 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34134 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
34135 type for 1-element load.
34137 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
34139 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
34141 2017-04-24 Martin Jambor <mjambor@suse.cz>
34143 PR tree-optimization/80293
34144 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
34145 char arrays not totally scalarizable if it is false.
34146 (analyze_all_variable_accesses): Pass correct value in the new
34147 parameter. Add a statistics counter.
34149 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
34151 PR middle-end/79931
34152 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
34154 2017-04-24 Richard Biener <rguenther@suse.de>
34156 PR tree-optimization/80494
34157 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
34158 out for complex types.
34160 2017-04-24 Richard Biener <rguenther@suse.de>
34162 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34163 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34164 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34165 (DFS): Adjust and never fail.
34166 (sccvn_dom_walker::fail): Remove.
34167 (sccvn_dom_walker::before_dom_children): Adjust.
34168 (run_scc_vn): Likewise and never fail.
34169 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34170 (pass_fre::execute): Likewise.
34172 2017-04-24 Richard Biener <rguenther@suse.de>
34174 PR tree-optimization/79725
34175 * tree-ssa-sink.c (statement_sink_location): Return whether
34176 failure reason was zero uses. Move that check later.
34177 (sink_code_in_bb): Deal with zero uses by removing the stmt
34180 2017-04-24 Richard Biener <rguenther@suse.de>
34183 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34184 pointer-based references.
34186 2017-04-24 Richard Biener <rguenther@suse.de>
34189 * pass_manager.h (pass_manager::operator new): Remove.
34190 (pass_manager::operator delete): Likewise.
34191 * passes.c (pass_manager::operator new): Remove.
34192 (pass_manager::operator delete): Likewise.
34193 (pass_manager::pass_manager): Zero individual pass members.
34195 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34198 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34199 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34200 Check "XEXP (src, 1)" operand here.
34201 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34202 Check "XEXP (src, 1)" operand here.
34203 (dimode_scalar_chain::make_vector_copies): Detect count register
34204 of a shift instruction. Zero extend count register from QImode
34205 to DImode to satisfy vector shift pattern count operand predicate.
34206 Substitute vector shift count operand with a DImode copy.
34207 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34210 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34212 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34213 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34214 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34215 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34216 (UNSPEC_NOREX_MEM): Remove definition.
34218 2017-04-21 Richard Biener <rguenther@suse.de>
34220 PR tree-optimization/79547
34221 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34222 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34223 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34224 without any constraints.
34226 2017-04-21 Richard Biener <rguenther@suse.de>
34228 PR tree-optimization/78847
34229 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34231 2017-04-21 Richard Biener <rguenther@suse.de>
34233 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34234 (build_distinct_type_copy): Likewise.
34235 (build_variant_type_copy): Likewise.
34236 * tree.c (build_qualified_type): Pass down mem-stat info.
34237 (build_distinct_type_copy): Likewise.
34238 (build_variant_type_copy): Likewise.
34240 2017-04-21 Richard Biener <rguenther@suse.de>
34242 PR tree-optimization/80237
34243 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34245 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34246 for a simplified result.
34248 2016-04-21 Richard Biener <rguenther@suse.de>
34250 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34251 sth as strict as a simple_iv but a chrec without symbols and an
34252 operand defined in the loop we are peeling (and not some subloop).
34253 (propagate_constants_for_unrolling): Propagate all constants.
34255 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34258 * config/i386/i386.c (print_reg): Remove assert for disalowed
34259 regno values, call output_operand_lossage instead.
34261 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34264 * config/i386/constraints.md (Yc): New register constraint.
34265 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34266 Use Yc constraint for alternative 2 of operand 0. Remove
34267 preferred_for_speed attribute.
34269 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34271 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34272 lastprivate clauses in SIMT case.
34274 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34276 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34278 2017-04-20 Richard Biener <rguenther@suse.de>
34280 PR tree-optimization/57796
34281 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34282 as N scalar stores.
34283 (vect_model_load_cost): Cost gathers as N scalar loads.
34285 2017-04-20 Richard Biener <rguenther@suse.de>
34287 * ggc-page.c (ggc_allocated_p): Rename to ...
34288 (safe_lookup_page_table_entry): ... this and return the lookup
34290 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34292 2017-04-20 Richard Biener <rguenther@suse.de>
34294 PR tree-optimization/80453
34295 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34296 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34297 from the conditions.
34298 (vn_phi_eq): Pass them down.
34299 (vn_phi_lookup): Record them.
34300 (vn_phi_insert): Likewise.
34302 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34304 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34305 uninitialized variable warning to avoid buffer overrun.
34307 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34310 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34311 is suppressed for '{ 0 }' in C.
34313 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34315 * BASE-VER: Set to 8.0.0.
34317 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34319 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34320 priority .init_array and .fini_array section with SECTION_NOTYPE
34323 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34325 PR middle-end/80423
34326 * tree.h (build_array_type): Add typeless_storage default argument.
34327 * tree.c (type_cache_hasher::equal): Also compare
34328 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34329 (build_array_type): Add typeless_storage argument, set
34330 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34332 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34333 (build_array_type): Likewise. Add typeless_storage argument.
34335 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34336 Jakub Jelinek <jakub@redhat.com>
34338 PR tree-optimization/80426
34339 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34340 operation on symbolic operands, also compute the overflow for the
34341 invariant part when the operation degenerates into a negation.
34343 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34346 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34347 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34350 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34352 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34355 * config/avr/avr.c (tree.h): Include it.
34356 (cgraph.h): Include it.
34357 (avr_encode_section_info): Don't warn for uninitialized progmem
34358 variable if it's just an alias.
34360 2017-04-19 Richard Biener <rguenther@suse.de>
34363 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34364 when needed by AutoPGO.
34366 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34369 * doc/lto.texi: Remove an extra 'that'.
34371 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34373 PR rtl-optimization/80429
34374 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34375 are only used in debug insns.
34377 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34378 Vladimir Makarov <vmakarov@redhat.com>
34380 * config/sparc/predicates.md (input_operand): Add comment. Return
34381 true for any memory operand when LRA is in progress.
34382 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34384 2017-04-18 Jeff Law <law@redhat.com>
34387 * mips.md ({return,simple_return}_internal): Do not overwrite
34390 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34392 PR tree-optimization/80443
34393 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34394 instead of adding 1, subtract -1 and similarly instead of subtracting
34397 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34399 PR rtl-optimization/80357
34400 * haifa-sched.c (tmp_bitmap): New variable.
34401 (model_recompute): Handle duplicate use records.
34402 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34403 (free_global_sched_pressure_data): Free it.
34405 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34408 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34409 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34410 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34411 instead of SYSTEM_HEADER_DIR.
34413 2017-04-18 Jeff Law <law@redhat.com>
34415 PR middle-end/80422
34416 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34417 predecessors after walking up the insn chain.
34419 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34422 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34423 sizetype type into debug info.
34425 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34428 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34429 unneeded test for TARGET_UPPER_REGS_SF.
34430 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34432 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34435 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34436 instead of gsi_after_labels.
34438 2017-04-18 Jeff Law <law@redhat.com>
34440 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34444 2017-04-13 Jeff Law <law@redhat.com>
34445 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34446 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34448 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34451 * config/avr/avr.c (intl.h): Include it.
34452 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34454 2017-04-18 Martin Liska <mliska@suse.cz>
34456 PR gcov-profile/78783
34457 * gcov-tool.c (gcov_output_files): Validate that destination
34458 file is either removed by the tool or by a user.
34460 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34461 Guy Benyei <guybe@mellanox.com>
34463 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34464 block, and do not negate it, the stored id is already negative.
34466 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34468 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34470 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34473 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34474 masks of options that should be turned off if the VSX vector
34475 options are turned off.
34476 (OTHER_P8_VECTOR_MASKS): Likewise.
34477 (OTHER_VSX_VECTOR_MASKS): Likewise.
34478 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34479 rs6000_disable_incompatible_switches to validate no type switches
34481 (rs6000_incompatible_switch): New function to disallow turning on
34482 other vector options if -mno-vsx, -mno-power8-vector, or
34483 -mno-power9-vector are specified.
34485 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34487 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34489 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34491 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34492 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34493 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34494 (ARG_POINTER_CFA_OFFSET): Likewise.
34496 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34498 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34499 conditions to take advantage of various optimizations.
34501 2017-04-13 Jeff Law <law@redhat.com>
34503 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34504 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34505 (zero_extendsidi2_dext): Likewise.
34507 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34510 * fold-const.c (fold_ternary_loc): Revert
34511 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34514 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34516 PR rtl-optimization/80343
34517 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34518 new scratch pseudo.
34520 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34523 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34524 to ubsan_encode_value.
34526 2017-04-13 Jeff Law <law@redhat.com>
34528 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34529 appearing in DEBUG_INSNs.
34531 2017-04-13 Martin Liska <mliska@suse.cz>
34533 PR gcov-profile/80413
34534 * gcov-io.c (gcov_write_string): Copy to buffer just when
34535 allocated size is greater than zero.
34537 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34540 * dwarf2out.c (decls_for_scope): Ignore declarations of
34541 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34543 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34546 * ipa-visibility.c (non_local_p): Fix typos.
34547 (localize_node): When localizing symbol in same comdat group,
34548 dissolve the group only when we know external symbols are going
34550 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34552 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34554 PR tree-optimization/79390
34555 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34556 order does not result in usable sequence, retry with reversed operand
34562 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34563 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34564 op0 instead of fold_convert_loc (loc, type, arg0).
34566 2017-04-12 Jeff Law <law@redhat.com>
34568 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34569 has a delay slot in the generated code.
34571 * config/cris/cris.md (cris_preferred_reload_class): Return
34572 GENNONACR_REGS rather than GENERAL_REGS.
34574 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34577 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34578 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34579 signedness of the result type.
34581 2017-04-12 Richard Biener <rguenther@suse.de>
34582 Jeff Law <law@redhat.com>
34584 PR tree-optimization/80359
34585 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34586 trim stores to TARGET_MEM_REFs.
34588 2017-04-12 Richard Biener <rguenther@suse.de>
34590 PR tree-optimization/79390
34591 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34592 threading case even more.
34594 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34597 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34598 for quad_address_p for TImode, instead of just not indexed_address.
34600 2017-04-12 Richard Biener <rguenther@suse.de>
34601 Bernd Edlinger <bernd.edlinger@hotmail.de>
34603 PR middle-end/79671
34604 * alias.c (component_uses_parent_alias_set_from): Handle
34605 TYPE_TYPELESS_STORAGE.
34606 (get_alias_set): Likewise.
34607 * tree-core.h (tree_type_common): Add typeless_storage flag.
34608 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34609 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34610 for types containing members with TYPE_TYPELESS_STORAGE.
34611 (place_field): Likewise.
34612 (layout_type): Likewise for ARRAY_TYPE.
34613 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34614 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34615 TYPE_TYPELESS_STORAGE.
34616 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34618 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34621 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34622 first argument to type.
34624 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34628 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34629 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34630 (rs6000_expand_binop_builtin): Likewise.
34631 (rs6000_expand_ternop_builtin): Likewise; also add missing
34632 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34633 vshasigma built-ins.
34634 * doc/extend.texi: Document that vec_xxpermdi's third argument
34635 must be a constant.
34637 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34639 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34640 Use shift_const cost parameter when calculating gain of STV shifts.
34642 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34644 PR rtl-optimization/70478
34645 * lra-constraints.c (process_alt_operands): Check memory for
34646 disfavoring memory insn operand.
34648 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34650 PR middle-end/80100
34651 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34652 left shift in unsigned HOST_WIDE_INT type.
34654 PR rtl-optimization/80385
34655 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34656 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34659 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34660 if they have any depend clauses.
34662 2017-04-11 Martin Liska <mliska@suse.cz>
34665 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34666 * ipa-split.c (split_function): Create a local comdat symbol
34667 if caller is in a comdat group.
34669 2017-04-11 Martin Liska <mliska@suse.cz>
34672 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34675 2017-04-11 Martin Sebor <msebor@redhat.com>
34677 PR middle-end/80364
34678 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34679 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34681 (directive::set_width, directive::set_precision, format_character):
34683 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34686 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34689 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34690 conflict, set target->arch_name instead of target->cpu_name.
34692 2017-04-11 Richard Biener <rguenther@suse.de>
34694 PR tree-optimization/80374
34695 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34696 build_zero_cst, remove fold_convertible_p check again.
34698 2017-04-11 Martin Liska <mliska@suse.cz>
34701 * ubsan.c (instrument_object_size): Do not instrument register
34704 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34707 * config/i386/i386-builtin-types.def
34708 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34709 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34710 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34711 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34712 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34713 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34714 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34715 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34716 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34717 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34718 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34719 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34720 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34721 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34722 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34723 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34724 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34725 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34726 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34727 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34728 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34729 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34730 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34731 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34732 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34733 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34734 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34735 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34736 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34737 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34738 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34739 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34740 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34741 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34742 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34743 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34744 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34745 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34746 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34747 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34748 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34749 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34750 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34751 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34752 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34753 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34755 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34756 flag to second_arg_count, handle 4 argument function type _COUNT
34757 aliases, handle second_arg_count on second argument rather than last.
34759 2017-04-10 Jeff Law <law@redhat.com>
34761 PR tree-optimization/80374
34762 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34763 record anything if we can not convert integer_zero_node to the
34766 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34769 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34770 Enhance special handling given to the TARGET_P9_MINMAX option in
34771 relation to certain other options.
34773 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34775 PR tree-optimization/80153
34776 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34777 remove POINTER_PLUS_EXPR's base part directly, rather than through
34780 2017-04-10 Richard Biener <rguenther@suse.de>
34781 Bin Cheng <bin.cheng@arm.com>
34783 PR tree-optimization/80153
34784 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34785 the first element of pointer type aff_tree. Build result expr in
34787 (add_elt_to_tree): Convert to type unconditionally. Remove other
34788 fold_convert calls.
34789 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34790 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34792 2017-04-10 Richard Biener <rguenther@suse.de>
34794 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34797 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34799 PR rtl-optimization/70478
34800 * lra-constraints.c (curr_small_class_check): New.
34801 (update_and_check_small_class_inputs): New.
34802 (process_alt_operands): Update curr_small_class_check. Disfavor
34803 alternative insn memory operands. Check available regs for small
34806 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34809 * config/mips/mips.opt (-mvirt): Update description.
34810 * doc/invoke.texi (-mvirt): Likewise.
34812 2017-04-10 Richard Biener <rguenther@suse.de>
34814 PR middle-end/80362
34815 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34816 looking for NEGATE_EXPR in -A / -B to A / B folding.
34818 2017-04-10 Martin Liska <mliska@suse.cz>
34820 PR gcov-profile/80224
34821 * gcov.c (print_usage): Fix usage string.
34822 (get_gcov_intermediate_filename): Remove.
34823 (output_gcov_file): Use both for normal and intermediate format.
34824 (generate_results): Do not initialize special file for
34825 intermediate format.
34827 2017-04-10 Richard Biener <rguenther@suse.de>
34829 PR tree-optimization/80304
34830 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34833 2017-04-10 Nathan Sidwell <nathan@acm.org>
34836 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34837 (rs6000_init_builtins): Use it.
34839 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34841 * config/arm/arm.md (<mrc>): Add mode to SET source.
34842 (<mrrc>): Likewise.
34844 2017-04-10 Richard Biener <rguenther@suse.de>
34846 PR middle-end/80344
34847 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34849 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34852 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34853 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34854 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34855 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34856 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34857 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34858 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34859 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34860 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34861 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34862 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34863 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34864 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34865 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34866 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34867 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34868 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34869 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34870 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34871 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34872 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34873 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34874 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34876 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34878 PR rtl-optimization/70478
34879 * lra-constraints.c: Reverse the last patch.
34881 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34883 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34884 Add comment for WCHAR_T.
34886 2017-04-08 Martin Liska <mliska@suse.cz>
34889 2017-04-07 Martin Liska <mliska@suse.cz>
34892 * ipa-split.c (split_function): Add function part to a same comdat
34895 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34898 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34900 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34902 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34903 (vec_pairing): ...to this.
34904 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34905 (rs6000_sched_init): Adjust for name change.
34906 (struct rs6000_sched_context): Likewise.
34907 (rs6000_init_sched_context): Likewise.
34908 (rs6000_set_sched_context): Likewise.
34910 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34916 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34918 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34919 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34920 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34922 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34924 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34926 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34928 PR rtl-optimization/70703
34929 * ira-color.c (update_conflict_hard_regno_costs): Use
34930 int64_t instead of HOST_WIDE_INT.
34932 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34934 PR rtl-optimization/70478
34935 * lra-constraints.c (process_alt_operands): Disfavor alternative
34936 insn memory operands.
34938 2017-04-07 Jeff Law <law@redhat.com>
34940 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34941 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34943 2017-04-07 Martin Liska <mliska@suse.cz>
34946 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34947 Show error message instead of an ICE.
34949 2017-04-07 Martin Liska <mliska@suse.cz>
34952 * ipa-split.c (split_function): Add function part to a same comdat
34955 2017-04-07 Richard Biener <rguenther@suse.de>
34957 PR middle-end/80341
34958 * tree.c (get_unwidened): Also handle ! for_type case for
34960 * convert.c (do_narrow): Split out from ...
34961 (convert_to_integer_1): ... here. Do not pass final truncation
34962 type to get_unwidened for TRUNC_DIV_EXPR.
34964 2017-04-07 Richard Biener <rguenther@suse.de>
34966 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34968 (aff_combination_const): Adjust.
34969 (aff_combination_scale): Likewise.
34970 (aff_combination_add_elt): Likewise.
34971 (aff_combination_add_cst): Likewise.
34972 (aff_combination_convert): Likewise.
34973 (add_elt_to_tree): Likewise. Remove unused argument.
34974 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34976 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34978 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34980 * config/arm/arm.c (arm_default_short_enums): Use
34981 ARM_DEFAULT_SHORT_ENUMS.
34982 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34984 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34987 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34988 members with redundant out-of-class redeclaration.
34990 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34993 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34994 * config/i386/i386.md (*zero_extendsidi2):
34995 Add (?*x,*x) and (?*v,*v) alternatives.
34997 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
35000 * config/i386/i386.c (ix86_expand_builtin)
35001 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
35002 mode from insn data. Convert operands to insn operand mode.
35003 Copy operands that don't satisfy insn predicate to a register.
35005 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
35007 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
35010 2017-04-06 Richard Biener <rguenther@suse.de>
35012 PR tree-optimization/80334
35013 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
35014 preserve alignment of accesses.
35016 2017-04-06 Richard Biener <rguenther@suse.de>
35018 PR tree-optimization/80262
35019 * tree-sra.c (build_ref_for_offset): Preserve address-space
35021 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
35022 Drop useless address-space information on MEM_REF offsets.
35024 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
35026 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
35028 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35030 PR rtl-optimization/70703
35031 * ira-color.c (update_conflict_hard_regno_costs): Use
35032 HOST_WIDE_INT instead of long.
35034 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
35037 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
35038 not defined for x86_64 target. Add -mmmx target option when __SSE2__
35040 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
35041 for x86_64 target. Handle -m3dnowa option.
35043 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35045 PR rtl-optimization/70703
35046 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
35047 (update_conflict_hard_regno_costs): Use long instead of unsigned
35048 arithmetic for cost calculation.
35050 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35051 Bernd Edlinger <bernd.edlinger@hotmail.de>
35054 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
35057 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
35060 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
35061 ptr_mode with Pmode throughout.
35062 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
35063 into probe_stack_range and use DImode.
35065 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
35068 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
35069 call_eh_return is true.
35071 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35073 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
35074 Initialize last_match_fntype_index.
35076 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35079 * tree-nvr.c: Include internal-fn.h.
35080 (pass_return_slot::execute): Ignore internal calls without
35083 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35084 Richard Biener <rguenther@suse.de>
35087 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
35088 captures used multiple times, except for the last use.
35089 * generic-match-head.c: Include gimplify.h.
35091 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35093 PR tree-optimization/79390
35094 * target.h (struct noce_if_info): Declare.
35095 * targhooks.h (default_noce_conversion_profitable_p): Declare.
35096 * target.def (noce_conversion_profitable_p): New target hook.
35097 * ifcvt.h (struct noce_if_info): New type, moved from ...
35098 * ifcvt.c (struct noce_if_info): ... here.
35099 (noce_conversion_profitable_p): Renamed to ...
35100 (default_noce_conversion_profitable_p): ... this. No longer
35102 (noce_try_store_flag_constants, noce_try_addcc,
35103 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
35104 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
35105 instead of noce_conversion_profitable_p.
35106 * config/i386/i386.c: Include ifcvt.h.
35107 (ix86_option_override_internal): Don't override
35108 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
35109 (ix86_noce_conversion_profitable_p): New function.
35110 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
35111 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
35112 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
35113 * doc/tm.texi: Regenerated.
35115 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35117 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
35120 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
35123 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
35124 instructions for small multiply cores.
35126 2017-04-04 Jeff Law <law@redhat.com>
35128 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
35130 (mips_expand_vec_perm_const): Initialize elements in orig_perm
35131 that are not set by the loop over the elements.
35133 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35136 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
35137 int mode, convert_modes it to mode as unsigned, otherwise use
35138 lowpart_subreg to mode rather than SImode.
35139 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
35140 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
35141 Use DImode instead of SImode for the shift count operand.
35142 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
35145 2017-04-04 Richard Biener <rguenther@suse.de>
35147 PR middle-end/80281
35148 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
35149 arithmetic done for the negate or the plus. Simplify.
35150 (A - (-B) -> A + B): Likewise.
35151 * fold-const.c (split_tree): Make sure to not negate pointers.
35153 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
35155 PR rtl-optimization/60818
35156 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
35157 a compare of comparisons with the thing compared if this results
35158 in a different machine mode.
35160 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35162 * alias.c (base_alias_check): Fix typo in comment.
35163 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35164 * cgraphunit.c (symbol_table::compile): Likewise.
35165 * collect2.c (maybe_run_lto_and_relink): Likewise.
35166 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35167 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35168 * config/avr/avr.c (avr_map_op_t): Likewise.
35169 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35170 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35171 * config/epiphany/epiphany.md (movcc): Likewise.
35172 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35173 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35175 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35176 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35177 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35178 * config/sh/sh.c (sh_rtx_costs): Likewise.
35179 * fold-const.c (fold_truth_andor): Likewise.
35180 * genautomata.c (collapse_flag): Likewise.
35181 * gengtype.h (struct type::u::s): Likewise.
35182 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35183 * input.c (FORMAT_AMOUNT): Likewise.
35184 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35185 (known_aggs_to_agg_replacement_list): Likewise.
35186 * ipa-inline-analysis.c: Likewise.
35187 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35188 * ipa-polymorphic-call.c
35189 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35190 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35191 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35193 * modulo-sched.c (apply_reg_moves): Likewise.
35194 * omp-expand.c (build_omp_regions_1): Likewise.
35195 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35196 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35197 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35198 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35199 * value-prof.c: Likewise.
35200 * var-tracking.c (val_reset): Likewise.
35202 2017-04-03 Richard Biener <rguenther@suse.de>
35204 PR tree-optimization/80275
35205 * fold-const.c (split_address_to_core_and_offset): Handle
35208 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35210 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35211 descriptors is at least equal to that of functions.
35213 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35215 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35217 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35220 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35221 (mov<IMOD4:mode>): New expander.
35222 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35224 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35226 PR rtl-optimization/79405
35227 * fwprop.c (propagations_left): New variable.
35228 (forward_propagate_into): Decrement it.
35229 (fwprop_init): Initialize it.
35230 (fw_prop): If the variable has reached zero, stop propagating.
35231 (fwprop_addr): Ditto.
35233 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35236 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35237 a FUNCTION_DECL, pass it as decl instead of origin to
35240 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35242 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35245 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35248 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35249 TARGET_VSX_SMALL_INTEGER.
35251 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35253 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35254 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35256 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35258 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35259 extraction from odd-numbered MSA register.
35261 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35263 PR middle-end/80173
35264 * expmed.c (store_bit_field_1): Don't attempt to create
35265 a word subreg out of hard registers wider than word if they
35266 have HARD_REGNO_NREGS of 1 for their mode.
35268 PR middle-end/80163
35269 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35270 conversions to integer types wider than word and pointer.
35273 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35274 (rtx_equal_for_cselib_p): Pass 0 to it.
35275 * cselib.c (cselib_hasher::equal): Likewise.
35276 (rtx_equal_for_cselib_1): Add depth argument. If depth
35277 is 128, don't look up VALUE locs and punt. Increment
35278 depth in recursive calls when walking VALUE locs.
35280 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35282 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35283 (make_gcov_file_name): Use the canonical path name for generating
35285 (read_line): Fix handling of files with ascii null bytes.
35287 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35289 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35290 to initialise a vector register instead
35291 of using a const_int.
35293 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35295 PR translation/80189
35296 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35297 diagnostic messages.
35299 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35302 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35303 (dfp_diex_<mode>): Update mode of operand 1.
35304 * doc/extend.texi (dxex, dxexq): Document change to return type.
35305 (diex, diexq): Document change to argument type.
35307 2017-03-30 Martin Jambor <mjambor@suse.cz>
35310 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35311 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35312 it reflects the signature changes performed at the callee side.
35313 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35314 to cgraph_build_function_type_skip_args.
35315 (build_function_decl_skip_args): Adjust call to the above function.
35317 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35320 * config/i386/sse.md
35321 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35322 register as dest whenever it is a MEM not rtx_equal_p to the
35323 corresponding dup operand, and when forcing into reg move the
35324 reg into the memory afterwards.
35325 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35326 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35327 for the force_reg mode.
35328 (avx512vl_vextractf128<mode>): Use register as dest either
35329 always when a MEM, or when it is a MEM not rtx_equal_p to the
35330 corresponding dup operand, or even not when it is a CONST_VECTOR
35331 depending on the mode and lo vs. hi.
35332 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35334 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35335 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35336 Likewise. Require that operands[2] is even.
35337 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35338 Remove extraneous parens. Require that operands[2] is a multiple
35340 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35341 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35342 disallow memory then.
35344 2017-03-30 Richard Biener <rguenther@suse.de>
35346 PR tree-optimization/77498
35347 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35348 to non-constants over backedges.
35350 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35352 PR rtl-optimization/80233
35353 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35354 as last_combined_insn. Do not test for BARRIER_P separately.
35356 2017-03-29 Andreas Schwab <schwab@suse.de>
35359 * calls.c (prepare_call_address): Convert funexp to Pmode before
35360 copying to temp reg.
35362 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35364 PR tree-optimization/80158
35365 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35366 Handle possible future case of more than one alternate
35368 (replace_rhs_if_not_dup): Likewise.
35369 (replace_one_candidate): Likewise.
35371 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35373 PR rtl-optimization/80193
35374 * ira.c (ira): Do not check allocation for LRA.
35376 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35378 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35379 (nvptx_output_simt_exit): Declare.
35380 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35381 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35382 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35383 (nvptx_declare_function_name): ...here. Emit declaration of local
35384 memory space buffer for omp_simt_enter insn.
35385 (nvptx_output_unisimt_switch): New.
35386 (nvptx_output_softstack_switch): New.
35387 (nvptx_output_simt_enter): New.
35388 (nvptx_output_simt_exit): New.
35389 * config/nvptx/nvptx.h (struct machine_function): New fields
35390 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35391 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35392 (UNSPECV_SIMT_EXIT): Ditto.
35393 (omp_simt_enter_insn): New insn.
35394 (omp_simt_enter): New expansion.
35395 (omp_simt_exit): New insn.
35396 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35398 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35399 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35400 (expand_GOMP_SIMT_EXIT): New.
35401 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35402 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35403 (GOMP_SIMT_EXIT): Ditto.
35404 * target-insns.def (omp_simt_enter): New insn.
35405 (omp_simt_exit): Ditto.
35406 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35408 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35409 (lower_rec_input_clauses): Likewise.
35410 (lower_lastprivate_clauses): Handle SIMT privatization.
35412 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35413 (ompdevlow_adjust_simt_enter): New.
35414 (find_simtpriv_var_op): New.
35415 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35416 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35418 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35419 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35420 (copy_decl_for_dup_finish): Ditto.
35422 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35424 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35427 * config/i386/i386.c (ix86_option_override_internal): Always
35428 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35430 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35432 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35434 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35436 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35437 mark new edge's irreducible flag accordign to it.
35438 (vect_do_peeling): Check loop preheader edge's irreducible flag
35439 and pass it to function slpeel_add_loop_guard.
35441 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35443 PR tree-optimization/80218
35444 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35445 Update block frequencies and counts.
35447 2017-03-28 Richard Biener <rguenther@suse.de>
35449 PR tree-optimization/78644
35450 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35451 of a simplification result we may not use it at all.
35453 2017-03-28 Richard Biener <rguenther@suse.de>
35456 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35457 without arguments, generate default definition of a SSA name.
35459 2017-03-28 Richard Biener <rguenther@suse.de>
35461 PR middle-end/80222
35462 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35463 TYPE_REF_CAN_ALIAS_ALL references.
35464 * fold-const.c (fold_indirect_ref_1): Likewise.
35466 2017-03-28 Martin Liska <mliska@suse.cz>
35469 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35470 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35472 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35473 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35475 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35476 (EXTRA_SPECS): Define.
35477 (SUBTARGET_EXTRA_SPECS): Likewise.
35478 (SUBTARGET_CPP_SPEC): Likewise.
35479 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35480 SUBTARGET_EXTRA_SPECS.
35481 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35483 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35485 * config/arc/simdext.md (vst64_insn): Update pattern.
35486 (vld32wh_insn): Likewise.
35487 (vld32wl_insn): Likewise.
35488 (vld64_insn): Likewise.
35489 (vld32_insn): Likewise.
35491 2017-03-28 Marek Polacek <polacek@redhat.com>
35494 * fold-const.c (fold_comparison): Use protected_set_expr_location
35495 instead of SET_EXPR_LOCATION.
35497 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35499 * tree.c (add_expr): Avoid name lookup warning.
35501 2017-03-27 Jeff Law <law@redhat.com>
35503 PR tree-optimization/80216
35504 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35505 function name. Limit recursion depth.
35506 (record_temporary_equivalences): Corresponding changes.
35508 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35510 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35513 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35516 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35518 * cfgcleanup.c (reg_note_cfa_p): New array.
35519 (insns_have_identical_cfa_notes): New function.
35520 (old_insns_match_p): Don't cross-jump in between /f
35521 and non-/f instructions. If both i1 and i2 are frame related,
35522 verify all CFA notes, their order and content.
35524 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35527 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35528 HImode and SImode with zero extend to DImode to one insn.
35529 (bswap<mode>2_extenddi): Likewise.
35530 (bswapsi2_extenddi): Likewise.
35531 (bswaphi2_extendsi): Likewise.
35532 (bswaphi2): Combine bswap HImode and SImode into one insn.
35533 Separate memory insns from swapping register.
35534 (bswapsi2): Likewise.
35535 (bswap<mode>2): Likewise.
35536 (bswaphi2_internal): Delete, no longer used.
35537 (bswapsi2_internal): Likewise.
35538 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35539 store, and gpr<-gpr swap insns.
35540 (bswap<mode>2_store): Likewise.
35541 (bswaphi2_reg): Register only splitter, combine with the splitter.
35542 (bswaphi2 splitter): Likewise.
35543 (bswapsi2_reg): Likewise.
35544 (bswapsi2 splitter): Likewise.
35545 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35546 the insns into load, store, and register/register insns.
35547 (bswapdi2_ldbrx): Likewise.
35548 (bswapdi2_load): Likewise.
35549 (bswapdi2_store): Likewise.
35550 (bswapdi2_reg): Likewise.
35552 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35554 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35555 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35557 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35560 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35562 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35563 special handling for target option conflicts between dform
35564 options (-mpower9-dform, -mpower9-dform-vector,
35565 -mpower9-dform-scalar) and -mno-direct-move.
35567 2017-03-27 Richard Biener <rguenther@suse.de>
35569 PR tree-optimization/80181
35570 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35572 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35574 * config/arc/predicates.md (move_double_src_operand): Replace the
35575 call to move_double_src_operand with a call to address_operand.
35577 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35579 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35580 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35581 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35583 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35585 * config/arc/predicates.md (long_immediate_loadstore_operand):
35586 Consider scaled addresses cases.
35588 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35590 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35591 restored when in interrupt.
35592 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35593 doesn't have delay slot.
35595 2017-03-27 Richard Biener <rguenther@suse.de>
35598 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35599 inlined thunk clones.
35601 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35604 * asan.c (instrument_derefs): Copy over last operand from
35605 original COMPONENT_REF to the new COMPONENT_REF with
35606 DECL_BIT_FIELD_REPRESENTATIVE.
35607 * ubsan.c (instrument_object_size): Likewise.
35609 2017-03-27 Richard Biener <rguenther@suse.de>
35611 PR tree-optimization/80170
35612 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35613 sure DR/SCEV didnt fold in constants we do not see when looking
35614 at the reference base alignment.
35616 2017-03-27 Richard Biener <rguenther@suse.de>
35618 PR middle-end/80171
35619 * gimple-fold.c (fold_ctor_reference): Properly guard against
35620 NULL return value from canonicalize_constructor_val.
35622 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35625 * config/i386/i386.c (ix86_expand_builtin)
35626 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35627 flags reg setting and flags reg using instructions.
35628 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35629 clobbering instructions to zero extend op2.
35631 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35633 * doc/install.texi (Configuration) <--with-aix-soname>:
35634 Update link to AIX ld.
35636 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35638 PR rtl-optimization/80160
35639 PR rtl-optimization/80159
35640 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35641 reg_alternate_class into account.
35643 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35646 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35647 to consider in curr_insn_transform.
35649 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35651 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35652 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35653 and emit_mode_inner.
35655 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35657 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35658 argument to the overloaded builtin variants. Use the new flag to
35659 deprecate certain builtin variants.
35660 * config/s390/s390-builtin-types.def: Add new builtin types.
35661 * config/s390/s390-builtins.h: Support new flags field for
35662 overloaded builtins.
35663 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35664 (s390_macro_to_expand): Enable vector float data type.
35665 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35666 builtins by incrementing the __VEC__ version number.
35667 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35669 (s390_resolve_overloaded_builtin): Emit error messages depending
35670 on the builtin flags.
35671 * config/s390/s390.c (s390_expand_builtin): Support additional
35672 flags argument. Change error message to match the messages
35673 emitted in s390-c.c.
35674 * config/s390/s390.md: New UNSPEC_* constants.
35675 (op_type): Add new instruction types.
35676 * config/s390/vecintrin.h: Add new builtins and test data class
35678 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35679 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35680 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35681 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35682 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35683 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35685 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35686 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35687 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35688 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35690 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35691 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35692 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35693 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35695 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35696 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35697 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35698 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35699 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35700 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35701 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35703 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35704 ("vec_scatter_element<V_HW_4:mode>_DI")
35705 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35706 ("vec_fpint<mode>", "vflls")
35707 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35708 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35709 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35710 ("*vec_cmphe<mode>_cc"): ... these.
35712 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35713 mode constant instead of magic value.
35715 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35717 * config/s390/s390.c (s390_expand_vec_compare): Support other
35718 vector floating point modes than just V2DF.
35719 (s390_expand_vcond): Likewise.
35720 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35721 (s390_cannot_change_mode_class): Prevent mode changes between TF
35722 and V1TF in vector registers.
35723 * config/s390/s390.md (DF, SF): New mode attributes.
35724 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35725 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35726 SFmode support for VRs.
35727 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35729 (VFT, VF_HW): New mode iterators.
35730 (vw, sdx): New mode attributes.
35731 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35732 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35733 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35734 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35735 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35736 also the new vector floating point modes. Renaming to ...
35738 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35739 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35740 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35741 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35742 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35743 ("vec_unordered<mode>"): ... these.
35745 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35746 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35747 ("*vec_extendv2df"): New insn definitions.
35749 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35751 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35752 ("mulditi3_2", "*muldi3_sign"): New patterns.
35753 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35754 rename the pattern definition.
35756 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35758 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35760 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35762 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35764 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35765 instruction if possible.
35766 * config/s390/vector.md (vec_halfnumelts): New mode
35768 ("*vec_vllezlf<mode>"): New pattern.
35770 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35772 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35773 ("popcountv4si2", "popcountv2di2"): Rename to ...
35774 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35775 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35777 ("popcount<mode>2_vxe"): New pattern.
35779 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35781 * common/config/s390/s390-common.c (processor_flags_table): Add
35783 * config.gcc: Add arch12.
35784 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35785 Default to arch12 for unknown CPU model numbers.
35786 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35787 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35788 PROCESSOR_max sanity check.
35789 * config/s390/s390-opts.h (enum processor_type): Add
35791 * config/s390/s390.c (processor_table): Add arch12.
35792 (s390_expand_builtin): Add check for B_VXE flag.
35793 (s390_issue_rate): Add PROCESSOR_ARCH12.
35794 (s390_get_sched_attrmask): Likewise.
35795 (s390_get_unit_mask): Likewise.
35796 (s390_sched_score): Enable z13 scheduling for arch12.
35797 (s390_sched_reorder): Likewise.
35798 (s390_sched_variable_issue): Likewise.
35799 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35801 (s390_tune_attr): Use z13 scheduling also for arch12.
35802 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35803 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35804 (TARGET_VXE_P): New macros.
35805 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35806 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35807 * config/s390/s390.opt: Add arch12 as processor_type.
35809 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35811 * config/s390/s390.md
35812 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35813 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35814 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35816 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35817 Rename expanders to ...
35819 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35820 ("fixuns_truncdddi2_emu"): ... these.
35822 ("fixuns_trunc<mode>si2_emu"): New expander.
35824 ("*fixuns_truncdfdi2_z13"): Rename to ...
35825 ("*fixuns_truncdfdi2_vx"): ... this.
35827 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35829 * config/s390/2964.md: Remove the single element vector compare
35830 instructions which are no longer used.
35831 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35833 (s390_canonicalize_comparison): Remove handling of DFmode
35835 (s390_expand_vec_compare_scalar): Remove function.
35836 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35837 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35839 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35841 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35843 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35845 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35846 will anyway by matched by mov<mode>_64dfp.
35848 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35850 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35851 vlef/vstef. Add missing operand to vleif.
35853 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35855 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35856 pair for all vector types with 64 bit elements.
35857 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35858 * config/s390/vector.md (V_HW_64): ... here.
35859 (V_128_NOSINGLE): New mode iterator.
35860 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35861 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35862 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35863 ("*vec_load_pairv2di"): Change to ...
35864 ("*vec_load_pair<mode>"): ... this one.
35866 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35868 * config/s390/constraints.md: Add comments.
35869 (jKK): Reject element sizes > 8 bytes.
35870 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35872 * config/s390/s390.md: Add the s_operand checks formerly in
35873 s390_split_ok_p to various splitters where they are still
35875 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35876 for 128 bit vectors. Plus two splitters.
35878 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35880 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35883 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35886 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35887 error if the boundary argument is not constant.
35889 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35891 PR rtl-optimization/80112
35892 * loop-doloop.c (doloop_condition_get): Don't check condition
35893 if cmp isn't SET with IF_THEN_ELSE src.
35895 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35897 PR tree-optimization/80158
35898 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35899 replacing a candidate statement, also replace it for the
35900 candidate's alternate interpretation.
35901 (replace_rhs_if_not_dup): Likewise.
35902 (replace_one_candidate): Likewise.
35904 2017-03-24 Richard Biener <rguenther@suse.de>
35906 PR tree-optimization/80167
35907 * graphite-isl-ast-to-gimple.c
35908 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35910 (translate_isl_ast_to_gimple::get_rename): Likewise.
35912 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35914 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35915 handling of certain combinations of target options, including the
35916 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35917 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35919 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35922 * config/arm/arm.md (*load_multiple): Add reload_completed to
35923 matching condition.
35925 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35926 Richard Biener <rguenth@suse.de>
35928 PR tree-optimization/79908
35929 PR tree-optimization/80136
35930 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35931 been cast away, gimplify_and_add suffices.
35933 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35935 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35937 2017-03-23 Richard Biener <rguenther@suse.de>
35939 PR tree-optimization/80032
35940 * gimplify.c (gimple_push_cleanup): Forced unconditional
35941 cleanups still have to go to the conditional_cleanups
35944 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35946 PR tree-optimization/80072
35947 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35949 (next_operand_entry_id): Change type to unsigned int.
35950 (sort_by_operand_rank): Make sure to return the right return value
35951 even if unsigned fields are bigger than INT_MAX.
35952 (struct oecount): Change cnt and id type to unsigned int.
35953 (oecount_hasher::equal): Formatting fix.
35954 (oecount_cmp): Make sure to return the right return value
35955 even if unsigned fields are bigger than INT_MAX.
35956 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35959 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35960 TREE_READONLY on result if writing it more than once.
35963 * doc/invoke.texi (-fsanitize=thread): Document that with
35964 -fnon-call-exceptions atomics are not able to throw
35968 * tsan.c: Include tree-eh.h.
35969 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35970 maybe_clean_or_replace_eh_stmt where needed.
35971 (instrument_memory_accesses): Add cfg_changed argument.
35972 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35973 if it returned true.
35974 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35976 PR rtl-optimization/63191
35977 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35978 wrapper function, moved the whole old content into ...
35979 (ix86_delegitimize_address_1): ... this. New inline function.
35980 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35981 true as last argument instead of ix86_delegitimize_address.
35983 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35985 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35986 cortexa57_branch_cost.
35988 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35990 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35992 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35995 * doc/md.texi (Constraints): Document wA constraint.
35996 * config/rs6000/constraints.md (wA): New.
35997 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35998 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35999 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
36000 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
36002 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
36005 * gimplify.c (is_oacc_declared): New function.
36006 (oacc_default_clause): Use it to set default flags for acc declared
36007 variables inside parallel regions.
36008 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
36009 declared variables.
36010 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
36011 declare attribute to any decl as necessary.
36013 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
36016 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
36017 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
36018 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
36019 (arm_arch_lpae): This.
36020 * config/arm/arm.c (arm_arch7ve): Rename into ...
36021 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
36022 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
36025 2017-03-22 Martin Liska <mliska@suse.cz>
36028 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
36029 error message instead of an ICE.
36031 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36033 * doc/extend.texi (6.11 Additional Floating Types): Revise.
36035 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
36037 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36039 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36042 2017-03-21 Martin Sebor <msebor@redhat.com>
36044 * doc/extend.texi: Use "cannot" instead of "can't."
36045 * doc/hostconfig.texi: Same.
36046 * doc/install.texi: Same.
36047 * doc/invoke.texi: Same.
36048 * doc/loop.texi: Same.
36049 * doc/md.texi: Same.
36050 * doc/objc.texi: Same.
36051 * doc/rtl.texi: Same.
36052 * doc/tm.texi: Same.
36053 * doc/tm.texi.in: Same.
36054 * doc/trouble.texi: Same.
36056 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
36059 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
36060 (collect_checksum_attributes): Set it.
36061 (die_checksum_ordered): Use it.
36063 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36065 PR tree-optimization/79908
36066 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
36067 change: For a VA_ARG whose LHS has been cast away, use
36068 force_gimple_operand to construct the side effects.
36070 2017-03-21 David Malcolm <dmalcolm@redhat.com>
36072 PR translation/80001
36073 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
36074 more amenable to translation.
36075 (oacc_loop_auto_partitions): Likewise.
36077 2017-03-21 Marek Polacek <polacek@redhat.com>
36078 Martin Sebor <msebor@redhat.com>
36080 PR tree-optimization/80109
36081 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
36082 on INTEGRAL_TYPE_P.
36084 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36085 Segher Boessenkool <segher@kernel.crashing.org>
36088 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
36089 check reg_used_between_p between insn and one of succ or succ2
36090 depending on if succ is artificial insn not inserted into insn
36093 2017-03-21 Martin Liska <mliska@suse.cz>
36095 PR gcov-profile/80081
36096 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
36097 * doc/gcc.texi: Include gcov-dump stuff.
36098 * doc/gcov-dump.texi: New file.
36100 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
36102 PR rtl-optimization/79150
36103 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
36104 conditional jump, if the jump is the last insn of the loop.
36106 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36107 Richard Biener <rguenth@suse.de>
36109 PR tree-optimization/79908
36110 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36111 been cast away, use force_gimple_operand to construct the side
36114 2017-03-21 Martin Liska <mliska@suse.cz>
36116 PR libfortran/79956
36117 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
36120 2017-03-21 Brad Spengler <spender@grsecurity.net>
36123 * plugin.c (htab_hash_plugin): New function.
36124 (add_new_plugin): Use it and adjust.
36125 (parse_plugin_arg_opt): Adjust.
36126 (init_one_plugin): Likewise.
36128 2017-03-21 Richard Biener <rguenther@suse.de>
36130 PR tree-optimization/80032
36131 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
36132 if set force the cleanup to happen unconditionally.
36133 (gimplify_target_expr): Push inserted clobbers with force_uncond
36134 to avoid them being removed by control-dependent DCE.
36136 2017-03-21 Richard Biener <rguenther@suse.de>
36138 PR tree-optimization/80122
36139 * tree-inline.c (copy_bb): Do not expans va-arg packs or
36140 va_arg_pack_len when the inlined call stmt requires pack
36142 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
36144 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36147 * tsan.c (instrument_builtin_call): If the memory model argument
36148 is not a constant, assume it is valid.
36151 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
36154 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36156 PR rtl-optimization/79910
36157 * combine.c (can_combine_p): Do not allow combining an I0 or I1
36158 if its dest is used by an insn before I2 (other than the combined
36159 insns themselves, which are properly handled already).
36161 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36164 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36166 * combine.c (record_used_regs): New static function.
36167 (try_combine): Handle situations where there is an additional
36168 instruction between I2 and I3 which needs to have a LOG_LINK
36172 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36174 * combine.c (try_combine): Delete redundant i1 test. Call
36175 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36177 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36180 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36181 alternatives 13/14.
36183 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36185 PR tree-optimization/80054
36186 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36187 the optimization if a PHI or any of its arguments is not dominated
36188 by the candidate's basis. Use gphi* rather than gimple* as
36190 (replace_profitable_candidates): Clean up a gimple* variable that
36191 should be a gphi* variable.
36193 2017-03-20 Martin Sebor <msebor@redhat.com>
36196 * doc/extend.texi (attribute constructor): Document present limitation.
36198 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36201 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36202 __POWER9_VECTOR__ #ifdef control, change template definition to
36203 use Power9-specific built-in function.
36204 (vec_any_eq): Likewise.
36205 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36206 to control outcomes from this test.
36207 (vector_ae_<mode>p): For VEC_F modes, likewise.
36209 2017-03-20 Ian Lance Taylor <iant@google.com>
36211 * config/i386/i386.c (ix86_function_regparm): Save an extra
36212 register for -fsplit-stack with DECL_STATIC_CHAIN.
36214 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36217 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36218 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36220 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36222 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36224 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36227 2017-03-20 Marek Polacek <polacek@redhat.com>
36230 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36232 2017-03-20 Richard Biener <rguenther@suse.de>
36234 PR tree-optimization/80113
36235 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36236 allocate extra SSA name for PHI def.
36237 (add_close_phis_to_outer_loops): Likewise.
36238 (add_close_phis_to_merge_points): Likewise.
36239 (copy_loop_close_phi_args): Likewise.
36240 (copy_cond_phi_nodes): Likewise.
36242 2017-03-20 Martin Liska <mliska@suse.cz>
36244 PR middle-end/79753
36245 * tree-chkp.c (chkp_build_returned_bound): Do not build
36246 returned bounds for a LHS that's not a BOUNDED_P type.
36248 2017-03-20 Martin Liska <mliska@suse.cz>
36252 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36253 COMPLEX_CST and VECTOR_CST.
36255 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36258 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36259 target operand. A new splitter adds the clobber statement in case
36260 the target operand is dead anyway.
36262 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36264 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36265 to age-old versions of binutils and glibc.
36267 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36269 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36271 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36273 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36275 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36277 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36278 requirement for binutils 2.13.
36280 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36282 * combine.c (try_combine): Delete redundant i1 test. Call
36283 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36285 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36287 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36288 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36290 <riscv64-*-elf>: Re-arrange section
36291 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36292 <riscv32-*-linux>: Likewise.
36293 <riscv64-*-elf>: Likewise
36294 <riscv64-*-linux>: Likewise.
36296 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36299 * aarch64.opt(verbose-cost-dump): Fix typo.
36301 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36304 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36305 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36307 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36309 * reload.c (find_reloads): When reloading a nonoffsettable address,
36310 use RELOAD_OTHER for it and its address reloads.
36312 PR rtl-optimization/79910
36313 * combine.c (record_used_regs): New static function.
36314 (try_combine): Handle situations where there is an additional
36315 instruction between I2 and I3 which needs to have a LOG_LINK
36318 2017-03-17 Jeff Law <law@redhat.com>
36320 PR tree-optimization/71437
36321 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36322 conditional in the hash table first.
36323 (vrp_dom_walker::before_dom_children): Extract condition from
36324 ASSERT_EXPR. Record condition, its inverion and any implied
36325 conditions as well.
36327 2017-03-17 Marek Polacek <polacek@redhat.com>
36328 Markus Trippelsdorf <markus@trippelsdorf.de>
36330 PR tree-optimization/80079
36331 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36334 2017-03-17 Richard Biener <rguenther@suse.de>
36336 PR middle-end/80075
36337 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36338 Properly verify the LHS before the RHS possibly claims to be
36340 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36343 2017-03-17 Martin Jambor <mjambor@suse.cz>
36345 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36346 (List of -O2 options): Likewise.
36347 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36350 2017-03-17 Tom de Vries <tom@codesourcery.com>
36352 * gcov-dump.c (print_usage): Print bug_report_url.
36354 2017-03-17 Richard Biener <rguenther@suse.de>
36356 PR middle-end/80050
36357 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36358 (parser::peek): Likewise.
36360 2017-03-17 Richard Biener <rguenther@suse.de>
36362 PR tree-optimization/80048
36363 * sese.c (free_sese_info): Properly release rename_map and
36364 copied_bb_map elements.
36366 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36368 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36369 Add linked-list forward and backlinks. Insert on
36370 construction, remove on destruction.
36371 (class pass_store_merging): Add m_stores_head field.
36372 (pass_store_merging::terminate_and_process_all_chains):
36373 Iterate over m_stores_head list.
36374 (pass_store_merging::terminate_all_aliasing_chains):
36376 (pass_store_merging::execute): Check for debug stmts first.
36377 Push new chains onto the m_stores_head stack.
36379 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36382 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36383 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36384 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36386 2017-03-16 Jeff Law <law@redhat.com>
36388 PR tree-optimization/71437
36389 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36390 member function. Implementation moved into after_dom_children
36391 member function and into the threader's thread_outgoing_edges
36393 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36394 some code into new thread_outgoing_edges.
36395 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36396 definition. Simplify marker handling (do it here). Assume we always
36397 have the available expression and the const/copies tables.
36398 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36400 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36401 * tree-vrp.c (equiv_stack): No longer file scoped.
36402 (vrp_dom_walker): New class.
36403 (vrp_dom_walker::before_dom_children): New member function.
36404 (vrp_dom_walker::after_dom_children): Likewise.
36405 (identify_jump_threads): Setup domwalker. Use it rather than
36406 walking edges in a random order by hand. Simplify setup/finalization.
36407 (finalize_jump_threads): Remove.
36408 (vrp_finalize): Do not call identify_jump_threads here.
36409 (execute_vrp): Do it here instead and call thread_through_all_blocks
36412 PR tree-optimization/71437
36413 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36415 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36417 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36418 (dom_opt_dom_walker::thread_across_edge): Remove
36419 handle_dominating_asserts argument. All callers changed.
36420 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36421 changes. Remove calls to lhs_of_dominating_assert. Other
36422 uses of handle_dominating_asserts turn into unconditional code
36423 (simplify_control_stmt_condition_1): Likewise.
36424 (simplify_control_stmt_condition): Likewise.
36425 (thread_through_normal_block, thread_across_edge): Likewise.
36426 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36427 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36428 object if it is not an SSA_NAME.
36429 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36430 before calling into the VRP specific simplifiers.
36431 (identify_jump_threads): Remove handle_dominating_asserts
36434 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36437 * tree-diagnostic.c (default_tree_printer): No longer static.
36438 * tree-diagnostic.h (default_tree_printer): New prototype.
36440 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36442 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36443 Change ins into fmov.
36445 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36447 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36448 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36449 Use h_con constraint for operand 1.
36450 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36451 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36453 2017-03-15 Jeff Law <law@redhat.com>
36455 PR tree-optimization/71437
36456 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36457 (record_temporary_equivalences): Use it.
36459 PR tree-optimization/71437
36460 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36461 tree-ssa-scopedtables.
36462 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36463 (record_conditions, record_cond, vuse_eq): Likewise.
36464 (record_edge_info): Adjust to API tweak of record_conditions.
36465 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36466 (record_temporary_equivalences, optimize_stmt): Likewise.
36467 (eliminate_redundant_computations): Likewise.
36468 (record_equivalences_from_stmt): Likewise.
36469 * tree-ssa-scopedtables.c: Include options.h and params.h.
36470 (vuse_eq): New function, moved from tree-ssa-dom.c
36471 (build_and_record_new_cond): Likewise.
36472 (record_conditions): Likewise. Accept vector of conditions rather
36473 than edge_equivalence structure for first argument.
36474 for the first argument.
36475 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36476 from tree-ssa-dom.c.
36477 (avail_exprs_stack::record_cond): Likewise.
36478 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36479 from tree-ssa-dom.c.
36480 (avail_exprs_stack): Add new member functions lookup_avail_expr
36482 (record_conditions): Declare.
36484 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36487 * lra-constraints.c (process_alt_operands): Increase reject for
36488 reloading an input/output operand.
36490 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36493 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36494 insns to convert from signed/unsigned char/short to IEEE 128-bit
36496 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36498 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36501 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36502 subreg of inner mode for values already in registers.
36504 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36506 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36507 iteration reg is used after the loop.
36509 2017-03-14 Martin Sebor <msebor@redhat.com>
36511 PR tree-optimization/79800
36512 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36513 precision in negative-positive range.
36514 (format_floating): Call non-const overload with adjusted precision.
36516 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36519 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36522 2017-03-14 Martin Sebor <msebor@redhat.com>
36524 PR middle-end/80020
36525 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36526 * builtins.def (aligned_alloc): Use it.
36529 * Makefile.in (GTFILES): Add calls.c.
36530 * calls.c: Include "gt-calls.h".
36532 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36534 PR rtl-optimization/79728
36535 * regs.h (struct target_regs): New field
36536 x_contains_allocatable_regs_of_mode.
36537 (contains_allocatable_regs_of_mode): New macro.
36538 * reginfo.c (init_reg_sets_1): Initialize it, and change
36539 contains_reg_of_mode so it includes global regs as well.
36540 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36541 rather than contains_regs_of_mode.
36543 2017-03-14 Martin Liska <mliska@suse.cz>
36545 * doc/invoke.texi: Document options that can't be combined with
36546 -fcheck-pointer-bounds.
36548 2017-03-14 Martin Liska <mliska@suse.cz>
36550 PR middle-end/79831
36551 * doc/invoke.texi (-Wchkp): Document the option.
36553 2017-03-14 Martin Liska <mliska@suse.cz>
36555 * Makefile.in: Install gcov-dump.
36557 2017-03-14 Martin Liska <mliska@suse.cz>
36559 * multiple_target.c (expand_target_clones): Bail out for
36560 an invalid attribute.
36562 2017-03-14 Richard Biener <rguenther@suse.de>
36564 * alias.c (struct alias_set_entry): Pack properly.
36565 * cfgloop.h (struct loop): Likewise.
36566 * cse.c (struct set): Likewise.
36567 * ipa-utils.c (struct searchc_env): Likewise.
36568 * loop-invariant.c (struct invariant): Likewise.
36569 * lra-remat.c (struct cand): Likewise.
36570 * recog.c (struct change_t): Likewise.
36571 * rtl.h (struct address_info): Likewise.
36572 * symbol-summary.h (function_summary): Likewise.
36573 * tree-loop-distribution.c (struct partition): Likewise.
36574 * tree-object-size.c (struct object_size_info): Likewise.
36575 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36576 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36577 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36578 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36579 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36580 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36581 (struct _stmt_vec_info): Likewise.
36583 2017-03-14 Martin Liska <mliska@suse.cz>
36586 * multiple_target.c (create_dispatcher_calls): Check that
36587 a target can create a function dispatcher.
36589 2017-03-14 Martin Liska <mliska@suse.cz>
36592 * multiple_target.c (expand_target_clones): Drop local.local
36593 flag for default implementation.
36595 2017-03-14 Richard Biener <rguenther@suse.de>
36597 PR tree-optimization/80030
36598 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36600 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36602 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36603 gcc_fallthrough() instead of __attribute__((fallthrough));
36605 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36607 * doc/gcc.texi: Remove "up" link to (DIR).
36608 * doc/gccint.texi: Ditto.
36610 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36612 * doc/install.texi (Specific) <avr>: Remove reference to
36615 2017-03-13 Jeff Law <law@redhat.com>
36617 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36618 attribute rather than comments.
36620 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36621 match_scratch operand is highest.
36623 2017-03-13 Martin Liska <mliska@suse.cz>
36625 PR middle-end/78339
36626 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36627 is a CHKP clone, use original declaration.
36629 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36631 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36632 (arc_conditional_register_usage): Use a different allocation order
36633 when optimizing for size.
36634 * common/config/arc/arc-common.c (arc_option_optimization_table):
36635 Section anchors default on when optimizing for size.
36637 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36639 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36641 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36643 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36644 * config/arc/arc.md (cpu_facility): Add cd variant.
36645 (*movqi_insn): Add code density variant.
36646 (*movhi_insn): Likewise.
36647 (*movqi_insn): Likewise.
36648 (*addsi3_mixed): Likewise.
36649 (subsi3_insn): Likewise.
36651 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36653 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36655 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36657 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36658 expressions with MINUS and UNARY ops.
36660 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36663 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36665 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36666 between vec_select and vector argument.
36667 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36668 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36669 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36670 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36671 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36672 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36674 2017-03-13 Richard Biener <rguenther@suse.de>
36677 * params.def (vect-max-peeling-for-alignment): Fix typo.
36679 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36681 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36682 issue that only occurred with binutils below 2.18.
36684 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36686 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36687 refer to binutils 2.11/2.12 minimum.
36689 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36691 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36692 ftp.kernel.org and simplify binutils requirement.
36694 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36696 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36698 (Optimize Options): Ditto. Also remove redundancy.
36700 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36702 PR translation/79848
36703 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36705 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36706 to G_ to avoid double translation.
36708 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36710 PR translation/79923
36711 * auto-profile.c (get_combined_location): Convert leading
36712 character of diagnostics to lower case and remove trailing period.
36713 (read_profile): Likewise for various diagnostics.
36714 * config/arm/arm.c (arm_option_override): Remove trailing period
36715 from various diagnostics.
36716 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36717 (msp430_expand_delay_cycles): Likewise.
36719 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36722 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36723 full command-line argument, rather than just "str".
36724 (aarch64_validate_march): Likewise.
36725 (aarch64_validate_mtune): Likewise.
36727 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36729 PR rtl-optimization/78911
36730 * lra-assigns.c (must_not_spill_p): New function.
36731 (spill_for): Use it.
36733 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36735 PR tree-optimization/79981
36736 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36737 ATOMIC_COMPARE_EXCHANGE ifn result.
36738 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36739 IFN_ATOMIC_COMPARE_EXCHANGE.
36741 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36744 * opts.c (parse_sanitizer_options): Add missing question mark to
36745 "did you mean" message.
36747 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36749 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36751 (VMULEUH_UNS): Likewise.
36752 (VMULOUB_UNS): Likewise.
36753 (VMULOUH_UNS): Likewise.
36754 * config/rs6000/rs6000.c (builtin_function_type): Remove
36755 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36757 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36760 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36761 x with result of extra_parsing_for_operand_code_0.
36762 (function_reader::extra_parsing_for_operand_code_0): Convert
36763 return type from void to rtx, returning x. When reading
36764 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36765 larger size containing struct block_symbol.
36767 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36769 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36770 -mfloat128-hardware without -m64.
36772 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36775 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36776 entries to the case statement that marks unsigned arguments to
36777 overloaded functions.
36779 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36781 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36782 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36784 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36787 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36788 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36790 2017-03-10 Martin Liska <mliska@suse.cz>
36794 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36795 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36798 2017-03-10 Olivier Hainque <hainque@adacore.com>
36800 * tree-switch-conversion (array_value_type): Start by resetting
36801 candidate type to it's main variant.
36803 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36805 PR rtl-optimization/79909
36806 * combine.c (try_combine): Use simplify_replace_rtx on individual
36807 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36808 of the whole CALL_INSN_FUNCTION_USAGE.
36810 PR tree-optimization/79972
36811 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36812 get_range_info on SSA_NAMEs. Formatting fixes.
36814 2017-03-10 Richard Biener <rguenther@suse.de>
36815 Jakub Jelinek <jakub@redhat.com>
36817 PR tree-optimization/77975
36818 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36819 edge to be constant.
36820 (get_val_for): For constant x return it. Formatting fix.
36821 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36822 would use the same bases as the current one.
36824 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36826 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36827 instead of vec_select for V1TImode.
36828 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36830 (VSX_LE_128): Add V1TI to this mode iterator.
36831 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36832 (*vsx_le_perm_store_<mode>): Likewise.
36833 (pre-reload splitter for VSX stores): Likewise.
36834 (post-reload splitter for VSX stores): Likewise.
36835 (*vsx_xxpermdi2_le_<mode>): Likewise.
36836 (*vsx_lxvd2x2_le_<mode>): Likewise.
36837 (*vsx_stxvd2x2_le_<mode>): Likewise.
36839 2017-03-09 Michael Eager <eager@eagercon.com>
36841 Correct failures with --enable-checking=yes,rtl.
36843 * config/microblaze/microblaze.c (microblaze_expand_shift):
36844 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36845 test for const0_rtx.
36846 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36847 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36849 2017-03-09 Richard Biener <rguenther@suse.de>
36851 PR tree-optimization/79977
36852 * graphite-scop-detection.c (scop_detection::merge_sese):
36853 Handle the case of extra exits to blocks dominating the entry.
36855 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36857 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36860 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36862 PR rtl-optimization/79949
36863 * lra-constraints.c (process_alt_operands): Check memory when
36864 trying to predict a cycle. Print about the overall increase.
36866 2017-03-09 Richard Biener <rguenther@suse.de>
36868 PR middle-end/79971
36869 * gimple-expr.c (useless_type_conversion_p): Preserve
36870 TYPE_SATURATING for fixed-point types.
36872 2017-03-09 Richard Biener <rguenther@suse.de>
36875 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36876 alignment of BLKmode params.
36878 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36881 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36882 (VALL_NO_V2Q): Likewise.
36884 * config/aarch64/aarch64-simd.md
36885 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36887 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36888 VALL_NO_V2Q mode iterator.
36889 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36891 2017-03-09 Martin Liska <mliska@suse.cz>
36893 PR tree-optimization/79631
36894 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36895 tree_int_cst_sign_bit just for INTEGER constants.
36897 2017-03-09 Martin Liska <mliska@suse.cz>
36901 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36904 2017-03-09 Marek Polacek <polacek@redhat.com>
36907 * tree.c (inchash::add_expr): Handle TREE_VEC.
36909 2017-03-09 Martin Liska <mliska@suse.cz>
36912 (chkp_narrow_size_and_offset): New function.
36913 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36914 (void chkp_parse_bit_field_ref): New function.
36915 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36916 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36918 2017-03-09 Martin Liska <mliska@suse.cz>
36921 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36922 (chkp_find_bounds_1): Remove gcc_unreachable.
36924 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36927 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36928 BUILT_IN_SYNC*, determine the access type from the size suffix and
36929 always build a MEM_REF with that type. Handle forgotten
36930 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36933 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36934 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36935 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36936 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36937 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36938 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36939 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36940 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36941 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36942 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36943 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36944 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36945 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36946 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36947 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36948 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36949 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36950 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36951 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36952 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36953 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36954 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36955 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36956 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36957 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36958 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36959 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36960 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36961 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36962 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36963 definitions outside of __OPTIMIZE__ guarded section.
36966 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36967 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36968 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36969 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36972 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36974 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36975 ("vfenez<mode>"): Add missing constraints.
36977 2017-03-08 Martin Sebor <msebor@redhat.com>
36980 * config/nds32/nds32.c (nds32_option_override):
36981 Fix misspelled diagnostic.
36983 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36986 * gimplify.c (gimplify_omp_for): Replace index var in outer
36987 taskloop statement with an artificial variable and add
36988 OMP_CLAUSE_PRIVATE clause for it.
36990 2017-03-08 Richard Biener <rguenther@suse.de>
36992 PR tree-optimization/79955
36993 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36994 for accesses that are completely outside of the variable.
36996 2017-03-08 Andrew Haley <aph@redhat.com>
36998 PR tree-optimization/79943
36999 * tree-ssa-loop-split.c (compute_new_first_bound): When
37000 calculating the new upper bound, (END-BEG) should be added, not
37003 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37005 * config/avr/avr.md (setmemhi): Make sure match_dup
37006 operand number comes before match_scratch.
37008 2017-03-08 Richard Biener <rguenther@suse.de>
37010 PR tree-optimization/79920
37011 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
37012 with ncopies == 1 to ...
37013 (vect_transform_slp_perm_load): ... here. Properly compute
37014 all element loads by iterating VF times over the group. Do
37015 not handle ncopies (computed in a broken way) in
37016 vect_create_mask_and_perm.
37018 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37021 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
37022 is a uniform vector, use uniform_vector_p return value instead of
37023 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
37025 2017-03-07 Marek Polacek <polacek@redhat.com>
37027 PR middle-end/79809
37028 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
37029 (alloca_call_type): Likewise.
37031 2017-03-07 Martin Liska <mliska@suse.cz>
37033 * gcov.c (process_args): Put comment to correct location.
37035 2017-03-07 Martin Liska <mliska@suse.cz>
37037 PR middle-end/68270
37038 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
37039 Use array_at_struct_end_p instead of DECL_CHAIN (field).
37040 (chkp_narrow_bounds_for_field): Likewise.
37041 (chkp_parse_array_and_component_ref): Pass one more argument to
37044 2017-03-07 Richard Biener <rguenther@suse.de>
37046 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
37049 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
37051 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
37052 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
37054 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37057 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
37058 to end of description.
37059 (PARAM_MAX_STORES_TO_MERGE): Likewise.
37061 2017-03-07 Jakub Jelinek <jakub@redhat.com>
37063 PR rtl-optimization/79901
37064 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
37066 (*avx512f_<code><mode>3<mask_name>): ... this.
37067 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
37068 iterator instead of VI8_AVX2_AVX512BW.
37070 PR rtl-optimization/79901
37071 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
37072 min/max expander, expand it using expand_vec_cond_expr.
37075 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
37078 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37081 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
37082 to void * for PCH reasons.
37083 * dwarf2out.c (output_loc_operands, output_die): Cast
37084 v.val_vec.array to unsigned char *.
37086 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
37089 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
37092 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
37094 PR rtl-optimization/79571
37095 * lra-constraints.c (process_alt_operands): Calculate static
37096 reject and subtract it from overall when only addresses will be
37099 2017-03-06 Julia Koval <julia.koval@intel.com>
37102 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
37103 incoming stack boundary to 128 for 64-bit targets.
37105 2017-03-06 Richard Biener <rguenther@suse.de>
37107 PR tree-optimization/79894
37108 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
37109 to NULL after folding it.
37111 2017-03-06 Richard Biener <rguenther@suse.de>
37113 PR tree-optimization/79824
37114 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
37115 check disabling peeling for gaps.
37117 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
37119 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
37120 attributes): Document gettimeofday.
37122 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37124 * config/s390/s390.c (s390_option_override_internal): Set
37125 PARAM_MIN_VECT_LOOP_BOUND
37127 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37129 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
37130 * config/s390/s390.md: Likewise.
37132 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37135 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
37136 (<avx2_avx512>_perm<mode>): Rename to ...
37137 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
37139 (<avx512>_perm<mode>_mask): Rename to ...
37140 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
37142 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
37143 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
37144 instead of VI8F_256_512.
37145 (avx512f_perm<mode>): New define_expand.
37146 (avx512f_perm<mode>_mask): Likewise.
37147 (avx512f_perm<mode>_1<mask_name>): New define_insn.
37148 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
37150 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37152 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
37153 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
37155 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
37157 2017-03-06 Martin Liska <mliska@suse.cz>
37160 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37161 when having a SSA NAME w/o VAR_DECL assigned to it.
37163 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37165 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37166 msa_dpsub_<su>_d): Fix MODE for vec_select.
37168 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37170 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37172 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37174 2017-03-06 Richard Biener <rguenther@suse.de>
37176 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37177 * plugin.c (register_plugin_info): Likewise.
37178 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37180 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37182 * config/i386/sse.md (sse_storehps, sse_storelps,
37183 avx_<castmode><avxsizesuffix>_<castmode>,
37184 avx512f_<castmode><avxsizesuffix>_<castmode>,
37185 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37186 in condition that at least one operand is not a MEM.
37188 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37190 PR middle-end/79805
37191 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37192 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37194 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37195 gimple_call_nothrow_p flag based on whether original builtin can throw.
37196 If it can, emit following stmts on the fallthrough edge.
37197 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37198 don't create new bb if inserting just debug stmts on the edge, try to
37199 insert them on the fallthru bb or just reset debug stmts.
37201 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37204 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37205 restore recog_data (including the operand rtxes inside it) around
37206 the call to get_insn_template.
37208 2017-03-03 Martin Sebor <msebor@redhat.com>
37210 PR tree-optimization/79699
37211 * context.c (context::~context): Free MPFR caches to avoid
37212 a memory leak on program exit.
37214 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37216 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37217 Use wide_int::ulow () instead of .elt (0).
37219 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37221 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37222 (*pushxf): Limit oF constraint to 32bit targets and add oC
37223 constraint for 64bit targets.
37224 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37225 (*pushdf): Change rmF constraint to rmC.
37227 2017-03-03 Martin Liska <mliska@suse.cz>
37229 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37230 Remove unused variable.
37232 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37235 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37236 is a memory operand, increase num_memory.
37237 (ix86_expand_args_builtin): Likewise.
37239 2017-03-03 Jan Hubicka <jh@suse.cz>
37242 * ipa-devirt.c (maybe_record_node): Properly handle
37243 __cxa_pure_virtual visibility.
37245 2017-03-03 Martin Liska <mliska@suse.cz>
37247 PR tree-optimization/79803
37248 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37250 (pass_loop_prefetch::execute): Disabled optimization if an
37251 assumption about L1 cache size is not met.
37253 2017-03-03 Martin Liska <mliska@suse.cz>
37255 PR rtl-optimization/79574
37256 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37257 (hash_scan_set): Likewise.
37258 (dump_hash_table): Likewise.
37259 (hoist_code): Likewise.
37261 2017-03-03 Richard Biener <rguenther@suse.de>
37263 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37264 in place of uhigh ().
37265 (fixed_convert_from_real): Likewise.
37267 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37270 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37272 2017-03-03 Richard Biener <rguenther@suse.de>
37274 PR middle-end/79818
37275 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37276 TYPE_OVERFLOW_UNDEFINED check.
37278 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37280 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37282 (vector_ae_<mode>_p): Likewise.
37283 (vector_nez_<mode>_p): Likewise.
37284 (vector_ne_v2di_p): Likewise.
37285 (vector_ae_v2di_p): Likewise.
37286 (vector_ne_<mode>_p): Likewise.
37287 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37289 (vsx_tsqrt<mode>2_fe): Likewise.
37291 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37294 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37296 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37298 PR rtl-optimization/79780
37299 * cprop.c (one_cprop_pass): When second and further conditional trap
37300 in a single basic block is turned into an unconditional trap, turn it
37301 into a deleted note to avoid RTL verification failures.
37303 2017-03-02 Richard Biener <rguenther@suse.de>
37305 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37307 2017-03-02 Richard Biener <rguenther@suse.de>
37309 PR tree-optimization/79345
37311 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37312 param and abort the walk, returning -1 if it is hit.
37313 (walk_aliased_vdefs): Take a limit param and pass it on.
37314 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37315 defaulting to 0 and return a signed int.
37316 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37317 (check_defs): New helper.
37318 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37319 about uninitialized memory.
37320 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37321 bogus uninitialized warning.
37322 (fixed_convert_from_real): Likewise.
37324 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37326 PR tree-optimization/66768
37327 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37328 iv_use if base object can't be determined.
37330 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37332 PR tree-optimization/79345
37333 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37334 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37335 (get_pattern_stats): Initialize it.
37336 * genemit.c (gen_expand): Verify match_scratch numbers come after
37337 match_operand/match_dup numbers.
37338 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37339 match_scratch numbers.
37340 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37342 * config/s390/s390.md (trunctdsd2): Likewise.
37344 2017-03-02 Richard Biener <rguenther@suse.de>
37346 * wide-int.h (wide_int_storage::operator=): Implement in terms
37349 2017-03-02 Richard Biener <rguenther@suse.de>
37351 PR tree-optimization/79777
37352 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37353 the to insert expression to sth existing.
37355 2017-03-01 Martin Sebor <msebor@redhat.com>
37357 PR middle-end/79692
37358 * gimple-ssa-sprintf.c
37359 (directive::known_width_and_precision): New function.
37360 (format_integer): Use it.
37361 (get_mpfr_format_length): Consider the full range of precision
37362 when computing %g output with the # flag. Set the likely byte
37363 count to 3 rather than 1 when precision is indeterminate.
37364 (format_floating): Correct the lower bound of precision.
37366 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37368 * doc/invoke.texi: Document default code model for 64-bit Linux.
37370 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37373 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37374 udiv rather than div since input pattern is unsigned.
37376 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37378 * config/i386/i386.c (print_reg): Warn for values of
37379 unsupported size in integer register.
37381 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37384 * config/rs6000/predicates.md (current_file_function_operand): Do
37385 not allow self calls to be local if the function is replaceable.
37387 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37390 * config/rs6000/altivec.h (vec_ctz and others): Change the
37391 preprocessor macro that controls conditional compilation from
37392 _ARCH_PWR9 to __POWER9_VECTOR__.
37393 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37394 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37395 control (instead of _ARCH_PWR9 control) so that template
37396 definition uses power9-specific function.
37397 (vec_any_eq): Likewise.
37398 (vec_all_ne): Change macro definition to use a power9-specific
37399 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37400 _ARCH_PWR9 control).
37401 (vec_any_eq) Likewise.
37402 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37403 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37404 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37405 support for xvcmpnedp instruction.
37406 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37407 macro expansion so that Power9 implementation of vec_all_ne does
37408 not use the AltiVec predicate framework.
37409 (VCMPNEH_P): Likewise.
37410 (VCMPNEW_P): Likewise.
37411 (VCMPNED_P): Likewise.
37412 (VCMPNEFP_P): Likewise.
37413 (VCMPNEDP_P): Likewise.
37414 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37415 implementation of vec_any_eq to not use AltiVec predicate
37417 (VCMPAEH_P): Likewise.
37418 (VCMPAEW_P): Likewise.
37419 (VCMPAED_P): Likewise.
37420 (VCMPAEFP_P): Likewise.
37421 (VCMPAEDP_P): Likewise.
37422 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37423 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37424 not use the AltiVec predicate framework.
37425 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37426 of vec_any_eq to not use AltiVec predicate framework.
37427 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37428 support for predefined __POWER9_VECTOR__ macro to indicate that
37429 Power9 instruction selection is enabled.
37430 (altivec_overloaded_builtins): Remove extraneous
37431 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37432 function argument types RS6000_BTI_bool_V16QI and
37433 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37434 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37435 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37436 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37437 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37438 Power9 for implementations of vec_cmpne. Change the signature for
37439 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37440 (representing vec_all_ne) to remove the previously described first
37441 argument of type RS6000_BTI_INTSI, as this was an artifact of
37442 reliance on the AltiVec predicate framework, which is no longer
37443 used in the implementation of these functions. Add
37444 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37445 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37446 since, unlike the AltiVec predicate framework implementation, we
37447 do not share function descriptors between vec_alle and vec_anyeq.
37448 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37449 set of modes that receive special treatment even when
37450 TARGET_P9_VECTOR is true. The special treatment emits code that
37451 does not depend on Power9 instructions.
37452 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37453 define_expand to not rely on AltiVec predicate framework.
37454 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37456 (vector_ne_v2di_p): Change this define_expand to not rely on
37457 AltiVec predicate framework.
37458 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37460 (vector_ne_<mode>_p): Change this define_expand to not rely on
37461 AltiVec predicate framework.
37462 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37464 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37465 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37466 define_insn pattern.
37467 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37468 define_insn pattern because the xvcmpne<VSs>. instruction is not
37470 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37471 instruction is not supported.
37473 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37475 * config/nvptx/nvptx.c: Include intl.h.
37477 2017-03-01 Martin Jambor <mjambor@suse.cz>
37480 * ipa-prop.h (ipa_bits): Removed field known.
37481 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37482 to pointers. Adjusted their comments to warn about their sharing.
37483 (ipcp_transformation_summary): Change bits to a vector of pointers.
37484 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37485 (ipa_get_ipa_bits_for_value): Declare.
37486 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37487 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37488 (ipa_bits_hash_table): Likewise.
37489 (ipa_vr_ggc_hash_traits): Likewise.
37490 (ipa_vr_hash_table): Likewise.
37491 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37492 being pointers and vr_known being removed.
37493 (ipa_set_jf_unknown): Likewise.
37494 (ipa_get_ipa_bits_for_value): New function.
37495 (ipa_set_jfunc_bits): Likewise.
37496 (ipa_get_value_range): New overloaded functions.
37497 (ipa_set_jfunc_vr): Likewise.
37498 (ipa_compute_jump_functions_for_edge): Use the above functions to
37499 construct bits and vr parts of jump functions.
37500 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37501 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37503 (ipcp_grow_transformations_if_necessary): Also allocate
37504 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37506 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37507 them. Fix too long lines.
37508 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37509 vr_known being removed.
37510 (ipa_read_jump_function): Use new setter functions to construct bits
37511 and vr parts of jump functions or set them to NULL.
37512 (write_ipcp_transformation_info): Adjust for bits being pointers.
37513 (read_ipcp_transformation_info): Likewise.
37514 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37516 Include gt-ipa-prop.h.
37517 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37519 (ipcp_store_bits_results): Likewise.
37520 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37521 Do not write to existing jump functions but use a temporary instead.
37523 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37526 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37527 attempt to use its first operand as BIT_FIELD_REF base.
37529 2017-03-01 Richard Biener <rguenther@suse.de>
37531 PR middle-end/79721
37532 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37533 interpolating formula in wrapping arithmetic.
37534 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37536 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37538 PR tree-optimization/79734
37539 * tree-vect-generic.c (expand_vector_condition): Optimize
37540 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37541 Handle VEC_COND_EXPR where comparison has different inner width from
37542 type's inner width.
37544 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37546 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37547 markup, and similar issues. Remove @opindex entries for things
37548 that aren't options. Add missing -mmpy-option entries.
37550 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37552 PR tree-optimization/79737
37553 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37554 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37555 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37556 instead of byte_size. Formatting fix.
37557 (shift_bytes_in_array_right): Formatting fix.
37559 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37562 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37563 condition on optimize for the leaf function test.
37565 2017-02-28 Martin Liska <mliska@suse.cz>
37568 * read-rtl-function.c (function_reader::handle_unknown_directive):
37569 Bail out when one uses -flto.
37571 2017-02-28 Martin Liska <mliska@suse.cz>
37573 * common.opt: Replace space with tabular for options of <number>
37575 * config/i386/i386.opt: Show <number> value for
37576 -mlarge-data-threshold.
37577 * opts.c (print_filtered_help): Do not display number in hexadecimal
37580 2017-02-28 Martin Liska <mliska@suse.cz>
37582 * common.opt: Fix --help=option -Q for options which are of
37585 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37587 * config/i386/i386.c (print_reg): Error out for values
37588 of 8-bit size in invalid integer register.
37590 2017-02-28 Martin Sebor <msebor@redhat.com>
37592 PR tree-optimization/79691
37593 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37595 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37598 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37599 gcc_unreachable with output_operand_lossage.
37601 2017-02-28 Richard Biener <rguenther@suse.de>
37603 PR tree-optimization/79740
37604 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37606 (visit_nary_op): Insert the nary into the hashtable if we
37607 pattern-matched sth.
37608 * tree-ssa-pre.c (eliminate_insert): Robustify.
37610 2017-02-28 Richard Biener <rguenther@suse.de>
37612 PR middle-end/79731
37613 * fold-const.c (decode_field_reference): Reject out-of-bound
37616 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37618 * config/i386/i386.c: Include intl.h.
37619 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37620 instead of just cond ? "..." : "...".
37621 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37622 * coverage.c (read_counts_file): Likewise.
37623 * omp-offload.c: Include intl.h.
37624 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37625 of just cond ? "..." : "...".
37626 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37627 of just cond ? "..." : "...".
37629 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37632 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37634 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37635 'tune for' CPU name.
37636 * config/arm/arm-cpu-data.h: Regenerated.
37638 2017-02-28 Richard Biener <rguenther@suse.de>
37640 PR tree-optimization/79732
37641 * tree-inline.c (expand_call_inline): Do not shadow var.
37643 2017-02-28 Richard Biener <rguenther@suse.de>
37645 PR tree-optimization/79723
37646 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37647 address-space properly.
37649 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37651 * doc/optinfo.texi (Optimization groups): Fix option used for
37653 * doc/invoke.texi (-fopt-info): Document "omp".
37654 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37655 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37656 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37657 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37658 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37660 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37662 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37663 OPTGROUP_OMP with "omp".
37665 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37668 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37669 for arithmetic shift of unsigned V2DI.
37671 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37673 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37674 arc/linux.h headers.
37675 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37676 (LINK_SPEC): Likewise.
37677 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37678 (EXTRA_SPECS): Likewise.
37679 (STARTFILE_SPEC): Likewise.
37680 (ENDFILE_SPEC): Likewise.
37681 (LIB_SPEC): Likewise.
37682 (TARGET_SDATA_DEFAULT): Likewise.
37683 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37684 (MULTILIB_DEFAULTS): Likewise.
37685 (DWARF2_UNWIND_INFO): Likewise.
37686 * config/arc/big.h: New file.
37687 * config/arc/elf.h: Likewise.
37688 * config/arc/linux.h: Likewise.
37689 * config/arc/t-uClibc: Remove.
37691 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37693 PR tree-optimization/77536
37694 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37695 (tree_transform_and_unroll_loop): Use above function to compute the
37696 estimated niter of unrolled loop and use it when scaling profile.
37697 Also use count info rather than frequency if it's non-zero.
37698 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37699 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37700 (vect_transform_loop): Call above function.
37702 2017-02-27 Richard Biener <rguenther@suse.de>
37704 PR tree-optimization/45397
37705 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37706 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37707 (visit_nary_op): Add pattern matching for CSEing sign-changed
37708 or truncated operations with wider ones.
37710 2017-02-27 Richard Biener <rguenther@suse.de>
37712 PR tree-optimization/79690
37713 * tree-vect-stmts.c (vectorizable_store): Use vector type
37714 built from the DR with address-space.
37716 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37718 * doc/invoke.texi (Optimize Options): Refine the description
37719 of asan-use-after-return.
37721 2017-02-25 Alan Modra <amodra@gmail.com>
37723 PR rtl-optimization/79584
37724 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37725 base, not ad->base_term, the reg within base. Remove assertion
37726 that ad->base == ad->base_term. Replace gen_int_mode using
37727 bogus mode with const0_rtx.
37729 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37731 PR middle-end/79396
37732 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37733 FMA_EXPR like tcc_binary or tcc_unary.
37735 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37738 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37740 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37741 (output_loc_operands): Handle DW_OP_call_ref and
37742 DW_OP_GNU_variable_value.
37743 (struct variable_value_struct): New type.
37744 (struct variable_value_hasher): Likewise.
37745 (variable_value_hash): New variable.
37746 (string_types): Remove.
37747 (copy_loc_descr): New function.
37748 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37749 (prepend_loc_descr_to_each): New function.
37750 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37751 instead of add_loc_descr_to_each if the first argument is single
37752 location list and the second has multiple.
37753 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37754 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37755 when looking for variable value which doesn't have other location info.
37756 (loc_list_from_tree): Formatting fix.
37757 (gen_array_type_die): Simplify DW_AT_string_length handling.
37758 (adjust_string_types): Remove.
37759 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37760 string_types. Call resolve_variable_values.
37761 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37762 (resolve_addr_in_expr): Likewise. Add A argument.
37763 (copy_deref_exprloc): Remove deref argument. Adjust for the
37764 original expression being DW_OP_GNU_variable_value with optionally
37765 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37766 optionally after it.
37767 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37768 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37769 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37770 (variable_value_hasher::hash, variable_value_hasher::equal): New
37772 (resolve_variable_value_in_expr, resolve_variable_value,
37773 resolve_variable_values, note_variable_value_in_expr,
37774 note_variable_value): New functions.
37775 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37778 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37781 * opts.h (handle_generated_option): Add GENERATED_P argument.
37782 * opts-common.c (handle_option): Adjust function comment.
37783 (handle_generated_option): Add GENERATED_P argument, pass it to
37785 (control_warning_option): Pass false to handle_generated_option
37787 * opts.c (maybe_default_option): Pass true to handle_generated_option
37789 * optc-gen.awk: Likewise.
37791 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37793 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37794 a REG, look at the REG it is a SUBREG of.
37795 (splitter for cmpeqsi_t): Ditto.
37797 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37799 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37800 the special USEs with the pattern of the insn, not the insn itself.
37802 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37805 * doc/invoke.texi: Document -mload-store-pairs.
37807 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37808 Sandra Loosemore <sandra@codesourcery.com>
37810 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37811 argument isn't a CONST_INT.
37812 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37813 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37814 (nios2_validate_compare): Bypass alternate compare logic if *op2
37815 is not a CONST_INT.
37816 (ldstwm_operation_p): Return false if first_base is not a REG or
37817 if first_offset is not a CONST_INT.
37819 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37821 * config/cris/cris.md: Use correct operand in a define_peephole2.
37823 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37825 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37827 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37829 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37830 this_insn if it is an INSN or JUMP_INSN.
37831 (force_offsettable): Look at base, not at addr.
37832 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37833 on things that aren't necessarily CONST_INTs.
37835 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37837 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37838 -mfpmath=sse is the default also for x86-32 targets with SSE2
37839 instruction set when @option{-ffast-math} is enabled
37841 2017-02-24 Jeff Law <law@redhat.com>
37843 PR rtl-optimizatoin/79286
37844 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37847 2017-02-24 Richard Biener <rguenther@suse.de>
37849 PR tree-optimization/79389
37850 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37853 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37855 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37856 function comment to reflect reality.
37857 (loop_exits_before_overflow): Fix typo in function description.
37859 2017-02-24 Richard Biener <rguenther@suse.de>
37861 PR tree-optimization/79389
37862 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37863 properly that a threading opportunity exists. Detect conditional
37864 copy/constant propagation opportunities.
37866 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37868 * config/visium/visium.md (type): Add trap.
37869 (b): New mode attribute.
37870 (*btst): Rename into...
37871 (*btst<mode>): ...this and adjust.
37872 (*cbranchsi4_btst_insn): Rename into...
37873 (*cbranch<mode>4_btst_insn): ...this and adjust.
37874 (trap): New define_insn.
37876 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37878 PR tree-optimization/79389
37879 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37880 (noce_reversed_cond_code): New function.
37881 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37882 reversed_comparison_code. Formatting fix.
37883 (noce_try_store_flag): Test rev_cond != NULL in addition to
37884 reversed_comparison_code.
37885 (noce_try_store_flag_constants): Likewise.
37886 (noce_try_store_flag_mask): Likewise.
37887 (noce_try_addcc): Use rev_cond if non-NULL instead of
37888 reversed_comparison_code.
37889 (noce_try_cmove_arith): Likewise. Formatting fixes.
37890 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37891 (noce_find_if_block): Initialize rev_cond.
37892 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37893 instead of false as last argument never attempt to reverse it
37896 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37898 PR tree-optimization/79663
37899 * tree-predcom.c (combine_chains): Process refs in reverse order
37900 only for ZERO length chains, and add explaining comment.
37902 2017-02-23 Jeff Law <law@redhat.com>
37904 PR tree-optimization/79578
37905 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37906 in call to operand_equal_p.
37908 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37911 * config/i386/cpuid.h: Fix another undefined behavior.
37913 2017-02-23 Richard Biener <rguenther@suse.de>
37915 PR tree-optimization/79683
37916 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37917 vector types for data-refs.
37919 2017-02-23 Martin Liska <mliska@suse.cz>
37921 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37923 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37925 PR middle-end/79665
37926 * internal-fn.c (get_range_pos_neg): Moved to ...
37927 * tree.c (get_range_pos_neg): ... here. No longer static.
37928 * tree.h (get_range_pos_neg): New prototype.
37929 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37930 are known to be in between 0 and signed maximum inclusive, try to
37931 expand both unsigned and signed divmod and use the cheaper one from
37934 2017-02-22 Jeff Law <law@redhat.com>
37936 PR tree-optimization/79578
37937 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37938 to compare base operands.
37940 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37943 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37944 gpc_reg_operand instead of fpr_reg_operand.
37946 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37948 * config/mips/mips.c (mips_return_in_memory): Force FP
37949 vector types to be returned in memory for o32 ABI.
37951 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37953 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37954 instead of DW_TAG_member for static data member declarations and don't
37955 set no_linkage_name for static inline data members.
37956 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37959 2017-02-22 Martin Liska <mliska@suse.cz>
37961 * doc/invoke.texi: Replace inequality signs with square brackets
37964 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37967 * lra-constraints.c (simplify_operand_subreg): Handle
37968 WORD_REGISTER_OPERATIONS targets.
37970 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37973 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37974 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37975 elimination by swapping fld*.
37977 2017-02-22 Richard Biener <rguenther@suse.de>
37979 PR tree-optimization/79673
37980 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37981 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37982 irrelevant address-space qualifiers and avoiding a
37983 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37985 2017-02-22 Richard Biener <rguenther@suse.de>
37987 PR tree-optimization/79666
37988 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37989 to not symbolically negate if that may introduce undefined
37992 2017-02-22 Martin Liska <mliska@suse.cz>
37995 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37996 * data-streamer-out.c (streamer_write_gcov_count_stream):
37998 * value-prof.c (stream_out_histogram_value): Make assert more
37999 precise based on type of counter.
38001 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
38004 * config/i386/i386.md (standard_x87sse_constant_load splitter):
38005 Use nonimmediate_operand instead of memory_operand for operand 1.
38006 (float-extend standard_x87sse_constant_load splitter): Ditto.
38008 2017-02-21 Jeff Law <law@redhat.com>
38010 PR tree-optimization/79621
38011 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
38012 blocks with edges to themselves.
38014 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38017 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
38018 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
38019 Use gimple_call_builtin_p.
38022 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
38023 on temporarily removed DEBUG_INSNs.
38025 PR tree-optimization/79649
38026 * tree-loop-distribution.c (classify_partition): Give up on
38027 non-generic address space loads/stores.
38029 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
38031 * doc/loop.texi (Loop manipulation): Remove nonexistent
38032 tree_ssa_loop_version from the documentation.
38033 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
38035 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38038 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
38039 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
38040 * config/rs6000/rs6000.c: Include except.h.
38041 (rs6000_expand_split_stack_prologue): Call
38042 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
38044 2017-02-21 Martin Jambor <mjambor@suse.cz>
38047 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
38048 have been analyzed.
38050 2017-02-21 Martin Jambor <mjambor@suse.cz>
38052 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
38053 for backward compatibility only.
38054 * doc/invoke.texi (Option Summary): Remove all references to
38055 -fipa-cp-alignment.
38057 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
38061 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38063 * lra-constraints.c (curr_insn_transform): Handle
38064 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38066 2017-02-21 Martin Liska <mliska@suse.cz>
38068 * config/i386/i386.opt: Replace -masm-dialect with -masm.
38070 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
38072 PR translation/79638
38073 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
38075 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
38078 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
38079 (arm_function_ok_for_sibcall): Return false for an indirect call by
38080 descriptor if all the argument registers are used.
38081 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
38082 alignment of the function.
38084 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38086 PR tree-optimization/61441
38087 * simplify-rtx.c (simplify_const_unary_operation): For
38088 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
38089 the sNaN unmodified.
38091 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
38093 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
38094 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
38095 instead of SYSTEM_HEADER_DIR.
38097 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
38098 Martin Liška <mliska@suse.cz>
38100 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
38101 Fix typos and grammar, use active voice, and clarify.
38103 2017-02-20 Marek Polacek <polacek@redhat.com>
38105 PR middle-end/79537
38106 * gimplify.c (gimplify_expr): Handle unused *&&L;.
38109 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
38111 2017-02-20 Jakub Jelinek <jakub@redhat.com>
38114 * config/i386/i386.c (ix86_expand_builtin): Handle
38115 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
38116 ix86_builtins_isa[fcode].isa as a requirement of those
38117 flags and any other flag in the bitmask.
38118 (ix86_init_mmx_sse_builtins): Use 0 instead of
38119 ~OPTION_MASK_ISA_64BIT as mask.
38120 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
38121 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
38122 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
38123 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
38125 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38128 * lra-constraints.c (split_reg): Check requested split mode
38129 is supported by the register.
38131 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38133 * lra-constraints.c (simplify_operand_subreg): Remove early
38136 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38139 * lra-constraints.c (curr_insn_transform): Tighten condition
38140 for converting SUBREG reloads from OP_OUT to OP_INOUT.
38142 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38145 * lra-constraints.c (curr_insn_transform): Handle
38146 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38148 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
38151 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
38153 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
38155 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
38158 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38161 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38163 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38166 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38168 * final.c (last_columnnum, override_columnnum): New variables.
38169 (final_start_function): Set last_columnnum, pass it to begin_prologue
38170 hook and pass 0 to dwarf2out_begin_prologue.
38171 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38172 to source_line debug hook.
38173 (notice_source_line): Compute last_columnnum and for debug_column_info
38174 return true on column changes.
38175 * debug.h (struct gcc_debug_hooks): Add column argument to
38176 source_line and begin_prologue hooks.
38177 (debug_nothing_int_charstar_int_bool): Remove prototype.
38178 (debug_nothing_int_int_charstar,
38179 debug_nothing_int_int_charstar_int_bool): New prototypes.
38180 (dwarf2out_begin_prologue): Add column argument.
38181 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38182 begin_prologue hooks.
38183 (debug_nothing_int_charstar_int_bool): Remove.
38184 (debug_nothing_int_int_charstar,
38185 debug_nothing_int_int_charstar_int_bool): New functions.
38186 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38187 through to dwarf2out_source_line.
38188 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38189 (dwarf2out_source_line): Add column argument, emit it if requested.
38190 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38192 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38193 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38194 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38195 through to dwarf2out_begin_prologue.
38196 (vmsdbgout_source_line): Add column argument, pass it through to
38197 dwarf2out_source_line.
38198 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38199 dbxout_source_line caller.
38200 (dbxout_source_line): Add column argument.
38202 * common.opt (gno-column-info, gcolumn-info): New options.
38203 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38204 (check_die): Also test for multiple DW_AT_decl_column attributes.
38205 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38206 DW_AT_decl_column if requested.
38207 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38209 (gen_variable_die): Likewise.
38210 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38211 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38214 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38215 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38216 (ix86_handle_option): Handle OPT_m3dnowa.
38217 * doc/invoke.texi (-m3dnowa): Document.
38218 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38219 -m3dnowa instead of -m3dnow -march=athlon.
38222 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38223 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38225 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38228 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38229 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38230 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38231 generator for vsx_xxpermdi_<mode>_be.
38232 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38233 force big-endian semantics.
38234 (vsx_xxpermdi_<mode>_be): New define_expand with same
38235 implementation as previous version of vsx_xxpermdi_<mode>.
38237 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38239 PR tree-optimization/79327
38240 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38241 variable, its initialization and use.
38243 2017-02-17 Julia Koval <julia.koval@intel.com>
38245 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38246 (OPTION_MASK_ISA_PKU_UNSET): New.
38247 (ix86_handle_option): Handle -mrdpid.
38248 * config/i386/cpuid.h (bit_RDPID): New.
38249 * config/i386/driver-i386.c (host_detect_local_cpu):
38250 Detect RDPID feature.
38251 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38252 * config/i386/i386-c.c (ix86_target_macros_internal):
38254 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38255 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38256 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38257 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38258 * config/i386/i386.md (define_insn "rdpid"): New.
38259 * config/i386/i386.opt Add -mrdpid.
38260 * config/i386/immintrin.h (_rdpid_u32): New.
38262 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38264 PR rtl-optimization/79541
38265 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38266 instead of transforming it into USE.
38268 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38270 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38271 If HONOR_SNANS (SFmode) force the input to a register.
38272 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38273 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38274 an frsp or similar insn.
38276 2017-02-17 Martin Liska <mliska@suse.cz>
38278 PR rtl-optimization/79577
38279 * params.def (selsched-max-sched-times): Increase minimum to 1.
38281 2017-02-17 Martin Liska <mliska@suse.cz>
38283 PR rtl-optimization/79574
38284 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38286 2017-02-17 Martin Liska <mliska@suse.cz>
38288 PR tree-optimization/79529
38289 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38290 ssa_defined_default_def_p to handle cases which are implicitly
38292 * tree-ssa.c (ssa_defined_default_def_p): New function.
38293 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38294 which are implicitly defined.
38295 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38297 2017-02-17 Richard Biener <rguenther@suse.de>
38299 PR middle-end/79576
38300 * params.def (max-ssa-name-query-depth): Limit to 10.
38302 2017-02-17 Richard Biener <rguenther@suse.de>
38304 PR tree-optimization/79552
38305 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38308 2017-02-17 Richard Biener <rguenther@suse.de>
38311 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38313 2017-02-17 Marek Polacek <polacek@redhat.com>
38315 PR middle-end/79536
38316 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38317 (fold_negate_expr): New wrapper.
38319 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38321 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38322 Correct terminology and de-emphasize pre-standard behavior.
38324 2017-02-16 Alan Modra <amodra@gmail.com>
38326 PR rtl-optimization/79286
38327 * ira.c (def_dominates_uses): New function.
38328 (update_equiv_regs): Don't create an equivalence for insns that
38329 may trap where the register def does not dominate the use.
38331 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38333 PR rtl-optimization/78127
38334 * lra.c (lra): Call lra_eliminate before finish the loop after
38337 2017-02-16 Richard Biener <rguenther@suse.de>
38339 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38341 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38342 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38343 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38344 (isl_val_int_from_wi): New function.
38345 (extract_affine_gmp): Rename to ...
38346 (extract_affine_wi): ... this, take a widest_int.
38347 (extract_affine_int): Just wrap extract_affine_wi.
38348 (add_param_constraints): Use isl_val_int_from_wi.
38349 (add_loop_constraints): Likewise, and extract_affine_wi.
38351 2017-02-15 Jeff Law <law@redhat.com>
38353 PR middle-end/79521
38354 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38355 ira_init_register_move_cost_if_necessary.
38357 2017-02-15 Martin Sebor <msebor@redhat.com>
38359 PR middle-end/32003
38360 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38361 removed in a prior commit.
38363 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38365 PR tree-optimization/79347
38366 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38367 counters during peeling.
38369 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38371 * Makefile.in (site.exp): Remove "set ISLVER".
38373 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38376 * real.c (real_from_integer): Call real_convert even for decimal.
38378 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38381 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38383 2017-02-14 Andrew Pinski <apinski@cavium.com>
38385 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38386 cores and change the partno/implementer to be correct.
38387 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38388 the 'B" as the implementer.
38389 * config/aarch64/aarch64-tune.md: Regenerate.
38391 2017-02-14 Carl Love <cel@us.ibm.com>
38393 * config/rs6000/rs6000.c: Add case statement entry to make the
38394 xvcvuxdsp built-in argument unsigned.
38395 * config/rs6000/vsx.md: Fix the source and return operand types so they
38396 match the instruction definitions from the ISA document. Fix typo
38397 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38400 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38403 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38404 member early_clobber_alts.
38405 * lra-lives.c (reg_early_clobber_p): New.
38406 (process_bb_lives): Use it.
38407 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38408 (debug_operand_data): Initialize early_clobber_alts.
38409 (setup_operand_alternative): Set up early_clobber_alts.
38410 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38411 alternatives to new_insn_reg.
38412 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38414 (lra_update_insn_regno_info): Pass the new arg.
38416 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38418 PR middle-end/79505
38419 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38420 (new_oacc_loop_raw): Don't clear already cleared fields.
38423 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38424 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38425 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38427 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38430 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38432 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38435 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38436 the extra instruction to the right place to store 128-bit constant
38439 2017-02-14 Martin Sebor <msebor@redhat.com>
38441 PR middle-end/79448
38442 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38443 warning for strings of unknown length.
38445 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38447 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38449 2017-02-14 Jeff Law <law@redhat.com>
38452 * ira-costs.c (scan_one_insn): Initialize register move costs
38453 for pseudos seen in USE/CLOBBER insns.
38455 PR tree-optimization/79095
38456 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38457 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38458 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38459 if the operands are known to be not equal, then the resulting range
38461 (intersect_ranges): If the new range is ~[0,0] and the old range is
38462 wide, then prefer ~[0,0].
38463 * tree-vrp.c (overflow_comparison_p_1): New function.
38464 (overflow_comparison_p): New function.
38465 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38466 if NAME is used in an overflow test.
38467 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38468 overflow check that can be expressed as an equality test, then adjust
38469 ops to be that equality test.
38471 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38473 * config/s390/s390-builtin-types.def: Remove flags argument.
38474 * config/s390/s390.c (s390_init_builtins): Likewise.
38476 2017-02-14 Martin Liska <mliska@suse.cz>
38478 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38479 vector. Fix trailing white spaces.
38481 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38483 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38486 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38488 PR rtl-optimization/68664
38489 * config/arm/arm.c (arm_sched_can_speculate_insn):
38490 New function. Declare prototype.
38491 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38493 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38495 PR rtl-optimization/68664
38496 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38498 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38500 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38502 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38503 max skip bytes for function, loop and jump.
38505 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38507 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38508 ABS_EXPR for gimple dump.
38510 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38513 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38515 PR tree-optimization/79408
38516 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38517 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38518 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38519 also if rhs1 is INTEGER_CST.
38521 2017-02-14 Richard Biener <rguenther@suse.de>
38523 PR middle-end/79432
38524 * tree-into-ssa.c (insert_phi_nodes): When the function can
38525 have abnormal edges rewrite SSA names with broken use-def
38526 dominance out of SSA and register them for PHI insertion.
38528 2017-02-13 Martin Sebor <msebor@redhat.com>
38530 PR middle-end/79496
38531 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38532 clearing info.nowrite flag when snprintf size argument is a range.
38534 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38536 * cprop.c (cprop_jump): Add missing space in string literal.
38537 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38538 (get_constraint_for_component_ref): Likewise.
38539 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38540 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38541 * lra-constraints.c (process_alt_operands): Likewise.
38542 * ipa-inline.c (inline_small_functions): Likewise.
38543 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38544 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38545 * trans-mem.c (diagnose_tm_1_op): Likewise.
38546 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38547 (grid_parallel_clauses_gridifiable): Likewise.
38549 * config/nvptx/mkoffload.c (process): Add space in between
38552 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38553 "MOD4_SSE_REGS" and "ALL_REGS".
38555 * spellcheck.c (test_data): Add , in between "foo" and "food".
38557 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38560 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38561 boundary crossing check and subsequent code generation agree.
38563 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38565 * config/aarch64/aarch64.c (has_memory_op): Delete.
38566 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38569 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38571 PR rtl-optimization/79388
38572 PR rtl-optimization/79450
38573 * combine.c (distribute_notes): When removing TEM_INSN for which
38574 corresponding dest has last value recorded, invalidate that last
38577 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38579 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38580 of explicit '@'. Add missing assembly comment marker on branch costs
38583 2017-02-13 Nathan Sidwell <nathan@acm.org>
38585 * gengtype-lex.l (<in_struct>): Add '/'.
38587 2017-02-13 Martin Liska <mliska@suse.cz>
38590 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38592 2017-02-13 Richard Biener <rguenther@suse.de>
38594 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38596 * configure: Re-generate.
38597 * config.in: Likewise.
38598 * graphite-dependences.c: Simplify as if
38599 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38600 * graphite-isl-ast-to-gimple.c: Likewise.
38601 * graphite-optimize-isl.c: Likewise.
38602 * graphite-poly.c: Likewise.
38603 * graphite-sese-to-poly.c: Likewise.
38604 * graphite.h: Likewise.
38605 * toplev.c: Include isl/version.h and use isl_version () for
38606 printing the ISL version.
38607 * doc/install.texi: Update ISL requirement.
38609 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38611 * doc/standards.texi (Standards): Update reference to
38614 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38616 * doc/extend.texi (Named Address Spaces): sourceware.org now
38618 * doc/install.texi (Binaries): Ditto.
38621 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38623 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38624 standard terminology "stringize"/"stringizing" throughout.
38625 * doc/cppinternals.texi: Likewise.
38627 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38629 * doc/extend.texi: Fix some spelling mistakes and typos.
38630 * doc/invoke.texi: Likewise.
38632 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38635 * params.def (inline-min-speedup) Change from 10 to 8.
38637 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38639 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38642 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38645 * ipa-inline-analysis.c (get_minimal_bb): New function.
38646 (record_modified): Use it.
38647 (remap_edge_change_prob): Handle also ancestor functions.
38649 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38651 * doc/contrib.texi (Contributors): Remove broken link into
38652 the Mauve CVS repository.
38654 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38656 PR middle-end/79454
38657 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38658 result computation whenever lhs doesn't have vector mode, not
38659 just when it has BLKmode.
38661 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38663 * doc/makefile.texi (profiledbootstrap): Refer to the
38664 installation instructions only in textual form.
38666 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38669 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38671 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38673 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38674 (Specific): Update mingw-w64 reference.
38676 (Specific): Remove broken link to Renesas RX processor.
38678 2017-02-10 Richard Biener <rguenther@suse.de>
38680 * toplev.c (process_options): Do not mention obsolete graphite
38681 options when printing sorry message about missing graphite support.
38682 Mention -floop-nest-optimize.
38684 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38686 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38687 (vtst_p16): Likewise.
38688 (vtstq_p8): Likewise.
38689 (vtstq_p16): Likewise.
38691 (vtstq_p64): Likewise.
38692 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38693 (vset_lane_p64): New.
38694 (vsetq_lane_p64): New.
38696 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38698 PR tree-optimization/79411
38699 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38700 stmt operands are SSA_NAMEs used in abnormal phis.
38701 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38704 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38707 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38710 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38712 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38714 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38716 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38720 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38721 not arbitrary TREE_CONSTANT.
38724 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38725 "omp declare target link" attribute unless is_global_var.
38726 * omp-offload.c (find_link_var_op): Likewise.
38728 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38729 Chung-Lin Tang <cltang@codesourcery.com>
38731 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38733 (gimplify_adjust_omp_clauses): Don't delete TILE.
38734 (gimplify_omp_for): Deal with TILE.
38735 * internal-fn.c (expand_GOACC_TILE): New function.
38736 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38738 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38739 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38741 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38742 avoid DIV for outermost collapse var.
38743 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38744 Remove out of date comments, fix whitespace.
38745 * omp-general.c (omp_extract_for_data): Deal with tiling.
38746 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38747 adjust OLF_DIM_BASE value.
38748 (struct omp_for_data): Add tiling field.
38749 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38750 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38751 for auto loops. Remove default auto determining, moved to
38752 oacc_loop_fixed_partitions.
38753 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38754 stmts, add e_mask field.
38755 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38756 (oacc_thread_numbers): Use oacc_dim_call.
38757 (oacc_xform_tile): New.
38758 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38759 (finish_oacc_loop): Adjust for ifns vector.
38760 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38761 add case for GOACC_TILE fns.
38762 (oacc_loop_xform_loop): Delete.
38763 (oacc_loop_process): Iterate over call list directly, and add
38764 handling for GOACC_TILE fns.
38765 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38767 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38768 vector partitioning to outer loops. Assign 2 partitions to loops
38769 when available. Add TILE handling.
38770 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38771 (execite_oacc_device_lower): Process GOACC_TILE fns,
38772 ignore unknown specs.
38773 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38774 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38775 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38777 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38779 * configure.ac (ACX_BUGURL): Update.
38780 * configure: Regenerate.
38782 2017-02-09 Richard Biener <rguenther@suse.de>
38784 PR tree-optimization/69823
38785 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38786 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38788 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38790 * config/arc/arc-c.def: Add __NPS400__ definition.
38791 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38792 (TARGET_NPS400): Define.
38794 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38796 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38798 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38799 pointer, arch_info.
38800 (arc_cpu_types): Fill the arch_info field with a pointer into the
38801 arc_arch_types table.
38802 (arc_selected_cpu): Declare.
38803 * config/arc/arc.c (arc_selected_cpu): Make global.
38804 (arc_selected_arch): Delete.
38805 (arc_base_cpu): Delete.
38806 (arc_override_options): Remove references to deleted variables,
38807 update access to arch information.
38808 (ARC_OPT): Update access to arch information.
38809 (ARC_OPTX): Likewise.
38810 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38811 (TARGET_ARC600): Update access to arch information.
38812 (TARGET_ARC601): Likewise.
38813 (TARGET_ARC700): Likewise.
38814 (TARGET_EM): Likewise.
38815 (TARGET_HS): Likewise.
38816 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38819 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38822 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38823 condition/operands for integer GE/LE/GEU/LEU operations.
38825 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38827 PR translation/79397
38828 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38831 2017-02-08 Martin Jambor <mjambor@suse.cz>
38834 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38835 whether allocation happened.
38836 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38837 nothing was allocated.
38839 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38841 PR tree-optimization/79408
38842 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38843 constant, but SSA_NAME with a known integer range, use the minimum
38844 of that range instead of op1 to determine if modulo can be replaced
38845 with its first operand.
38847 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38849 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38851 2017-02-08 Richard Biener <rguenther@suse.de>
38853 PR tree-optimization/71824
38854 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38855 Check all loops contained in the merged region.
38857 2017-02-07 Andrew Pinski <apinski@cavium.com>
38859 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38861 2017-02-07 Andrew Pinski <apinski@cavium.com>
38863 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38864 (thunderxt88): Likewise.
38865 (thunderxt81): Disable LSE and change v8.1 to v8.
38866 (thunderxt83): Likewise.
38868 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38869 Richard Biener <rguenther@suse.de>
38871 PR middle-end/79399
38872 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38873 type from int to size_t.
38874 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38876 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38878 PR rtl-optimization/79386
38879 * cprop.c (bypass_conditional_jumps): Initialize
38880 bypass_last_basic_block already before splitting bbs after
38881 unconditional traps...
38882 (bypass_conditional_jumps): ... rather than here.
38885 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38886 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38887 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38888 fix -masm=intel patterns.
38890 2017-02-07 Richard Biener <rguenther@suse.de>
38892 PR tree-optimization/79256
38893 PR middle-end/79278
38894 * builtins.c (get_object_alignment_2): Use min_align_of_type
38895 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38896 and ADJUST_FIELD_ALIGN.
38898 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38900 * doc/tm.texi: Regenerate.
38901 * stor-layout.c (layout_decl): Adjust.
38902 (update_alignment_for_field): Likewise.
38903 (place_field): Likewise.
38904 (min_align_of_type): Likewise.
38905 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38906 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38907 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38908 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38909 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38910 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38911 * config/i386/i386.c (x86_field_alignment): Likewise.
38912 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38913 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38914 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38915 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38916 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38917 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38921 2017-01-30 Richard Biener <rguenther@suse.de>
38923 PR tree-optimization/79256
38924 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38925 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38928 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38930 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38931 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38932 builtins to SImode and emit a zero-extend, if necessary.
38934 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38936 * docs/invoke.texi (RISC-V Options): Alphabetize.
38938 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38940 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38943 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38945 * config/riscv/riscv.c: New file.
38946 * common/config/riscv/riscv-common.c: Likewise.
38947 * config.gcc: Likewise.
38948 * config/riscv/constraints.md: Likewise.
38949 * config/riscv/elf.h: Likewise.
38950 * config/riscv/generic.md: Likewise.
38951 * config/riscv/linux.h: Likewise.
38952 * config/riscv/multilib-generator: Likewise.
38953 * config/riscv/peephole.md: Likewise.
38954 * config/riscv/pic.md: Likewise.
38955 * config/riscv/predicates.md: Likewise.
38956 * config/riscv/riscv-builtins.c: Likewise.
38957 * config/riscv/riscv-c.c: Likewise.
38958 * config/riscv/riscv-ftypes.def: Likewise.
38959 * config/riscv/riscv-modes.def: Likewise.
38960 * config/riscv/riscv-opts.h: Likewise.
38961 * config/riscv/riscv-protos.h: Likewise.
38962 * config/riscv/riscv.h: Likewise.
38963 * config/riscv/riscv.md: Likewise.
38964 * config/riscv/riscv.opt: Likewise.
38965 * config/riscv/sync.md: Likewise.
38966 * config/riscv/t-elf-multilib: Likewise.
38967 * config/riscv/t-linux: Likewise.
38968 * config/riscv/t-linux-multilib: Likewise.
38969 * config/riscv/t-riscv: Likewise.
38970 * configure.ac: Likewise.
38971 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38972 Waterman as RISC-V maintainers.
38973 * doc/install.texi: Add RISC-V entries.
38974 * doc/invoke.texi: Add RISC-V options section.
38975 * doc/md.texi: Add RISC-V constraints section.
38976 * configure: Regenerated.
38978 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38981 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38982 false values to be constant vectors with all 0 or all 1 bits set.
38983 (vcondu<mode><mode>): Likewise.
38984 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38986 (fpmask_comparison_operator): Update comment.
38987 (vecint_comparison_operator): New predicate.
38988 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38989 vector conditionals when the true and false values are constant
38990 vectors with all 0 bits or all 1 bits set.
38992 2017-02-06 Martin Sebor <msebor@redhat.com>
38994 PR tree-optimization/79376
38995 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38997 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38999 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
39000 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
39001 to simplify split condition.
39003 2017-02-06 Jakub Jelinek <jakub@redhat.com>
39005 * omp-expand.c (oxpand_omp_atomic_fetch_op,
39006 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
39009 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
39011 PR rtl-optimization/68664
39012 * target.def (can_speculate_insn): New hook.
39013 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
39014 * doc/tm.texi: Regenerate.
39015 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
39016 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
39017 (rs6000_sched_can_speculate_insn): New function.
39019 2017-02-06 Jakub Jelinek <jakub@redhat.com>
39021 PR tree-optimization/79284
39022 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
39023 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
39024 vectorizable_mask_load_store, vectorizable_operation,
39025 vect_is_simple_cond, get_same_sized_vectype): Use it instead
39026 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
39027 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
39028 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
39029 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
39030 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
39031 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
39032 is_gimple_assign (stmt). Replace another such test with
39033 is_gimple_assign (stmt).
39035 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39038 * config/avr/avr.c (rtl-iter.h): Include it.
39039 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
39040 (avr_legitimate_combined_insn): ...and implementation.
39042 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
39044 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
39045 * config/s390/s390.c (s390_const_operand_ok)
39046 (s390_canonicalize_comparison, s390_extract_part)
39047 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
39048 (s390_contiguous_bitmask_p, s390_rtx_costs)
39049 (legitimize_pic_address): Likewise.
39050 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
39051 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
39052 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
39053 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
39054 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
39056 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39058 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
39059 REGNO($0) == REGNO($1).
39061 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39063 * config/s390/linux.h(SIZE_TYPE): Add comment.
39065 2017-02-06 Julian Brown <julian@codesourcery.com>
39066 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39067 Virendra Pathak <virendra.pathak@broadcom.com>
39069 * config/aarch64/aarch64-cores.def: Change the scheduler
39071 * config/aarch64/aarch64.md: Include thunderx2t99.md.
39072 * config/aarch64/thunderx2t99.md: New file.
39074 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39076 * doc/standards.texi (Go Language): Update link to language
39079 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39081 * tree-eh.c (lower_resx): Sanitize profile.
39082 (cleanup_empty_eh_move_lp): Likewise.
39084 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39087 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
39089 * cfgloopmanip.h (loop_version): Update prototype.
39090 * modulo-sched.c (sms_schedule): Update call of loop_version.
39091 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
39092 * tree-parloops.c (gen_parallel_loop): Likewise.
39093 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
39094 * tree-ssa-loop-split.c (split_loop): Likewise.
39095 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
39096 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
39098 2017-02-05 Martin Liska <mliska@suse.cz>
39101 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
39103 (print_operand_address): Initialize a struct to zero.
39105 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39107 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
39108 garbage collector only in textual form.
39110 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39112 * doc/extend.texi (x86 specific memory model extensions for
39113 transactional memory): Simplify a phrase.
39115 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
39118 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
39119 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
39120 (atomic_storedi_1): Likewise.
39122 2017-02-04 Jakub Jelinek <jakub@redhat.com>
39124 PR tree-optimization/79338
39125 * tree-parloops.c (gather_scalar_reductions): Don't call
39126 vect_analyze_loop_form for loop->inner before destroying loop's
39129 2017-02-03 Martin Sebor <msebor@redhat.com>
39131 PR tree-optimization/79327
39132 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
39133 when precision has resulted in leading zeros.
39134 (format_integer): Adjust the likely counter to assume an unknown
39135 argument that may be zero is non-zero.
39137 2017-02-03 Jason Merrill <jason@redhat.com>
39140 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
39141 avoid copying non-taken branch.
39143 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39145 PR tree-optimization/79340
39146 * tree-vect-loop.c (vectorizable_reduction): Release
39147 vec_defs elements after safe_splicing them into other vectors.
39150 PR tree-optimization/79327
39151 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
39152 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
39154 (format_integer): Use wide_int_to_tree instead of build_int_cst
39155 + to_?hwi. If argmin is NULL, just set argmin and argmax to
39156 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
39157 of shortest and longest sequence.
39159 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39161 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39162 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39164 2017-02-03 Walter Lee <walt@tilera.com>
39167 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39168 after initial stackframe link reg save.
39169 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39171 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39174 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39175 wu for stxssp alternative.
39177 2017-02-03 Martin Sebor <msebor@redhat.com>
39179 PR tree-optimization/79352
39180 * gimple-fold.c (get_range_strlen): Add argument.
39181 (get_range_strlen): Change return type to bool.
39182 (get_maxval_strlen): Pass in a dummy argument.
39183 * gimple-fold.h (get_range_strlen): Change return type to bool.
39184 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39185 * tree.h (array_at_struct_end_p): Add argument.
39186 * tree.c (array_at_struct_end_p): Handle it.
39188 2017-02-03 Martin Liska <mliska@suse.cz>
39191 * multiple_target.c (create_dispatcher_calls): Redirect edge
39192 from a caller of a dispatcher.
39193 (expand_target_clones): Make the clones local.
39194 (ipa_target_clone): Do both target clones and resolvers.
39195 (ipa_dispatcher_calls): Remove the pass.
39196 (pass_dispatcher_calls::gate): Likewise.
39197 (make_pass_dispatcher_calls): Likewise.
39198 * passes.def (pass_target_clone): Put as very first IPA early
39201 2017-02-03 Martin Liska <mliska@suse.cz>
39203 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39204 in case of a function with ifunc attribute.
39206 2017-02-03 Martin Liska <mliska@suse.cz>
39208 * cgraph.c (cgraph_node::dump): Dump function version info.
39209 * symtab.c (symtab_node::dump_base): Add missing new line.
39211 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39213 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39214 (ifcombine_ifandif): Use it.
39216 2017-02-03 Martin Liska <mliska@suse.cz>
39218 * doc/invoke.texi: Document default value for
39219 use-after-scope-direct-emission-threshold.
39221 2017-02-03 Martin Liska <mliska@suse.cz>
39223 PR tree-optimization/79339
39224 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39225 (format_floating): Likewise.
39227 2017-02-03 Martin Liska <mliska@suse.cz>
39230 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39232 (ipa_node_params_t::remove): Likewise.
39233 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39234 initialization from removed ipa_node_params_t::insert.
39235 (ipa_node_params::~ipa_node_params): Move from removed
39236 ipa_node_params_t::release.
39237 * symbol-summary.h (symbol_summary::m_released): New member.
39238 Do not release a summary twice. Do not allow to call finalizer
39239 for types of a summary that live in GGC memory.
39241 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39243 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39246 2017-02-02 Martin Sebor <msebor@redhat.com>
39248 PR middle-end/79275
39249 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39250 (format_string): Tighten up the range of output for non-constant
39251 strings and correct the expected range for wide non-constant strings.
39253 2017-02-02 Martin Sebor <msebor@redhat.com>
39255 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39257 PR middle-end/32003
39258 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39260 (-fdump-tree-@var): Add to index and document how to come up
39261 with pass-specific option and dump file names.
39262 (-fdump-passes): Clarify where to look for output.
39264 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39266 PR middle-end/77445
39267 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39268 statistics of the analyzed path; allow threading for speed when
39269 any of BBs along the path are optimized for speed.
39271 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39273 PR middle-end/78468
39274 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39275 settings of the virtual registers.
39278 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39280 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39281 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39284 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39286 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39287 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39289 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39291 * config/s390/s390.md: Add missing comments with the expanded
39293 * config/s390/vector.md: Likewise.
39294 * config/s390/vx-builtins.md: Likewise.
39296 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39299 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39300 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39301 conditions on a single line.
39303 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39305 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39306 __S390_VX__ to __VX__.
39308 2017-02-01 Andrew Pinski <apinski@cavium.com>
39310 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39311 stmt_info to record_stmt_cost.
39312 (vect_get_known_peeling_cost): Pass stmt_info if known to
39314 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39315 cpu_vector_cost field into
39316 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39317 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39318 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39319 splitting of scalar_stmt_cost and vec_stmt_cost.
39320 (thunderx_vector_cost): Likewise.
39321 (cortexa57_vector_cost): LIkewise.
39322 (exynosm1_vector_cost): Likewise.
39323 (xgene1_vector_cost): Likewise.
39324 (thunderx2t99_vector_cost): Improve after the splitting of the two
39326 (aarch64_builtin_vectorization_cost): Update for the splitting of
39327 scalar_stmt_cost and vec_stmt_cost.
39329 2017-02-01 Torvald Riegel <triegel@redhat.com>
39330 Richard Henderson <rth@redhat.com>
39332 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39333 conditional on existance of a fast atomic load.
39334 * optabs-query.c (can_atomic_load_p): New function.
39335 * optabs-query.h (can_atomic_load_p): Declare it.
39336 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39337 no fast atomic load is available for the particular size of access.
39338 (expand_atomic_compare_and_swap): Likewise.
39339 (expand_atomic_load): Likewise.
39340 (expand_atomic_store): Likewise.
39341 (expand_atomic_fetch_op): Likewise.
39342 * testsuite/lib/target-supports.exp
39343 (check_effective_target_sync_int_128): Remove x86 because it provides
39344 no fast atomic load.
39345 (check_effective_target_sync_int_128_runtime): Likewise.
39347 2017-02-01 Richard Biener <rguenther@suse.de>
39349 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39350 (graphite_transform_loops): Provide opt-info for optimized nests.
39351 * tree-parloop.c (parallelize_loops): Provide opt-info for
39352 parallelized loops.
39354 2017-02-01 Richard Biener <rguenther@suse.de>
39356 PR middle-end/79315
39357 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39358 was not set before.
39360 2017-02-01 Richard Biener <rguenther@suse.de>
39362 PR tree-optimization/71824
39363 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39364 Verify the loops are valid in the merged SESE region.
39365 (scop_detection::can_represent_loop_1): Check analyzing the
39366 evolution of the number of iterations in the region succeeds.
39368 2017-01-31 Ian Lance Taylor <iant@golang.org>
39370 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39371 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39373 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39375 PR preprocessor/79210
39376 * input.c (get_substring_ranges_for_loc): Replace line_width
39377 assertion with error-handling.
39379 2017-01-31 Richard Biener <rguenther@suse.de>
39381 PR tree-optimization/77318
39382 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39383 (create_pw_aff_from_tree): Take loop parameter.
39384 (add_condition_to_pbb): Pass loop of the condition to
39385 create_pw_aff_from_tree.
39387 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39389 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39390 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39392 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39396 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39398 (convert_int_to_float128): Likewise.
39399 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39400 (convert_int_to_float128): Likewise.
39401 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39402 (UNSPEC_IEEE128_CONVERT): Likewise.
39403 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39404 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39405 Use local variables for IBM extended format.
39406 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39407 (fix_trunc<mode>si2_fprs): Likewise.
39408 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39409 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39410 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39411 to know that we can now have integers of all sizes in vector
39413 (fix<uns>_<mode>di2_hw): Likewise.
39414 (float<uns>_<mode>si2_hw): Likewise.
39415 (fix_<mode>si2_hw): Likewise.
39416 (fixuns_<mode>si2_hw): Likewise.
39417 (float<uns>_<mode>di2_hw): Likewise.
39418 (float_<mode>di2_hw): Likewise.
39419 (float_<mode>si2_hw): Likewise.
39420 (floatuns_<mode>di2_hw): Likewise.
39421 (floatuns_<mode>si2_hw): Likewise.
39422 (xscvqp<su>wz_<mode>): Delete, no longer used.
39423 (xscvqp<su>dz_<mode>): Likewise.
39424 (xscv<su>dqp_<mode>): Likewise.
39425 (ieee128_mfvsrd_64bit): Likewise.
39426 (ieee128_mfvsrd_32bit): Likewise.
39427 (ieee128_mfvsrwz): Likewise.
39428 (ieee128_mtvsrw): Likewise.
39429 (ieee128_mtvsrd_64bit): Likewise.
39430 (ieee128_mtvsrd_32bit): Likewise.
39432 2017-01-31 Martin Liska <mliska@suse.cz>
39435 * ipa-prop.c (ipa_free_all_node_params): Call release method
39436 instead of ~sumbol_summary to not to trigger double times
39439 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39441 PR tree-optimization/71691
39442 * bitmap.h (class auto_bitmap): New.
39443 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39444 is_maybe_undefined instead of ssa_undefined_value_p.
39446 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39448 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39449 __S390_ARCH_LEVEL__ to __ARCH__.
39451 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39453 PR tree-optimization/79267
39454 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39455 if should_remove_lhs_p is true.
39457 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39460 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39461 (add_alignment_attribute): New.
39462 (base_type_die): Add alignment attribute.
39463 (subrange_type_die): Likewise.
39464 (modified_type_die): Likewise.
39465 (gen_array_type_die): Likewise.
39466 (gen_descr_array_type_die: Likewise.
39467 (gen_enumeration_type_die): Likewise.
39468 (gen_subprogram_die): Likewise.
39469 (gen_variable_die): Likewise.
39470 (gen_field_die): Likewise.
39471 (gen_ptr_to_mbr_type_die): Likewise.
39472 (gen_struct_or_union_type_die): Likewise.
39473 (gen_subroutine_type_die): Likewise.
39474 (gen_typedef_die): Likewise.
39475 (base_type_cmp): Compare alignment attribute.
39477 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39480 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39481 (setb_unsigned) New pattern for setb with CCUNS.
39482 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39483 subfc./subfe sequence to avoid overflow problems. Generate a
39484 shorter sequence with cmpld/setb for power9.
39485 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39486 for generating subfc. instruction.
39487 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39488 now uses this instruction.
39490 2017-01-30 Ian Lance Taylor <iant@google.com>
39493 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39494 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39496 2017-01-30 Martin Sebor <msebor@redhat.com>
39498 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39499 Move constant to the right of a relational operator.
39500 (get_mpfr_format_length, format_character, format_string): Ditto.
39501 (should_warn_p, maybe_warn): Same.
39503 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39505 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39508 * asan.c (get_translation_unit_decl): Remove function.
39509 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39511 2017-01-30 Martin Liska <mliska@suse.cz>
39513 PR gcov-profile/79259
39514 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39515 -fprofile-generate.
39517 2017-01-30 Martin Liska <mliska@suse.cz>
39520 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39521 Initialize variables with NULL value.
39523 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39526 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39528 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39530 2017-01-30 Richard Biener <rguenther@suse.de>
39532 PR tree-optimization/79276
39533 * tree-vrp.c (process_assert_insertions): Properly adjust common
39534 when removing a duplicate.
39536 2017-01-30 Richard Biener <rguenther@suse.de>
39538 PR tree-optimization/79256
39539 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39540 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39542 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39544 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39547 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39548 ("*r<noxa>sbg_<mode>_sll_bitmask")
39549 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39550 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39551 Use contiguous_bitmask_nowrap_operand.
39553 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39556 * config/rs6000/altivec.h (vec_xl): Revise #define.
39557 (vec_xst): Likewise.
39559 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39561 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39563 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39565 PR rtl-optimization/79194
39566 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39567 traps before call to bypass_conditional_jumps.
39569 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39571 PR tree-optimization/71374
39572 * lra-constraints.c (check_conflict_input_operands): New.
39573 (match_reload): Use it.
39575 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39578 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39579 account to calculate conflict_set.
39581 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39583 PR rtl-optimization/78559
39584 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39585 other_insn in combine.
39587 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39589 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39590 uint16_type_node for BT_UINT16.
39592 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39594 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39595 "RTL Tests" to menu.
39596 (GIMPLE Tests): New node.
39597 (RTL Tests): New node.
39599 2017-01-27 Richard Biener <rguenther@suse.de>
39601 PR tree-optimization/79245
39602 * tree-loop-distribution.c (distribute_loop): Apply cost
39603 modeling also to detected patterns.
39605 2017-01-27 Richard Biener <rguenther@suse.de>
39607 PR tree-optimization/71433
39608 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39609 (compare_assert_loc): New function.
39610 (process_assert_insertions): Sort and optimize assert locations
39611 to remove duplicates and push down identical assertions on
39612 edges to their destination block.
39614 2017-01-27 Richard Biener <rguenther@suse.de>
39616 PR tree-optimization/79244
39617 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39618 out SSA names even if abnormal.
39620 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39622 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39623 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39624 instead of MPFR_RNDN.
39626 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39629 * arm.c (arm_option_override): Don't call build_target_option_node
39630 until after doing all option overrides.
39631 (arm_valid_target_attribute_tree): Likewise.
39633 2017-01-27 Martin Liska <mliska@suse.cz>
39635 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39636 for {cd}tors and C++ {cd}tors.
39638 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39640 * config/s390/s390.md ("*setmem_long_and")
39641 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39643 2017-01-26 Martin Sebor <msebor@redhat.com>
39645 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39648 2017-01-26 Martin Sebor <msebor@redhat.com>
39650 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39651 HAVE_DFmode before using XFmode or DFmode.
39652 (parse_directive): Avoid using the z length modifier to avoid
39653 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39655 PR middle-end/78703
39656 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39657 to accept adjustment as an array.
39658 (get_int_range): New function.
39659 (struct directive): Make width and prec arrays.
39660 (directive::set_width, directive::set_precision): Call get_int_range.
39661 (format_integer, format_floating): Handle width and precision ranges.
39662 (format_string, parse_directive): Same.
39664 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39667 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39668 comdat_type_p set, just clone them, but keep the children in the
39672 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39673 which have direct callers with -fvar-tracking-assignments enabled
39675 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39676 inside of type units.
39678 2017-01-26 Martin Sebor <msebor@redhat.com>
39680 PR middle-end/78703
39681 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39683 (struct format_result): Replace number_chars, number_chars_min,
39684 and number_chars_max with a single member of struct result_range.
39686 (format_result::operator+=): Adjust.
39687 (struct fmtresult): Remove bounded. Handle likely and unlikely
39689 (fmtresult::adjust_for_width_or_precision): New function.
39690 (fmtresult:type_max_digits): New function.
39691 (bytes_remaining): Handle likely and unlikely counters.
39692 (min_bytes_remaining): Remove.
39693 (format_percent): Simplify.
39694 (format_integer, format_floating): Set likely and unlikely counters.
39695 (get_string_length, format_character, format_string): Same.
39696 (format_plain, should_warn_p): New function.
39697 (maybe_warn): Call should_warn_p. Update diagnostic messages
39698 and handle those for all directives, including plain strings.
39699 (format_directive): Handle likely and unlikely counters.
39700 Remove unnecessary quoting from diagnostics. Add an informational
39702 (add_bytes): Remove.
39703 (pass_sprintf_length::compute_format_length): Simplify.
39704 (try_substitute_return_value): Handle likely and unlikely counters.
39706 2017-01-26 Carl Love <cel@us.ibm.com>
39708 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39709 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39711 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39714 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39715 endianess for subregs into account.
39716 * lra-constraints.c (lra_constraints): Do risky transformations
39717 always on the first iteration.
39718 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39719 last_call_used_reg_set.
39720 (process_bb_lives): Define and use last_call_used_reg_set.
39721 * lra.c (lra): Always continue after lra_constraints on the first
39724 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39726 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39728 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39730 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39732 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39733 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39734 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39735 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39736 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39737 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39738 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39739 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39740 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39742 2017-01-26 Marek Polacek <polacek@redhat.com>
39745 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39746 for the third operand.
39748 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39750 PR middle-end/79236
39751 * omp-low.c (struct omp_context): Add simt_stmt field.
39752 (scan_omp_for): Return omp_context *.
39753 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39754 context to the _simt_ SIMD stmt.
39755 (lower_omp_for): For combined SIMD with sibling _simt_
39756 SIMD, make sure to use the same decls in _looptemp_
39757 clauses as in the sibling.
39759 2017-01-26 David Sherwood <david.sherwood@arm.com>
39761 PR middle-end/79212
39762 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39765 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39768 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39769 emit fld b; fld a; if possible.
39771 * brig-builtins.def: Update copyright years.
39772 * config/arm/arm_acle_builtins.def: Update copyright years.
39774 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39777 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39778 constraint instead of o for the stxsd instruction.
39780 2017-01-25 Carl Love <cel@us.ibm.com>
39782 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39783 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39785 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39787 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39789 2017-01-25 Richard Biener <rguenther@suse.de>
39791 PR tree-optimization/69264
39792 * target.def (vector_alignment_reachable): Improve documentation.
39793 * doc/tm.texi: Regenerate.
39794 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39796 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39797 earlier changes with respect to TYPE_USER_ALIGN.
39798 (vector_alignment_reachable_p): Likewise. Improve dumping.
39800 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39803 * config/arm/arm.md (xordi3): Force constant operand into a register
39806 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39808 * doc/invoke.texi (-fstore-merging): Correct default optimization
39809 levels at which it is enabled.
39810 (-O): Move -fstore-merging from list to...
39813 2017-01-25 Richard Biener <rguenther@suse.de>
39816 * omp-expand.c: Include debug.h.
39817 (expand_omp_taskreg): Make sure to generate early debug before
39818 outlining anything from a function.
39819 (expand_omp_target): Likewise.
39820 (grid_expand_target_grid_body): Likewise.
39822 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39825 * asan.c (get_translation_unit_decl): New function.
39826 (asan_add_global): Extract modules file name from globals
39827 TRANSLATION_UNIT_DECL name.
39829 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39832 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39833 for long calls with APCS frame and VFP.
39835 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39837 * cfg.c (original_copy_tables_initialized_p): New function.
39838 * cfg.h (original_copy_tables_initialized_p): New decl.
39839 * cfgrtl.c (relink_block_chain): Guard the call to
39840 free_original_copy_tables with a call to
39841 original_copy_tables_initialized_p.
39842 * cgraph.h (symtab_node::native_rtl_p): New decl.
39843 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39844 (symtab_node::needed_p): Don't assert for early assembly output
39845 for __RTL functions.
39846 (cgraph_node::finalize_function): Set "force_output" for __RTL
39848 (cgraph_node::analyze): Bail out early for __RTL functions.
39849 (analyze_functions): Update assertion to support __RTL functions.
39850 (cgraph_node::expand): Bail out early for __RTL functions.
39851 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39853 * function.h (struct function): Update comment for field
39855 * gimple-expr.c: Include "tree-pass.h".
39856 (gimple_has_body_p): Return false for __RTL functions.
39857 * Makefile.in (OBJS): Add run-rtl-passes.o.
39858 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39860 (gcc::pass_manager::get_clean_slate): New accessor.
39861 * passes.c: Include "insn-addr.h".
39862 (should_skip_pass_p): Add logging. Update logic for running
39863 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39864 property-provider override so it is only done for gimple passes.
39866 (skip_pass): New function.
39867 (execute_one_pass): Call skip_pass when skipping passes.
39868 * read-md.c (md_reader::read_char): Support filtering
39869 the input to a subset of line numbers.
39870 (md_reader::md_reader): Initialize fields
39871 m_first_line and m_last_line.
39872 (md_reader::read_file_fragment): New function.
39873 * read-md.h (md_reader::read_file_fragment): New decl.
39874 (md_reader::m_first_line): New field.
39875 (md_reader::m_last_line): New field.
39876 * read-rtl-function.c (function_reader::create_function): Only
39877 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39878 curr_properties. Set DECL_INITIAL to a dummy block.
39879 (read_rtl_function_body_from_file_range): New function.
39880 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39882 * run-rtl-passes.c: New file.
39883 * run-rtl-passes.h: New file.
39885 2017-01-24 Jeff Law <law@redhat.com>
39887 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39890 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39892 PR tree-optimization/79159
39893 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39894 (record_nonwrapping_iv): Improve boundary using above function if no
39895 value range information.
39897 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39898 Martin Jambor <mjambor@suse.cz>
39900 * brig-builtins.def: New file.
39901 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39902 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39903 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39904 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39905 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39906 * builtin-types.def (BT_INT8): New.
39907 (BT_INT16): Likewise.
39908 (BT_UINT8): Likewise.
39909 (BT_UINT16): Likewise.
39910 (BT_FN_ULONG): Likewise.
39911 (BT_FN_UINT_INT): Likewise.
39912 (BT_FN_UINT_ULONG): Likewise.
39913 (BT_FN_UINT_LONG): Likewise.
39914 (BT_FN_UINT_PTR): Likewise.
39915 (BT_FN_ULONG_PTR): Likewise.
39916 (BT_FN_INT8_FLOAT): Likewise.
39917 (BT_FN_INT16_FLOAT): Likewise.
39918 (BT_FN_UINT32_FLOAT): Likewise.
39919 (BT_FN_UINT16_FLOAT): Likewise.
39920 (BT_FN_UINT8_FLOAT): Likewise.
39921 (BT_FN_UINT64_FLOAT): Likewise.
39922 (BT_FN_UINT16_UINT32): Likewise.
39923 (BT_FN_UINT32_UINT16): Likewise.
39924 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39925 (BT_FN_INT_PTR_INT): Likewise.
39926 (BT_FN_UINT_PTR_UINT): Likewise.
39927 (BT_FN_LONG_PTR_LONG): Likewise.
39928 (BT_FN_ULONG_PTR_ULONG): Likewise.
39929 (BT_FN_VOID_UINT64_UINT64): Likewise.
39930 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39931 (BT_FN_INT8_INT8_INT8): Likewise.
39932 (BT_FN_INT16_INT16_INT16): Likewise.
39933 (BT_FN_INT_INT_INT): Likewise.
39934 (BT_FN_UINT_FLOAT_UINT): Likewise.
39935 (BT_FN_FLOAT_UINT_UINT): Likewise.
39936 (BT_FN_ULONG_UINT_UINT): Likewise.
39937 (BT_FN_ULONG_UINT_PTR): Likewise.
39938 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39939 (BT_FN_UINT_UINT_UINT): Likewise.
39940 (BT_FN_VOID_UINT_PTR): Likewise.
39941 (BT_FN_UINT_UINT_PTR: Likewise.
39942 (BT_FN_UINT32_UINT64_PTR): Likewise.
39943 (BT_FN_INT_INT_UINT_UINT): Likewise.
39944 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39945 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39946 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39947 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39948 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39949 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39950 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39951 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39952 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39953 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39954 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39955 * doc/frontends.texi: List BRIG FE.
39956 * doc/install.texi (Testing): Add BRIG tesring requirements.
39957 * doc/invoke.texi (Overall Options): Mention BRIG.
39958 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39960 2017-01-24 Richard Biener <rguenther@suse.de>
39962 PR translation/79208
39963 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39965 2017-01-24 Martin Jambor <mjambor@suse.cz>
39968 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39969 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39970 and known_contexts.
39972 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39974 PR middle-end/79123
39975 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39976 casts from signed to unsigned really don't have a range.
39978 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39980 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39981 GMP_RNDx for compatiblity.
39983 2017-01-24 Martin Liska <mliska@suse.cz>
39986 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39987 that would prevent us to call alloca with -1 as argument.
39989 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39991 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39992 Avoid -Wformat-security warning.
39994 2017-01-23 Andrew Pinski <apinski@cavium.com>
39996 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39999 2017-01-23 Martin Sebor <msebor@redhat.com>
40001 PR middle-end/78703
40002 * gimple-ssa-sprintf.c (warn_level): New global.
40003 (format_integer): Use it here and throughout the rest of the file.
40004 Use the same switch to compute sign as base.
40005 (maybe_warn): New function.
40006 (format_directive): Factor out warnings into maybe_warn.
40007 Add debugging output. Use warn_level.
40008 (add_bytes): Use warn_level.
40009 (pass_sprintf_length::compute_format_length): Add debugging output.
40010 (try_substitute_return_value): Same.
40011 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
40013 PR middle-end/78703
40014 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
40015 (struct fmtresult, format_integer, format_floating): Adjust.
40016 (fmtresult::fmtresult): Set max correctly in two argument ctor.
40017 (get_string_length, format_string,format_directive): Same.
40018 (pass_sprintf_length::compute_format_length): Same.
40019 (try_substitute_return_value): Simplify slightly.
40021 PR middle-end/78703
40022 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
40023 (fmtresult::operator+=): Outlined.
40024 (struct fmtresult): Add ctors.
40025 (struct conversion_spec): Rename...
40026 (struct directive): ...to this. Add and remove data members.
40027 (directive::set_width, directive::set_precision): New functions.
40028 (format_percent): Use fmtresult ctor.
40029 (get_width_and_precision): Remove.
40030 (format_integer): Make naming changes. Avoid computing width and
40032 (format_floating): Same. Adjust indentation.
40033 (format_character, format_none): New functions.
40034 (format_string): Moved character handling to format_character.
40035 (format_directive): Remove arguments, change return type.
40036 (parse_directive): New function.
40037 (pass_sprintf_length::compute_format_length): Move directive
40038 parsing to parse_directive.
40040 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40042 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
40043 (assign_assembler_name_if_needed): ... this.
40044 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
40045 (assign_assembler_name_if_needed): ... this.
40046 (free_lang_data_in_cgraph): Adjust callers.
40047 * cgraphunit.c (cgraph_node::analyze): Likewise.
40048 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
40051 2017-01-23 Richard Biener <rguenther@suse.de>
40053 PR tree-optimization/79088
40054 PR tree-optimization/79188
40055 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
40056 resetting loop bounds after last path deletion. Reset loop
40057 bounds of the target loop, make code match the comments.
40058 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
40059 Make sure loops need no fixups.
40061 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
40063 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
40064 exponent support with double type for first argument.
40065 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
40066 type returned by __builtin_vec_extract_sig,
40067 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
40068 functions from "vector int" to "vector unsigned int" or from
40069 "vector long long int" to "vector unsigned long long int".
40070 Changed type returned by __builtin_vec_extract_exp,
40071 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
40072 functions from "vector int" to "vector unsigned int" or from
40073 "vector long long int" to "vector unsigned long long int".
40074 Changed return type of __builtin_vec_test_data_class,
40075 __builtin_vec_test_data_class_sp, and
40076 __builtin_vec_test_data_class_dp from "vector int" to
40077 "vector bool int" or from "vector long long int" to "vector bool
40078 long long int" and changed second argument type from "unsigned
40079 int" to "int". Added new overloaded function forms "vector float
40080 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
40081 "vector float __builtin_vec_insert_exp_sp (vector float, vector
40082 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
40083 double, vector unsigned long long int)" and "vector double
40084 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
40085 long int)". Changed return type of
40086 __builtin_scalar_test_data_class and
40087 __builtin_scalar_test_data_class_sp and
40088 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
40089 int" and changed second argument from "unsigned int" to "int".
40090 Changed type returned by __builtin_scalar_test_neg,
40091 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
40092 from "int" to "bool int". Added new overloaded function form
40093 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
40094 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
40095 exponent double-precision with floating point first argument.
40096 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
40097 documentation of scalar_test_data_class, scalar_test_neg,
40098 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
40099 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
40100 vec_test_data_class built-in functions to reflect refinements in
40101 their type signatures.
40103 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
40105 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
40107 (aarch64_elf_asm_destructor): Likewise.
40109 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
40111 PR rtl-optimization/78634
40112 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
40113 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
40114 * ifcvt.c (noce_try_cmove): Add missing cost check.
40116 PR rtl-optimization/71724
40117 * combine.c (if_then_else_cond): Look for situations where it is
40118 beneficial to undo the work of one of the recursive calls.
40120 2017-01-23 Bin Cheng <bin.cheng@arm.com>
40122 PR tree-optimization/70754
40123 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
40124 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
40125 combined stmt before it if not NULL.
40126 (combine_chains): Process refs reversely and compute dominance point
40129 2017-01-23 Martin Liska <mliska@suse.cz>
40131 PR tree-optimization/79196
40132 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
40133 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
40135 (strlen_optimize_stmt): Call the renamed function.
40137 2017-01-23 Michael Matz <matz@suse.de>
40139 PR tree-optimization/78384
40140 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
40142 2017-01-23 Richard Biener <rguenther@suse.de>
40144 PR tree-optimization/79186
40145 * tree-vrp.c (register_new_assert_for): Make sure we've seen
40146 both incoming edges before moving an assert.
40148 2017-01-23 Martin Jambor <mjambor@suse.cz>
40150 * ipa-prop.c (load_from_param_1): Removed.
40151 (load_from_unmodified_param): Bits from load_from_param_1 put back
40153 (load_from_param): Removed.
40154 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
40155 with stmt. Reverted back to use of load_from_unmodified_param.
40157 2017-01-23 Martin Jambor <mjambor@suse.cz>
40160 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40161 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40162 field a pointer to garbage collected vector, mark lattices and
40163 ipcp_orig_node with GTY((skip)).
40164 (ipa_get_param_count): Adjust to descriptors being a pointer.
40165 (ipa_get_param): Likewise.
40166 (ipa_get_type): Likewise.
40167 (ipa_get_param_move_cost): Likewise.
40168 (ipa_set_param_used): Likewise.
40169 (ipa_get_controlled_uses): Likewise.
40170 (ipa_set_controlled_uses): Likewise.
40171 (ipa_is_param_used): Likewise.
40172 (ipa_node_params_t): Move into garbage collector. New methods insert
40174 (ipa_node_params_sum): Annotate wth GTY(()).
40175 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40177 (ipa_load_from_parm_agg): Adjust declaration.
40178 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40179 * ipa-profile.c (ipa_profile): Likewise.
40180 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40181 (ipa_populate_param_decls): Make descriptors parameter garbage
40183 (ipa_dump_param): Adjust to descriptors being a pointer.
40184 (ipa_alloc_node_params): Likewise.
40185 (ipa_initialize_node_params): Likewise.
40186 (load_from_param_1): Make descriptors parameter garbage collected.
40187 (load_from_unmodified_param): Likewise.
40188 (load_from_param): Likewise.
40189 (ipa_load_from_parm_agg): Likewise.
40190 (ipa_node_params::~ipa_node_params): Removed.
40191 (ipa_free_all_node_params): Remove call to delete operator.
40192 (ipa_node_params_t::insert): New.
40193 (ipa_node_params_t::remove): Likewise.
40194 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40195 copy known_csts and known_contexts vectors.
40196 (ipa_read_node_info): Adjust to descriptors being a pointer.
40197 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40199 (ipcp_transform_function): Make descriptors variable garbage
40202 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40204 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40205 * config/i386/avx512dqintrin.h: Ditto.
40206 * config/i386/avx512fintrin.h: Ditto.
40207 * config/i386/i386.c: Handle new builtins.
40208 * config/i386/i386-builtin.def: Add new builtins.
40209 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40210 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40212 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40213 Martin Liska <mliska@suse.cz>
40215 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40216 * asan.c (asan_expand_poison_ifn): Support stores and use
40217 appropriate ASAN report function.
40218 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40219 * internal-fn.def (ASAN_POISON_USE): Declare.
40220 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40221 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40222 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40223 ASAN_POISON calls w/o LHS.
40224 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40225 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40226 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40227 * gimplify.c (asan_poison_variables): Add attribute
40228 use_after_scope_memory to variables that really needs to live
40230 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40231 having the attribute.
40233 2017-01-23 Martin Liska <mliska@suse.cz>
40235 * asan.c (create_asan_shadow_var): New function.
40236 (asan_expand_poison_ifn): Likewise.
40237 * asan.h (asan_expand_poison_ifn): New declaration.
40238 * internal-fn.c (expand_ASAN_POISON): Likewise.
40239 * internal-fn.def (ASAN_POISON): New builtin.
40240 * sanopt.c (pass_sanopt::execute): Expand
40241 asan_expand_poison_ifn.
40242 * tree-inline.c (copy_decl_for_dup_finish): Make function
40244 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40245 * tree-ssa.c (is_asan_mark_p): New function.
40246 (execute_update_addresses_taken): Rewrite local variables
40247 (identified just by use-after-scope as addressable) into SSA.
40249 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40251 * doc/install.texi (Specific): opensource.apple.com uses https
40252 now. Remove trailing slash.
40254 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40256 * README.Portability: Remove note on an Irix compatibility issue.
40258 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40260 * gcov.c (INCLUDE_ALGORITHM): Define.
40261 (INCLUDE_VECTOR): Define.
40262 No longer include <vector> and <algorithm> directly.
40264 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40266 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40268 * doc/invoke.texi (Code Gen Options): Ditto.
40270 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40273 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40275 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40277 rtl-optimization/79125
40278 * cprop.c (local_cprop_pass): Handle cases where we make an
40279 unconditional trap.
40281 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40285 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40286 read from, for big endian.
40288 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40290 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40291 register pauth builtins for LP64 only.
40293 2017-01-20 Marek Polacek <polacek@redhat.com>
40296 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40299 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40301 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40303 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40304 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40305 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40307 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40310 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40311 in big-endian mode when they are not a single duplicated value.
40313 2017-01-20 Richard Biener <rguenther@suse.de>
40315 * BASE-VER: Bump to 7.0.1.
40317 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40319 * omp-low.c (omplow_simd_context): New struct. Use it...
40320 (lower_rec_simd_input_clauses): ...here and...
40321 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40322 references to idx, lane, max_vf, is_simt.
40324 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40326 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40329 2017-01-20 Martin Jambor <mjambor@suse.cz>
40331 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40332 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40334 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40335 (GTFILES): Rename hsa.c to hsa-common.c.
40336 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40337 * hsa-dump.c: Likewise.
40338 * hsa-gen.c: Likewise.
40339 * hsa-regalloc.c: Likewise.
40340 * ipa-hsa.c: Likewise.
40341 * omp-expand.c: Likewise.
40342 * omp-low.c: Likewise.
40343 * toplev.c: Likewise.
40345 2017-01-20 Marek Polacek <polacek@redhat.com>
40348 * doc/invoke.texi: Document -Wduplicated-branches.
40349 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40350 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40351 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40352 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40353 return 0 only when not OEP_LEXICOGRAPHIC.
40354 (fold_build_cleanup_point_expr): Use the expression
40355 location when building CLEANUP_POINT_EXPR.
40356 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40357 * tree.c (add_expr): Handle error_mark_node.
40359 2017-01-20 Martin Liska <mliska@suse.cz>
40362 * tree-profile.c (init_ic_make_global_vars): Do not call
40364 (gimple_init_gcov_profiler): Likewise.
40366 2017-01-20 Martin Liska <mliska@suse.cz>
40369 * cgraph.h (maybe_create_reference): Remove argument and
40371 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40373 * ipa-cp.c (create_specialized_node): Likewise.
40374 * symtab.c (symtab_node::maybe_create_reference): Handle
40375 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40377 2017-01-20 Martin Liska <mliska@suse.cz>
40379 * read-rtl-function.c (function_reader::create_function): Use
40380 build_decl instread of build_decl_stat.
40382 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40384 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40385 * config/i386/avx512dqintrin.h: Ditto.
40386 * config/i386/avx512fintrin.h: Ditto.
40387 * config/i386/i386-builtin-types.def: Add new types.
40388 * config/i386/i386.c: Handle new types.
40389 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40390 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40391 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40392 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40393 (__builtin_ia32_kshiftridi): New.
40394 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40396 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40400 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40401 define to rs6000_init_stack_protect_guard.
40402 (rs6000_init_stack_protect_guard): New function.
40404 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40405 Yunqiang Su <yunqiang.su@imgtec.com>
40407 * config.gcc (supported_defaults): Add madd4.
40408 (with_madd4): Add validation.
40409 (all_defaults): Add madd4.
40410 * config/mips/mips.opt (mmadd4): New option.
40411 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40413 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40415 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40416 (ISA_HAS_FUSED_MADD4): Likewise.
40417 * doc/invoke.texi (-mmadd4): Document the new option.
40418 * doc/install.texi (--with-madd4): Document the new option.
40420 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40422 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40423 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40424 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40425 (aarch64_init_pauth_hint_builtins): New.
40426 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40427 (aarch64_expand_builtin): Expand new builtins.
40429 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40431 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40432 * combine-stack-adj.c (no_unhandled_cfa): Handle
40433 REG_CFA_TOGGLE_RA_MANGLE.
40434 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40435 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40436 info for return address signing.
40437 (aarch64_expand_epilogue): Likewise.
40439 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40441 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40442 * config/aarch64/aarch64-protos.h
40443 (aarch64_return_address_signing_enabled): New declaration.
40444 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40446 (aarch64_expand_prologue): Sign return address before it's pushed onto
40448 (aarch64_expand_epilogue): Authenticate return address fetched from
40450 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40451 (aarch64_attributes): New function attributes for "sign-return-address".
40452 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40453 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40454 ("*do_return"): Generate combined instructions according to key index.
40455 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40456 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40458 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40459 * config/aarch64/aarch64.opt (msign-return-address=): New.
40460 * doc/extend.texi (AArch64 Function Attributes): Documents
40461 "sign-return-address=".
40462 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40464 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40466 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40467 overall option summary.
40469 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40471 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40472 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40473 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40474 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40476 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40478 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40479 -mpower9-minmax by default for -mcpu=power9.
40480 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40481 128-bit floating point.
40483 2017-01-20 Alan Modra <amodra@gmail.com>
40485 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40486 optimizing for size.
40488 2017-01-20 Alan Modra <amodra@gmail.com>
40491 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40492 for strcmp and strncmp from corresponding builtin decl.
40494 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40496 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40497 instead of i386/rtems-64.h.
40498 * config/i386/rtems-64.h: Remove.
40500 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40504 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40506 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40508 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40510 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40511 Change int to HOST_WIDE_INT.
40512 * config/aarch64/aarch64-protos.h
40513 (aarch64_simd_gen_const_vector_dup): Likewise.
40514 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40516 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40518 * langhooks-def.h (lhd_type_for_size): New decl.
40519 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40520 * langhooks.c (lhd_type_for_size): New function, taken from
40523 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40525 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40526 define_bypass for CR latency.
40527 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40528 (power9-alu2): Add define_bypass for CR latency.
40530 (power9-mul): Update insn latency.
40531 (power9-mul-compare): Update insn latency, bypass latency and remove
40534 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40536 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40538 * config/aarch64/aarch64.md
40539 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40540 aarch64_nopcrelative_literal_loads.
40541 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40543 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40545 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40546 TARGET_LOONGSON_3A.
40547 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40549 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40552 * config.gcc (supported_defaults): Add lxc1-sxc1.
40553 (with_lxc1_sxc1): Add validation.
40554 (all_defaults): Add lxc1-sxc1.
40555 * config/mips/mips.opt (mlxc1-sxc1): New option.
40556 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40558 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40559 __mips_no_lxc1_sxc1.
40560 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40561 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40562 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40564 2017-01-19 Richard Biener <rguenther@suse.de>
40566 PR tree-optimization/72488
40567 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40568 sure to restore SSA info.
40569 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40571 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40573 PR rtl-optimization/79121
40574 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40575 of the inner type when shifting an extended value.
40577 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40580 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40581 interposable aliases.
40583 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40586 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40587 Use the evmergelohi instruction.
40588 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40589 (mov_sitf_e500_subreg8_2_be): Likewise.
40590 (mov_sitf_e500_subreg12_2_le): Likewise.
40591 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40592 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40593 (mov_sitf_e500_subreg8_2_le): Likewise.
40594 (mov_sitf_e500_subreg12_2_be): Likewise.
40596 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40598 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40599 attribute from vecsimple to vecperm.
40600 (altivec_vbpermq2): Likewise.
40602 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40605 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40607 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40608 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40609 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40610 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40611 case where N arg is SIZE_MAX.
40612 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40613 (cmpstrsi): Add pattern.
40615 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40617 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40618 __builtin_vec_revb builtins.
40619 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40620 built-in functions to support generation of the ISA 3.0 XXBR<x>
40621 vector byte reverse instructions.
40622 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40623 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40624 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40625 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40626 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40627 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40628 (P9V_BUILTIN_VEC_REVB): Likewise.
40629 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40630 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40631 (p9_xxbrq_v16qi): Likewise.
40632 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40633 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40634 (p9_xxbrh_v8hi): Likewise.
40635 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40636 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40637 vec_revb built-in functions.
40639 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40641 PR rtl-optimization/78952
40642 * config/i386/i386.md (any_extract): New code iterator.
40643 (*insvqi_2): Use any_extract for source operand.
40644 (*insvqi_3): Use any_shiftrt for source operand.
40646 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40648 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40650 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40652 2017-01-18 Matthias Klose <doko@ubuntu.com>
40654 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40656 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40658 * config/rs6000/altivec.h (vec_bperm): Change #define.
40659 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40660 (altivec_vbpermq2): New define_insn.
40661 (altivec_vbpermd): Likewise.
40662 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40663 function interface.
40664 (VBPERMD): Likewise.
40665 (VBPERM): New polymorphic function interface.
40666 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40667 Add entries for P9V_BUILTIN_VEC_VBPERM.
40668 * doc/extend.texi: Add interfaces for vec_bperm.
40670 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40672 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40673 first letter of error messages.
40674 (s390_resolve_overloaded_builtin): Likewise.
40675 * config/s390/s390.c (s390_expand_builtin): Likewise.
40676 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40677 (s390_valid_target_attribute_inner_p): Likewise.
40678 * config/s390/s390.md ("tabort"): Likewise.
40680 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40682 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40683 (ISA_AVOID_DIV_HILO): New macro.
40684 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40685 (ISA_HAS_DDIV): Likewise.
40687 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40689 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40691 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40693 * doc/invoke.texi (fabi-version): Spelling fix.
40695 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40698 * doc/invoke.texi (fabi-version): Mention mangling fix for
40701 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40704 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40706 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40709 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40710 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40712 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40714 (stack_protect_test): Ditto.
40715 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40716 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40718 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40719 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40720 -mstack-protector-guard-offset=.
40721 (RS/6000 and PowerPC Options): Ditto.
40723 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40725 * config/i386/i386.h (MASK_CLASS_P): New define.
40726 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40727 there are no registers from different register sets also when
40728 mask registers are used. Update function comment.
40729 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40730 to (*k/*r) and (*k/*km) alternatives.
40732 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40734 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40735 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40736 (EH_RETURN_HANDLER_RTX): New define.
40737 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40738 Force frame pointer in EH return functions.
40739 (aarch64_expand_epilogue): Add barrier for eh_return.
40740 (aarch64_final_eh_return_addr): Remove.
40741 (aarch64_eh_return_handler_rtx): New function.
40742 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40744 (aarch64_eh_return_handler_rtx): New prototype.
40746 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40748 * config/rs6000/altivec.h (vec_rlmi): New #define.
40749 (vec_vrlnm): Likewise.
40750 (vec_rlnm): Likewise.
40751 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40752 (UNSPEC_VRLNM): Likewise.
40753 (VIlong): New mode iterator.
40754 (altivec_vrl<VI_char>mi): New define_insn.
40755 (altivec_vrl<VI_char>nm): Likewise.
40756 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40758 (VRLDNM): Likewise.
40759 (RLNM): New polymorphic function entry.
40760 (VRLWMI): New monomorphic function entry.
40761 (VRLDMI): Likewise.
40762 (RLMI): New polymorphic function entry.
40763 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40764 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40765 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40768 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40771 * dwarf2out.c (field_byte_offset): Restore the
40772 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40773 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40774 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40777 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40780 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40782 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40785 * dwarf2out.c (add_data_member_location_attribute): For constant
40786 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40787 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40788 DW_AT_byte_size attributes.
40790 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40792 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40793 after forcing to constant memory when the code model is medium.
40795 2017-01-17 Julia Koval <julia.koval@intel.com>
40798 * config/i386/avx512fintrin.h
40799 (_mm512_i32gather_ps): Change __addr type to void const*.
40800 (_mm512_mask_i32gather_ps): Ditto.
40801 (_mm512_i32gather_pd): Ditto.
40802 (_mm512_mask_i32gather_pd): Ditto.
40803 (_mm512_i64gather_ps): Ditto.
40804 (_mm512_mask_i64gather_ps): Ditto.
40805 (_mm512_i64gather_pd): Ditto.
40806 (_mm512_mask_i64gather_pd): Ditto.
40807 (_mm512_i32gather_epi32): Ditto.
40808 (_mm512_mask_i32gather_epi32): Ditto.
40809 (_mm512_i32gather_epi64): Ditto.
40810 (_mm512_mask_i32gather_epi64): Ditto.
40811 (_mm512_i64gather_epi32): Ditto.
40812 (_mm512_mask_i64gather_epi32): Ditto.
40813 (_mm512_i64gather_epi64): Ditto.
40814 (_mm512_mask_i64gather_epi64): Ditto.
40815 (_mm512_i32scatter_ps): Change __addr type to void*.
40816 (_mm512_mask_i32scatter_ps): Ditto.
40817 (_mm512_i32scatter_pd): Ditto.
40818 (_mm512_mask_i32scatter_pd): Ditto.
40819 (_mm512_i64scatter_ps): Ditto.
40820 (_mm512_mask_i64scatter_ps): Ditto.
40821 (_mm512_i64scatter_pd): Ditto.
40822 (_mm512_mask_i64scatter_pd): Ditto.
40823 (_mm512_i32scatter_epi32): Ditto.
40824 (_mm512_mask_i32scatter_epi32): Ditto.
40825 (_mm512_i32scatter_epi64): Ditto.
40826 (_mm512_mask_i32scatter_epi64): Ditto.
40827 (_mm512_i64scatter_epi32): Ditto.
40828 (_mm512_mask_i64scatter_epi32): Ditto.
40829 (_mm512_i64scatter_epi64): Ditto.
40830 (_mm512_mask_i64scatter_epi64): Ditto.
40831 * config/i386/avx512pfintrin.h
40832 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40833 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40834 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40835 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40836 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40837 (_mm512_prefetch_i32scatter_ps): Ditto.
40838 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40839 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40840 (_mm512_prefetch_i64scatter_pd): Ditto.
40841 (_mm512_prefetch_i64scatter_ps): Ditto.
40842 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40843 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40844 * config/i386/avx512vlintrin.h
40845 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40846 (_mm_mmask_i32gather_ps): Ditto.
40847 (_mm256_mmask_i32gather_pd): Ditto.
40848 (_mm_mmask_i32gather_pd): Ditto.
40849 (_mm256_mmask_i64gather_ps): Ditto.
40850 (_mm_mmask_i64gather_ps): Ditto.
40851 (_mm256_mmask_i64gather_pd): Ditto.
40852 (_mm_mmask_i64gather_pd): Ditto.
40853 (_mm256_mmask_i32gather_epi32): Ditto.
40854 (_mm_mmask_i32gather_epi32): Ditto.
40855 (_mm256_mmask_i32gather_epi64): Ditto.
40856 (_mm_mmask_i32gather_epi64): Ditto.
40857 (_mm256_mmask_i64gather_epi32): Ditto.
40858 (_mm_mmask_i64gather_epi32): Ditto.
40859 (_mm256_mmask_i64gather_epi64): Ditto.
40860 (_mm_mmask_i64gather_epi64): Ditto.
40861 (_mm256_i32scatter_ps): Change __addr type to void*.
40862 (_mm256_mask_i32scatter_ps): Ditto.
40863 (_mm_i32scatter_ps): Ditto.
40864 (_mm_mask_i32scatter_ps): Ditto.
40865 (_mm256_i32scatter_pd): Ditto.
40866 (_mm256_mask_i32scatter_pd): Ditto.
40867 (_mm_i32scatter_pd): Ditto.
40868 (_mm_mask_i32scatter_pd): Ditto.
40869 (_mm256_i64scatter_ps): Ditto.
40870 (_mm256_mask_i64scatter_ps): Ditto.
40871 (_mm_i64scatter_ps): Ditto.
40872 (_mm_mask_i64scatter_ps): Ditto.
40873 (_mm256_i64scatter_pd): Ditto.
40874 (_mm256_mask_i64scatter_pd): Ditto.
40875 (_mm_i64scatter_pd): Ditto.
40876 (_mm_mask_i64scatter_pd): Ditto.
40877 (_mm256_i32scatter_epi32): Ditto.
40878 (_mm256_mask_i32scatter_epi32): Ditto.
40879 (_mm_i32scatter_epi32): Ditto.
40880 (_mm_mask_i32scatter_epi32): Ditto.
40881 (_mm256_i32scatter_epi64): Ditto.
40882 (_mm256_mask_i32scatter_epi64): Ditto.
40883 (_mm_i32scatter_epi64): Ditto.
40884 (_mm_mask_i32scatter_epi64): Ditto.
40885 (_mm256_i64scatter_epi32): Ditto.
40886 (_mm256_mask_i64scatter_epi32): Ditto.
40887 (_mm_i64scatter_epi32): Ditto.
40888 (_mm_mask_i64scatter_epi32): Ditto.
40889 (_mm256_i64scatter_epi64): Ditto.
40890 (_mm256_mask_i64scatter_epi64): Ditto.
40891 (_mm_i64scatter_epi64): Ditto.
40892 (_mm_mask_i64scatter_epi64): Ditto.
40893 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40894 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40895 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40896 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40897 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40898 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40899 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40900 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40901 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40902 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40903 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40904 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40905 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40906 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40907 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40908 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40909 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40910 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40911 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40912 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40913 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40914 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40915 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40916 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40917 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40918 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40919 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40920 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40921 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40922 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40923 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40924 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40925 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40926 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40927 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40928 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40929 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40930 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40931 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40932 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40933 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40934 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40935 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40936 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40937 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40938 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40939 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40940 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40941 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40942 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40943 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40944 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40945 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40946 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40947 definitions accordingly.
40949 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40950 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40953 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40956 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40959 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40960 hard regs for inner regno.
40962 2017-01-17 Martin Liska <mliska@suse.cz>
40965 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40966 assumption and add comment.
40968 2017-01-17 Nathan Sidwell <nathan@acm.org>
40970 * ipa-visibility.c (localize_node): New function, broken out of ...
40971 (function_and_variable_visibility): ... here. Call it.
40973 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40975 PR middle-end/77445
40976 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40977 correctly set frequency of oudgoing edge.
40978 (duplicate_thread_path): Fix profile updating.
40980 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40983 * configure.ac: Add GCC_BASE_VER.
40984 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40985 version from BASE-VER file.
40986 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40987 (gcc.o): Depend on $(BASEVER).
40988 * common.opt (dumpfullversion): New option.
40989 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40990 * doc/invoke.texi: Document -dumpfullversion.
40991 * doc/install.texi: Document --with-gcc-major-version-only.
40992 * configure: Regenerated.
40994 2017-01-17 Richard Biener <rguenther@suse.de>
40996 PR tree-optimization/71433
40997 * tree-vrp.c (register_new_assert_for): Merge same asserts
40998 on all incoming edges.
40999 (process_assert_insertions_for): Handle insertions at the
41002 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
41004 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
41005 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
41007 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
41010 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
41013 2017-01-17 Alan Modra <amodra@gmail.com>
41016 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
41017 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
41018 symbolic stack limit when pic.
41020 2017-01-16 Martin Sebor <msebor@redhat.com>
41022 PR tree-optimization/78608
41023 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
41025 2017-01-16 Jeff Law <law@redhat.com>
41028 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
41029 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
41030 for several include directories that may be relative to sysroot.
41031 * config/i386/x-mingw32 (gplus_includedir): Define.
41032 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
41033 (native_system_includedir): Likewise.
41034 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
41035 override if TARGET_SYSTEM_ROOT is defined.
41036 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
41038 PR tree-optimization/79090
41039 PR tree-optimization/33562
41040 PR tree-optimization/61912
41041 PR tree-optimization/77485
41042 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
41043 and computed trims into the dump file.
41045 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
41047 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
41049 2017-01-16 Jakub Jelinek <jakub@redhat.com>
41052 * gimplify.c (gimplify_init_constructor): If want_value and
41053 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
41057 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
41058 sequence. Formatting fixes.
41059 (doloop_optimize): Formatting fixes.
41062 * gcc.c (debug_level_greater_than_spec_func): New function.
41063 (static_spec_functions): Add debug-level-gt spec function.
41064 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
41066 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41067 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41068 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
41069 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
41070 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
41071 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
41073 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
41075 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
41076 QImode fixups to general and mask registers only.
41078 2017-01-16 Carl Love <cel@us.ibm.com>
41080 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
41081 for built-in functions
41082 vector signed char vec_nabs (vector signed char)
41083 vector signed short vec_nabs (vector signed short)
41084 vector signed int vec_nabs (vector signed int)
41085 vector signed long long vec_nabs (vector signed long long)
41086 vector float vec_nabs (vector float)
41087 vector double vec_nabs (vector double)
41088 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
41090 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
41091 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
41092 * doc/extend.texi: Update the documentation file for the new built-in
41095 2017-01-16 Martin Sebor <msebor@redhat.com>
41097 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
41100 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41102 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
41103 UNSPEC_VSX__XXSPLTD to require special splat handling.
41105 2017-01-16 David Malcolm <dmalcolm@redhat.com>
41108 * system.h: Poison strndup.
41110 2017-01-16 Alan Modra <amodra@gmail.com>
41113 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
41116 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
41118 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
41120 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
41122 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
41123 call recog here. Assert that INSN_CODE (insn) is non-negative.
41125 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41128 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
41130 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
41131 in the currently scheduled RTL fragment.
41133 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41135 PR rtl-optimization/78751
41136 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
41139 2017-01-14 Jeff Law <law@redhat.com>
41141 PR tree-optimization/79090
41142 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
41143 variable length stores.
41144 (compute_trims): Delete dead assignment to *trim_tail.
41145 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
41148 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
41150 PR rtl-optimization/78626
41151 PR rtl-optimization/78727
41152 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
41153 of a block, and split such blocks after everything else is finished.
41155 2017-01-14 Alan Modra <amodra@gmail.com>
41158 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
41159 target legitimate_combined_insn.
41160 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41161 (rs6000_legitimate_combined_insn): New function.
41162 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41164 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41165 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41166 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41168 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41170 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41172 2017-01-13 Jeff Law <law@redhat.com>
41174 PR tree-optimization/33562
41175 PR tree-optimization/61912
41176 PR tree-optimization/77485
41177 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41179 (delete_dead_assignment): Likewise.
41180 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41181 statement to delete_dead_call and delete_dead_assignment.
41183 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41186 * substring-locations.c (format_warning_va): Strengthen case 1 so
41187 that both endpoints of the substring must be within the format
41188 range for just the substring to be printed.
41190 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41192 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41193 * config/i386/i386.c (ix86_target_string): Add missing options
41194 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41195 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41196 flags_other and ix86_target_other to flags2_other. Display unknown
41198 (ix86_valid_target_attribute_inner_p): Add missing options and
41199 reorder options by implied ISAs, as in ix86_target_string.
41201 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41203 * hash-table.h (hash_table::too_empty_p): New function.
41204 (hash_table::expand): Use it.
41205 (hash_table::traverse): Likewise.
41206 (hash_table::empty_slot): Use sizeof (value_type) instead of
41207 sizeof (PTR) to convert bytes to elements. Shrink the table
41208 if the current size is excessive for the current number of
41211 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41213 * ira-costs.c (record_reg_classes): Break from the inner loop
41214 early once alt_fail is known to be true. Update outer loop
41215 handling accordingly.
41217 2017-01-13 Jeff Law <law@redhat.com>
41219 * tree-ssa-dse.c (decrement_count): New function.
41220 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41221 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41222 when we know the partially dead statement is a mem* function.
41224 PR tree-optimization/61912
41225 PR tree-optimization/77485
41226 * tree-ssa-dse.c: Include expr.h.
41227 (maybe_trim_constructor_store): New function.
41228 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41230 PR tree-optimization/33562
41231 PR tree-optimization/61912
41232 PR tree-optimization/77485
41233 * doc/invoke.texi: Document new dse-max-object-size param.
41234 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41235 * tree-ssa-dse.c: Include params.h.
41236 (dse_store_status): New enum.
41237 (initialize_ao_ref_for_dse): New, partially extracted from
41239 (valid_ao_ref_for_dse, normalize_ref): New.
41240 (setup_live_bytes_from_ref, compute_trims): Likewise.
41241 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41242 (maybe_trim_partially_dead_store): Likewise.
41243 (maybe_trim_complex_store): Likewise.
41244 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41245 Track what bytes live from the original store. Return tri-state
41246 for dead, partially dead or live.
41247 (dse_dom_walker): Add constructor, destructor and new private members.
41248 (delete_dead_call, delete_dead_assignment): New extracted from
41250 (dse_optimize_stmt): Make a member of dse_dom_walker.
41251 Use initialize_ao_ref_for_dse.
41253 PR tree-optimization/33562
41254 PR tree-optimization/61912
41255 PR tree-optimization/77485
41256 * sbitmap.h (bitmap_count_bits): Prototype.
41257 (bitmap_clear_range, bitmap_set_range): Likewise.
41258 * sbitmap.c (bitmap_clear_range): New function.
41259 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41261 2017-01-13 Martin Liska <mliska@suse.cz>
41264 * function.c (set_cfun): Add new argument force.
41265 * function.h (set_cfun): Likewise.
41266 * ipa-inline-transform.c (inline_call): Use the function when
41267 strict alising from is dropped for function we inline to.
41269 2017-01-13 Richard Biener <rguenther@suse.de>
41271 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41272 for dumping GIMPLE INTEGER_CSTs.
41274 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41276 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41277 to 201112L since C++17.
41279 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41282 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41283 if -fsanitize=kernel-address is present.
41285 2017-01-13 Richard Biener <rguenther@suse.de>
41287 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41288 as _Literal ( type ) number in case usual suffixes do not
41289 preserve all information.
41291 2017-01-13 Richard Biener <rguenther@suse.de>
41293 PR tree-optimization/77283
41294 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41295 and ssa-iterators.h.
41296 (is_feasible_trace): Implement a cost model based on joiner
41299 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41302 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41303 char or short to __float128/_Float128 directly.
41305 2017-01-12 Martin Sebor <msebor@redhat.com>
41307 to -Wformat-overflow.
41308 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41309 (min_bytes_remaining): Same.
41310 (get_string_length): Same.
41311 (format_string): Same.
41312 (format_directive): Same.
41314 (pass_sprintf_length::handle_gimple_call): Same.
41316 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41318 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41319 info.nowrite calls with no lhs that can't throw. Return bool
41320 whether gsi_remove has been called or not.
41321 (pass_sprintf_length::handle_gimple_call): Return bool whether
41322 try_substitute_return_value called gsi_remove. Formatting fix.
41323 (pass_sprintf_length::execute): Don't use gsi_remove if
41324 handle_gimple_call returned true.
41327 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41328 be removed due to side-effects, don't remove following barrier nor
41329 turn the successor edge into fallthru edge.
41331 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41334 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41335 element-reversing loads and stores as not swappable.
41337 2017-01-12 Nathan Sidwell <nathan@acm.org>
41338 Nicolai Stange <nicstange@gmail.com>
41340 * combine.c (try_combine): Don't ignore result of overlap checking
41341 loop. Combine overlap & asm check into single loop.
41343 2017-01-12 Richard Biener <rguenther@suse.de>
41345 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41346 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41348 2017-01-12 Richard Biener <rguenther@suse.de>
41350 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41351 and TS_TARGET_OPTION directly derive from TS_BASE.
41352 * tree-core.h (tree_optimization_option): Derive from tree_base.
41353 (tree_target_option): Likewise.
41355 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41357 * config/i386/i386.c (memory_address_length): Increase len
41358 only when rip_relative_addr_p returns false.
41360 2017-01-11 Julia Koval <julia.koval@intel.com>
41362 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41363 (OPTION_MASK_ISA_SGX_SET): New.
41364 (ix86_handle_option): Handle OPT_msgx.
41365 * config.gcc: Added sgxintrin.h.
41366 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41367 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41368 * config/i386/i386.c (ix86_target_string): Add -msgx.
41370 (ix86_option_override_internal): Handle new options.
41371 (ix86_valid_target_attribute_inner_p): Add sgx.
41372 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41373 * config/i386/i386.opt: Add msgx.
41374 * config/i386/sgxintrin.h: New file.
41375 * config/i386/x86intrin.h: Add sgxintrin.h.
41377 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41380 * fold-const.c (maybe_nonzero_address): Return 1 for function
41382 (tree_single_nonzero_warnv_p): Don't handle function local objects
41386 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41389 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41392 * opts.c: Include "spellcheck.h"
41393 (struct string_fragment): New struct.
41394 (struct edit_distance_traits<const string_fragment &>): New
41396 (get_closest_sanitizer_option): New function.
41397 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41399 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41401 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41403 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41404 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41405 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41406 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41407 for initial die_offset if dwarf_split_debug_info.
41408 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41409 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41411 (output_skeleton_debug_sections): Formatting fix. Use
41412 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41413 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41415 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41417 * config/arm/cortex-a53.md: Add bypasses for
41418 cortex_a53_r2f_cvt.
41419 (cortex_a53_r2f): Only use for transfers.
41420 (cortex_a53_f2r): Likewise.
41421 (cortex_a53_r2f_cvt): Add reservation for conversions.
41422 (cortex_a53_f2r_cvt): Likewise.
41424 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41426 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41427 to all inlined functions, change static to extern.
41429 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41432 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41434 (arm_assemble_integer): Likewise.
41436 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41438 * config.gcc: Use new awk script to check CPU, FPU and architecture
41439 parameters for --with-... options.
41440 * config/arm/parsecpu.awk: New file
41441 * config/arm/arm-cpus.in: New file.
41442 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41444 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41446 * config/arm/t-arm: Update dependency rules.
41447 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41448 of processing .def files.
41449 * config/arm/genopt.sh: Deleted.
41450 * config/arm/gentune.sh: Deleted.
41451 * config/arm/arm-cores.def: Deleted.
41452 * config/arm/arm-arches.def: Deleted.
41453 * config/arm/arm-fpus.def: Deleted.
41454 * config/arm/arm-tune.md: Regenerated.
41455 * config/arm/arm-tables.opt: Regenerated.
41456 * config/arm/arm-cpu.h: New generated file.
41457 * config/arm/arm-cpu-data.h: New generated file.
41458 * config/arm/arm-cpu-cdata.h: New generated file.
41460 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41463 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41465 (input_varpool_node): Unpack dynamically_initialized bit.
41467 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41469 PR rtl-optimization/79032
41470 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41471 the alignment of the adjusted memory reference against that of MODE,
41472 instead of the alignment of the original memory reference.
41474 2017-01-11 Martin Jambor <mjambor@suse.cz>
41476 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41478 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41479 decorated functions.
41481 2017-01-11 Richard Biener <rguenther@suse.de>
41483 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41484 set range/nonnull info for PHI results. Do not set it on
41485 stmts marked for removal.
41487 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41489 * expr.c (store_field): In the bitfield case, fetch the return value
41490 from the registers before applying a single big-endian adjustment.
41491 Always do a final load for a BLKmode value not larger than a word.
41493 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41496 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41497 that we correctly handle column numbers greater than
41498 LINE_MAP_MAX_COLUMN_NUMBER.
41500 2017-01-10 Martin Sebor <msebor@redhat.com>
41502 PR middle-end/78245
41503 * gimple-ssa-sprintf.c (get_destination_size): Call
41504 {init,fini}object_sizes.
41505 * tree-object-size.c (addr_object_size): Adjust.
41506 (pass_through_call): Adjust.
41507 (pass_object_sizes::execute): Adjust.
41508 * tree-object-size.h (fini_object_sizes): Declare.
41510 2017-01-10 Martin Sebor <msebor@redhat.com>
41512 PR tree-optimization/78775
41513 * builtins.c (get_size_range): Move...
41514 * calls.c: ...to here.
41515 (alloc_max_size): Accept zero argument.
41516 (operand_signed_p): Remove.
41517 (maybe_warn_alloc_args_overflow): Call get_size_range.
41518 * calls.h (get_size_range): Declare.
41520 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41522 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41523 from TI's devices.csv file as of September 2016.
41524 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41526 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41528 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41529 * doc/invoke.texi: Likewise.
41530 * doc/md.texi: Likewise.
41531 * doc/objc.texi: Likewise.
41533 2017-01-10 Joshua Conner <joshconner@google.com>
41535 * config/arm/fuchsia-elf.h: New file.
41536 * config/fuchsia.h: New file.
41537 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41538 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41540 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41542 2016-01-10 Richard Biener <rguenther@suse.de>
41544 PR tree-optimization/79034
41545 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41546 Propagate out degenerate PHIs in the joiner.
41548 2017-01-10 Martin Liska <mliska@suse.cz>
41550 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41551 (sort_congruence_classes_by_decl_uid): Likewise.
41552 (sort_congruence_class_groups_by_decl_uid): Likewise.
41553 (sem_item_optimizer::merge_classes): Sort class, groups in these
41554 classes and members in the groups by DECL_UID of declarations.
41555 This would make merge operations stable.
41557 2017-01-10 Martin Liska <mliska@suse.cz>
41559 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41560 usage of m_classes_vec.
41561 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41562 (sem_item_optimizer::get_group_by_hash): Likewise.
41563 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41564 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41565 (sem_item_optimizer::verify_classes): Likewise.
41566 (sem_item_optimizer::process_cong_reduction): Likewise.
41567 (sem_item_optimizer::dump_cong_classes): Likewise.
41568 (sem_item_optimizer::merge_classes): Likewise.
41569 * ipa-icf.h (congruence_class_hash): Rename from
41570 congruence_class_group_hash. Remove declaration of m_classes_vec.
41572 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41574 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41575 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41576 * config.gcc: Add avx512vpopcntdqintrin.h.
41577 * config/i386/avx512vpopcntdqintrin.h: New.
41578 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41579 * config/i386/i386-builtin-types.def: Add new types.
41580 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41581 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41582 __builtin_ia32_vpopcountq_v8di_mask): New.
41583 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41584 __AVX512VPOPCNTDQ__.
41585 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41586 (PTA_AVX512VPOPCNTDQ): Define.
41587 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41588 TARGET_AVX512VPOPCNTDQ_P): Define.
41589 * config/i386/i386.opt: Add mavx512vpopcntdq.
41590 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41591 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41593 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41595 PR middle-end/77484
41596 * predict.def (PRED_CALL): Set to 67.
41598 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41600 * expr.c (store_field): In the bitfield case, if the value comes from
41601 a function call and is of an aggregate type returned in registers, do
41602 not modify the field mode; extract the value in all cases if the mode
41603 is BLKmode and the size is not larger than a word.
41605 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41608 * config/i386/cpuid.h: Fix undefined behavior.
41610 2017-01-04 Jeff Law <law@redhat.com>
41612 PR tree-optimization/79007
41613 PR tree-optimization/67955
41614 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41615 conservative for pt.null when flag_non_call_exceptions is on.
41617 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41619 PR translation/79019
41620 PR translation/79020
41621 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41622 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41623 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41625 * config/avr/avr.opt (maccumulate-args): Likewise.
41626 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41627 * common.opt (freport-bug): Likewise.
41628 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41629 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41630 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41631 translatable string.
41632 * config/i386/i386.c (function_value_32): Likewise.
41633 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41634 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41636 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41637 * common/config/msp430/msp430-common.c (msp430_handle_option):
41639 * symtab.c (symtab_node::verify_base): Likewise.
41640 * opts.c (set_debug_level): Likewise.
41641 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41642 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41643 missing whitespace to translatable strings.
41644 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41645 * config/sh/superh.h: Likewise.
41646 * config/i386/xopintrin.h: Likewise.
41647 * config/i386/znver1.md: Likewise.
41648 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41649 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41650 * double-int.h (struct double_int): Likewise.
41651 * double-int.c (div_and_round_double): Likewise.
41652 * wide-int.cc: Likewise.
41653 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41654 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41655 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41656 (crossjumps_occurred): ... this.
41657 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41660 PR tree-optimization/78899
41661 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41662 returning bool return struct loop *, NULL for failure and the new
41664 (versionable_outer_loop_p): Don't version outer loop if it has
41665 dont_vectorized bit set.
41666 (tree_if_conversion): When versioning outer loop, ensure
41667 tree_if_conversion is performed also on the inner loop of the
41668 non-vectorizable outer loop copy.
41669 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41670 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41671 prevent vectorization of it.
41672 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41673 the outer loop vectorization of the non-scalar version is attempted
41674 before vectorization of the inner loop in scalar version. If
41675 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41676 vectorization of its inner loop.
41677 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41678 has 2 inner loops, rename also on edges from bb whose single pred
41679 is outer_loop->header. Fix typo in function comment.
41681 2017-01-09 Martin Sebor <msebor@redhat.com>
41684 * asan.c (asan_emit_stack_protection): Increase local buffer size
41685 to avoid snprintf truncation warning.
41687 2017-01-09 Andrew Pinski <apinski@cavium.com>
41689 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41690 to reference thunderx2t99 for the tuning structure
41691 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41693 (thunderx2t99_extra_costs): This.
41694 * config/aarch64/aarch64-tune.md: Regenerate.
41695 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41696 (vulcan_addrcost_table): This.
41697 (vulcan_regmove_cost): Rename to ...
41698 (thunderx2t99_regmove_cost): This.
41699 (vulcan_vector_cost): Rename to ...
41700 (thunderx2t99_vector_cost): this.
41701 (vulcan_branch_cost): Rename to ...
41702 (thunderx2t99_branch_cost): This.
41703 (vulcan_tunings): Rename to ...
41704 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41705 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41707 2017-01-09 Martin Jambor <mjambor@suse.cz>
41711 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41712 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41713 (propagate_vr_accross_jump_function): Use the above function for all
41714 value range computations for pass-through jump functions and type
41715 converasion from explicit value range values.
41716 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41717 parameters from TYPE_ARG_TYPES.
41718 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41719 (ipa_write_node_info): Stream type of the actual argument.
41720 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41722 2017-01-09 Martin Liska <mliska@suse.cz>
41725 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41726 (lookup_compiler): Do not show error message with have_E.
41728 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41730 PR tree-optimization/78938
41731 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41732 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41733 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41734 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41737 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41739 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41742 2017-01-09 Richard Biener <rguenther@suse.de>
41744 PR tree-optimization/78997
41745 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41746 name condition properly.
41748 2017-01-09 Richard Biener <rguenther@suse.de>
41751 * dwarf2out.c (is_cxx): New overload with context.
41752 (is_naming_typedef_decl): Use it.
41754 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41756 * invoke.texi (Option Summary): Correct spacing in option lists
41757 and add line breaks to fix over-long lines.
41759 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41761 PR middle-end/17660
41763 * extend.texi (Common Variable Attributes): Add xref to GCC
41764 Internals manual to explain mode attribute keywords.
41766 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41769 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41770 and Preprocessor Options.
41771 (Options for Linking): Document -pthread here....
41772 (RS/6000 and PowerPC Options): ...not here.
41773 (Solaris 2 Options): ...or here.
41774 * doc/cppopts.texi: Document -pthread.
41776 2017-01-08 Martin Sebor <msebor@redhat.com>
41778 PR middle-end/77708
41779 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41780 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41781 New member functions.
41782 (format_directive): Used them.
41784 (pass_sprintf_length::handle_gimple_call): Same.
41785 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41786 to avoid truncation for any argument.
41787 (extract_affine_mul): Same.
41788 * tree.c (get_file_function_name): Same.
41790 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41792 PR middle-end/77484
41793 * predict.def (PRED_INDIR_CALL): Set to 86.
41795 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41797 PR preprocessor/54124
41798 * doc/cppopts.texi: Reformat -d subtable to list the full name
41799 of the options. Add cross-reference to the docs for the general
41800 compiler -d options.
41801 * doc/invoke.texi (Developer Options): Add cross-reference to the
41802 preprocessor-specific -d option documentation.
41804 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41806 PR preprocessor/13498
41807 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41808 redudant material, and reflect new command-line options.
41809 (System Headers): Likewise.
41811 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41813 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41814 -isystem, and -idirafter. Copy-edit.
41815 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41816 default for -ftrack-macro-expansion. Delete obsolete and
41817 badly-formatted implementation details about -fdebug-cpp output.
41818 * doc/cppwarnopts.texi: Copy-edit.
41820 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41823 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41824 that the transition from a max line width >= 1<<10 to narrower
41825 lines works correctly.
41827 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41829 * doc/options.texi (PerFunction): New.
41830 * opt-functions.awk (switch_flags): Map both Optimization and
41831 PerFunction to CL_OPTIMIZATION.
41832 * opth-gen.awk: Test for PerFunction flag along with
41834 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41835 it only when the latter is present. Skip those that don't in
41836 the hash function generator.
41837 * common.opt (fvar-tracking): Mark as PerFunction instead of
41839 (fvar-tracking-assignments): Likewise.
41840 (fvar-tracking-assignments-toggle): Likewise.
41841 (fvar-tracking-uninit): Likewise.
41843 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41845 PR translation/79018
41846 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41849 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41852 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41853 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41854 TARGET_LONG_JUMP_TABLE_OFFSETS.
41855 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41856 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41857 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41858 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41859 * config/m68k/m68k.md (tablejump expander): Likewise.
41860 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41861 TARGET_LONG_JUMP_TABLE_OFFSETS.
41862 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41863 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41865 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41866 David Holsgrove <david.holsgrove@xilinx.com>
41868 * common/config/microblaze/microblaze-common.c
41869 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41870 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41872 * config/microblaze/microblaze.c (microblaze_must_save_register)
41873 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41875 (microblaze_eh_return): New function.
41876 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41877 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41878 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41879 * config/microblaze/microblaze.md (eh_return): New pattern.
41881 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41883 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41884 GCC_DIAGNOSTIC_STRINGIFY): Define.
41886 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41888 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41890 * config/arm/arm.md (<mcrr>): New.
41892 * config/arm/arm.c (arm_arch5te): New.
41893 (arm_option_override): Set arm_arch5te.
41894 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41896 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41897 (arm_mcrr_qualifiers): ... this. New.
41898 (MRRC_QUALIFIERS): Define to...
41899 (arm_mrrc_qualifiers): ... this. New.
41900 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41902 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41903 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41904 (MRRCI, mrrc, MRRC): New.
41905 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41906 VUNSPEC_MRRC2): New.
41908 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41910 * config/arm/arm.md (<mcr>): New.
41912 * config/arm/arm.c (arm_coproc_builtin_available): Add
41913 support for mcr, mrc, mcr2 and mrc2.
41914 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41915 (arm_mcr_qualifiers): ... this. New.
41916 (MRC_QUALIFIERS): Define to ...
41917 (arm_mrc_qualifiers): ... this. New.
41918 (MCR_QUALIFIERS): Define to ...
41919 (arm_mcr_qualifiers): ... this. New.
41920 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41922 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41923 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41924 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41925 VUNSPEC_MRC2): New.
41927 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41929 * config/arm/arm.md (*ldc): New.
41933 * config/arm/arm.c (arm_coproc_builtin_available): Add
41934 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41935 (arm_coproc_ldc_stc_legitimate_address): New.
41936 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41937 'qualifier_const_pointer'.
41938 (LDC_QUALIFIERS): Define to...
41939 (arm_ldc_qualifiers): ... this. New.
41940 (STC_QUALIFIERS): Define to...
41941 (arm_stc_qualifiers): ... this. New.
41942 * config/arm/arm-protos.h
41943 (arm_coproc_ldc_stc_legitimate_address): New.
41944 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41945 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41946 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41947 stc2, stcl, stc2l): New.
41948 * config/arm/constraints.md (Uz): New.
41949 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41950 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41951 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41952 VUNSPEC_STC2L): New.
41954 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41956 * config/arm/arm.md (<cdp>): New.
41957 * config/arm/arm.c (neon_const_bounds): Rename this ...
41958 (arm_const_bounds): ... this.
41959 (arm_coproc_builtin_available): New.
41960 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41961 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41962 (CDP_QUALIFIERS): Define to...
41963 (arm_cdp_qualifiers): ... this. New.
41965 (arm_expand_builtin_args): Add case for 6 arguments.
41966 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41967 (arm_const_bounds): ... this.
41968 (arm_coproc_builtin_available): New.
41969 * config/arm/arm_acle.h (__arm_cdp): New.
41971 * config/arm/arm_acle_builtins.def (cdp): New.
41973 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41974 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41975 'arm_const_bounds'.
41976 * config/arm/types.md (coproc): New.
41977 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41978 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41979 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41980 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41982 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41984 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41985 (UBINOP_QUALIFIERS): New.
41987 (acle_builtin_data): New. Change comment.
41988 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41989 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41990 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41991 arm_acle_builtins.def.
41992 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41993 (arm_init_acle_builtins): New.
41994 (CRC32_BUILTIN): Remove.
41995 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41996 crc32cb, crc32ch and crc32cw.
41997 (arm_init_crc32_builtins): Remove.
41998 (arm_init_builtins): Use arm_init_acle_builtins rather
41999 than arm_init_crc32_builtins.
42000 (arm_expand_acle_builtin): New.
42001 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
42002 * config/arm/arm_acle_builtins.def: New.
42004 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42006 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
42007 (arm_builtin_datum): ... this.
42008 (arm_init_neon_builtin): Rename to ...
42009 (arm_init_builtin): ... this. Add a new parameters PREFIX
42010 and USE_SIG_IN_NAME.
42011 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
42012 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
42013 'arm_builtin_datum'.
42014 (arm_init_vfp_builtins): Likewise.
42015 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
42016 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
42017 (arm_expand_neon_args): Rename to ...
42018 (arm_expand_builtin_args): ... this. Rename builtin_arg
42019 enum values and differentiate between ARG_BUILTIN_MEMORY
42020 and ARG_BUILTIN_NEON_MEMORY.
42021 (arm_expand_neon_builtin_1): Rename to ...
42022 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
42023 values, arm_expand_builtin_args and add bool parameter NEON.
42024 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
42025 (arm_expand_vfp_builtin): Likewise.
42026 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
42028 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42030 PR middle-end/77484
42031 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
42032 * predict.c (tree_estimate_probability_bb): Reverse direction of
42033 polymorphic call predictor.
42035 2017-01-06 David Malcolm <dmalcolm@redhat.com>
42037 * passes.c (execute_one_pass): Split out pass-skipping logic into...
42038 (determine_pass_name_match): ...this new function and...
42039 (should_skip_pass_p): ...this new function.
42041 2017-01-06 Nathan Sidwell <nathan@acm.org>
42043 * ipa-visibility.c (function_and_variable_visibility): Reformat
42044 comments and long lines. Remove extrneous if.
42045 * symtab.c (symtab_node::make_decl_local): Fix code format.
42046 (symtab_node::set_section_for_node): Fix comment typo.
42048 2017-01-06 Martin Liska <mliska@suse.cz>
42051 * lra-constraints.c: Rename invariant to lra_invariant.
42052 * predict.c (set_even_probabilities): Initialize e to NULL.
42054 2017-01-05 Martin Sebor <msebor@redhat.com>
42056 PR tree-optimization/78910
42057 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
42058 (format_integer): Correct off-by-one error in the handling
42059 of precision with negative numbers in signed conversions..
42061 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
42063 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
42065 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42067 PR tree-optimization/71016
42068 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
42069 factor_out_conditional_conversion. Formatting fix.
42070 (factor_out_conditional_conversion): Add cond_stmt argument.
42071 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
42072 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
42075 2017-01-05 David Malcolm <dmalcolm@redhat.com>
42077 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
42078 read-rtl-function.o, and selftest-rtl.o.
42079 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
42080 (selftest::aarch64_test_loading_full_dump): New function.
42081 (selftest::aarch64_run_selftests): New function.
42082 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
42083 selftest::aarch64_run_selftests.
42084 * config/i386/i386.c
42085 (selftest::ix86_test_loading_dump_fragment_1): New function.
42086 (selftest::ix86_test_loading_call_insn): New function.
42087 (selftest::ix86_test_loading_full_dump): New function.
42088 (selftest::ix86_test_loading_unspec): New function.
42089 (selftest::ix86_run_selftests): Call the new functions.
42090 * emit-rtl.c (maybe_set_max_label_num): New function.
42091 * emit-rtl.h (maybe_set_max_label_num): New decl.
42092 * function.c (instantiate_decls): Guard call to
42093 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
42094 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
42096 * gensupport.c (gen_reader::gen_reader): Pass "false"
42097 for new "compact" param of rtx_reader.
42098 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
42099 rather than an empty string for NULL strings.
42100 * read-md.c: Potentially include config.h rather than bconfig.h.
42101 Wrap include of errors.h with #ifdef GENERATOR_FILE.
42102 (have_error): New global, copied from errors.c.
42103 (md_reader::read_name): Rename to...
42104 (md_reader::read_name_1): ...this, adding "out_loc" param,
42105 and converting "missing name or number" to returning false, rather
42107 (md_reader::read_name): Reimplement in terms of read_name_1.
42108 (md_reader::read_name_or_nil): New function.
42109 (md_reader::read_string): Handle "(nil)" by returning NULL.
42110 (md_reader::md_reader): Add new param "compact".
42111 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
42112 (md_reader::read_file): New method.
42113 * read-md.h (md_reader::md_reader): Add new param "compact".
42114 (md_reader::read_file): New method.
42115 (md_reader::is_compact): New accessor.
42116 (md_reader::read_name): Convert return type from void to file_location.
42117 (md_reader::read_name_or_nil): New decl.
42118 (md_reader::read_name_1): New decl.
42119 (md_reader::m_compact): New field.
42120 (noop_reader::noop_reader): Pass "false" for new "compact" param
42122 (rtx_reader::rtx_reader): Add new "compact" param.
42123 (rtx_reader::read_rtx_operand): Make virtual and convert return
42124 type from void to rtx.
42125 (rtx_reader::read_until): New decl.
42126 (rtx_reader::handle_any_trailing_information): New virtual function.
42127 (rtx_reader::postprocess): New virtual function.
42128 (rtx_reader::finalize_string): New virtual function.
42129 (rtx_reader::m_in_call_function_usage): New field.
42130 (rtx_reader::m_reuse_rtx_by_id): New field.
42131 * read-rtl-function.c: New file.
42132 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
42133 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
42134 (selftest::verify_three_block_rtl_cfg): New decl.
42135 * read-rtl-function.h: New file.
42136 * read-rtl.c: Potentially include config.h rather than bconfig.h.
42137 For host, include function.h, memmodel.h, and emit-rtl.h.
42138 (one_time_initialization): New function.
42139 (struct compact_insn_name): New struct.
42140 (compact_insn_names): New array.
42141 (find_code): Handle insn codes in compact dumps.
42142 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
42143 (bind_subst_iter_and_attr): Likewise.
42144 (add_condition_to_string): Likewise.
42145 (add_condition_to_rtx): Likewise.
42146 (apply_attribute_uses): Likewise.
42147 (add_current_iterators): Likewise.
42148 (apply_iterators): Likewise.
42149 (initialize_iterators): Guard usage of apply_subst_iterator with
42150 #ifdef GENERATOR_FILE.
42151 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
42152 (md_reader::read_mapping): Likewise.
42153 (add_define_attr_for_define_subst): Likewise.
42154 (add_define_subst_attr): Likewise.
42155 (read_subst_mapping): Likewise.
42156 (check_code_iterator): Likewise.
42157 (rtx_reader::read_rtx): Likewise. Move one-time initialization
42159 (one_time_initialization): New function.
42160 (rtx_reader::read_until): New method.
42161 (read_flags): New function.
42162 (parse_reg_note_name): New function.
42163 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42164 Handle reuse_rtx ids.
42165 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42166 Add parsing support for RTL dumps, mirroring the special-cases in
42167 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42168 values, and calling handle_any_trailing_information.
42169 (rtx_reader::read_rtx_operand): Convert return type from void
42170 to rtx, returning return_rtx. Handle case 'e'. Call
42171 finalize_string on XSTR and XTMPL fields.
42172 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42173 "(nil)" values were omitted. Call the postprocess vfunc on the
42175 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42176 class ctor. Initialize m_in_call_function_usage. Call
42177 one_time_initialization.
42178 * rtl-tests.c (selftest::test_uncond_jump): Call
42179 set_new_first_and_last_insn.
42180 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42181 * selftest-rtl.c: New file.
42182 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42183 (selftest::get_insn_by_uid): New decl.
42184 * selftest-run-tests.c (selftest::run_tests): Call
42185 read_rtl_function_c_tests.
42186 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42187 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42190 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42192 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42193 operands in a special way. Assert that pos+len <= mode precision.
42195 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42197 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42198 3 argument Alias with unlimited for the negative form.
42199 (fno-vect-cost-model): Removed.
42201 2017-01-05 Martin Liska <mliska@suse.cz>
42203 * hsa-gen.c (gen_hsa_divmod): New function.
42204 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42206 2017-01-05 Martin Liska <mliska@suse.cz>
42209 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42212 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42214 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42215 small constant length operands.
42217 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42219 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42220 between loop iterations.
42222 2017-01-05 Martin Liska <mliska@suse.cz>
42225 * gimplify.c (gimplify_decl_expr): Compare to
42226 asan_poisoned_variables instread of checking flags.
42227 (gimplify_target_expr): Likewise.
42228 (gimplify_expr): Likewise.
42229 (gimplify_function_tree): Conditionally initialize
42230 asan_poisoned_variables.
42232 2017-01-04 Jeff Law <law@redhat.com>
42234 PR tree-optimizatin/78812
42235 * rtl.h (contains_mem_rtx_p): Prototype.
42236 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42237 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42238 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42239 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42240 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42242 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42244 * input.c (assert_char_at_range): Default-initialize actual_range.
42246 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42248 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42249 to match the caller.
42251 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42253 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42254 insns after final jump in test to emit dummy move.
42256 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42258 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42259 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42261 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42263 * multiple_target.c (create_dispatcher_calls): Init e_next.
42264 * tree-ssa-loop-split.c (split_loop): Init border.
42265 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42268 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42273 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42274 (altivec_register_operand): Do not return true if the operand
42275 contains a SUBREG mixing SImode and SFmode.
42276 (vsx_register_operand): Likewise.
42277 (vsx_reg_sfsubreg_ok): New predicate.
42278 (vfloat_operand): Do not return true if the operand contains a
42279 SUBREG mixing SImode and SFmode.
42280 (vint_operand): Likewise.
42281 (vlogical_operand): Likewise.
42282 (gpc_reg_operand): Likewise.
42283 (int_reg_operand): Likewise.
42284 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42285 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42286 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42288 (rs6000_emit_move_si_sf_subreg): New helper function.
42289 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42290 fixup SUBREGs involving SImode and SFmode.
42291 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42292 numbers for the new peephole2 optimization.
42293 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42294 the GLIBC math library that do AND/IOR/XOR operations on single
42295 precision floating point.
42296 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42297 target macros to say whether we need to avoid SUBREGs mixing
42299 (TARGET_ALLOW_SF_SUBREG): Likewise.
42300 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42301 (UNSPEC_SI_FROM_SF): Likewise.
42302 (iorxor): Change spacing.
42303 (and_ior_xor): New iterator for AND, IOR, and XOR.
42304 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42305 (movdi_from_sf_zero_ext): Likewise.
42306 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42307 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42308 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42309 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42310 (fms<mode>4): Likewise.
42311 (fnma<mode>4): Likewise.
42312 (fnms<mode>4): Likewise.
42313 (nfma<mode>4): Likewise.
42314 (nfms<mode>4): Likewise.
42316 2017-01-04 Marek Polacek <polacek@redhat.com>
42319 * doc/invoke.texi: Document -Wpointer-compare.
42321 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42323 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42326 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42327 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42330 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42333 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42334 documentation of the powerpc_popcntb_ok attribute.
42335 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42336 code to issue warning messages if a requested CPU configuration is
42337 not supported by the binary (assembler and loader) toolchain.
42338 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42339 made to define a built-in function that has been disabled.
42340 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42341 made to define a built-in function that has been disabled.
42342 (altivec_init_builtins): Add comment explaining why definition
42343 of the DST built-in functions is not preceded by an assertion
42344 check. Add assertions to prevent ICE if attempts are made to
42345 define an altivec predicate or an abs* built-in function that has
42347 (htm_init_builtins): Add comment explaining why definition of the
42348 htm built-in functions is not preceded by an assertion check.
42350 2017-01-04 Jeff Law <law@redhat.com>
42352 PR tree-optimizatin/67955
42353 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42354 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42355 the points-to solution does not include pt_null. Use DECL_PT_UID
42358 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42360 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42361 Use gen_int_mode instead of gen_lopwart for const_int operands.
42363 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42365 PR tree-optimization/71563
42366 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42367 out of range value - has low bits known to be zero.
42369 2017-01-04 Alan Modra <amodra@gmail.com>
42371 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42372 * configure: Regenerate.
42373 * config.in: Regenerate.
42375 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42378 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42379 a substring of the message, but strcmp with the whole message. Ifdef
42380 ENABLE_NLS, translate the message first using dgettext.
42382 2017-01-03 Jeff Law <law@redhat.com>
42384 PR tree-optimizatin/78856
42385 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42386 (mark_threaded_blocks): Remove code to truncate thread paths that
42387 cross multiple loop headers. Instead invalidate the cached loop
42388 iteration information and handle case of a thread path walking
42389 into an irreducible region.
42391 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42394 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42395 assertions. Add support for doing the signbit if the IEEE 128-bit
42396 floating point value is in a GPR.
42397 * config/rs6000/rs6000.md (Fsignbit): Delete.
42398 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42399 Update the length attribute if the value is in a GPR.
42400 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42401 the sign or zero extension instruction, since the value is always 0/1.
42402 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42405 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42406 extracting SImode to a GPR register so that we can generate a
42407 store, limit the vector to be in a traditional Altivec register
42408 for the vextuwrx instruction.
42410 2017-01-03 Ian Lance Taylor <iant@google.com>
42412 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42414 2017-01-03 Martin Sebor <msebor@redhat.com>
42416 PR tree-optimization/78696
42417 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42418 precision. Use MPFR for %f for greater fidelity. Correct handling
42420 (pass_sprintf_length::compute_format_length): Set width and precision
42421 specified by asrerisk to void_node for vararg functions.
42422 (try_substitute_return_value): Adjust dump output.
42424 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42426 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42428 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42430 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42431 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42432 -mlra/-mno-lra was passed to the compiler.
42434 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42436 PR rtl-optimization/65618
42437 * emit-rtl.c (try_split): Move initialization of "before" and
42438 "after" to just before the call to emit_insn_after_setloc.
42440 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42442 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42444 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42446 * dwarf2out.c (gen_enumeration_type_die): When
42447 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42449 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42451 PR tree-optimization/78965
42452 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42453 Change first argument from const call_info & to call_info &. For %n
42454 set info.nowrite to false.
42456 PR middle-end/78901
42457 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42458 possibly throwing calls.
42460 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42461 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42462 and fns handling, rather than in a separate case SSA_NAME.
42464 2017-01-02 Jeff Law <law@redhat.com>
42466 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42467 fixes for first_period and second_period variables.
42469 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42472 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42473 (*insvqi_1): New insn pattern.
42474 (*insvqi_1_mem_rex64): Ditto.
42475 (*insvqi_2): Ditto.
42476 (*insvqi_3): Rename from *insvqi.
42478 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42480 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42482 * doc/cfg.texi (Edges): Remove reference to Java.
42483 (Maintaining the CFG): Ditto.
42485 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42487 PR middle-end/77674
42488 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42489 transparent aliases.
42491 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42493 PR middle-end/77484
42494 * predict.def (PRED_CALL): Update hitrate.
42495 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42496 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42497 into direct/indirect/polymorphic variants.
42499 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42501 Update copyright years.
42503 * gcc.c (process_command): Update copyright notice dates.
42504 * gcov-dump.c (print_version): Ditto.
42505 * gcov.c (print_version): Ditto.
42506 * gcov-tool.c (print_version): Ditto.
42507 * gengtype.c (create_file): Ditto.
42508 * doc/cpp.texi: Bump @copying's copyright year.
42509 * doc/cppinternals.texi: Ditto.
42510 * doc/gcc.texi: Ditto.
42511 * doc/gccint.texi: Ditto.
42512 * doc/gcov.texi: Ditto.
42513 * doc/install.texi: Ditto.
42514 * doc/invoke.texi: Ditto.
42516 Copyright (C) 2017 Free Software Foundation, Inc.
42518 Copying and distribution of this file, with or without modification,
42519 are permitted in any medium without royalty provided the copyright
42520 notice and this notice are preserved.