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 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
6 * target.def (autovectorize_vector_sizes): Return the vector sizes
7 by pointer, using vector_sizes rather than a bitmask.
8 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
9 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
10 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
12 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
13 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
14 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
15 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
16 * omp-general.c (omp_max_vf): Likewise.
17 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
19 * tree-vect-loop.c (vect_analyze_loop): Likewise.
20 * tree-vect-slp.c (vect_slp_bb): Likewise.
21 * doc/tm.texi: Regenerate.
22 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
24 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
25 the vector size as a poly_uint64 rather than an unsigned int.
26 (current_vector_size): Change from an unsigned int to a poly_uint64.
27 (get_vectype_for_scalar_type): Update accordingly.
28 * tree.h (build_truth_vector_type): Take the size and number of
29 units as a poly_uint64 rather than an unsigned int.
30 (build_vector_type): Add a temporary overload that takes
31 the number of units as a poly_uint64 rather than an unsigned int.
32 * tree.c (make_vector_type): Likewise.
33 (build_truth_vector_type): Take the number of units as a poly_uint64
34 rather than an unsigned int.
36 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
37 Alan Hayward <alan.hayward@arm.com>
38 David Sherwood <david.sherwood@arm.com>
40 * target.def (get_mask_mode): Take the number of units and length
41 as poly_uint64s rather than unsigned ints.
42 * targhooks.h (default_get_mask_mode): Update accordingly.
43 * targhooks.c (default_get_mask_mode): Likewise.
44 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
45 * doc/tm.texi: Regenerate.
47 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
48 Alan Hayward <alan.hayward@arm.com>
49 David Sherwood <david.sherwood@arm.com>
51 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
52 * omp-general.c (omp_max_vf): Likewise.
53 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
54 (expand_omp_simd): Handle polynomial safelen.
55 * omp-low.c (omplow_simd_context): Add a default constructor.
56 (omplow_simd_context::max_vf): Change from int to poly_uint64.
57 (lower_rec_simd_input_clauses): Update accordingly.
58 (lower_rec_input_clauses): Likewise.
60 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
61 Alan Hayward <alan.hayward@arm.com>
62 David Sherwood <david.sherwood@arm.com>
64 * tree-vectorizer.h (vect_nunits_for_cost): New function.
65 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
66 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
67 (vect_analyze_slp_cost): Likewise.
68 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
69 (vect_model_load_cost): Likewise.
71 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
72 Alan Hayward <alan.hayward@arm.com>
73 David Sherwood <david.sherwood@arm.com>
75 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
76 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
77 from an unsigned int * to a poly_uint64_pod *.
78 (calculate_unrolling_factor): New function.
79 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
81 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
82 Alan Hayward <alan.hayward@arm.com>
83 David Sherwood <david.sherwood@arm.com>
85 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
86 from an unsigned int to a poly_uint64.
87 (_loop_vec_info::slp_unrolling_factor): Likewise.
88 (_loop_vec_info::vectorization_factor): Change from an int
90 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
91 (vect_get_num_vectors): New function.
92 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
93 (vect_get_num_copies): Use vect_get_num_vectors.
94 (vect_analyze_data_ref_dependences): Change max_vf from an int *
96 (vect_analyze_data_refs): Change min_vf from an int * to a
98 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
99 than an unsigned HOST_WIDE_INT.
100 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
101 (vect_analyze_data_ref_dependence): Change max_vf from an int *
102 to an unsigned int *.
103 (vect_analyze_data_ref_dependences): Likewise.
104 (vect_compute_data_ref_alignment): Handle polynomial vf.
105 (vect_enhance_data_refs_alignment): Likewise.
106 (vect_prune_runtime_alias_test_list): Likewise.
107 (vect_shift_permute_load_chain): Likewise.
108 (vect_supportable_dr_alignment): Likewise.
109 (dependence_distance_ge_vf): Take the vectorization factor as a
110 poly_uint64 rather than an unsigned HOST_WIDE_INT.
111 (vect_analyze_data_refs): Change min_vf from an int * to a
113 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
114 vfm1 as a poly_uint64 rather than an int. Make the same change
115 for the returned bound_scalar.
116 (vect_gen_vector_loop_niters): Handle polynomial vf.
117 (vect_do_peeling): Likewise. Update call to
118 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
119 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
121 * tree-vect-loop.c (vect_determine_vectorization_factor)
122 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
123 (vect_get_known_peeling_cost): Likewise.
124 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
125 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
126 (vect_transform_loop): Likewise. Use the lowest possible VF when
127 updating the upper bounds of the loop.
128 (vect_min_worthwhile_factor): Make static. Return an unsigned int
130 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
131 polynomial unroll factors.
132 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
133 (vect_make_slp_decision): Likewise.
134 (vect_supported_load_permutation_p): Likewise, and polynomial
136 (vect_analyze_slp_cost): Handle polynomial vf.
137 (vect_slp_analyze_node_operations): Likewise.
138 (vect_slp_analyze_bb_1): Likewise.
139 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
140 than an unsigned HOST_WIDE_INT.
141 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
142 (vectorizable_load): Handle polynomial vf.
143 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
145 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
147 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
148 Alan Hayward <alan.hayward@arm.com>
149 David Sherwood <david.sherwood@arm.com>
151 * match.pd: Handle bit operations involving three constants
152 and try to fold one pair.
154 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
156 * tree-vect-loop-manip.c: Include gimple-fold.h.
157 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
158 niters_maybe_zero parameters. Handle other cases besides a step of 1.
159 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
160 Add a path that uses a step of VF instead of 1, but disable it
162 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
163 and niters_no_overflow parameters. Update calls to
164 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
165 Create a new SSA name if the latter choses to use a ste other
166 than zero, and return it via niters_vector_mult_vf_var.
167 * tree-vect-loop.c (vect_transform_loop): Update calls to
168 vect_do_peeling, vect_gen_vector_loop_niters and
169 slpeel_make_loop_iterate_ntimes.
170 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
171 (vect_gen_vector_loop_niters): Update declarations after above changes.
173 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
175 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
176 128-bit round to integer instructions.
177 (ceil<mode>2): Likewise.
178 (btrunc<mode>2): Likewise.
179 (round<mode>2): Likewise.
181 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
183 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
184 unaligned VSX load/store on P8/P9.
185 (expand_block_clear): Allow the use of unaligned VSX
188 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
190 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
192 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
193 swap associated with both a load and a store.
195 2018-01-02 Andrew Waterman <andrew@sifive.com>
197 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
198 * config/riscv/riscv.md (clear_cache): Use it.
200 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
202 * web.c: Remove out-of-date comment.
204 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
206 * expr.c (fixup_args_size_notes): Check that any existing
207 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
208 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
209 (emit_single_push_insn): ...here.
211 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
213 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
214 (const_vector_encoded_nelts): New function.
215 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
216 (const_vector_int_elt, const_vector_elt): Declare.
217 * emit-rtl.c (const_vector_int_elt_1): New function.
218 (const_vector_elt): Likewise.
219 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
222 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
224 * expr.c: Include rtx-vector-builder.h.
225 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
226 directly on the tree encoding.
227 (const_vector_from_tree): Likewise.
228 * optabs.c: Include rtx-vector-builder.h.
229 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
230 sequence of "u" values.
231 * vec-perm-indices.c: Include rtx-vector-builder.h.
232 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
233 directly on the vec_perm_indices encoding.
235 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
237 * doc/rtl.texi (const_vector): Describe new encoding scheme.
238 * Makefile.in (OBJS): Add rtx-vector-builder.o.
239 * rtx-vector-builder.h: New file.
240 * rtx-vector-builder.c: Likewise.
241 * rtl.h (rtx_def::u2): Add a const_vector field.
242 (CONST_VECTOR_NPATTERNS): New macro.
243 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
244 (CONST_VECTOR_DUPLICATE_P): Likewise.
245 (CONST_VECTOR_STEPPED_P): Likewise.
246 (CONST_VECTOR_ENCODED_ELT): Likewise.
247 (const_vec_duplicate_p): Check for a duplicated vector encoding.
248 (unwrap_const_vec_duplicate): Likewise.
249 (const_vec_series_p): Check for a non-duplicated vector encoding.
250 Say that the function only returns true for integer vectors.
251 * emit-rtl.c: Include rtx-vector-builder.h.
252 (gen_const_vec_duplicate_1): Delete.
253 (gen_const_vector): Call gen_const_vec_duplicate instead of
254 gen_const_vec_duplicate_1.
255 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
256 (gen_const_vec_duplicate): Use rtx_vector_builder.
257 (gen_const_vec_series): Likewise.
258 (gen_rtx_CONST_VECTOR): Likewise.
259 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
260 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
261 Build a new vector rather than modifying a CONST_VECTOR in-place.
262 (handle_special_swappables): Update call accordingly.
263 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
264 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
265 Build a new vector rather than modifying a CONST_VECTOR in-place.
266 (handle_special_swappables): Update call accordingly.
268 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
270 * simplify-rtx.c (simplify_const_binary_operation): Use
271 CONST_VECTOR_ELT instead of XVECEXP.
273 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
275 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
276 the selector elements to be different from the data elements
277 if the selector is a VECTOR_CST.
278 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
279 ssizetype for the selector.
281 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
283 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
284 before testing each element individually.
285 * tree-vect-generic.c (lower_vec_perm): Likewise.
287 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
289 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
290 * selftest-run-tests.c (selftest::run_tests): Call it.
291 * vector-builder.h (vector_builder::operator ==): New function.
292 (vector_builder::operator !=): Likewise.
293 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
294 (vec_perm_indices::all_from_input_p): New function.
295 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
296 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
297 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
298 instead of reading the VECTOR_CST directly. Detect whether both
299 vector inputs are the same before constructing the vec_perm_indices,
300 and update the number of inputs argument accordingly. Use the
301 utility functions added above. Only construct sel2 if we need to.
303 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
305 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
306 the broadcast of the low byte.
307 (expand_mult_highpart): Use an explicit encoding for the permutes.
308 * optabs-query.c (can_mult_highpart_p): Likewise.
309 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
310 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
311 (vectorizable_bswap): Likewise.
312 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
313 explicit encoding for the power-of-2 permutes.
314 (vect_permute_store_chain): Likewise.
315 (vect_grouped_load_supported): Likewise.
316 (vect_permute_load_chain): Likewise.
318 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
320 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
321 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
322 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
323 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
324 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
325 (vect_gen_perm_mask_any): Likewise.
327 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
329 * int-vector-builder.h: New file.
330 * vec-perm-indices.h: Include int-vector-builder.h.
331 (vec_perm_indices): Redefine as an int_vector_builder.
332 (auto_vec_perm_indices): Delete.
333 (vec_perm_builder): Redefine as a stand-alone class.
334 (vec_perm_indices::vec_perm_indices): New function.
335 (vec_perm_indices::clamp): Likewise.
336 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
337 (vec_perm_indices::new_vector): New function.
338 (vec_perm_indices::new_expanded_vector): Update for new
339 vec_perm_indices class.
340 (vec_perm_indices::rotate_inputs): New function.
341 (vec_perm_indices::all_in_range_p): Operate directly on the
342 encoded form, without computing elided elements.
343 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
344 encoding. Update for new vec_perm_indices class.
345 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
346 the given vec_perm_builder.
347 (expand_vec_perm_var): Update vec_perm_builder constructor.
348 (expand_mult_highpart): Use vec_perm_builder instead of
349 auto_vec_perm_indices.
350 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
351 vec_perm_indices instead of auto_vec_perm_indices. Use a single
352 or double series encoding as appropriate.
353 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
354 vec_perm_indices instead of auto_vec_perm_indices.
355 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
356 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
357 (vect_permute_store_chain): Likewise.
358 (vect_grouped_load_supported): Likewise.
359 (vect_permute_load_chain): Likewise.
360 (vect_shift_permute_load_chain): Likewise.
361 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
362 (vect_transform_slp_perm_load): Likewise.
363 (vect_schedule_slp_instance): Likewise.
364 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
365 (vectorizable_mask_load_store): Likewise.
366 (vectorizable_bswap): Likewise.
367 (vectorizable_store): Likewise.
368 (vectorizable_load): Likewise.
369 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
370 vec_perm_indices instead of auto_vec_perm_indices. Use
371 tree_to_vec_perm_builder to read the vector from a tree.
372 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
373 vec_perm_builder instead of a vec_perm_indices.
374 (have_whole_vector_shift): Use vec_perm_builder and
375 vec_perm_indices instead of auto_vec_perm_indices. Leave the
376 truncation to calc_vec_perm_mask_for_shift.
377 (vect_create_epilog_for_reduction): Likewise.
378 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
379 from auto_vec_perm_indices to vec_perm_indices.
380 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
381 instead of changing individual elements.
382 (aarch64_vectorize_vec_perm_const): Use new_vector to install
383 the vector in d.perm.
384 * config/arm/arm.c (expand_vec_perm_d::perm): Change
385 from auto_vec_perm_indices to vec_perm_indices.
386 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
387 instead of changing individual elements.
388 (arm_vectorize_vec_perm_const): Use new_vector to install
389 the vector in d.perm.
390 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
391 Update vec_perm_builder constructor.
392 (rs6000_expand_interleave): Likewise.
393 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
394 (rs6000_expand_interleave): Likewise.
396 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
398 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
399 to qimode could truncate the indices.
400 * optabs.c (expand_vec_perm_var): Likewise.
402 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
404 * Makefile.in (OBJS): Add vec-perm-indices.o.
405 * vec-perm-indices.h: New file.
406 * vec-perm-indices.c: Likewise.
407 * target.h (vec_perm_indices): Replace with a forward class
409 (auto_vec_perm_indices): Move to vec-perm-indices.h.
410 * optabs.h: Include vec-perm-indices.h.
411 (expand_vec_perm): Delete.
412 (selector_fits_mode_p, expand_vec_perm_var): Declare.
413 (expand_vec_perm_const): Declare.
414 * target.def (vec_perm_const_ok): Replace with...
415 (vec_perm_const): ...this new hook.
416 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
417 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
418 * doc/tm.texi: Regenerate.
419 * optabs.def (vec_perm_const): Delete.
420 * doc/md.texi (vec_perm_const): Likewise.
421 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
422 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
423 expand_vec_perm for constant permutation vectors. Assert that
424 the mode of variable permutation vectors is the integer equivalent
425 of the mode that is being permuted.
426 * optabs-query.h (selector_fits_mode_p): Declare.
427 * optabs-query.c: Include vec-perm-indices.h.
428 (selector_fits_mode_p): New function.
429 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
430 is defined, instead of checking whether the vec_perm_const_optab
431 exists. Use targetm.vectorize.vec_perm_const instead of
432 targetm.vectorize.vec_perm_const_ok. Check whether the indices
433 fit in the vector mode before using a variable permute.
434 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
435 vec_perm_indices instead of an rtx.
436 (expand_vec_perm): Replace with...
437 (expand_vec_perm_const): ...this new function. Take the selector
438 as a vec_perm_indices rather than an rtx. Also take the mode of
439 the selector. Update call to shift_amt_for_vec_perm_mask.
440 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
441 Use vec_perm_indices::new_expanded_vector to expand the original
442 selector into bytes. Check whether the indices fit in the vector
443 mode before using a variable permute.
444 (expand_vec_perm_var): Make global.
445 (expand_mult_highpart): Use expand_vec_perm_const.
446 * fold-const.c: Includes vec-perm-indices.h.
447 * tree-ssa-forwprop.c: Likewise.
448 * tree-vect-data-refs.c: Likewise.
449 * tree-vect-generic.c: Likewise.
450 * tree-vect-loop.c: Likewise.
451 * tree-vect-slp.c: Likewise.
452 * tree-vect-stmts.c: Likewise.
453 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
455 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
456 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
457 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
458 (aarch64_vectorize_vec_perm_const): ...this new function.
459 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
460 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
461 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
462 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
463 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
464 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
465 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
467 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
468 check for NEON modes.
469 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
470 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
471 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
472 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
474 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
475 the old VEC_PERM_CONST conditions.
476 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
477 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
478 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
479 (ia64_vectorize_vec_perm_const_ok): Merge into...
480 (ia64_vectorize_vec_perm_const): ...this new function.
481 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
482 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
483 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
484 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
485 * config/mips/mips.c (mips_expand_vec_perm_const)
486 (mips_vectorize_vec_perm_const_ok): Merge into...
487 (mips_vectorize_vec_perm_const): ...this new function.
488 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
489 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
490 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
491 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
492 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
493 (rs6000_expand_vec_perm_const): Delete.
494 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
496 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
497 (altivec_expand_vec_perm_const_le): Take each operand individually.
498 Operate on constant selectors rather than rtxes.
499 (altivec_expand_vec_perm_const): Likewise. Update call to
500 altivec_expand_vec_perm_const_le.
501 (rs6000_expand_vec_perm_const): Delete.
502 (rs6000_vectorize_vec_perm_const_ok): Delete.
503 (rs6000_vectorize_vec_perm_const): New function.
504 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
505 an element count and rtx array.
506 (rs6000_expand_extract_even): Update call accordingly.
507 (rs6000_expand_interleave): Likewise.
508 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
509 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
510 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
511 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
512 (rs6000_expand_vec_perm_const): Delete.
513 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
514 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
515 (altivec_expand_vec_perm_const_le): Take each operand individually.
516 Operate on constant selectors rather than rtxes.
517 (altivec_expand_vec_perm_const): Likewise. Update call to
518 altivec_expand_vec_perm_const_le.
519 (rs6000_expand_vec_perm_const): Delete.
520 (rs6000_vectorize_vec_perm_const_ok): Delete.
521 (rs6000_vectorize_vec_perm_const): New function. Remove stray
522 reference to the SPE evmerge intructions.
523 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
524 an element count and rtx array.
525 (rs6000_expand_extract_even): Update call accordingly.
526 (rs6000_expand_interleave): Likewise.
527 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
528 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
530 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
532 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
534 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
535 vector mode and that that mode matches the mode of the data
537 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
538 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
539 directly using expand_vec_perm_1 when forcing selectors into
541 (expand_vec_perm_var): New function, split out from expand_vec_perm.
543 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
545 * optabs-query.h (can_vec_perm_p): Delete.
546 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
547 * optabs-query.c (can_vec_perm_p): Split into...
548 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
549 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
550 particular selector is valid.
551 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
552 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
553 (vect_grouped_load_supported): Likewise.
554 (vect_shift_permute_load_chain): Likewise.
555 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
556 (vect_transform_slp_perm_load): Likewise.
557 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
558 (vectorizable_bswap): Likewise.
559 (vect_gen_perm_mask_checked): Likewise.
560 * fold-const.c (fold_ternary_loc): Likewise. Don't take
561 implementations of variable permutation vectors into account
562 when deciding which selector to use.
563 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
564 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
565 with a false third argument.
566 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
567 to test whether the constant selector is valid and can_vec_perm_var_p
568 to test whether a variable selector is valid.
570 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
572 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
573 * optabs-query.c (can_vec_perm_p): Likewise.
574 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
575 instead of vec_perm_indices.
576 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
577 (vect_gen_perm_mask_checked): Likewise,
578 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
579 (vect_gen_perm_mask_checked): Likewise,
581 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
583 * optabs-query.h (qimode_for_vec_perm): Declare.
584 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
585 (qimode_for_vec_perm): ...this new function.
586 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
588 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
590 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
591 does not have a conditional at the top.
593 2018-01-02 Richard Biener <rguenther@suse.de>
595 * ipa-inline.c (big_speedup_p): Fix expression.
597 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
600 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
603 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
607 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
608 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
609 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
610 cond_taken_branch_cost 3->4.
612 2018-01-01 Jakub Jelinek <jakub@redhat.com>
614 PR tree-optimization/83581
615 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
616 TODO_cleanup_cfg if any changes have been made.
619 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
620 convert_modes if target mode has the right side, but different mode
624 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
625 last argument when extracting from CONCAT. If either from_real or
626 from_imag is NULL, use expansion through memory. If result is not
627 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
628 the parts directly to inner mode, if even that fails, use expansion
632 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
633 check for bswap in mode rather than HImode and use that in expand_unop
636 2017-12-31 Jakub Jelinek <jakub@redhat.com>
639 * config/i386/i386.c (ix86_attribute_table): Remove excess
640 initializer for "shared" attribute.
642 2017-12-30 Tom de Vries <tom@codesourcery.com>
645 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
647 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
648 offload_funcs entries that no longer have a corresponding cgraph_node.
649 Mark the remaining ones as DECL_PRESERVE_P.
650 (output_lto): Call prune_offload_funcs.
652 2017-12-30 Jakub Jelinek <jakub@redhat.com>
654 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
655 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
656 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
657 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
658 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
659 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
660 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
661 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
662 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
663 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
664 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
665 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
666 vaesenc_<mode>, vpclmulqdq_<mode>,
667 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
669 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
671 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
672 _Float<N> and _Float<N>X variants for rounding built-in
674 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
675 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
676 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
677 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
678 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
679 * builtins.c (mathfn_built_in_2): Likewise.
680 * internal-fn.def (CEIL): Likewise.
682 (NEARBYINT): Likewise.
686 * convert.c (convert_to_integer_1): Likewise.
687 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
688 (integer_valued_real_call_p): Likewise.
689 * fold-const-call.c (fold_const_call_ss): Likewise.
690 * gencfn-macros.c (print_case_cfn): Change CFN and operator
691 printers to take a const char * suffix instead of a bool.
692 (print_define_operator_list): Likewise.
693 (fltall_suffixes): New list of suffixes, that include the
694 traditional suffixes as well as all of the _Float<N> and
696 (main): For _Float<N> and _Float<N>X functions, emit both
697 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
698 has the _Float<N> and _Float<N>X case names or operators. The
699 <name>_ALL variant has both the traditional and the
700 _Float<N>/_Float<N>X case names or operators.
701 * match.pd (COPYSIGN optimizations): Provide optimizations for
702 _Float<N> and _Float<N>X types where possible.
703 (MIN/MAX optimizations): Likewise.
704 (sqrt optimizations): Likewise.
705 (rounding optimizations): Likewise.
707 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
709 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
710 gen_int_mode rather than GEN_INT.
711 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
713 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
715 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
716 instead of CONSTANT_P.
717 (gen_vec_series): Likewise.
718 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
720 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
722 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
725 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
727 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
728 (aarch64_ld1x2<VDC:mode>): Likewise.
729 (aarch64_simd_ld1<mode>_x2): Likewise.
730 (aarch64_simd_ld1<mode>_x2): Likewise.
731 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
732 (vld1_s8_x2): Likewise.
733 (vld1_u16_x2): Likewise.
734 (vld1_s16_x2): Likewise.
735 (vld1_u32_x2): Likewise.
736 (vld1_s32_x2): Likewise.
737 (vld1_u64_x2): Likewise.
738 (vld1_s64_x2): Likewise.
739 (vld1_f16_x2): Likewise.
740 (vld1_f32_x2): Likewise.
741 (vld1_f64_x2): Likewise.
742 (vld1_p8_x2): Likewise.
743 (vld1_p16_x2): Likewise.
744 (vld1_p64_x2): Likewise.
745 (vld1q_u8_x2): Likewise.
746 (vld1q_s8_x2): Likewise.
747 (vld1q_u16_x2): Likewise.
748 (vld1q_s16_x2): Likewise.
749 (vld1q_u32_x2): Likewise.
750 (vld1q_s32_x2): Likewise.
751 (vld1q_u64_x2): Likewise.
752 (vld1q_s64_x2): Likewise.
753 (vld1q_f16_x2): Likewise.
754 (vld1q_f32_x2): Likewise.
755 (vld1q_f64_x2): Likewise.
756 (vld1q_p8_x2): Likewise.
757 (vld1q_p16_x2): Likewise.
758 (vld1q_p64_x2): Likewise.
760 2017-12-27 Martin Liska <mliska@suse.cz>
762 PR tree-optimization/83552
763 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
764 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
766 2017-12-27 Tom de Vries <tom@codesourcery.com>
768 * config/nvptx/nvptx.c (nvptx_option_override): Disable
769 -gstatement-frontiers.
771 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
773 PR rtl-optimization/83513
774 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
775 before priority comparison.
777 2017-12-25 Jakub Jelinek <jakub@redhat.com>
780 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
781 ix86_isa_flags2 to ix86_isa_flags.
782 * config/i386/i386-c.c (ix86_target_macros_internal): Test
783 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
784 isa_flags rather than isa_flags2.
785 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
786 and -mavx512bitalg from isa2_opts to isa_opts.
787 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
788 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
789 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
791 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
792 OPTION_MASK_ISA_AVX512F ored with another option similarly to
793 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
794 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
795 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
796 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
797 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
798 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
799 builtins from bdesc_args2 to bdesc_args section.
800 (__builtin_ia32_compressstoreuqi512_mask,
801 __builtin_ia32_compressstoreuhi512_mask,
802 __builtin_ia32_compressstoreuqi256_mask,
803 __builtin_ia32_expandloadqi512_mask,
804 __builtin_ia32_expandloadqi512_maskz,
805 __builtin_ia32_expandloadhi512_mask,
806 __builtin_ia32_expandloadhi512_maskz,
807 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
808 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
809 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
810 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
811 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
812 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
813 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
814 __builtin_ia32_vpshldv_v32hi_maskz,
815 __builtin_ia32_vpopcountb_v64qi_mask,
816 __builtin_ia32_vpopcountw_v32hi_mask,
817 __builtin_ia32_vpshufbitqmb512_mask,
818 __builtin_ia32_vpshufbitqmb256_mask): Add
819 " | OPTION_MASK_ISA_AVX512BW".
820 (__builtin_ia32_expandloadqi256_mask,
821 __builtin_ia32_expandloadqi256_maskz,
822 __builtin_ia32_vpopcountb_v32qi_mask): Add
823 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
824 (__builtin_ia32_expandloadhi256_mask,
825 __builtin_ia32_expandloadhi256_maskz,
826 __builtin_ia32_expandloadqi128_mask,
827 __builtin_ia32_expandloadqi128_maskz,
828 __builtin_ia32_expandloadhi128_mask,
829 __builtin_ia32_expandloadhi128_maskz,
830 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
831 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
832 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
833 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
834 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
835 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
836 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
837 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
838 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
839 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
840 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
841 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
842 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
843 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
844 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
845 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
846 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
847 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
848 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
849 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
850 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
851 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
852 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
853 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
854 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
855 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
856 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
857 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
858 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
859 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
860 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
861 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
862 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
863 __builtin_ia32_vpopcountw_v8hi_mask): Add
864 " | OPTION_MASK_ISA_AVX512VL".
865 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
866 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
867 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
868 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
869 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
870 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
871 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
872 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
873 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
874 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
875 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
876 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
877 _mm_mask_bitshuffle_epi64_mask): Likewise.
878 * common/config/i386/i386-common.c
879 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
880 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
881 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
882 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
883 OPTION_MASK_ISA_AVX512BITALG_UNSET.
884 (OPTION_MASK_ISA2_AVX512F_UNSET,
885 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
886 (ix86_handle_option): For -mno-general-regs-only, clear from
887 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
888 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
889 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
890 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
891 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
892 for moving from ix86_isa_flags2 to ix86_isa_flags.
894 2017-12-23 Jakub Jelinek <jakub@redhat.com>
897 * fold-const.c (struct contains_label_data): New type.
898 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
899 inside of a SWITCH_BODY seen during the walk.
900 (contains_label_p): Use walk_tree instead of
901 walk_tree_without_duplicates, prepare data for contains_label_1 and
904 2017-12-22 Jakub Jelinek <jakub@redhat.com>
907 * tree-iterator.c (alloc_stmt_list): Start with cleared
908 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
912 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
913 enable avx512f explicitly in #pragma GCC target.
914 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
915 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
916 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
917 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
918 __builtin_ia32_vpdpbusds_v8si_mask,
919 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
920 __builtin_ia32_vpdpbusds_v4si_mask,
921 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
922 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
923 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
924 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
925 __builtin_ia32_vpdpwssds_v8si_mask,
926 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
927 __builtin_ia32_vpdpwssds_v4si_mask,
928 __builtin_ia32_vpdpwssds_v4si_maskz): Use
929 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
930 just OPTION_MASK_ISA_AVX512VNNI.
932 2017-12-22 Martin Jambor <mjambor@suse.cz>
935 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
938 2017-12-22 Julia Koval <julia.koval@intel.com>
939 Sebastian Peryt <sebastian.peryt@intel.com>
941 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
942 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
943 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
944 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
945 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
946 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
947 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
948 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
949 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
950 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
951 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
952 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
953 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
954 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
955 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
956 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
957 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
958 _mm256_maskz_popcnt_epi64): New intrinsics.
959 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
960 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
961 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
962 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
963 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
964 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
965 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
966 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
967 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
968 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
969 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
970 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
971 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
972 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
973 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
974 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
975 __builtin_ia32_vpshufbitqmb256_mask,
976 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
977 * config/i386/i386-c.c (__AVX512BITALG__): New.
978 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
979 (ix86_valid_target_attribute_inner_p): Ditto.
980 (ix86_expand_args_builtin): Handle new types.
981 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
982 * config/i386/i386.opt: Add -mavx512bitalg.
983 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
984 avx512bitalgintrin.h.
985 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
986 (vpopcount<mode><mask_name>): Add more types.
987 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
988 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
990 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
992 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
993 Or in OPTION_MASK_ISA_AVX512F_SET.
994 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
995 OPTION_MASK_ISA_AVX512VNNI_UNSET.
996 (ix86_handle_option): Adjust for
997 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
998 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
1000 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
1001 isa_flag instead of isa_flag2.
1002 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
1003 isa_opts2 to isa_opts.
1004 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
1007 2017-12-22 Mike Stump <mikestump@comcast.net>
1008 Eric Botcazou <ebotcazou@adacore.com>
1010 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
1012 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1015 * config/i386/i386.c (ix86_function_arg_boundary): Return
1016 PARM_BOUNDARY for TYPE_EMPTY_P types.
1019 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
1020 if navail is >= dir.len.
1022 2017-12-21 Steve Ellcey <sellcey@cavium.com>
1024 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
1027 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1029 PR rtl-optimization/80747
1030 PR rtl-optimization/83512
1031 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
1032 succ edge from ENTRY, copy partition from e->dest to the newly
1034 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
1035 ENTRY, use BB_PARTITION of its successor block as current_partition.
1036 Don't copy partition when splitting succ edge from ENTRY.
1038 PR tree-optimization/83523
1039 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
1040 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
1041 (convert_mult_to_fma): Likewise.
1043 PR tree-optimization/83521
1044 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
1045 gimple_build_assign without code on result of
1046 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
1047 a VIEW_CONVERT_EXPR.
1049 2017-12-21 Andrew Pinski <apinski@cavium.com>
1050 Steve Ellcey <sellcey@cavium.com>
1052 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
1053 multi-arch for ilp32.
1055 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
1058 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1059 constraints to operand 2.
1060 (*ashl<mode>3_mask_1): Ditto.
1061 (*<shift_insn><mode>3_mask): Ditto.
1062 (*<shift_insn><mode>3_mask_1): Ditto.
1063 (*<rotate_insn><mode>3_mask): Ditto.
1064 (*<rotate_insn><mode>3_mask_1): Ditto.
1066 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1068 * reorg.c (make_return_insns): Reemit each insn with its own location.
1070 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1072 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1073 generation for cases where splatting a value is not useful.
1074 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1075 across a vec_duplicate and a paradoxical subreg forming a vector
1076 mode to a vec_concat.
1078 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1080 * combine.c (simplify_set): Do not transform subregs to zero_extends
1081 if the destination is not a scalar int mode.
1083 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1086 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1087 shared zero if the input has overflowed.
1089 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1091 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1092 support for Cortex-A55 and Cortex-A75.
1094 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1096 * common/config/arm/arm-common.c (compare_opt_names): Add function
1097 comment. Use strcmp instead of manual loop.
1099 2017-12-21 Martin Liska <mliska@suse.cz>
1101 PR gcov-profile/83509
1102 * gcov-dump.c (dump_gcov_file): Do not read info about
1103 support_unexecuted_blocks for gcda files.
1105 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1107 PR rtl-optimization/82973
1108 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1109 (valid_for_const_vector_p): ... this.
1110 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1111 (valid_for_const_vector_p): ... this. Adjust function comment.
1112 (gen_vec_duplicate): Adjust caller.
1113 * optabs.c (expand_vector_broadcast): Likewise.
1114 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1115 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1117 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1118 instead of GET_CODE == CONST_FIXED.
1119 (simplify_subreg): Use CONST_FIXED_P macro instead of
1120 GET_CODE == CONST_FIXED.
1123 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1124 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1125 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1126 (ix86_option_override_internal): Adjust for
1127 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1128 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1129 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1130 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1131 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1132 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1133 def_builtin2 for CET builtins.
1134 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1135 up formatting in IX86_BUILTIN_RDPID code.
1136 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1137 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1138 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1139 ix86_isa_flags2 to ix86_isa_flags.
1140 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1142 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1143 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1144 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1146 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1147 Or in OPTION_MASK_ISA_AVX512F_SET.
1148 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1149 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1150 (ix86_handle_option): Adjust for
1151 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1152 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
1154 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1155 Alan Hayward <alan.hayward@arm.com>
1156 David Sherwood <david.sherwood@arm.com>
1158 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
1159 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1160 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1161 Track polynomial offsets.
1163 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1164 Alan Hayward <alan.hayward@arm.com>
1165 David Sherwood <david.sherwood@arm.com>
1167 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1168 Treat drb->init as a poly_int. Fail if its misalignment wrt
1169 vector_alignment isn't known.
1171 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1172 Alan Hayward <alan.hayward@arm.com>
1173 David Sherwood <david.sherwood@arm.com>
1175 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1177 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1178 (vect_loop_versioning): Take the loop versioning threshold as a
1180 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1181 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1182 versioning_threshold.
1183 (vect_analyze_loop_2): Compute the loop versioning threshold
1184 whenever loop versioning is needed, and store it in the new
1185 field rather than combining it with the cost model threshold.
1186 (vect_transform_loop): Update call to vect_loop_versioning.
1187 Try to combine the loop versioning and cost thresholds here.
1189 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1190 Alan Hayward <alan.hayward@arm.com>
1191 David Sherwood <david.sherwood@arm.com>
1193 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1194 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1195 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1196 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1197 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1198 (group_compare_offset): Update accordingly.
1199 (split_small_address_groups_p): Likewise.
1200 (record_use): Take addr_offset as a poly_uint64 rather than
1201 an unsigned HOST_WIDE_INT.
1202 (strip_offset): Return the offset as a poly_uint64 rather than
1203 an unsigned HOST_WIDE_INT.
1204 (record_group_use, split_address_groups): Track polynomial offsets.
1205 (add_iv_candidate_for_use): Likewise.
1206 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1207 than a HOST_WIDE_INT.
1208 (strip_offset_1): Return the offset as a poly_int64 rather than
1211 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1212 Alan Hayward <alan.hayward@arm.com>
1213 David Sherwood <david.sherwood@arm.com>
1215 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1216 rather than a HOST_WIDE_INT.
1217 * tree.c (get_binfo_at_offset): Likewise.
1219 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1220 Alan Hayward <alan.hayward@arm.com>
1221 David Sherwood <david.sherwood@arm.com>
1223 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1224 rather than a HOST_WIDE_INT.
1225 * tree-sra.c (build_ref_for_offset): Likewise.
1227 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1228 Alan Hayward <alan.hayward@arm.com>
1229 David Sherwood <david.sherwood@arm.com>
1231 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1233 * tree.c (mem_ref_offset): Likewise.
1234 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1235 * builtins.c (get_object_alignment_2): Likewise.
1236 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1237 * gimple-fold.c (get_base_constructor): Likewise.
1238 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1239 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1241 * ipa-polymorphic-call.c
1242 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1243 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1244 (get_ancestor_addr_info): Likewise.
1245 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1246 * match.pd: Likewise.
1247 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1248 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1249 * tree-eh.c (tree_could_trap_p): Likewise.
1250 * tree-object-size.c (addr_object_size): Likewise.
1251 * tree-ssa-address.c (copy_ref_info): Likewise.
1252 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1253 (indirect_refs_may_alias_p): Likewise.
1254 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1255 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1256 (non_rewritable_mem_ref_base): Likewise.
1257 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1258 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1259 * varasm.c (decode_addr_const): Likewise.
1261 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1262 Alan Hayward <alan.hayward@arm.com>
1263 David Sherwood <david.sherwood@arm.com>
1265 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1266 offsets for MEM_REFs.
1268 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1269 Alan Hayward <alan.hayward@arm.com>
1270 David Sherwood <david.sherwood@arm.com>
1272 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1273 as a poly_uint64 rather than an unsigned int.
1274 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1276 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1277 Alan Hayward <alan.hayward@arm.com>
1278 David Sherwood <david.sherwood@arm.com>
1280 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1282 (decode_addr_const): Update accordingly.
1284 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1285 Alan Hayward <alan.hayward@arm.com>
1286 David Sherwood <david.sherwood@arm.com>
1288 * tree.h (bit_field_size, bit_field_offset): New functions.
1289 * hsa-gen.c (gen_hsa_addr): Use them.
1290 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1291 (simplify_vector_constructor): Likewise.
1292 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1293 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1294 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1295 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1298 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1299 Alan Hayward <alan.hayward@arm.com>
1300 David Sherwood <david.sherwood@arm.com>
1302 * expr.h (emit_group_load, emit_group_load_into_temps)
1303 (emit_group_store): Take the size as a poly_int64 rather than an int.
1304 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1305 (emit_group_load_into_temp, emit_group_store): Likewise.
1307 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1308 Alan Hayward <alan.hayward@arm.com>
1309 David Sherwood <david.sherwood@arm.com>
1311 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1312 an unsigned int to a poly_uint64.
1313 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1314 sizes as poly_uint64s rather than unsigned ints.
1315 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1318 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1319 Alan Hayward <alan.hayward@arm.com>
1320 David Sherwood <david.sherwood@arm.com>
1322 * lra-constraints.c (emit_inc): Change inc_amount from an int
1325 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1326 Alan Hayward <alan.hayward@arm.com>
1327 David Sherwood <david.sherwood@arm.com>
1329 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1331 (add_stack_var, stack_var_cmp, partition_stack_vars)
1332 (dump_stack_var_partition): Update accordingly.
1333 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1334 than a HOST_WIDE_INT.
1335 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1336 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1337 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1338 rather than a HOST_WIDE_INT.
1340 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1341 Alan Hayward <alan.hayward@arm.com>
1342 David Sherwood <david.sherwood@arm.com>
1344 * target.def (return_pops_args): Treat both the input and output
1345 sizes as poly_int64s rather than HOST_WIDE_INTS.
1346 * targhooks.h (default_return_pops_args): Update accordingly.
1347 * targhooks.c (default_return_pops_args): Likewise.
1348 * doc/tm.texi: Regenerate.
1349 * emit-rtl.h (incoming_args): Change pops_args, size and
1350 outgoing_args_size from int to poly_int64_pod.
1351 * function.h (expr_status): Change x_pending_stack_adjust and
1352 x_stack_pointer_delta from int to poly_int64.
1353 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1354 (ARGS_SIZE_RTX): Update accordingly.
1355 * calls.c (highest_outgoing_arg_in_use): Change from int to
1357 (stack_usage_watermark, stored_args_watermark): New variables.
1358 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1359 (emit_call_1): Change the stack_size and rounded_stack_size
1360 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1362 (save_fixed_argument_area): Check stack_usage_watermark.
1363 (initialize_argument_information): Change old_pending_adj from
1364 a HOST_WIDE_INT * to a poly_int64_pod *.
1365 (compute_argument_block_size): Return the size as a poly_int64
1367 (finalize_must_preallocate): Track polynomial argument sizes.
1368 (compute_argument_addresses): Likewise.
1369 (internal_arg_pointer_based_exp): Track polynomial offsets.
1370 (mem_overlaps_already_clobbered_arg_p): Rename to...
1371 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1372 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1373 Check stored_args_used_watermark.
1374 (load_register_parameters): Update accordingly.
1375 (check_sibcall_argument_overlap_1): Likewise.
1376 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1377 args size as a poly_int64 rather than an int. Return a bool
1378 indicating whether the optimization was possible and return
1379 the new adjustment by reference.
1380 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1381 Update stored_args_watermark.
1382 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1383 (expand_call): Likewise. Maintain stack_usage_watermark and
1384 stored_args_watermark. Update calls to
1385 combine_pending_stack_adjustment_and_call.
1386 (emit_library_call_value_1): Handle polynomial argument sizes.
1387 Call stack_region_maybe_used_p and mark_stack_region_used.
1388 Maintain stack_usage_watermark.
1389 (store_one_arg): Likewise. Update call to
1390 mem_overlaps_already_clobbered_arg_p.
1391 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1393 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1394 * config/microblaze/microblaze.c (microblaze_function_prologue):
1396 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1397 TARGET_RETURN_POPS_ARGS interface.
1398 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1400 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1401 * config/i386/i386.c (ix86_return_pops_args): Update for new
1402 TARGET_RETURN_POPS_ARGS interface.
1403 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1404 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1405 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1406 TARGET_RETURN_POPS_ARGS interface.
1407 * config/vax/vax.c (vax_return_pops_args): Likewise.
1408 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1409 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1410 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1411 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1412 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1413 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1414 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1415 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1416 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1417 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1418 and x_stack_pointer_delta from int to poly_int64.
1419 * dojump.c (do_pending_stack_adjust): Update accordingly.
1420 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1421 stack_pointer_deltas.
1422 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1423 (pad_to_arg_alignment): Track polynomial offsets.
1424 (assign_parm_find_stack_rtl): Likewise.
1425 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1426 * toplev.c (output_stack_usage): Update reference to
1427 current_function_pushed_stack_size.
1429 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1430 Alan Hayward <alan.hayward@arm.com>
1431 David Sherwood <david.sherwood@arm.com>
1433 * function.c (in_arg_offset, var_offset, dynamic_offset)
1434 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1435 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1436 rather than a HOST_WIDE_INT.
1437 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1438 (instantiate_virtual_regs_in_insn): Likewise.
1440 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1441 Alan Hayward <alan.hayward@arm.com>
1442 David Sherwood <david.sherwood@arm.com>
1444 * rtl.h (get_args_size, add_args_size_note): New functions.
1445 (find_args_size_adjust): Return a poly_int64 rather than a
1447 (fixup_args_size_notes): Likewise. Make the same change to the
1448 end_args_size parameter.
1449 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1450 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1451 * calls.c (emit_call_1): Likewise.
1452 * explow.c (adjust_stack_1): Likewise.
1453 * cfgcleanup.c (old_insns_match_p): Update use of
1454 find_args_size_adjust.
1455 * combine.c (distribute_notes): Track polynomial arg sizes.
1456 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1457 end_true_args_size, beg_delay_args_size and end_delay_args_size
1458 from HOST_WIDE_INT to poly_int64.
1459 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1460 than a HOST_WIDE_INT.
1461 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1462 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1463 polynomial arg sizes.
1464 * emit-rtl.c (try_split): Use get_args_size.
1465 * recog.c (peep2_attempt): Likewise.
1466 * reload1.c (reload_as_needed): Likewise.
1467 * expr.c (find_args_size_adjust): Return the adjustment as a
1468 poly_int64 rather than a HOST_WIDE_INT.
1469 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1470 to a poly_int64 and change the return type in the same way.
1471 (emit_single_push_insn): Track polynomial arg sizes.
1473 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1474 Alan Hayward <alan.hayward@arm.com>
1475 David Sherwood <david.sherwood@arm.com>
1477 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1478 from HOST_WIDE_INT to poly_int64.
1479 * expr.c (push_block, emit_push_insn): Likewise.
1481 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1482 Alan Hayward <alan.hayward@arm.com>
1483 David Sherwood <david.sherwood@arm.com>
1485 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1487 (get_frame_size): Return the size as a poly_int64 rather than a
1489 (frame_offset_overflow): Take the offset as a poly_int64 rather
1490 than a HOST_WIDE_INT.
1491 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1492 (assign_stack_temp): Likewise for the size.
1493 * function.c (get_frame_size): Return a poly_int64 rather than
1495 (frame_offset_overflow): Take the offset as a poly_int64 rather
1496 than a HOST_WIDE_INT.
1497 (try_fit_stack_local): Take the start, length and size as poly_int64s
1498 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1499 rather than a HOST_WIDE_INT.
1500 (add_frame_space): Take the start and end as poly_int64s rather than
1502 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1503 (assign_stack_temp): Likewise for the size.
1504 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1506 (find_temp_slot_from_address): Handle polynomial offsets.
1507 (combine_temp_slots): Likewise.
1508 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1510 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1511 poly_int64 rather than a HOST_WIDE_INT.
1512 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1513 than a HOST_WIDE_INT.
1514 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1515 polynomial frame offsets.
1516 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1517 as a poly_int64 rather than an int.
1518 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1519 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1520 * config/v850/v850.c (compute_frame_size): Likewise.
1521 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1522 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1523 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1524 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1525 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1526 rather than a HOST_WIDE_INT.
1527 * explow.c (get_dynamic_stack_base): Likewise.
1528 * final.c (final_start_function): Use the constant lower bound
1529 of the frame size for -Wframe-larger-than.
1530 * ira.c (do_reload): Adjust for new get_frame_size return type.
1531 * lra.c (lra): Likewise.
1532 * reload1.c (reload): Likewise.
1533 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1534 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1535 * rtlanal.c (get_initial_register_offset): Return the offset as
1536 a poly_int64 rather than a HOST_WIDE_INT.
1538 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1539 Alan Hayward <alan.hayward@arm.com>
1540 David Sherwood <david.sherwood@arm.com>
1542 * reload1.c (elim_table): Change initial_offset, offset and
1543 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1544 (offsets_at): Change the target array's element type from
1545 HOST_WIDE_INT to poly_int64_pod.
1546 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1547 (elimination_costs_in_insn, update_eliminable_offsets)
1548 (verify_initial_elim_offsets, set_offsets_for_label)
1549 (init_eliminable_invariants): Update after above changes.
1551 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1552 Alan Hayward <alan.hayward@arm.com>
1553 David Sherwood <david.sherwood@arm.com>
1555 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1556 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1557 (decomposition): Change start and end from HOST_WIDE_INT
1559 (decompose, immune_p): Update accordingly.
1560 (find_inc_amount): Return a poly_int64 rather than an int.
1561 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1564 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1565 Alan Hayward <alan.hayward@arm.com>
1566 David Sherwood <david.sherwood@arm.com>
1568 * tree.h (get_inner_reference): Return the bitsize and bitpos
1569 as poly_int64_pods rather than HOST_WIDE_INT.
1570 * fold-const.h (ptr_difference_const): Return the pointer difference
1571 as a poly_int64_pod rather than a HOST_WIDE_INT.
1572 * expr.c (get_inner_reference): Return the bitsize and bitpos
1573 as poly_int64_pods rather than HOST_WIDE_INT.
1574 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1576 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1577 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1578 (optimize_bit_field_compare): Update call to get_inner_reference.
1579 (decode_field_reference): Likewise.
1580 (fold_unary_loc): Track polynomial offsets and sizes.
1581 (split_address_to_core_and_offset): Return the bitpos as a
1582 poly_int64_pod rather than a HOST_WIDE_INT.
1583 (ptr_difference_const): Likewise for the pointer difference.
1584 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1585 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1586 * dbxout.c (dbxout_expand_expr): Likewise.
1587 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1588 (loc_list_from_tree_1, fortran_common): Likewise.
1589 * gimple-laddress.c (pass_laddress::execute): Likewise.
1590 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1591 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1592 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1593 * tree-affine.c (tree_to_aff_combination): Likewise.
1594 (get_inner_reference_aff): Likewise.
1595 * tree-data-ref.c (split_constant_offset_1): Likewise.
1596 (dr_analyze_innermost): Likewise.
1597 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1598 * tree-sra.c (ipa_sra_check_caller): Likewise.
1599 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1600 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1601 (instrument_bool_enum_load, instrument_object_size): Likewise.
1602 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1603 to get_inner_reference.
1604 * hsa-gen.c (gen_hsa_addr): Likewise.
1605 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1606 * tsan.c (instrument_expr): Likewise.
1607 * match.pd: Update call to ptr_difference_const.
1609 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1610 Alan Hayward <alan.hayward@arm.com>
1611 David Sherwood <david.sherwood@arm.com>
1613 * fold-const.c (fold_comparison): Track sizes and offsets as
1614 poly_int64s rather than HOST_WIDE_INTs when folding address
1617 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1618 Alan Hayward <alan.hayward@arm.com>
1619 David Sherwood <david.sherwood@arm.com>
1621 * expr.h (get_bit_range): Return the bitstart and bitend as
1622 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1623 as a poly_int64 rather than a HOST_WIDE_INT.
1624 * expr.c (get_bit_range): Likewise.
1625 (expand_assignment): Update call accordingly.
1626 * fold-const.c (optimize_bit_field_compare): Likewise.
1628 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1629 Alan Hayward <alan.hayward@arm.com>
1630 David Sherwood <david.sherwood@arm.com>
1632 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1633 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1634 as an optional final parameter.
1635 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1636 (aarch64_legitimate_address_p): Likewise.
1637 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1638 instead of an rtx code.
1639 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1640 (aarch64_legitimate_address_hook_p): Likewise.
1641 (aarch64_print_ldpstp_address): Likewise.
1642 (aarch64_print_operand_address): Likewise.
1643 (aarch64_address_cost): Likewise.
1644 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1645 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1646 (aarch64_mem_pair_lanes_operand): Likewise.
1648 2017-12-20 Richard Biener <rguenther@suse.de>
1650 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1651 update_stmt_if_modified.
1653 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1655 PR tree-optimization/83491
1656 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1657 before walking uses. Improve coding style and comments.
1659 2017-12-20 Tom de Vries <tom@codesourcery.com>
1661 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1663 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1666 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1668 (pass_ipa_free_fn_summary): Add small_p private data member,
1669 initialize to false in the ctor.
1670 (pass_ipa_free_fn_summary::clone,
1671 pass_ipa_free_fn_summary::set_pass_param,
1672 pass_ipa_free_fn_summary::gate): New methods.
1673 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1674 | TODO_dump_symtab if small_p.
1675 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1676 entry and add another instance of the pass with false parm after
1678 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1679 ipa_free_fn_summary here.
1681 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1683 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1685 2017-12-20 Martin Sebor <msebor@redhat.com>
1688 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1689 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1690 (expand_builtin_strncmp): Same.
1692 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1695 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1696 allowed debug stmts before labels.
1697 (expand_gimple_basic_block): Likewise.
1698 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1699 * gimple-iterator.h (gsi_after_labels): Likewise.
1700 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1701 rename reused variable, and simplify using gsi_move_before.
1702 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1703 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1704 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1705 (gimple_verify_flow_info, gimple_block_label): Likewise.
1706 (make_blocks): Move debug markers after adjacent labels.
1707 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1708 allowed debug insns outside blocks.
1709 * df-scan.c (df_insn_delete): Likewise.
1710 * lra-constraints.c (update_ebb_live_info): Likewise.
1711 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1712 (vt_initialize, delete_vta_debug_insns): Likewise.
1713 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1715 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1716 Alan Hayward <alan.hayward@arm.com>
1717 David Sherwood <david.sherwood@arm.com>
1719 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1720 (round_up_to_byte_boundary): Likewise.
1721 * expr.h (get_bit_range): Add temporary shim.
1722 * gimple-ssa-store-merging.c (store_operand_info): Change the
1723 bitsize, bitpos, bitregion_start and bitregion_end fields from
1724 unsigned HOST_WIDE_INT to poly_uint64.
1725 (merged_store_group): Likewise load_align_base.
1726 (compatible_load_p, compatible_load_p): Update accordingly.
1727 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1728 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1729 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1730 bitregion_start and bitregion_end as poly_uint64s rather than
1731 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1732 (handled_load): Take the bitsize, bitpos,
1733 bitregion_start and bitregion_end as poly_uint64s rather than
1734 unsigned HOST_WIDE_INTs.
1735 (pass_store_merging::process_store): Update call to
1736 mem_valid_for_store_merging.
1738 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1739 Alan Hayward <alan.hayward@arm.com>
1740 David Sherwood <david.sherwood@arm.com>
1742 * builtins.c (get_object_alignment_2): Track polynomial offsets
1743 and sizes. Update the alignment handling.
1745 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1746 Alan Hayward <alan.hayward@arm.com>
1747 David Sherwood <david.sherwood@arm.com>
1749 * tree.h (get_inner_reference): Add a version that returns the
1750 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1751 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1752 the case in which bitpos is not associated with the first byte.
1754 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1755 Alan Hayward <alan.hayward@arm.com>
1756 David Sherwood <david.sherwood@arm.com>
1758 * tree-affine.h (get_inner_reference_aff): Return the size as a
1760 * tree-affine.c (get_inner_reference_aff): Likewise.
1761 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1762 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1764 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1765 Alan Hayward <alan.hayward@arm.com>
1766 David Sherwood <david.sherwood@arm.com>
1768 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1769 HOST_WIDE_INT rather than a poly_int64.
1771 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1772 Alan Hayward <alan.hayward@arm.com>
1773 David Sherwood <david.sherwood@arm.com>
1775 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1776 HOST_WIDE_INT to poly_int64_pod.
1777 (perform_symbolic_merge): Update accordingly.
1778 (bswap_replace): Likewise.
1780 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1781 Alan Hayward <alan.hayward@arm.com>
1782 David Sherwood <david.sherwood@arm.com>
1784 * tree-affine.h (aff_tree::offset): Change from widest_int
1786 (wide_int_ext_for_comb): Delete.
1787 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1788 constants as poly_widest_int rather than widest_int.
1789 (aff_combination_constant_multiple_p): Return the multiplier
1790 as a poly_widest_int.
1791 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1793 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1794 version static and add an overload for poly_widest_int.
1795 (aff_combination_const, aff_combination_add_cst)
1796 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1797 the constants as poly_widest_int rather than widest_int.
1798 (tree_to_aff_combination): Generalize INTEGER_CST case to
1800 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1801 (aff_combination_add_product, aff_combination_mult): Handle
1803 (aff_combination_constant_multiple_p): Return the multiplier
1804 as a poly_widest_int.
1805 * tree-predcom.c (determine_offset): Return the offset as a
1807 (split_data_refs_to_components, suitable_component_p): Update
1809 (valid_initializer_p): Update call to
1810 aff_combination_constant_multiple_p.
1811 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1812 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1813 as a poly_int64 rather than a HOST_WIDE_INT.
1814 (get_address_cost): Handle polynomial offsets.
1815 (iv_elimination_compare_lt): Likewise.
1816 (rewrite_use_nonlinear_expr): Likewise.
1818 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1819 Alan Hayward <alan.hayward@arm.com>
1820 David Sherwood <david.sherwood@arm.com>
1822 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1823 as a poly_int64_pod rather than a HOST_WIDE_INT.
1824 (get_addr_base_and_unit_offset): Likewise.
1825 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1826 (get_addr_base_and_unit_offset): Likewise.
1827 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1828 to poly_int64 in example.
1829 * fold-const.c (fold_binary_loc): Update call to
1830 get_addr_base_and_unit_offset.
1831 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1832 (maybe_canonicalize_mem_ref_addr): Likewise.
1833 (gimple_fold_stmt_to_constant_1): Likewise.
1834 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1836 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1837 * match.pd: Likewise.
1838 * omp-low.c (lower_omp_target): Likewise.
1839 * tree-sra.c (build_ref_for_offset): Likewise.
1840 (build_debug_ref_for_model): Likewise.
1841 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1842 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1843 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1844 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1845 (constant_pointer_difference): Likewise.
1846 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1847 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1848 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1849 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1850 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1851 (set_ssa_val_to): Likewise.
1852 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1853 (maybe_diag_stxncpy_trunc): Likewise.
1854 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1855 * tree.c (build_simple_mem_ref_loc): Likewise.
1856 (array_at_struct_end_p): Likewise.
1858 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1859 Alan Hayward <alan.hayward@arm.com>
1860 David Sherwood <david.sherwood@arm.com>
1862 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1863 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1864 (get_ref_base_and_extent_hwi): Declare.
1865 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1866 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1867 (get_ref_base_and_extent_hwi): New function.
1868 * cfgexpand.c (expand_debug_expr): Update call to
1869 get_ref_base_and_extent.
1870 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1871 * gimple-fold.c (get_base_constructor): Return the offset as a
1872 poly_int64_pod rather than a HOST_WIDE_INT.
1873 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1874 * ipa-polymorphic-call.c
1875 (ipa_polymorphic_call_context::set_by_invariant)
1876 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1877 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1878 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1879 rather than get_ref_base_and_extent.
1880 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1881 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1882 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1884 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1885 call to get_ref_base_and_extent.
1886 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1887 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1888 (stmt_kills_ref_p): Likewise.
1889 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1890 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1892 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1893 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1894 when calling native_encode_expr.
1895 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1896 call to get_ref_base_and_extent.
1897 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1898 get_ref_base_and_extent.
1899 * var-tracking.c (track_expr_p): Likewise.
1901 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1902 Alan Hayward <alan.hayward@arm.com>
1903 David Sherwood <david.sherwood@arm.com>
1905 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1906 HOST_WIDE_INT to poly_int64_pod.
1907 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1908 polynomail parameter offsets.
1910 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1911 Alan Hayward <alan.hayward@arm.com>
1912 David Sherwood <david.sherwood@arm.com>
1914 * gengtype.c (main): Handle poly_int64_pod.
1915 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1916 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1917 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1918 from HOST_WIDE_INT to poly_int64_pod.
1919 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1920 (copy_cfa): New function.
1921 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1922 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1923 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1924 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1925 (initial_return_save): Treat offsets as poly_ints.
1926 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1927 offset is nonconstant.
1928 (reg_save): Take the offset as a poly_int64. Fall back to
1929 DW_CFA_expression for nonconstant offsets.
1930 (queue_reg_save): Take the offset as a poly_int64.
1931 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1933 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1934 Alan Hayward <alan.hayward@arm.com>
1935 David Sherwood <david.sherwood@arm.com>
1937 * rtl.h (operand_subword, operand_subword_force): Take the offset
1938 as a poly_uint64 an unsigned int.
1939 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1941 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1942 Alan Hayward <alan.hayward@arm.com>
1943 David Sherwood <david.sherwood@arm.com>
1945 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1946 'p' format code. Use INT_LIST rather than SUBREG as the example of
1947 a code with an XINT and an XEXP. Remove the implication that
1948 accessing an rtx field using XINT is expected to work.
1949 * rtl.def (SUBREG): Change format from "ei" to "ep".
1950 * rtl.h (rtunion::rt_subreg): New field.
1951 (XCSUBREG): New macro.
1952 (SUBREG_BYTE): Use it.
1953 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1954 Update constructor accordingly.
1955 (subreg_shape::operator ==): Update accordingly.
1956 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1957 than an unsigned int.
1958 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1959 a poly_uint64 rather than an unsigned int.
1960 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1961 than an unsigned int.
1962 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1963 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1964 an unsigned int. Take the sizes as poly_uint64s.
1965 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1966 an unsigned int. Take the shift as a poly_uint64 rather than
1968 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1969 as a poly_uint64 rather than an unsigned int.
1970 (simplify_subreg_regno): Likewise.
1971 (byte_lowpart_offset): Return the memory offset as a poly_int64
1973 (subreg_memory_offset): Likewise. Take the subreg offset as a
1974 poly_uint64 rather than an unsigned int.
1975 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1976 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1977 poly_uint64 rather than an unsigned int.
1978 * rtl.c (rtx_format): Describe 'p' in comment.
1979 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1980 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1981 offset as a poly_uint64 rather than an unsigned int.
1982 (byte_lowpart_offset): Return the memory offset as a poly_int64
1984 (subreg_memory_offset): Likewise. Take the subreg offset as a
1985 poly_uint64 rather than an unsigned int.
1986 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1987 mode sizes as poly_uint64s rather than unsigned ints. Return a
1988 poly_uint64 rather than an unsigned int.
1989 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1990 (copy_insn_1): Handle 'p'.
1991 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1992 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1993 an unsigned int. Return the shift in the same way.
1994 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1996 (subreg_size_offset_from_lsb): Take the sizes and shift as
1997 poly_uint64s rather than unsigned ints. Return the offset as
1999 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
2000 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
2002 * rtlhash.c (add_rtx): Handle 'p'.
2003 * genemit.c (gen_exp): Likewise.
2004 * gengenrtl.c (type_from_format, gendef): Likewise.
2005 * gensupport.c (subst_pattern_match, get_alternatives_number)
2006 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
2007 (subst_dup): Likewise.
2008 * gengtype.c (adjust_field_rtx_def): Likewise.
2009 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
2010 (match_pattern_2): Likewise.
2011 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
2012 (rtx_test::subreg_field): New function.
2013 (operator ==, safe_to_hoist_p, transition_parameter_type)
2014 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
2015 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
2016 * genpeep.c (match_rtx): Likewise.
2017 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
2018 (rtx_writer::print_rtx_operand): Handle 'p'.
2019 (print_value): Handle SUBREG.
2020 * read-rtl.c (apply_int_iterator): Likewise.
2021 (rtx_reader::read_rtx_operand): Handle 'p'.
2022 * alias.c (rtx_equal_for_memref_p): Likewise.
2023 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
2024 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
2026 * calls.c (expand_call): Likewise.
2027 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
2028 (make_extraction, gen_lowpart_for_combine): Likewise.
2029 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
2031 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
2032 rather than an unsigned int. Treat subreg offsets as poly_ints.
2033 (exp_equiv_p): Handle 'p'.
2034 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
2035 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
2036 * dse.c (find_shift_sequence): Likewise.
2037 * dwarf2out.c (rtl_for_decl_location): Likewise.
2038 * expmed.c (extract_low_bits): Likewise.
2039 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
2040 (expand_expr_real_2): Likewise.
2041 * final.c (alter_subreg): Likewise.
2042 (leaf_renumber_regs_insn): Handle 'p'.
2043 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
2044 Treat subreg offsets as poly_ints.
2045 * fwprop.c (forward_propagate_and_simplify): Likewise.
2046 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
2047 * ira.c (get_subreg_tracking_sizes): Likewise.
2048 * ira-conflicts.c (go_through_subreg): Likewise.
2049 * ira-lives.c (process_single_reg_class_operands): Likewise.
2050 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
2051 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
2052 as a poly_uint64 rather than an unsigned int.
2053 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
2054 subreg offsets as poly_ints.
2055 * lra-constraints.c (operands_match_p): Handle 'p'.
2056 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
2057 * lra-spills.c (assign_mem_slot): Likewise.
2058 * postreload.c (move2add_valid_value_p): Likewise.
2059 * recog.c (general_operand, indirect_operand): Likewise.
2060 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2061 (copyprop_hardreg_forward_1): Likewise.
2062 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2063 (record_subregs_of_mode): Likewise.
2064 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2065 * reload.c (operands_match_p): Handle 'p'.
2066 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2067 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2068 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2069 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2070 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2071 (simplify_const_poly_int_tests<N>::run): Likewise.
2072 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2073 a poly_uint64 rather than an unsigned int.
2074 * valtrack.c (debug_lowpart_subreg): Likewise.
2075 * var-tracking.c (var_lowpart): Likewise.
2076 (loc_cmp): Handle 'p'.
2078 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2079 Alan Hayward <alan.hayward@arm.com>
2080 David Sherwood <david.sherwood@arm.com>
2082 * ira.c (get_subreg_tracking_sizes): New function.
2083 (init_live_subregs): Take an integer size rather than a register.
2084 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2085 to init_live_subregs.
2087 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2088 Alan Hayward <alan.hayward@arm.com>
2089 David Sherwood <david.sherwood@arm.com>
2091 * expr.c (store_constructor_field): Change bitsize from a
2092 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2093 HOST_WIDE_INT to a poly_int64.
2094 (store_constructor): Change size from a HOST_WIDE_INT to
2096 (store_field): Likewise bitsize and bitpos.
2098 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2099 Alan Hayward <alan.hayward@arm.com>
2100 David Sherwood <david.sherwood@arm.com>
2102 * expmed.h (store_bit_field): Change bitregion_start and
2103 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2104 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2105 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2106 (store_fixed_bit_field, store_split_bit_field): Likewise.
2107 * expr.c (store_constructor_field, store_field): Likewise.
2108 (optimize_bitfield_assignment_op): Likewise. Make the same change
2109 to bitsize and bitpos.
2110 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2111 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2112 same change in the constructor arguments.
2113 (get_best_mode): Change bitregion_start and bitregion_end from
2114 unsigned HOST_WIDE_INT to poly_uint64.
2115 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2116 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2118 (bit_field_mode_iterator::next_mode): Update for new types
2119 of m_bitregion_start and m_bitregion_end.
2120 (get_best_mode): Change bitregion_start and bitregion_end from
2121 unsigned HOST_WIDE_INT to poly_uint64.
2123 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2124 Alan Hayward <alan.hayward@arm.com>
2125 David Sherwood <david.sherwood@arm.com>
2127 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2128 accepts poly_uint64 offsets.
2129 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2130 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2131 * expmed.c (lowpart_bit_field_p): Likewise.
2132 (extract_bit_field_as_subreg): New function, split out from...
2133 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2134 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2135 extractions, check that BITSIZE matches the size of the extracted
2136 value and that BITNUM is an exact multiple of that size.
2137 If all else fails, try forcing the value into memory if
2138 BITNUM is variable, and adjusting the address so that the
2139 offset is constant. Split the part that can only handle constant
2140 bitsize and bitnum out into...
2141 (extract_integral_bit_field): ...this new function.
2142 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2143 rather than unsigned HOST_WIDE_INTs.
2145 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2146 Alan Hayward <alan.hayward@arm.com>
2147 David Sherwood <david.sherwood@arm.com>
2149 * expmed.h (store_bit_field): Take bitsize and bitnum as
2150 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2151 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2152 that returns the byte size.
2153 (store_bit_field_1): Take bitsize and bitnum as
2154 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2155 to simple_mem_bitfield_p. Split the part that can only handle
2156 constant bitsize and bitnum out into...
2157 (store_integral_bit_field): ...this new function.
2158 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2159 than unsigned HOST_WIDE_INTs.
2160 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2162 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2163 Alan Hayward <alan.hayward@arm.com>
2164 David Sherwood <david.sherwood@arm.com>
2166 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2167 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2169 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2170 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2171 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2172 offset as a poly_int64 rather than an int.
2173 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2174 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2175 * lra-constraints.c (equiv_address_substitution): Track offsets
2177 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2178 (curr_insn_transform): Handle the new form of sp_offset.
2179 * lra-eliminations.c (lra_elim_table): Change previous_offset
2180 and offset from HOST_WIDE_INT to poly_int64.
2181 (print_elim_table, update_reg_eliminate): Update accordingly.
2182 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2183 (get_elimination): Update accordingly.
2184 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2185 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2186 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2187 poly_int64 offsets generally.
2188 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2189 (mark_not_eliminable, init_elimination): Update accordingly.
2190 (remove_reg_equal_offset_note): Return a bool and pass the new
2191 offset back by pointer as a poly_int64.
2192 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2193 rather than a HOST_WIDE_INT.
2194 (do_remat): Track offsets poly_int64s.
2195 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2197 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2198 Alan Hayward <alan.hayward@arm.com>
2199 David Sherwood <david.sherwood@arm.com>
2201 * rtl.h (mem_attrs): Add a default constructor. Change size and
2202 offset from HOST_WIDE_INT to poly_int64.
2203 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2204 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2205 (widen_memory_access): Take the sizes and offsets as poly_int64s
2206 rather than HOST_WIDE_INTs.
2207 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2208 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2210 (adjust_offset_for_component_ref): Change the offset from a
2211 HOST_WIDE_INT to a poly_int64.
2212 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2213 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2214 * dce.c (find_call_stack_args): Likewise.
2215 * dse.c (record_store): Likewise.
2216 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2217 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2218 * read-rtl-function.c (test_loading_mem): Likewise.
2219 * rtlanal.c (may_trap_p_1): Likewise.
2220 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2221 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2222 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2223 (mem_attrs::mem_attrs): New function.
2224 (set_mem_attributes_minus_bitpos): Change bitpos from a
2225 HOST_WIDE_INT to poly_int64.
2226 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2227 (clear_mem_offset, clear_mem_size, change_address)
2228 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2229 initialize mem_attrs.
2230 (set_mem_offset, set_mem_size, adjust_address_1)
2231 (adjust_automodify_address_1, offset_address, widen_memory_access):
2232 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2234 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2235 Alan Hayward <alan.hayward@arm.com>
2236 David Sherwood <david.sherwood@arm.com>
2238 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2239 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2240 rather than 0 to represent an unknown size. Assert that the size
2241 is known when the mode isn't BLKmode.
2242 (may_trap_p_1): Use -1 for unknown sizes.
2243 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2245 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2246 Alan Hayward <alan.hayward@arm.com>
2247 David Sherwood <david.sherwood@arm.com>
2249 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2250 to poly_int64. Update commentary for positions_needed.large.
2251 (read_info_type): Change offset and width from HOST_WIDE_INT
2253 (set_usage_bits): Likewise.
2254 (canon_address): Return the offset as a poly_int64 rather than
2255 a HOST_WIDE_INT. Use strip_offset_and_add.
2256 (set_all_positions_unneeded, any_positions_needed_p): Use
2257 positions_needed.large to track stores with non-constant widths.
2258 (all_positions_needed_p): Likewise. Take the offset and width
2259 as poly_int64s rather than ints. Assert that rhs is nonnull.
2260 (record_store): Cope with non-constant offsets and widths.
2261 Nullify the rhs of an earlier store if we can't tell which bytes
2263 (find_shift_sequence): Take the access_size and shift as poly_int64s
2265 (get_stored_val): Take the read_offset and read_width as poly_int64s
2266 rather than HOST_WIDE_INTs.
2267 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2268 non-constant offsets and widths.
2270 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2271 Alan Hayward <alan.hayward@arm.com>
2272 David Sherwood <david.sherwood@arm.com>
2274 * inchash.h (inchash::hash::add_poly_int): New function.
2275 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2276 Use poly_int64 rather than HOST_WIDE_INT.
2277 (ao_ref::max_size_known_p): New function.
2278 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2279 rather than HOST_WIDE_INT.
2280 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2281 to temporaries until its interface is adjusted to match.
2282 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2283 (aliasing_component_refs_p, decl_refs_may_alias_p)
2284 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2285 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2286 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2288 * alias.c (ao_ref_from_mem): Likewise.
2289 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2290 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2291 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2292 (maybe_trim_complex_store, maybe_trim_constructor_store)
2293 (live_bytes_read, dse_classify_store): Likewise.
2294 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2295 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2296 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2297 (vn_reference_lookup_3): Likewise.
2298 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2300 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2301 Alan Hayward <alan.hayward@arm.com>
2302 David Sherwood <david.sherwood@arm.com>
2304 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2305 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2306 instead of ranges_overlap_p.
2308 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2309 Alan Hayward <alan.hayward@arm.com>
2310 David Sherwood <david.sherwood@arm.com>
2312 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2313 sizes as poly_int64s rather than HOST_WIDE_INTs.
2315 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2316 Alan Hayward <alan.hayward@arm.com>
2317 David Sherwood <david.sherwood@arm.com>
2319 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2320 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2321 * gimple-fold.c (fold_ctor_reference): Likewise.
2323 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2324 Alan Hayward <alan.hayward@arm.com>
2325 David Sherwood <david.sherwood@arm.com>
2327 * target.def (dwarf_poly_indeterminate_value): New hook.
2328 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2329 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2330 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2331 * doc/tm.texi: Regenerate.
2332 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2333 offset as a poly_int64.
2334 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2335 offset as a poly_int64.
2336 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2337 Take the offset as a poly_int64.
2338 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2339 (frame_pointer_fb_offset): Change to a poly_int64.
2340 (int_loc_descriptor): Take the offset as a poly_int64. Use
2341 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2342 (based_loc_descr): Take the offset as a poly_int64.
2343 Use strip_offset_and_add to handle (plus X (const)).
2344 Use new_reg_loc_descr instead of an open-coded version of the
2345 previous implementation.
2346 (mem_loc_descriptor): Handle CONST_POLY_INT.
2347 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2348 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2350 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2351 Alan Hayward <alan.hayward@arm.com>
2352 David Sherwood <david.sherwood@arm.com>
2354 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2355 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2356 * inchash.h (inchash::hash::add_poly_hwi): New function.
2357 * gengtype.c (main): Register poly_int64.
2358 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2359 offset as a poly_int.
2360 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2361 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2362 offset as a poly_int64.
2363 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2364 * print-rtl.c (print_poly_int): New function.
2365 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2367 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2369 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2370 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2371 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2372 Enforce track_offset_p for parts of a PARALLEL.
2373 (vt_add_function_parameter): Use const_offset for the final
2374 offset to track. Use get_tracked_reg_offset for the parts
2377 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2378 Alan Hayward <alan.hayward@arm.com>
2379 David Sherwood <david.sherwood@arm.com>
2381 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2382 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2383 * doc/tm.texi: Regenerate.
2384 * hooks.h (hook_bool_uint_uint_true): Delete.
2385 (hook_bool_puint64_puint64_true): Declare.
2386 * hooks.c (hook_bool_uint_uint_true): Delete.
2387 (hook_bool_puint64_puint64_true): New function.
2388 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2389 instead of unsigned ints.
2390 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2391 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2393 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2394 Alan Hayward <alan.hayward@arm.com>
2395 David Sherwood <david.sherwood@arm.com>
2397 * optabs.h (expand_operand): Add an int_value field.
2398 (create_expand_operand): Add an int_value parameter and use it
2399 to initialize the new expand_operand field.
2400 (create_integer_operand): Replace with a declaration of a function
2401 that accepts poly_int64s. Move the implementation to...
2402 * optabs.c (create_integer_operand): ...here.
2403 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2404 the mode preserves the value of int_value, instead of calling
2405 const_int_operand on the rtx. Use gen_int_mode to generate
2408 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2409 Alan Hayward <alan.hayward@arm.com>
2410 David Sherwood <david.sherwood@arm.com>
2412 * dumpfile.h (dump_dec): Declare.
2413 * dumpfile.c (dump_dec): New function.
2414 * pretty-print.h (pp_wide_integer): Turn into a function and
2415 declare a poly_int version.
2416 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2418 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2419 Alan Hayward <alan.hayward@arm.com>
2420 David Sherwood <david.sherwood@arm.com>
2422 * doc/generic.texi (POLY_INT_CST): Document.
2423 * tree.def (POLY_INT_CST): New tree code.
2424 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2425 * tree-core.h (tree_poly_int_cst): New struct.
2426 (tree_node): Add a poly_int_cst field.
2427 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2428 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2429 instead of a wide_int_ref.
2430 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2432 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2433 instead of an unsigned HOST_WIDE_INT.
2434 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2435 (ptrdiff_tree_p): Declare.
2436 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2437 extern inline implementations if the target doesn't use POLY_INT_CST.
2438 (poly_int_tree_p): New function.
2439 (wi::unextended_tree): New class.
2440 (wi::int_traits <unextended_tree>): New override.
2441 (wi::extended_tree): Add a default constructor.
2442 (wi::extended_tree::get_tree): New function.
2443 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2444 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2445 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2446 (wi::tree_to_poly_wide_ref): New typedefs.
2447 (wi::ints_for): Provide overloads for extended_tree and
2449 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2450 (wi::to_wide): New functions.
2451 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2452 * tree.c (poly_int_cst_hasher): New struct.
2453 (poly_int_cst_hash_table): New variable.
2454 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2455 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2457 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2458 (init_ttree): Initialize poly_int_cst_hash_table.
2459 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2460 a poly_int64 instead of a HOST_WIDE_INT.
2461 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2462 instead of an unsigned HOST_WIDE_INT.
2463 (wide_int_to_tree): Rename to...
2464 (wide_int_to_tree_1): ...this.
2465 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2466 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2467 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2468 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2469 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2470 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2472 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2473 (streamer_read_tree_body): Likewise.
2474 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2475 (streamer_write_tree_body): Likewise.
2476 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2477 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2478 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2479 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2480 * gimple-expr.h (is_gimple_constant): Likewise.
2481 * gimplify.c (maybe_with_size_expr): Likewise.
2482 * print-tree.c (print_node): Likewise.
2483 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2484 * tree-pretty-print.c (dump_generic_node): Likewise.
2485 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2486 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2487 * tree-vrp.c (compare_values_warnv): Likewise.
2488 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2489 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2490 (force_expr_to_var_cost): Likewise.
2491 * tree-ssa-loop.c (for_each_index): Likewise.
2492 * fold-const.h (build_invariant_address, size_int_kind): Take a
2493 poly_int64 instead of a HOST_WIDE_INT.
2494 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2495 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2497 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2498 (int_const_binop_2): New function, split out from...
2499 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2500 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2501 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2502 * gimple-ssa-strength-reduction.c (slsr_process_add)
2503 (slsr_process_mul): Check for INTEGER_CSTs before using them
2505 * stor-layout.c (bits_from_bytes): New function.
2506 (bit_from_pos): Use it.
2507 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2508 by BITS_PER_UNIT to get the TYPE_SIZE.
2509 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2510 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2512 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2513 Alan Hayward <alan.hayward@arm.com>
2514 David Sherwood <david.sherwood@arm.com>
2516 * doc/rtl.texi (const_poly_int): Document. Also document the
2517 rtl sharing behavior.
2518 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2519 * rtl.h (const_poly_int_def): New struct.
2520 (rtx_def::u): Add a cpi field.
2521 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2522 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2523 (wi::rtx_to_poly_wide_ref): New typedef
2524 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2525 (poly_int_rtx_p): New functions.
2526 (trunc_int_for_mode): Declare a poly_int64 version.
2527 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2528 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2530 (strip_offset): Declare.
2531 (strip_offset_and_add): New function.
2532 * rtl.def (CONST_POLY_INT): New rtx code.
2533 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2534 (shared_const_p): Use poly_int_rtx_p.
2535 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2537 (gen_int_shift_amount): Likewise.
2538 * emit-rtl.c (const_poly_int_hasher): New class.
2539 (const_poly_int_htab): New variable.
2540 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2541 (const_poly_int_hasher::hash): New function.
2542 (const_poly_int_hasher::equal): Likewise.
2543 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2544 (immed_wide_int_const): Rename to...
2545 (immed_wide_int_const_1): ...this and make static.
2546 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2547 instead of a wide_int_ref.
2548 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2549 (gen_lowpart_common): Handle CONST_POLY_INT.
2550 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2551 * cselib.c (cselib_hash_rtx): Likewise.
2552 * dwarf2out.c (const_ok_for_output_1): Likewise.
2553 * expr.c (convert_modes): Likewise.
2554 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2555 * rtlhash.c (add_rtx): Likewise.
2556 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2557 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2558 Handle existing CONST_POLY_INT rtxes.
2559 * expmed.h (expand_shift): Take a poly_int64 instead of a
2561 * expmed.c (expand_shift): Likewise.
2562 * rtlanal.c (strip_offset): New function.
2563 (commutative_operand_precedence): Give CONST_POLY_INT the same
2564 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2566 * rtl-tests.c (const_poly_int_tests): New struct.
2567 (rtl_tests_c_tests): Use it.
2568 * simplify-rtx.c (simplify_const_unary_operation): Handle
2570 (simplify_const_binary_operation): Likewise.
2571 (simplify_binary_operation_1): Fold additions of symbolic constants
2572 and CONST_POLY_INTs.
2573 (simplify_subreg): Handle extensions and truncations of
2575 (simplify_const_poly_int_tests): New struct.
2576 (simplify_rtx_c_tests): Use it.
2577 * wide-int.h (storage_ref): Add default constructor.
2578 (wide_int_ref_storage): Likewise.
2579 (trailing_wide_ints): Use GTY((user)).
2580 (trailing_wide_ints::operator[]): Add a const version.
2581 (trailing_wide_ints::get_precision): New function.
2582 (trailing_wide_ints::extra_size): Likewise.
2584 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2585 Alan Hayward <alan.hayward@arm.com>
2586 David Sherwood <david.sherwood@arm.com>
2588 * emit-rtl.h (gen_int_shift_amount): Declare.
2589 * emit-rtl.c (gen_int_shift_amount): New function.
2590 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2592 * calls.c (shift_return_value): Likewise.
2593 * cse.c (fold_rtx): Likewise.
2594 * dse.c (find_shift_sequence): Likewise.
2595 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2596 (expand_shift, expand_smod_pow2): Likewise.
2597 * lower-subreg.c (shift_cost): Likewise.
2598 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2599 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2600 (expand_vec_perm_var): Likewise.
2601 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2602 (simplify_binary_operation_1): Likewise.
2603 * combine.c (try_combine, find_split_point, force_int_to_mode)
2604 (simplify_shift_const_1, simplify_shift_const): Likewise.
2605 (change_zero_ext): Likewise. Use simplify_gen_binary.
2607 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2609 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2611 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2613 * doc/invoke.texi (ARM Options): Document accepted extension options
2614 for -march=armv8.3-a.
2616 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2619 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2620 or --with-float=softfp, set the default CPU to arm10e.
2622 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2624 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2625 * config/visium/predicates.md (const_shift_operand): Likewise.
2626 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2627 (visium_legitimize_reload_address): Likewise.
2629 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2632 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2634 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2637 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2639 2017-12-20 Julia Koval <julia.koval@intel.com>
2641 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2642 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2643 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2644 * config.gcc: Include vpclmulqdqintrin.h.
2645 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2646 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2648 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2649 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2650 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2651 * config/i386/i386.c (isa2_opts): Add -mcx16.
2652 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2653 (ix86_option_override_internal): Move mcx16 to flags2.
2654 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2655 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2656 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2657 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2658 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2659 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2660 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2661 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2662 * doc/invoke.texi: Add -mvpclmulqdq.
2664 2017-12-20 Tom de Vries <tom@codesourcery.com>
2667 * config/i386/i386.c (ix86_static_chain): Move
2668 DECL_STATIC_CHAIN test ...
2669 * calls.c (rtx_for_static_chain): ... here. New function.
2670 * calls.h (rtx_for_static_chain): Declare.
2671 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2672 instead of targetm.calls.static_chain.
2673 * df-scan.c (df_get_entry_block_def_set): Same.
2675 2017-12-19 Tom de Vries <tom@codesourcery.com>
2677 PR tree-optimization/83493
2678 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2681 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2683 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2684 inform with hardcoded english plural handling.
2686 2017-12-18 Jeff Law <law@redhat.com>
2688 PR tree-optimization/83477
2689 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2690 a non-virtual PHI, always push a new range.
2692 2017-12-19 Martin Sebor <msebor@redhat.com>
2695 * builtins.c (compute_objsize): Handle non-constant offsets.
2697 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2699 PR tree-optimization/83444
2700 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2701 character load case, if get_stridx on MEM_REF's operand doesn't
2702 look usable, retry with get_addr_stridx.
2704 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2707 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2708 (variable_tracking_main_1): Keep markers even when VTA fails.
2711 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2712 even if there are markers before it.
2713 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2715 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2717 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2718 typical order conditions.
2719 * sel-sched.c (extract_new_fences_from): Likewise.
2720 * config/visium/constraints.md (J, K, L): Likewise.
2721 * config/visium/predicates.md (const_shift_operand): Likewise.
2722 * config/visium/visium.c (visium_legitimize_address,
2723 visium_legitimize_reload_address): Likewise.
2724 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2725 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2726 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2727 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2728 SET_DUMP_DETAIL): Likewise.
2729 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2730 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2731 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2732 avr_set_core_architecture, avr_set_current_function,
2733 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2734 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2735 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2736 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2737 avr_map_decompose, avr_fold_builtin): Likewise.
2738 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2739 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2740 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2741 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2742 * config/m32c/m32c.c (m32c_conditional_register_usage,
2743 m32c_address_cost): Likewise.
2744 * config/m32c/predicates.md (shiftcount_operand,
2745 longshiftcount_operand): Likewise.
2746 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2747 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2748 can_use_cdx_ldstw): Likewise.
2749 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2750 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2752 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2753 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2754 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2755 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2756 * config/vax/vax.c (adjacent_operands_p): Likewise.
2757 * config/ft32/constraints.md (L, b, KA): Likewise.
2758 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2760 * cfgexpand.c (expand_stack_alignment): Likewise.
2761 * gcse.c (insert_expr_in_table): Likewise.
2762 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2763 * cgraphunit.c (cgraph_node::expand): Likewise.
2764 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2765 * emit-rtl.c (add_insn): Likewise.
2766 * input.c (dump_location_info): Likewise.
2767 * passes.c (NEXT_PASS): Likewise.
2768 * read-rtl-function.c (parse_note_insn_name,
2769 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2771 * sched-rgn.c (sched_rgn_init): Likewise.
2772 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2773 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2774 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2776 * explow.c (eliminate_constant_term): Likewise.
2777 * final.c (leaf_renumber_regs_insn): Likewise.
2778 * cfgrtl.c (print_rtl_with_bb): Likewise.
2779 * genhooks.c (emit_init_macros): Likewise.
2780 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2781 * tree-data-ref.c (conflict_fn): Likewise.
2782 * selftest.c (assert_streq): Likewise.
2783 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2784 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2785 fold_binary_loc, multiple_of_p): Likewise.
2786 * reload.c (push_reload, find_equiv_reg): Likewise.
2787 * et-forest.c (et_nca, et_below): Likewise.
2788 * dbxout.c (dbxout_symbol_location): Likewise.
2789 * reorg.c (relax_delay_slots): Likewise.
2790 * dojump.c (do_compare_rtx_and_jump): Likewise.
2791 * gengtype-parse.c (type): Likewise.
2792 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2793 simplify_const_relational_operation): Likewise.
2794 * reload1.c (do_output_reload): Likewise.
2795 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2796 * gengtype.c (type_for_name): Likewise.
2797 * gimple-ssa-sprintf.c (format_directive): Likewise.
2799 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2802 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2803 accessing it. Adjust comment.
2805 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2808 * predict.c (zero_one_minusone): New function.
2809 (apply_return_prediction): Avoid return prediction for functions
2810 returning only -1, 0 and 1 values, unless they only return -1 and 0
2813 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2815 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2816 fall-through warning.
2817 (arc_compute_frame_size): Remove unused variables.
2818 (arc_print_operand): Fix fprintif format.
2819 (arc_can_follow_jump): Clean fall-through warning.
2821 2017-12-19 Marek Polacek <polacek@redhat.com>
2824 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2827 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2829 * config/arc/arc.c (overriderregs): New variable.
2830 (arc_override_options): Track fixed/call saved/call options.
2831 (arc_conditional_register_usage): Check against overrideregs
2832 variable whenever we change register properties.
2834 2017-12-19 Nathan Sidwell <nathan@acm.org>
2836 * opts.c (finish_options): Don't prefix dump_base_name if it
2837 already contains directories.
2839 2017-12-19 Martin Liska <mliska@suse.cz>
2841 PR rtl-optimization/82675
2842 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2843 more element in sbitmap.
2845 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2847 * gimplify.c (gimplify_expr): Use error_operand_p.
2849 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2852 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2855 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2857 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2858 the jump is to a label.
2860 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2862 PR tree-optimization/83444
2863 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2868 * ipa-inline.c (flatten_remove_node_hook): New function.
2869 (ipa_inline): Keep only nodes with flatten attribute at the end of
2870 the array in the order from ipa_reverse_postorder, only walk that
2871 portion of array for flattening, if there is more than one such
2872 node, temporarily register a removal hook and ignore removed nodes.
2874 PR tree-optimization/80631
2875 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2876 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2877 IFN_REDUC_MAX or IFN_REDUC_MIN.
2879 2017-12-18 Jeff Law <law@redhat.com>
2881 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2882 record symbolic equivalences from backedges in the CFG.
2885 2017-11-19 Jeff Law <law@redhat.com>
2887 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2888 of degenerates resulting from ignoring an edge.
2890 2017-12-18 Martin Sebor <msebor@redhat.com>
2893 PR tree-optimization/78450
2894 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2895 (handle_builtin_strlen): Call it.
2897 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2899 PR rtl-optimization/83424
2900 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2902 2017-12-18 Marek Polacek <polacek@redhat.com>
2905 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2906 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2909 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2911 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2912 (umaddsidi4, umaddsidi_split): Likewise.
2914 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2916 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2919 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2922 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2924 2017-12-18 Richard Biener <rguenther@suse.de>
2926 PR tree-optimization/81877
2927 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2928 (outermost_indep_loop): Adjust.
2929 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2930 (can_sm_ref_p): Adjust.
2932 2017-12-18 Richard Biener <rguenther@suse.de>
2935 * tree.c (array_at_struct_end_p): Return true if the underlying
2936 object has space for at least one element in excess of what
2937 the array domain specifies.
2939 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2941 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2942 example to avoid overfull hbox.
2943 * doc/invoke.texi (Option Summary): Add missing @gol.
2944 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2947 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2948 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2950 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2951 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2952 (AArch64 Options): Document -mverbose-cost-dump.
2953 (ARM Options): Likewise, plus -mflip-thumb.
2955 2017-12-17 Martin Sebor <msebor@redhat.com>
2958 * gimple-ssa-warn-restrict.c
2959 (builtin_memref::offset_out_of_bounds): Correct the handling of
2962 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2964 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2966 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2968 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2970 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2972 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2975 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2977 * doc/invoke.texi: Fix some typos.
2979 2017-12-16 Martin Sebor <msebor@redhat.com>
2981 PR tree-optimization/78918
2982 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2983 * builtins.c (check_sizes): Rename...
2984 (check_access): ...to this. Rename function arguments for clarity.
2985 (check_memop_sizes): Adjust names.
2986 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2987 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2988 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2989 (check_strncat_sizes, expand_builtin_strncat): Same.
2990 (expand_builtin_strncpy, expand_builtin_memset): Same.
2991 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2992 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2993 (maybe_emit_sprintf_chk_warning): Same.
2994 (expand_builtin_strcpy): Adjust.
2995 (expand_builtin_stpcpy): Same.
2996 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2997 in pointer-checking forms of memcpy, memmove, and mempcpy.
2998 (gcall_to_tree_minimal, max_object_size): Define new functions.
2999 * builtins.h (max_object_size): Declare.
3000 * calls.c (alloc_max_size): Call max_object_size instead of
3001 hardcoding ssizetype limit.
3002 (get_size_range): Handle new argument.
3003 * calls.h (get_size_range): Add a new argument.
3004 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
3005 * doc/invoke.texi (-Wrestrict): Adjust, add example.
3006 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
3008 (gimple_fold_builtin_memory_chk): Same.
3009 (gimple_fold_builtin_stxcpy_chk): New function.
3010 * gimple-ssa-warn-restrict.c: New source.
3011 * gimple-ssa-warn-restrict.h: New header.
3012 * gimple.c (gimple_build_call_from_tree): Propagate location.
3013 * passes.def (pass_warn_restrict): Add new pass.
3014 * tree-pass.h (make_pass_warn_restrict): Declare.
3015 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
3017 (handle_builtin_strcat): Same.
3018 (strlen_optimize_stmt): Rename...
3019 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
3020 stpncpy, strncpy, and their checking forms.
3022 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
3024 PR rtl-optimization/82849
3025 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
3026 and get_max_loop_iterations_int.
3028 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3029 Alan Hayward <alan.hayward@arm.com>
3030 David Sherwood <david.sherwood@arm.com>
3032 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
3033 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
3034 size as a poly_uint64.
3035 (mode_for_vector, mode_for_int_vector): Take the number of vector
3036 elements as a poly_uint64.
3037 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
3038 size as a poly_uint64.
3039 (mode_for_vector, mode_for_int_vector): Take the number of vector
3040 elements as a poly_uint64.
3042 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3043 Alan Hayward <alan.hayward@arm.com>
3044 David Sherwood <david.sherwood@arm.com>
3046 * machmode.h (MACRO_MODE): New macro.
3047 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
3048 * targhooks.c (default_libcall_value, default_secondary_reload)
3049 (default_memory_move_cost, default_register_move_cost)
3050 (default_class_max_nregs): Likewise.
3052 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3053 Alan Hayward <alan.hayward@arm.com>
3054 David Sherwood <david.sherwood@arm.com>
3056 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
3057 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3059 * genautomata.c (main): Likewise.
3060 * genconditions.c (write_header): Likewise.
3061 * genemit.c (main): Likewise.
3062 * genextract.c (print_header): Likewise.
3063 * genopinit.c (main): Likewise.
3064 * genoutput.c (output_prologue): Likewise.
3065 * genpeep.c (main): Likewise.
3066 * genpreds.c (write_insn_preds_c): Likewise.
3067 * genrecog.c (writer_header): Likewise.
3068 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3069 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3070 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3071 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3072 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3073 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3074 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3075 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3076 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3077 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3078 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3079 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3080 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3081 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3082 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3083 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3084 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3085 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3086 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3087 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3088 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3089 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3090 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3091 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3092 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3093 * config/darwin.c (IN_TARGET_CODE): Likewise.
3094 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3095 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3096 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3097 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3098 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3099 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3100 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3101 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3102 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3103 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3104 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3105 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3106 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3107 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3108 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3109 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3110 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3111 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3112 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3113 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3114 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3115 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3116 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3117 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3118 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3119 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3120 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3121 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3122 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3123 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3124 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3125 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3126 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3127 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3128 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3129 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3130 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3131 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3132 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3133 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3134 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3135 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3136 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3137 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3138 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3139 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3140 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3141 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3142 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3143 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3144 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3145 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3146 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3147 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3148 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3149 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3150 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3151 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3152 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3153 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3154 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3155 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3156 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3157 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3158 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3159 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3160 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3161 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3162 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3163 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3164 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3165 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3166 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3167 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3168 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3169 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3170 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3171 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3172 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3173 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3174 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3175 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3176 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3177 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3178 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3179 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3180 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3181 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3182 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3183 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3184 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3185 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3186 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3187 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3188 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3189 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3190 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3191 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3192 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3193 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3194 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3195 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3196 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3197 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3198 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3199 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3201 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3202 Alan Hayward <alan.hayward@arm.com>
3203 David Sherwood <david.sherwood@arm.com>
3205 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3208 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3209 Alan Hayward <alan.hayward@arm.com>
3210 David Sherwood <david.sherwood@arm.com>
3212 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3213 * doc/md.texi (vec_series@var{m}): Document.
3214 * tree.def (VEC_SERIES_EXPR): New tree code.
3215 * tree.h (build_vec_series): Declare.
3216 * tree.c (build_vec_series): New function.
3217 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3218 * tree-pretty-print.c (dump_generic_node): Likewise.
3219 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3220 * tree-inline.c (estimate_operator_cost): Likewise.
3221 * expr.c (expand_expr_real_2): Likewise.
3222 * optabs-tree.c (optab_for_tree_code): Likewise.
3223 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3224 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3225 * expmed.c (make_tree): Handle VEC_SERIES.
3226 * optabs.def (vec_series_optab): New optab.
3227 * optabs.h (expand_vec_series_expr): Declare.
3228 * optabs.c (expand_vec_series_expr): New function.
3229 * tree-vect-generic.c (expand_vector_operations_1): Check that
3230 the operands also have vector type.
3232 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3233 Alan Hayward <alan.hawyard@arm.com>
3234 David Sherwood <david.sherwood@arm.com>
3236 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3237 (VEC_COND_EXPR): Add missing @tindex.
3238 * doc/md.texi (vec_duplicate@var{m}): Document.
3239 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3240 * tree.c (build_vector_from_val): Add stubbed-out handling of
3241 variable-length vectors, using VEC_DUPLICATE_EXPR.
3242 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3243 * cfgexpand.c (expand_debug_expr): Likewise.
3244 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3245 * tree-inline.c (estimate_operator_cost): Likewise.
3246 * tree-pretty-print.c (dump_generic_node): Likewise.
3247 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3248 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3249 (test_vec_duplicate_folding): New function.
3250 (fold_const_c_tests): Call it.
3251 * optabs.def (vec_duplicate_optab): New optab.
3252 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3253 * optabs.h (expand_vector_broadcast): Declare.
3254 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3255 vec_duplicate_optab.
3256 * expr.c (store_constructor): Try using vec_duplicate_optab for
3258 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3260 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3263 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3264 div/mod latencies a bit.
3266 2017-12-15 Jeff Law <law@redhat.com>
3268 PR tree-optimization/36550
3269 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3270 (mark_threaded_blocks): Rewrite code to avoid block copying when
3271 optimizing for size. Don't pessimize blocks which will be
3272 copied, but all the statements will be dead.
3274 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3276 PR tree-optimization/81165
3277 * tree-ssa-threadupdate.c (uses_in_bb): New.
3278 (estimate_threading_killed_stmts): New.
3279 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3280 * tree-ssa-threadedge.c
3281 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3284 2017-12-15 Jeff Law <law@redhat.com>
3286 PR tree-optimization/83410
3287 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3288 threads when parallelizing loops.
3290 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3292 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3294 * config/alpha/alpha.c (vms_attribute_table): Swap
3295 affects_type_identity and handler fields, adjust comments.
3296 * config/mips/mips.c (mips_attribute_table): Likewise.
3297 * config/visium/visium.c (visium_attribute_table): Likewise.
3298 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3299 * config/microblaze/microblaze.c (microblaze_attribute_table):
3301 * config/spu/spu.c (spu_attribute_table): Likewise.
3302 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3303 * config/arc/arc.c (arc_attribute_table): Likewise.
3304 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3305 * config/v850/v850.c (v850_handle_interrupt_attribute,
3306 v850_handle_data_area_attribute): Formatting fixes.
3307 (v850_attribute_table): Swap affects_type_identity and handler
3308 fields, adjust comments.
3309 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3310 * config/arm/arm.c (arm_attribute_table): Likewise.
3311 * config/avr/avr.c (avr_attribute_table): Likewise.
3312 * config/s390/s390.c (s390_attribute_table): Likewise.
3313 * config/sh/sh.c (sh_attribute_table): Likewise.
3314 * config/i386/i386.c (ix86_handle_cconv_attribute,
3315 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3316 (ix86_attribute_table): Swap affects_type_identity and handler
3317 fields, adjust comments.
3318 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3319 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3320 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3321 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3322 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3323 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3324 * config/rx/rx.c (rx_attribute_table): Likewise.
3325 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3326 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3327 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3328 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3329 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3330 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3331 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3332 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3333 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3334 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3335 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3337 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3339 (register_scoped_attributes, decl_attributes): Formatting fixes.
3341 PR tree-optimization/83269
3342 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3343 subtraction in arg0's type if type is signed and arg0 is unsigned.
3347 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3348 convert? on @0 instead of convert. Check type of @1, not @0.
3349 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3350 convert? on @0 instead of convert. Check type of @1, not @0.
3351 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3352 only optimize if either both @1 and @2 types are narrower
3353 precision, or both are wider or equal precision, and in the former
3354 case only if both have undefined overflow.
3356 2017-12-15 Richard Biener <rguenther@suse.de>
3359 * internal-fn.def (IFN_NOP): Add.
3360 * internal-fn.c (expand_NOP): Do nothing.
3361 * lto-streamer-in.c (input_function): Instead of removing
3362 sanitizer calls replace them with IFN_NOP calls.
3364 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3365 Alan Hayward <alan.hayward@arm.com>
3366 David Sherwood <david.sherwood@arm.com>
3368 * dse.c (store_info, read_info_type): Replace begin and end with
3370 (print_range): New function.
3371 (set_all_positions_unneeded, any_positions_needed_p)
3372 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3374 (record_store): Likewise. Optimize the case in which all positions
3376 (get_stored_val): Replace read_begin and read_end with read_offset
3378 (replace_read): Update call accordingly.
3380 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3382 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3383 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3384 (loop_cand::loop_cand): Initialize above members.
3385 (loop_cand::supported_operations): Delete.
3386 (loop_cand::can_interchange_p): Inline above function.
3387 (loop_cand::classify_simple_reduction): Record number of constant
3388 initialized simple reductions.
3389 (should_interchange_loops): New parameters. Check stmt cost of loops
3391 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3392 Update call to should_interchange_loops.
3393 (should_interchange_loop_nest): Update call to
3394 should_interchange_loops.
3396 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3399 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3400 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3401 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3403 2017-12-15 Julia Koval <julia.koval@intel.com>
3405 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3406 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3408 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3409 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3410 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3412 2017-12-15 Julia Koval <julia.koval@intel.com>
3414 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3415 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3416 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3417 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3418 _mm_aesenc_epi128): New intrinsics.
3420 2017-12-15 Julia Koval <julia.koval@intel.com>
3422 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3423 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3425 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3426 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3427 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3429 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3431 * gimple-ssa-strength-reduction.c (analyze_increments):
3432 Distinguish replacement costs for constant strides from those for
3435 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3437 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3439 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3441 * doc/invoke.texi: Document -Wcast-function-type.
3442 * recog.h (stored_funcptr): Change signature.
3443 * tree-dump.c (dump_node): Avoid warning.
3444 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3446 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3449 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3452 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3454 PR tree-optimization/83312
3455 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3456 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3457 "val" param, and to cope with arbitrary basic blocks.
3458 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3459 handle NULL_TREE for "val", dropping "bb" param.
3460 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3461 drop "bb" param. Handle NULL_TREE for "val".
3462 (find_case_label_for_value): Make "switch_stmt" param const.
3463 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3465 (vrp_prop::check_all_array_refs): Reimplement as...
3466 (check_array_bounds_dom_walker::before_dom_children): ...this new
3467 vfunc. Replace linear search through BB block list, excluding
3468 those with non-executable in-edges via dominator walk.
3470 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3472 * config/arm/arm.opt (mverbose-cost-dump): New option.
3473 * config/arm/arm.c (arm_rtx_costs): Use it.
3475 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3478 * reload1.c (emit_input_reload_insns): Skip debug markers.
3480 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3482 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3486 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3487 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3488 * config/sh/sh.c: Skip debug insns besides notes.
3489 * config/sh/sh.md: Likewise.
3490 * config/sh/sh_treg_combine.cc: Likewise.
3491 * config/sh/sync.md: Likewise.
3493 2017-12-14 Tom de Vries <tom@codesourcery.com>
3495 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3496 Add item for weak_undefined.
3498 2017-12-14 Richard Biener <rguenther@suse.de>
3500 PR tree-optimization/67842
3501 * sese.h (bb_in_region): Remove #if 0'ed code.
3503 2017-12-14 Richard Biener <rguenther@suse.de>
3505 PR tree-optimization/83326
3506 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3507 may_be_zero parameter and handle it by not marking the first
3508 peeled copy as not exiting the loop.
3509 (try_peel_loop): Likewise.
3510 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3511 to handle the case of constant or zero iterations and perform
3512 loop header copying on-the-fly.
3514 2017-12-14 Richard Biener <rguenther@suse.de>
3516 PR tree-optimization/83418
3517 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3518 Instead of asserting we don't get unfolded comparisons deal with them.
3520 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3523 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3524 basic blocks. Assert debug bind insns don't appear outside of bbs,
3525 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3528 PR tree-optimization/83198
3529 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3530 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3531 value if arg is a REAL_CST with incompatible type.
3533 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3534 Bin Cheng <bin.cheng@arm.com>
3537 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3539 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3542 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3544 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3545 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3547 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3548 Alan Hayward <alan.hayward@arm.com>
3549 David Sherwood <david.sherwood@arm.com>
3551 * poly-int.h: New file.
3552 * poly-int-types.h: Likewise.
3553 * coretypes.h: Include them.
3554 (POLY_INT_CONVERSION): Define.
3555 * target.def (estimated_poly_value): New hook.
3556 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3557 * doc/tm.texi: Regenerate.
3558 * doc/poly-int.texi: New file.
3559 * doc/gccint.texi: Include it.
3560 * doc/rtl.texi: Describe restrictions on subreg modes.
3561 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3562 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3563 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3564 * targhooks.h (default_estimated_poly_value): Declare.
3565 * targhooks.c (default_estimated_poly_value): New function.
3566 * target.h (estimated_poly_value): Likewise.
3567 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3568 (wi::unary_traits): Delete.
3569 (wi::binary_traits::signed_shift_result_type): Define for
3570 offset_int << HOST_WIDE_INT, etc.
3571 (generic_wide_int::operator <<=): Define for all types and use
3572 wi::lshift instead of <<.
3573 (wi::hwi_with_prec): Add a default constructor.
3574 (wi::ints_for): New class.
3575 (operator <<): Define for all wide-int types.
3576 (operator /): New function.
3577 (operator %): Likewise.
3578 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3579 (ASSERT_MAYBE_NE_AT): New macros.
3581 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3582 Dominik Vogt <vogt@linux.vnet.ibm.com>
3585 * emit-rtl.c (init_emit): Remove ??? comment.
3586 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3587 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3588 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3589 alignment of 3 virtual registers to BITS_PER_WORD.
3591 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3593 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3595 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3596 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3598 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3601 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3603 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3605 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3606 with the exception of debug begin stmt markers appear before
3610 * final.c (rest_of_handle_final): Call variable_tracking_main only
3611 if !flag_var_tracking.
3613 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3614 Jakub Jelinek <jakub@redhat.com>
3618 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3619 labels debug stmts that can only appear after labels.
3621 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3623 PR rtl-optimization/82398
3624 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3625 EXPR_USEFULNESS in priority comparison.
3627 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3629 PR rtl-optimization/83393
3630 * combine.c (move_deaths): If reg_stat points to a too new insn in
3631 last_death, do not use it: find the proper insn instead.
3633 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3636 * config/rs6000/rs6000-string.c (expand_block_compare,
3637 expand_strn_compare): Fix set_mem_size() calls.
3639 2017-12-12 Jeff Law <law@redhat.com>
3641 PR tree-optimization/83298
3642 PR tree-optimization/83362
3643 PR tree-optimization/83383
3644 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3645 push_value_range a public interface. Add new argument to
3646 record_ranges_from_stmt.
3647 * gimple-ssa-evrp-analyze.c
3648 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3649 Update comments. Handle recording temporary equivalences.
3650 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3651 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3652 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3653 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3654 gimple-ssa-evrp-analyze.h.
3655 (record_temporary_equivalences_from_phis): Add new argument. When
3656 the PHI arg is an SSA_NAME, set the result's range to the range
3658 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3659 from statements too.
3660 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3661 Pass it down to children as needed.
3662 (thread_outgoing_edges): Likewise.
3663 (thread_across_edge): Likewise. Push/pop range state as needed.
3664 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3666 2017-12-12 Julia Koval <julia.koval@intel.com>
3668 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3669 (PTA_CANNONLAKE): Remove PTA_CLWB.
3671 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3674 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3675 (vcondv2div2df): Likewise.
3676 (vconduv2dfv2di): Likewise.
3678 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3680 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3681 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3682 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3683 ATTR_MATHFN_FPROUNDING_ERRNO.
3685 2017-12-12 Richard Biener <rguenther@suse.de>
3687 PR tree-optimization/83385
3688 * graphite-scop-detection.c (get_order, order): Remove.
3689 (bb_to_rpo): New global.
3691 (build_scops): Sort pbbs in RPO order.
3693 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3695 * combine.c (simplify_set): Do not transform subregs to zero_extends
3696 if the destination mode is a vector mode.
3698 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3700 PR tree-optimization/83359
3701 * tree-cfg.h (fold_loop_internal_call): Declare.
3702 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3703 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3704 (find_loop_dist_alias): New function.
3705 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3706 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3709 PR tree-optimization/80631
3710 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3711 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3712 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3713 hardcoding zero as the value if COND_EXPR is never true. For
3714 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3715 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3716 hardcoding MAX_EXPR as the reduction operation.
3717 (is_nonwrapping_integer_induction): Allow negative step.
3718 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3719 vect_create_epilog_for_reduction, if no value is suitable, don't
3720 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3722 2017-12-12 Richard Biener <rguenther@suse.de>
3724 PR tree-optimization/81889
3725 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3726 range info from the non-wrapping IV instead of just the range
3729 2017-12-12 Julia Koval <julia.koval@intel.com>
3731 * config.gcc: Add vaesintrin.h.
3732 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3734 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3735 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3737 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3738 * config/i386/immintrin.h: Include vaesintrin.h.
3739 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3740 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3741 _mm_aesdec_epi128): New intrinsics.
3743 2017-12-12 Julia Koval <julia.koval@intel.com>
3745 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3746 OPTION_MASK_ISA_VAES_UNSET): New.
3747 (ix86_handle_option): Handle -mvaes.
3748 * config/i386/cpuid.h: Define bit_VAES.
3749 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3750 * config/i386/i386-c.c (__VAES__): New.
3751 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3752 (ix86_valid_target_attribute_inner_p): Ditto.
3753 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3754 * config/i386/i386.opt: Add -mvaes.
3755 * doc/invoke.texi: Ditto.
3757 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3759 * debug.h (gcc_debug_hooks): Add inline_entry.
3760 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3761 * debug.c (do_nothing_debug_hooks): Likewise.
3762 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3763 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3764 (dwarf2_lineno_debug_hooks): Likewise.
3766 * common.opt (gstatement-frontiers): New, setting
3767 debug_nonbind_markers_p.
3768 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3769 * toplev.c (process_options): Autodetect value for debug statement
3771 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3772 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3774 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3775 markers. Integrate source bind into debug stmt expand loop.
3776 (pass_expand::execute): Check debug marker limit. Avoid deep
3777 TER and expand debug locations for debug bind insns only.
3778 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3780 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3781 * final.c (reemit_insn_block_notes): Take current block from
3782 nonbind markers. Declare note where it's first set.
3783 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3784 begin stmt markers if enabled.
3785 (notice_source_line): Handle nonbind markers. Fail if their
3786 location is unknown or that of builtins.
3787 (rest_of_handle_final): Convert begin stmt markers to notes if
3788 var-tracking didn't run.
3789 (rest_of_clean_state): Skip begin stmt markers.
3790 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3792 * function.c (allocate_struct_function): Set begin_stmt_markers.
3793 * function.h (struct function): Add debug_marker_count counter
3794 and debug_nonbind_markers flag.
3795 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3796 * gimple-low.c (lower_function_body): Adjust
3797 debug_nonbind_markers.
3798 (lower_stmt): Drop or skip gimple debug stmts.
3799 (lower_try_catch): Skip debug stmts.
3800 * gimple.c (gimple_build_debug_begin_stmt): New.
3801 (gimple_copy): Increment debug_marker_count if copying one.
3802 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3803 * gimplify.c (rexpr_location): New.
3804 (rexpr_has_location): New.
3805 (warn_switch_unreachable_r): Handle gimple debug stmts.
3806 (shortcut_cond_r): Call expr_location.
3808 (find_goto_label): New.
3809 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3811 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3813 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3814 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3815 (LANG_HOOKS_INITIALIZER): ... this.
3816 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3817 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3819 (update_ebb_live_info): Skip debug insn markers.
3820 * lra.c (debug_insn_static_data): Rename to...
3821 (debug_bind_static_data): ... this.
3822 (debug_marker_static_data): New.
3823 (lra_set_insn_recog_data): Select one of the above depending
3825 (lra_update_isn_regno_info): Don't assume debug insns have
3827 (push_insns): Skip debug insns.
3828 * lto-streamer-in.c (input_function): Drop debug stmts
3829 depending on active options. Adjust debug_nonbind_markers.
3830 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3831 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3832 begin stmt marker notes.
3833 (print_insn): Likewise.
3834 * recog.c (extract_insn): Recognize rtl for debug markers.
3835 * rtl.def (DEBUG_MARKER): New.
3836 * tree-inline.c: Include params.h.
3837 (remap_gimple_stmt): Handle nonbind markers.
3838 (maybe_move_debug_stmts_to_successors): Likewise.
3839 (copy_debug_stmt): Likewise.
3840 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3841 markers regardless of no side effects.
3842 (tsi_link_before): Don't update container's side effects when adding
3843 a begin stmt marker.
3844 (tsi_link_after): Likewise.
3845 (expr_first): Skip begin stmt markers.
3846 (expr_last): Likewise.
3847 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3848 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3849 Disregard nonbind markers.
3850 * tree.c (make_node_stat): Don't set side effects for begin stmt
3852 (build1_stat): Likewise.
3853 * tree.def (DEBUG_BEGIN_STMT): New.
3854 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3855 * var-tracking.c (delete_debug_insns): Renamed to...
3856 (delete_vta_debug_insns): ... this.
3857 (reemit_marker_as_note): New.
3858 (vt_initialize): Reemit markers.
3859 (delete_vta_debug_insns): Likewise.
3860 (vt_debug_insns_local): Reemit or delete markers.
3861 (variable_tracking_main_1): Likewise.
3862 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3863 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3864 (gimple_debug_nonbind_marker_p): New.
3865 (gimple_build_debug_bind): Adjust.
3866 (gimple_build_debug_begin_stmt): New.
3867 * doc/invoke.texi (max-debug-marker-count): New param.
3868 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3869 (debug_parameter_ref, debug_marker): New.
3870 (NOTE_INSN_BEGIN_STMT): New.
3871 (DEBUG_INSN): Describe begin stmt markers.
3873 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3874 without debug insns, we wouldn't, but clean up debug insns
3875 after a control flow insn nevertheless.
3877 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3878 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3879 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3880 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3881 (rtl_tidy_fallthru_edge): Likewise.
3882 (rtl_verify_fallthru): Likewise.
3883 (rtl_verify_bb_layout): Likewise.
3884 (skip_insns_after_block): Likewise.
3885 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3886 * dwarf2out.c: Include print-rtl.h.
3887 (dwarf2out_next_real_insn): New.
3888 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3889 Dump debug binds in asm comments.
3890 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3891 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3892 callers to use gsi_start_nondebug_bb instead.
3893 (gsi_after_labels): Skip gimple debug stmts.
3894 (gsi_start_nondebug): New.
3895 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3896 (proper_loop_form_for_interchange): Adjust.
3897 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3898 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3899 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3900 (collect_fallthrough_labels): Likewise.
3901 (should_warn_for_implicit_fallthrough): Likewise.
3902 (warn_implicit_fallthrough_r): Likewise.
3903 (expand_FALLTHROUGH_r): Likewise.
3904 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3905 (graphite_copy_stmts_from_block): Skip nonbind markers.
3906 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3907 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3908 * jump.c (clean_barriers): Skip debug insns.
3909 * omp-expand.c (expand_parallel_call): Skip debug insns.
3910 (expand_task_call): Likewise.
3911 (remove_exit_barrier): Likewise.
3912 (expand_omp_taskreg): Likewise.
3913 (expand_omp_for_init_counts): Likewise.
3914 (expand_omp_for_generic): Likewise.
3915 (expand_omp_for_static_nochunk): Likewise.
3916 (expand_omp_for_static_chunk): Likewise.
3917 (expand_omp_simd): Likewise.
3918 (expand_omp_taskloop_for_outer): Likewise.
3919 (expand_omp_taskloop_for_inner): Likewise.
3920 (expand_oacc_for): Likewise.
3921 (expand_omp_sections): Likewise.
3922 (expand_omp_single): Likewise.
3923 (expand_omp_synch): Likewise.
3924 (expand_omp_atomic_load): Likewise.
3925 (expand_omp_atomic_store): Likewise.
3926 (expand_omp_atomic_fetch_op): Likewise.
3927 (expand_omp_atomic_pipeline): Likewise.
3928 (expand_omp_atomic_mutex): Likewise.
3929 (expand_omp_target): Likewise.
3930 (grid_expand_omp_for_loop): Likewise.
3931 (grid_expand_target_grid_body): Likewise.
3932 (build_omp_regions_1): Likewise.
3933 * omp-low.c (check_combined_parallel): Skip debug stmts.
3934 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3935 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3936 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3938 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3939 (make_edges): Likewise.
3940 (cleanup_dead_labels): Likewise.
3941 (gimple_can_merge_blocks_p): Likewise.
3942 (stmt_starts_bb_p): Likewise.
3943 (gimple_block_label): Likewise.
3944 (gimple_redirect_edge_and_branch): Likewise.
3945 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3947 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3949 * tree-pretty-print (print_declaration): Omit initializer in slim
3951 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3953 (eliminate_unnecessary_stmts): Stabilize block removal order.
3954 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3955 * var-tracking.c (get_first_insn): New.
3956 (vt_emit_notes): Call it.
3957 (vt_initialize): Walk any insns before the first BB.
3958 (delete_debug_insns): Likewise.
3960 * gimple.h (enum gimple_debug_subcode): Add
3961 GIMPLE_DEBUG_BEGIN_STMT.
3962 (gimple_debug_begin_stmt_p): New.
3963 (gimple_debug_nonbind_marker_p): New.
3964 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3965 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3966 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3967 * insn-notes.def (BEGIN_STMT): New.
3968 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3969 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3970 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3971 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3972 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3973 (INSN_DEBUG_MARKER_KIND): New.
3974 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3975 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3976 (INSN_VAR_LOCATION_PTR): New.
3977 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3979 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3980 deps only if debug bind insns are enabled.
3981 (pass_expand::execute): Avoid deep TER and expand
3982 debug locations for debug bind insns only.
3983 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3984 debug stmts special handling down to debug bind stmts.
3985 * combine.c (try_combine): Narrow debug insns special handling
3986 down to debug bind insns.
3987 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3988 Narrow debug insns preexisting special handling down to debug
3990 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3991 handling down to debug bind insns.
3992 * function.c (instantiate_virtual_regs): Skip debug markers,
3993 adjust handling of debug binds.
3994 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3995 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3996 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3997 insns to debug bind insns.
3998 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3999 special handling of debug stmts to debug bind stmts.
4000 * ipa-split.c (split_function): Likewise.
4001 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
4002 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
4004 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
4005 * regrename.c (build_def_use): Likewise.
4006 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
4007 (pass_cprop_hardreg): Narrow special casing of debug insns to
4009 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
4010 * reload1.c (reload): Likewise.
4011 * sese.c (sese_insert_phis_for_liveouts): Narrow special
4012 casing of debug stmts to debug bind stmts.
4013 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
4014 * ssa-iterators.h (num_imm_uses): Likewise.
4015 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
4016 debug stmts to debug bind stmts.
4017 * tree-inline.c (tree_function_versioning): Narrow special casing
4018 of debug stmts to debug bind stmts.
4019 * tree-loop-distribution.c (generate_loops_for_partition):
4020 Narrow special casing of debug stmts to debug bind stmts.
4021 * tree-sra.c (analyze_access_subtree): Narrow special casing
4022 of debug stmts to debug bind stmts.
4023 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
4024 stmts to debug bind stmts.
4025 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
4026 casing of debug stmts to debug bind stmts.
4027 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
4028 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
4029 casing of debug stmts to debug bind stmts.
4030 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4032 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
4033 debug stmts to debug bind stmts.
4034 (gimple_replace_ssa_lhs): Likewise.
4035 (insert_debug_temp_for_var_def): Likewise.
4036 (insert_debug_temps_for_defs): Likewise.
4037 (reset_debug_uses): Likewise.
4038 * tree-ssanames.c (release_ssa_name_fn): Likewise.
4039 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
4040 (adjust_debug_stmts): Likewise.
4041 (adjust_phi_and_debug_stmts): Likewise.
4042 (vect_do_peeling): Likewise.
4043 * tree-vect-loop.c (vect_transform_loop): Likewise.
4044 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
4045 * var-tracking.c (adjust_mems): Narrow special casing of debug
4046 insns to debug bind insns.
4047 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
4048 (compute_bb_dataflow, vt_find_locations): Likewise.
4049 (vt_expand_loc, emit_notes_for_changes): Likewise.
4050 (vt_init_cfa_base): Likewise.
4051 (vt_emit_notes): Likewise.
4052 (vt_initialize): Likewise.
4053 (vt_finalize): Likewise.
4055 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
4056 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
4057 (next_nonnote_nondebug_insn_bb): New.
4058 (prev_nonnote_nondebug_insn_bb): New.
4059 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4060 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4061 (prev_nonnote_nondebug_insn_bb): Declare.
4062 (next_nonnote_nondebug_insn_bb): Declare.
4063 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4064 * cfgrtl.c (get_last_bb_insn): Likewise.
4065 * lra.c (push_insns): Likewise.
4067 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4070 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4072 * selftest.c (selftest::test_runner::test_runner): New ctor.
4073 (selftest::test_runner::~test_runner): New dtor.
4074 * selftest.h (class selftest::test_runner): New class.
4076 2017-12-11 Carl Love <cel@us.ibm.com>
4078 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4079 vec_extract_fp32_from_shortl]): Add #defines.
4080 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4081 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4082 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4083 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4084 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4085 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4086 * doc/extend.texi: Add documentation for the added builtins.
4088 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4090 PR rtl-optimization/80693
4091 PR rtl-optimization/81019
4092 PR rtl-optimization/81020
4093 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4094 are not mentioned in i3. Place the REG_UNUSED note on i2,
4095 possibly modified to REG_DEAD, if it did not originate in i3.
4097 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4099 * recog.c (store_data_bypass_p_1): New function.
4100 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4101 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4103 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4105 PR rtl-optimization/83361
4106 * ifcvt.c (if_convert): Call fixup_partitions.
4108 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4110 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4111 early folding of splat_u{8,16,32}.
4113 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4115 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4116 output_operand_lossage first argument with capital letter.
4117 (aarch64_override_options): Don't start error and sorry first argument
4118 with capital letter.
4120 2017-12-11 Andi Kleen <ak@linux.intel.com>
4122 PR gcov-profile/83355
4123 * auto-profile.c (string_table::get_index_by_decl): Don't
4124 recurse when abstract origin points to itself.
4126 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4128 PR tree-optimization/83320
4129 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4130 (prune_datarefs_not_in_loop): Ditto.
4132 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4134 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4136 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4138 PR tree-optimization/83337
4139 * gimple-loop-interchange.cc (compute_access_stride): Handle
4140 bitfield DRs properly.
4142 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4144 PR tree-optimization/83338
4145 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4146 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4147 vector of unsigned integers to vector of signed integers.
4149 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4151 PR rtl-optimization/83317
4152 * lra-constraints.c (process_address_1): Add insn code check.
4154 2017-12-08 Michael Matz <matz@suse.de>
4156 Fix PR tree-optimization/83323
4157 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4158 head-controlled loops and loop BBs.
4159 * common.opt (funroll-and-jam): Remove, instead ...
4160 (floop-unroll-and-jam): ... reuse this option.
4161 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4162 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4163 (-floop-unroll-and-jam): ... this option.
4165 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4167 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4168 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4169 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4171 2017-12-08 Richard Biener <rguenther@suse.de>
4174 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4175 handle accesses outside of zero-sized vars.
4177 2017-12-08 Martin Jambor <mjambor@suse.cz>
4179 PR tree-optimization/83141
4180 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4181 test for MEM_REFs implicitely changing types with padding. Remove
4183 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4185 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4187 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4189 * config/avr/avr.c (avr_attribute_table): Likewise.
4190 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4191 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4192 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4193 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4194 of struct attribute_spec.
4195 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4197 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4200 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4202 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4205 * profile-count.c (profile_count::from_gcov_type): Move from
4206 profile-count.h; handle overflow.
4207 * profile-count.h (profile_count::from_gcov_type): Move offline.
4209 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4211 PR rtl-optimization/83304
4212 * combine.c (move_deaths): If we do not know where a register died,
4215 2017-12-08 Richard Biener <rguenther@suse.de>
4217 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4218 Provide -fopt-info-loop feedback when we interchange in a nest.
4220 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4222 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4223 for armv6 ARM CPU IDs.
4225 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4227 * common/config/arm/arm-common.c: Include <algorithm>.
4228 (INCLUDE_VECTOR): Define.
4229 (compare_opt_names): New function.
4230 (arm_rewrite_selected_arch): Only strip out extensions that can be
4231 expressed through -mfpu. Sort the remaining extensions
4234 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4236 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4237 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4238 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4239 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4241 2017-06-08 Tristan Gingold <gindold@adacore.com>
4244 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4245 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4247 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4249 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4250 of struct attribute_spec.
4252 2017-12-08 Julia Koval <julia.koval@intel.com>
4254 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4255 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4256 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4257 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4258 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4259 _mm_maskz_dpwssds_epi32): Ditto.
4261 2017-12-08 Richard Biener <rguenther@suse.de>
4263 PR tree-optimization/81303
4264 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4265 conditions try to create a comparison vector type matching
4266 the data vector type.
4267 (vectorizable_condition): Adjust.
4268 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4269 Leave invariant conditions alone in case we can vectorize those.
4271 2017-12-08 Julia Koval <julia.koval@intel.com>
4273 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4274 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4275 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4276 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4277 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4279 2017-12-08 Julia Koval <julia.koval@intel.com>
4281 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4282 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4283 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4284 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4285 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4286 _mm_maskz_dpbusds_epi32): New intrinsics.
4288 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4290 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4292 (ld<bh>io_signed): Likewise.
4293 (st<bhw>io): Likewise.
4294 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4295 SMALL_INT12 constant integer operand.
4297 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4299 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4301 (nios2_large_constant_allowed): ...this. Adjust uses.
4302 (nios2_plus_symbolic_constant_p): Rename to...
4303 (nios2_plus_large_constant_p): ...this. Adjust uses.
4304 (nios2_legitimate_address_p): Correct CONST_INT handling.
4305 (nios2_symbolic_memory_operand_p): Rename to...
4306 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4307 (nios2_large_constant_p): Check for large constant integers too.
4308 (nios2_split_large_constant): Handle constant integers.
4309 (nios2_split_symbolic_memory_operand): Rename to...
4310 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4311 (nios2_legitimize_constant_address): Handle constant integers.
4312 (r0rel_constant_p): Handle small constant integers.
4313 (nios2_print_operand_address): Handle r0-relative integer addresses.
4314 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4315 * config/nios2/nios2.md: Adjust for renamed functions.
4317 2017-12-07 Andrew Waterman <andrew@sifive.com>
4319 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4320 (TARGET_HAVE_SRODATA_SECTION): New define.
4321 (riscv_select_section): New function.
4323 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4324 Alexander Monakov <amonakov@ispras.ru>
4325 Jakub Jelinek <jakub@redhat.com>
4328 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4330 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4332 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4333 a bool success value. Don't call output_operand_lossage here.
4334 (aarch64_print_ldpstp_address): Return a bool success value.
4335 (aarch64_print_operand_address): Call output_addr_const if
4336 aarch64_print_address_internal fails.
4337 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4338 'y'; call output_operand_lossage instead. Call output_operand_lossage
4339 if aarch64_print_ldpstp_address fails.
4341 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4343 * tree-vector-builder.h
4344 (tree_vector_builder::binary_encoded_nelts): Declare.
4345 * tree-vector-builder.c
4346 (tree_vector_builder::binary_encoded_nelts): New function.
4347 * fold-const.c (negate_expr_p): Likewise.
4348 (operand_equal_p, fold_checksum_tree): Likewise.
4349 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4350 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4351 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4352 (uniform_vector_p): Likewise.
4353 * varasm.c (const_hash_1, compare_constant): Likewise.
4354 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4355 (valid_lattice_transition): Operate directly on the VECTOR_CST
4357 * ipa-icf.c: Include tree-vector-builder.h.
4358 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4359 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4361 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4363 * tree.c (build_vector): Delete.
4364 * tree.h (build_vector): Make static and move into the self-testing
4367 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4369 * vector-builder.h (vector_builder::derived): New const overload.
4370 (vector_builder::elt): New function.
4371 * tree-vector-builder.h (tree_vector_builder::type): New function.
4372 (tree_vector_builder::apply_step): Declare.
4373 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4375 * gimple-fold.h (tree_vector_builder): Declare.
4376 (gimple_build_vector): Take a tree_vector_builder instead of a
4377 type and vector of elements.
4378 * gimple-fold.c (gimple_build_vector): Likewise.
4379 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4381 (get_initial_defs_for_reduction): Likewise.
4382 (vectorizable_induction): Likewise.
4384 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4386 * tree-vector-builder.h
4387 (tree_vector_builder::new_binary_operation): Declare.
4388 * tree-vector-builder.c
4389 (tree_vector_builder::new_binary_operation): New function.
4390 * fold-const.c (fold_relational_const): Use it.
4391 (const_binop): Likewise. Check that both input vectors have
4392 the same number of elements, thus excluding things like WIDEN_SUM.
4393 Check whether it is possible to operate directly on the encodings
4396 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4398 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4399 new_unary_operation, operating only on the encoded elements.
4400 (const_unop): Likewise.
4401 (exact_inverse): Likewise.
4402 (distributes_over_addition_p): New function.
4403 (const_binop): Use tree_vector_builder and new_unary_operation
4404 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4405 on the encoded elements unless the encoding is strided and the
4406 operation does not distribute over addition.
4407 (fold_convert_const): Use tree_vector_builder and
4408 new_unary_operation. Operate only on the encoded elements
4409 for truncating integer conversions, or for non-stepped encodings.
4411 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4413 * config/sparc/sparc.c: Include tree-vector-builder.h.
4414 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4415 * expmed.c: Include tree-vector-builder.h.
4416 (make_tree): Use tree_vector_builder instead of build_vector.
4417 * fold-const.c: Include tree-vector-builder.h.
4418 (const_binop): Use tree_vector_builder instead of build_vector.
4419 (const_unop): Likewise.
4420 (native_interpret_vector): Likewise.
4421 (fold_vec_perm): Likewise.
4422 (fold_ternary_loc): Likewise.
4423 * gimple-fold.c: Include tree-vector-builder.h.
4424 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4426 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4427 (simplify_vector_constructor): Use tree_vector_builder instead
4429 * tree-vect-generic.c: Include tree-vector-builder.h.
4430 (add_rshift): Use tree_vector_builder instead of build_vector.
4431 (expand_vector_divmod): Likewise.
4432 (optimize_vector_constructor): Likewise.
4433 * tree-vect-loop.c: Include tree-vector-builder.h.
4434 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4435 of build_vector. Explicitly use a stepped encoding for
4437 * tree-vect-slp.c: Include tree-vector-builder.h.
4438 (vect_get_constant_vectors): Use tree_vector_builder instead
4440 (vect_transform_slp_perm_load): Likewise.
4441 (vect_schedule_slp_instance): Likewise.
4442 * tree-vect-stmts.c: Include tree-vector-builder.h.
4443 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4444 (vect_gen_perm_mask_any): Likewise.
4445 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4446 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4448 (build_vector_from_val): Likewise. Explicitly use a duplicate
4451 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4453 * doc/generic.texi (VECTOR_CST): Describe new representation of
4455 * vector-builder.h: New file.
4456 * tree-vector-builder.h: Likewise.
4457 * tree-vector-builder.c: Likewise.
4458 * Makefile.in (OBJS): Add tree-vector-builder.o.
4459 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4460 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4461 (tree_vector): Change the number of elements to
4462 vector_cst_encoded_nelts.
4463 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4464 (VECTOR_CST_ELTS): Delete.
4465 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4466 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4467 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4468 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4469 (VECTOR_CST_ENCODED_ELT): Likewise.
4470 (vector_cst_encoded_nelts): New function.
4471 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4472 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4473 (vector_cst_int_elt, vector_cst_elt): Declare.
4474 * tree.c: Include tree-vector-builder.h.
4475 (tree_code_size): Abort if passed VECTOR_CST.
4476 (tree_size): Update for new VECTOR_CST layout.
4477 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4478 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4479 (build_vector): Use tree_vector_builder.
4480 (vector_cst_int_elt, vector_cst_elt): New functions.
4481 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4482 encoded elements and then create the vector in the canonical form.
4483 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4484 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4485 (tree_c_tests): Call test_vector_cst_patterns.
4486 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4488 (hash_tree): Likewise.
4489 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4490 (streamer_write_tree_header): Likewise.
4491 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4492 (streamer_alloc_tree): Likewise. Update call to make_vector.
4493 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4495 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4497 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4498 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4499 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4501 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4502 Richard Biener <rguenther@suse.de>
4504 PR tree-optimization/81303
4505 * Makefile.in (gimple-loop-interchange.o): New object file.
4506 * common.opt (floop-interchange): Reuse the option from graphite.
4507 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4508 -floop-interchange and mention it for -O3.
4509 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4510 * gimple-loop-interchange.cc: New file.
4511 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4512 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4513 * passes.def (pass_linterchange): New pass.
4514 * timevar.def (TV_LINTERCHANGE): New time var.
4515 * tree-pass.h (make_pass_linterchange): New declaration.
4516 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4517 interchange. Record IV before/after increment in new parameters.
4518 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4519 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4521 (check_reduction_path): ...New function here.
4522 * tree-vectorizer.h (check_reduction_path): New declaration.
4524 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4527 PR rtl-optimization/80818
4528 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4529 always early clobbered.
4530 * lra-lives.c (process_bb_lives): Check input hard regs for early
4531 clobbered non-operand hard reg.
4533 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4536 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4537 types_compatible_p, just that TYPE_MODE is the same.
4539 2017-12-07 Martin Sebor <msebor@redhat.com>
4542 * attribs.c (empty_attribute_table): Initialize new member of
4543 struct attribute_spec.
4544 (decl_attributes): Add argument. Handle mutually exclusive
4545 combinations of attributes.
4546 (selftests::test_attribute_exclusions): New function.
4547 (selftests::attribute_c_tests): Ditto.
4548 * attribs.h (decl_attributes): Add default argument.
4549 * selftest.h (attribute_c_tests): Declare.
4550 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4551 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4553 * doc/extend.texi (Common Function Attributes): Update const and pure.
4554 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4555 of struct attribute_spec.
4556 * config/arc/arc.c (arc_attribute_table): Same.
4557 * config/arm/arm.c (arm_attribute_table): Same.
4558 * config/avr/avr.c ( avr_attribute_table): Same.
4559 * config/bfin/bfin.c (bfin_attribute_table): Same.
4560 * config/cr16/cr16.c (cr16_attribute_table): Same.
4561 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4562 * config/h8300/h8300.c (h8300_attribute_table): Same.
4563 * config/i386/i386.c (ix86_attribute_table): Same.
4564 * config/ia64/ia64.c (ia64_attribute_table): Same.
4565 * config/m32c/m32c.c (m32c_attribute_table): Same.
4566 * config/m32r/m32r.c (m32r_attribute_table): Same.
4567 * config/m68k/m68k.c (m68k_attribute_table): Same.
4568 * config/mcore/mcore.c (mcore_attribute_table): Same.
4569 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4570 * config/mips/mips.c (mips_attribute_table): Same.
4571 * config/msp430/msp430.c (msp430_attribute_table): Same.
4572 * config/nds32/nds32.c (nds32_attribute_table): Same.
4573 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4574 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4575 * config/rl78/rl78.c (rl78__attribute_table): Same.
4576 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4577 * onfig/rx/rx.c (rx_attribute_table): Same.
4578 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4579 * config/sh/sh.c (sh_attribute_table): Same.
4580 * config/sparc/sparc.c (sparc_attribute_table): Same.
4581 * config/spu/spu.c (spu_attribute_table): Same.
4582 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4583 * config/v850/v850.c (v850_attribute_table): Same.
4584 * config/visium/visium.c (visium_attribute_table): Same.
4586 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4589 * config/arm/arm.c (INCLUDE_STRING): Define.
4590 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4591 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4594 2017-12-07 Michael Matz <matz@suse.de>
4596 Add unroll and jam pass
4598 * gimple-loop-jam.c: New file.
4599 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4600 * common.opt (funroll-and-jam): New option.
4601 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4602 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4603 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4604 * passes.def: Add pass_loop_jam.
4605 * timevar.def (TV_LOOP_JAM): Add.
4606 * tree-pass.h (make_pass_loop_jam): Declare.
4607 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4608 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4609 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4610 to flow_loop_tree_node_add.
4611 (duplicate_subloops, copy_loops_to): Append to sibling list.
4612 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4613 * doc/invoke.texi (-funroll-and-jam): Document new option.
4614 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4616 2017-12-07 Richard Biener <rguenther@suse.de>
4618 PR tree-optimization/83296
4619 PR tree-optimization/67769
4620 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4621 flow sensitive info in an unrelated BB.
4622 (value_replacement): Use reset_flow_sensitive_info.
4623 (minmax_replacement): Reset flow sensitive info on the def
4624 we move. Do not reset flow sensitive info in the whole BB
4625 we move the stmt to.
4626 (abs_replacement): Likewise.
4628 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4631 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4632 rs6000_cpu to the given -mcpu=, or to the default processor.
4634 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4636 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4637 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4638 instead of rs6000_cpu_attr.
4639 (rs6000_adjust_cost): Ditto.
4640 (is_microcoded_insn): Ditto.
4641 (rs6000_adjust_priority): Ditto.
4642 (rs6000_issue_rate): Ditto.
4643 (rs6000_use_sched_lookahead): Ditto.
4644 (rs6000_use_sched_lookahead_guard): Ditto.
4645 (rs6000_sched_reorder): Ditto.
4646 (force_new_group): Ditto.
4647 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4648 (group_ending_nop): Ditto.
4650 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4652 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4653 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4654 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4655 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4657 (rs6000_reassociation_width): Ditto.
4658 (rs6000_emit_epilogue): Ditto.
4659 (rs6000_adjust_cost): Ditto.
4660 (is_microcoded_insn): Ditto.
4661 (is_cracked_insn): Ditto.
4662 (rs6000_adjust_priority): Ditto.
4663 (rs6000_sched_reorder): Ditto.
4664 (rs6000_sched_reorder2): Ditto.
4665 (insn_must_be_first_in_group): Ditto.
4666 (insn_must_be_last_in_group): Ditto.
4667 (rs6000_register_move_cost): Ditto.
4668 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4671 2017-12-07 Julia Koval <julia.koval@intel.com>
4673 * config.gcc: Add vaesintrin.h.
4674 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4676 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4677 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4679 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4680 * config/i386/immintrin.h: Include vaesintrin.h.
4681 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4682 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4683 _mm_aesdec_epi128): New intrinsics.
4685 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4687 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4688 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4689 c-family/c-spellcheck.cc.
4690 (best_macro_match::best_macro_match): Likewise.
4692 (struct edit_distance_traits<cpp_hashnode *>): Move to
4693 c-family/c-spellcheck.h.
4694 (class best_macro_match): Likewise.
4696 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4698 PR tree-optimization/83293
4699 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4700 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4701 might insert into empty bb.
4704 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4705 simplify for plus with :c added, and pointer_plus without that.
4706 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4707 with undefined overflow and the conversion is not widening,
4708 perform negation in utype and only convert to type afterwards.
4709 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4710 simplify for plus with :c added, and pointer_plus without that.
4711 If type is integral with undefined overflow and the conversion is
4712 not widening, perform minus in utype and only convert to type
4713 afterwards. Move the last pointer_diff_expr simplify into the
4716 2017-12-06 Martin Sebor <msebor@redhat.com>
4718 PR tree-optimization/82646
4719 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4720 strncpy, not maxlen.
4722 2017-12-06 Martin Sebor <msebor@redhat.com>
4724 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4727 PR tree-optimization/83075
4728 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4729 strncat/strncpy don't change length of source string.
4731 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4734 2017-11-29 Martin Aberg <maberg@gaisler.com>
4736 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4737 to prevent b2bst errata sequence.
4738 (sqrtdf2_fix): Likewise.
4740 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4742 PR tree-optimization/81945
4743 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4744 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4745 to dest_cfun has orig_loop_num set, either remap it to the new
4746 loop number if the loop got moved too, or clear it.
4748 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4750 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4752 (thunderx2t99_nothing): New insn reservation.
4753 (thunderx2t99_mrs): New insn reservation.
4754 (thunderx2t99_multiple): New insn reservation.
4755 (thunderx2t99_alu_basi): Add bfx to reservation.
4756 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4758 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4761 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4763 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4765 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4766 tree-ssa-pre.c::remove_dead_inserted_code.
4767 * tree-ssa-dce.h: New file.
4768 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4769 (remove_dead_inserted_code): Move and rename to function
4770 tree-ssa-dce.c::simple_dce_from_worklist.
4771 (pass_pre::execute): Update use.
4773 2017-12-05 Richard Biener <rguenther@suse.de>
4775 PR tree-optimization/83277
4776 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4777 to code-gen liveout vars.
4779 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4781 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4782 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4783 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4786 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4788 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4790 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4791 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4794 2017-12-05 Richard Biener <rguenther@suse.de>
4796 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4797 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4798 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4799 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4801 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4803 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4805 2017-12-05 Martin Liska <mliska@suse.cz>
4806 Jakub Jelinek <jakub@redhat.com>
4808 * doc/invoke.texi: Document the options.
4809 * flag-types.h (enum sanitize_code): Add
4810 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4811 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4812 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4813 * opts.c: Define new sanitizer options.
4814 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4815 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4817 2017-12-05 Julia Koval <julia.koval@intel.com>
4819 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4820 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4821 (ix86_handle_option): Handle -mavx512vnni.
4822 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4823 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4824 * config/i386/i386-c (__AVX512VNNI__): New.
4825 * config/i386/i386.c (ix86_target_string): Handle new option.
4826 (ix86_valid_target_attribute_inner_p): Handle new option.
4827 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4828 * config/i386/i386.opt (mavx512vnni): New option.
4830 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4833 * athlon.md: Disable for generic.
4834 * haswell.md: Enable for generic.
4835 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4836 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4838 (ix86_adjust_cost): Move generic to haswell path.
4840 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4842 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4843 instead of MEM_P in a couple more places. Fix formatting issues.
4845 2017-12-04 Jim Wilson <jimw@sifive.com>
4847 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4848 instead of GP_REG_LAST-1.
4849 (riscv_adjust_libcall_cfi_prologue): Likewise.
4850 (riscv_adjust_libcall_cri_epilogue): Likewise.
4851 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4854 2017-12-04 Luis Machado <luis.machado@linaro.org>
4856 * ipa-pure-const.c (check_decl): Add missing newline.
4857 (state_from_flags): Likewise.
4859 2017-12-04 Jeff Law <law@redhat.com>
4861 PR tree-optimizatin/78496
4862 * gimple-ssa-evrp-analyze.h
4863 (evrp_range_analyzer::get_vr_values): Simplify.
4864 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4865 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4866 and gimple-ssa-evrp-analyze.h.
4867 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4868 (simplify_stmt_for_jump_threading): Copy a blob of code from
4869 tree-vrp.c to use ranges to simplify statements.
4870 (dom_opt_dom_walker::before_dom_children): Call
4871 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4872 (dom_opt_dom_walker::after_dom_children): Similarly for
4873 evrp_range_analyzer::leave.
4874 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4877 * gimple-ssa-evrp-analyze.c
4878 (evrp_range_analyzer::extract_range_from_stmt): Always use
4879 vr_values::update_value_range so preexisting range info is
4880 medged with new range info, even if the new range is VR_VARYING.
4882 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4884 * combine.c: Adjust comment.
4885 (use_crosses_set_p): Delete.
4886 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4887 (try_combine): Ditto.
4889 2017-12-04 Richard Biener <rguenther@suse.de>
4891 PR tree-optimization/83255
4892 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4893 Re-add zero-iteration check.
4895 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4897 PR rtl-optimization/83245
4898 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4899 hard registers as earlyclobber, also if not in an asm.
4901 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4905 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4908 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4909 rs6000_cpu based on cpu_index, not tune_index.
4911 2017-12-04 Richard Biener <rguenther@suse.de>
4913 PR tree-optimization/83238
4914 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4915 code match comment, rejecting invalid SESE regions.
4917 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4919 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4920 require base operand is a REG_POINTER prior to reload on targets
4921 with non-equivalent space registers.
4923 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4925 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4926 (update_specialized_profile): Fix updating of counts.
4927 (perhaps_add_new_callers): Likewise.
4929 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4932 * x86-tune.def: Remove obsolette FIXMEs.
4933 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4934 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4935 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4937 (X86_TUNE_PAD_RETURNS): Disable for generic.
4939 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4941 PR tree-optimization/83170
4942 PR tree-optimization/83241
4943 * gimple-ssa-store-merging.c
4944 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4945 gimple_vuse (ins_stmt) in case it has changed.
4946 (imm_store_chain_info::output_merged_store): Likewise.
4948 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4952 * tree-cfg.c (pass_warn_function_return::execute): Handle
4953 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4954 with BUILTINS_LOCATION.
4958 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4959 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4960 instead of DECL_MODE.
4962 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4963 last argument to rtx pointer.
4964 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4965 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4966 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4967 depending on the chosen ISAs.
4968 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4969 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4970 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4972 * config/i386/sse.md (mov<mode>_internal): Likewise.
4973 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4975 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4977 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4979 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4981 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4982 parameters from prototype.
4983 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4984 parameters. Don't print a comment.
4985 (emit_fusion_gpr_load): Adjust.
4986 (emit_fusion_load_store): Adjust.
4987 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4988 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4989 comment on the second line.
4991 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4994 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4995 rs6000_cpu based on cpu_index, not tune_index.
4997 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4999 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
5000 which_alternative instead of which_alternative + 1.
5001 (output_asm_insn): Print an extra tab if the template is short.
5003 2017-12-01 Jim Wilson <jimw@sifive.com>
5005 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
5007 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
5008 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
5010 * doc/tm.texi: Regenerate.
5012 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
5015 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
5016 whether we can allocate pseudos before trying to fix an address.
5017 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
5018 memory address is indexed or indirect.
5019 (floatuns_<mode>si2_hw2): Likewise.
5021 2017-12-01 Jason Merrill <jason@redhat.com>
5023 * Makefile.in (TAGS): Add c-family/*.cc.
5025 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
5027 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
5028 (call_value_insn): Likewise.
5029 (sibcall_insn): Likewise.
5030 (sibcall_value_insn): Likewise.
5031 (movsi_aarch64): Likewise.
5032 (movdi_aarch64): Likewise.
5033 (add_losym_): Likewise.
5034 (ldr_got_small_): Likewise.
5035 (ldr_got_small_sidi): Likewise.
5036 (ldr_got_small_28k_): Likewise.
5037 (ldr_got_small_28k_sidi): Likewise.
5038 * config/aarch64/aarch64.c (aarch64_print_address_internal):
5039 Move output_addr_const to symbolic case. Add error check.
5041 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5043 * config/s390/predicates.md (plus16_Q_operand): New predicate.
5044 * config/s390/s390.md: Disable MVC merging peephole if it would
5045 disable operand forwarding.
5046 (new peephole2): Split MVCs if it would turn them into up to 2
5049 2017-12-01 Richard Biener <rguenther@suse.de>
5051 PR tree-optimization/83232
5052 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
5053 detection of same access. Instead of breaking the group here
5054 do not consider the duplicate. Add comment explaining real fix.
5056 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
5058 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5060 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5062 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5064 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5066 * function.h (struct function): Remove cilk_frame_decl,
5067 is_cilk_function and calls_cilk_spawn fields.
5068 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5070 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5071 cilk_elemental field.
5072 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5073 * target.def: Adjust comment.
5074 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5075 Don't test cilk_elemental.
5077 PR tree-optimization/83233
5078 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5079 bswap_stat name for the struct.
5082 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5083 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5084 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5085 SWITCH_BREAK_LABEL_P set on the label.
5086 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5087 added for default case if it was missing and not all cases covered.
5088 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5089 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5091 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5092 to avoid -Wimplicit-fallthrough warning.
5093 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5094 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5096 PR tree-optimization/83221
5097 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5099 (init_reassoc): Formatting fix.
5102 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5103 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5107 * tree-cfg.c: Include asan.h.
5108 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5110 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5112 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5114 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5116 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5117 vpcmpeqd instruction.
5119 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5121 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5122 registers type for 128bit mode.
5124 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5126 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5128 * gengtype-state.c (read_state_pair): Likewise.
5129 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5130 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5131 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5132 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5133 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5134 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5135 * builtins.c (fold_builtin_3): Likewise.
5136 * graphite-scop-detection.c
5137 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5138 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5141 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5144 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5145 parameter. Return true if ignore_decl_rtl_set_p is true and other
5146 conditions are satisfied.
5147 * asan.h (asan_protect_global): Add new parameter.
5148 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5149 to asan_protect_global calls.
5151 2017-11-30 Jim Wilson <jimw@sifive.com>
5153 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5154 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5155 -mstrict-align, add info on default value. Delete redundant lines for
5156 -mabi. Add missing -mexplicit-relocs docs.
5158 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5160 * config/arc/arc.md (trap): New pattern.
5162 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5164 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5165 instruction to end into a delay slot.
5166 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5167 can be placed into a delay slot against reg_note.
5169 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5171 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5172 labels number of usages.
5174 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5176 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5178 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5180 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5183 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5184 multiplication by power of 2 constant into two shifts + comparison.
5186 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5189 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5191 2017-11-30 Richard Biener <rguenther@suse.de>
5193 PR tree-optimization/83202
5194 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5195 allow_peel argument and guard peeling.
5196 (canonicalize_loop_induction_variables): Likewise.
5197 (canonicalize_induction_variables): Pass false.
5198 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5199 peeling from cunrolli.
5201 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5203 * combine.c (try_combine): Print a message to dump file whenever
5204 I0, I1, or I2 cannot be combined into I3.
5206 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5208 PR rtl-optimization/83156
5209 PR rtl-optimization/82621
5210 * combine.c (try_combine): Don't split an I2 if one of the dests is
5211 set again before I3. Allow unused dests.
5213 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5215 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5217 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5219 PR rtl-optimization/80818
5220 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5221 recursively. Use insn code for clobber.
5222 (lra_set_insn_recog_data): Pass the new arg to
5223 collect_non_operand_hard_regs.
5224 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5226 (lra_update_insn_regno_info): Pass insn to
5227 add_regs_to_insn_regno_info.
5229 2017-11-29 Jim Wilson <jimw@sifive.com>
5230 Andrew Waterman <andrew@sifive.com>
5232 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5233 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5234 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5235 use SINGLE_SHIFT_COST.
5236 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5237 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5239 2017-11-29 Julia Koval <julia.koval@intel.com>
5241 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5242 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5243 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5244 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5245 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5246 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5247 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5248 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5249 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5250 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5251 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5252 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5253 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5254 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5255 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5256 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5257 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5258 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5259 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5260 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5261 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5262 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5263 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5264 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5265 __builtin_ia32_vpshldv_v2di_mask,
5266 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5267 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5268 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5270 2017-11-29 Julia Koval <julia.koval@intel.com>
5272 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5273 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5274 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5275 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5276 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5277 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5278 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5279 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5280 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5281 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5282 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5283 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5284 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5285 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5286 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5287 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5288 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5289 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5290 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5291 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5292 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5294 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5296 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5297 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5300 2017-11-29 Martin Aberg <maberg@gaisler.com>
5302 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5303 to prevent b2bst errata sequence.
5304 (sqrtdf2_fix): Likewise.
5306 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5308 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5309 (div_sqrt_insn_p): New function.
5310 (sparc_do_work_around_errata): Insert NOP instructions to
5311 prevent sequences that could trigger the TN-0013 errata for
5312 certain LEON3 processors.
5313 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5314 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5315 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5316 (in_branch_delay): Prevent div and sqrt in delay slot if
5318 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5320 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5322 * config/sparc/sparc.c (atomic_insn_p): New function.
5323 (sparc_do_work_around_errata): Insert NOP instructions to
5324 prevent sequences that could trigger the TN-0010 errata for
5326 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5327 instruction referable in atomic_insns_p.
5329 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5331 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5332 (atomic_compare_and_swap_leon3_1): Likewise.
5335 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5337 * config/sparc/sparc.c (fpop_insn_p): New function.
5338 (sparc_do_work_around_errata): Insert NOP instructions to
5339 prevent sequences that could trigger the TN-0012 errata for
5341 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5342 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5343 (in_branch_annul_delay): Prevent floating-point instructions
5344 in delay slot of annulled integer branch.
5346 2017-11-29 Richard Biener <rguenther@suse.de>
5348 PR tree-optimization/83202
5349 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5351 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5352 nodes vectorized to the same stmts multiple times.
5353 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5354 (vect_analyze_slp_instance): Adjust.
5355 (scalar_stmts_to_slp_tree_map_t): New typedef.
5356 (vect_schedule_slp_instance): Add a map recording the SLP node
5357 representing the vectorized stmts for a set of scalar stmts.
5358 Avoid code-generating redundancies.
5359 (vect_schedule_slp): Allocate map and pass it down.
5361 2017-11-29 Nathan Sidwell <nathan@acm.org>
5364 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5367 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5369 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5370 cases for printing LDP/STP memory addresses.
5371 (aarch64_print_address_internal): Renamed from
5372 aarch64_print_operand_address, added parameter, add Pmode check.
5373 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5374 (aarch64_print_operand_address): Indirect to
5375 aarch64_print_address_internal.
5376 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5377 'y' operand output specifier.
5379 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5382 * tree.c (build_simple_mem_ref_loc): Handle
5383 get_addr_base_and_unit_offset returning a MEM_REF.
5386 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5389 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5392 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5394 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5396 (layout::move_to_column): Likewise.
5398 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5401 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5402 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5404 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5405 Martin Jambor <mjambor@suse.cz>
5408 * tree.h (expr_type_first_operand_type_p): Declare
5409 * tree.c (expr_type_first_operand_type_p): New function.
5410 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5411 (ipa_value_from_jfunc): Adjust declaration.
5412 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5413 Use it as result type for arithmetics, unless it is NULL in which case
5414 be more conservative.
5415 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5416 ipa_get_jf_pass_through_result.
5417 (propagate_vals_across_pass_through): Likewise.
5418 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5419 is to propagate_vals_across_pass_through.
5420 (propagate_constants_across_call): Pass PARM_TYPE to
5421 propagate_scalar_across_jump_function.
5422 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5423 ipa_value_from_jfunc.
5424 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5425 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5426 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5427 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5428 target_type, pass it to ipa_value_from_jfunc.
5429 (update_indirect_edges_after_inlining): Pass parameter type to
5430 try_make_edge_direct_simple_call.
5432 2017-11-28 Jeff Law <law@redhat.com>
5434 * gimple-ssa-evrp-analyze.c
5435 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5436 refine ranges if scev_initialized_p returns true.
5437 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5439 2017-11-28 Julia Koval <julia.koval@intel.com>
5441 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5442 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5443 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5444 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5445 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5446 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5447 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5448 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5449 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5450 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5451 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5452 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5453 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5454 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5455 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5456 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5457 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5458 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5459 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5460 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5461 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5462 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5464 2017-11-28 Julia Koval <julia.koval@intel.com>
5466 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5467 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5468 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5469 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5470 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5471 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5472 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5473 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5474 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5475 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5476 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5477 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5478 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5479 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5480 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5481 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5482 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5483 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5484 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5485 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5486 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5487 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5488 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5489 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5490 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5491 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5492 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5493 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5494 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5495 __builtin_ia32_vpshld_v2di_mask): New builtins.
5496 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5497 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5499 2017-11-28 Richard Biener <rguenther@suse.de>
5501 PR tree-optimization/80776
5502 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5504 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5506 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5507 If the incoming edge is an effective fallthru because the other
5508 edge only reaches a __builtin_unreachable () then record ranges
5509 derived from the controlling condition in SSA info.
5510 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5511 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5513 2017-11-28 Olivier Hainque <hainque@adacore.com>
5515 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5516 on mingw build hosts.
5518 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5521 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5523 2017-11-28 Richard Biener <rguenther@suse.de>
5526 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5527 copies generated from memcpy use a character array as reference
5530 2017-11-28 Julia Koval <julia.koval@intel.com>
5531 Sebastian Peryt <sebastian.peryt@intel.com>
5533 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5534 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5535 cilk-common.o, cilk.h, cilk-common.c): Remove.
5537 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5538 * builtins.c (is_builtin_name): Remove cilkplus condition.
5539 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5540 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5541 cilk-builtins.def, cilkplus.def): Remove.
5542 * cif-code.def (CILK_SPAWN): Remove.
5543 * cilk-builtins.def: Delete.
5544 * cilk-common.c: Ditto.
5546 * cilkplus.def: Ditto.
5547 * config/darwin.h (fcilkplus): Delete.
5548 * cppbuiltin.c: Ditto.
5549 * doc/extend.texi: Remove cilkplus doc.
5550 * doc/generic.texi: Ditto.
5551 * doc/invoke.texi: Ditto.
5552 * doc/passes.texi: Ditto.
5553 * gcc.c (fcilkplus): Remove.
5554 * gengtype.c (cilk.h): Remove.
5555 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5557 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5559 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5561 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5562 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5563 cilkplus conditions.
5564 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5565 inline_read_section): Ditto.
5566 * ipa-inline-analysis.c (cilk.h): Remove.
5567 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5568 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5569 append_linker_options): Remove condition for fcilkplus.
5570 * lto/lto-lang.c (cilk.h): Remove.
5571 (lto_init): Remove condition for fcilkplus.
5572 * omp-expand.c (expand_cilk_for_call): Delete.
5573 (expand_omp_taskreg, expand_omp_for_static_chunk,
5574 expand_omp_for): Remove cilkplus
5576 (expand_cilk_for): Delete.
5577 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5578 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5579 execute_lower_omp, diagnose_sb_0): Ditto.
5580 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5581 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5582 * tree-nested.c: Ditto.
5583 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5584 (dump_generic_node): Ditto.
5585 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5586 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5588 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5590 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5592 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5593 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5594 for vector mode and !TARGET_SIMD.
5596 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5598 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5600 * tree.h (SWITCH_LABELS): Remove.
5601 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5602 assert SWITCH_BODY is non-NULL.
5603 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5605 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5607 PR tree-optimization/80788
5608 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5609 has TREE_OVERFLOW set, call drop_tree_overflow.
5611 2017-11-28 Richard Biener <rguenther@suse.de>
5613 PR tree-optimization/83158
5614 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5616 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5619 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5620 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5622 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5624 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5625 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5627 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5630 * varasm.c (compare_constant): Take the mode of the constants into
5631 account when comparing floating point constants.
5633 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5635 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5636 from explicit instantiation of debug_helper.
5637 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5639 2017-11-27 Richard Biener <rguenther@suse.de>
5641 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5644 2017-11-27 Richard Biener <rguenther@suse.de>
5646 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5647 (type_hash_canon): Also clear int_cst_hash_table entry for
5649 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5651 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5653 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5654 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5655 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5657 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5659 * hash-map.h (gt_cleare_cache): Avoid UB.
5661 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5663 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5664 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5665 (decide_unroll_runtime_iterations): Likewise.
5666 (decide_unroll_stupid): Likewise.
5668 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5671 * config/i386/i386.md: Add a loop with incssp.
5673 2017-11-27 Martin Jambor <mjambor@suse.cz>
5675 PR tree-optimization/81248
5676 * tree-sra.c (splice_param_accesses): Remove size check.
5677 (decide_one_param_reduction): Fix size check.
5678 * gimple-pretty-print.c (dump_profile): Silence warning.
5679 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5681 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5684 * dbxout.c (lastlineno): New variable.
5685 (dbx_debug_hooks): Use dbxout_switch_text_section as
5686 switch_text_section debug hook.
5687 (dbxout_function_end): Switch to current_function_section
5688 rather than function_section. If crtl->has_bb_partition,
5689 output just one N_FUN, depending on in_cold_section_p.
5690 (dbxout_source_line): Remember last lineno in lastlineno.
5691 (dbxout_switch_text_section): New function.
5692 (dbxout_function_decl): Adjust dbxout_block caller.
5693 (dbx_block_with_cold_children): New function.
5694 (dbxout_block): Return true if any LBRAC/RBRAC have been
5695 emitted. Use dbx_block_with_cold_children at depth == 0
5696 in second partition. Add PARENT_BLOCKNUM argument, pass
5697 it optionally adjusted to children. Output LBRAC/RBRAC
5698 around recursive call only if the block is in the current
5699 partition, if not and anything was output, emit empty
5701 * final.c (final_scan_insn): Compute cold_function_name
5702 before calling switch_text_section debug hook. Call
5703 that hook even if dwarf2out_do_frame if not emitting
5707 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5708 TREE_READONLY decls.
5710 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5712 PR rtl-optimization/82488
5713 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5715 2017-11-26 Julia Koval <julia.koval@intel.com>
5717 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5719 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5722 2017-11-26 Julia Koval <julia.koval@intel.com>
5724 * config/i386/driver-i386.c (host_detect_local_cpu):
5725 Detect skylake-avx512.
5727 2017-11-26 Julia Koval <julia.koval@intel.com>
5729 * config.gcc: Add -march=cannonlake.
5730 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5731 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5732 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5733 (PTA_CANNONLAKE): New.
5734 (processor_target_table): Add cannonlake.
5735 (ix86_option_override_internal): Ditto.
5736 (fold_builtin_cpu): Ditto.
5737 (get_builtin_code_for_version): Handle cannonlake.
5738 (M_INTEL_COREI7_CANNONLAKE): New.
5739 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5740 * doc/invoke.texi: Add -march=cannonlake.
5742 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5744 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5745 (try_init_one_plugin): Alternative implementation for MinGW.
5746 * Makefile.in (plugin_implib): New.
5747 (gengtype-lex.c): Fix broken AIX workaround.
5748 * configure: Regenerate.
5749 * doc/plugins.texi: Document support for MinGW.
5751 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5753 PR rtl-optimization/81553
5754 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5755 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5756 is a shift where C1 has different mode than the whole shift, use C1's
5757 mode for MULT rather than the shift's mode.
5760 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5761 builtins not enabled in the currently selected ISA.
5763 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5765 PR tree-optimization/71026
5766 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5767 (insert_reciprocals): Change to insert reciprocals before a division
5768 by a square and to insert the square of a reciprocal.
5769 (execute_cse_reciprocals_1): Change to consider division by a square.
5770 (register_division_in): Add importance parameter.
5772 2017-11-24 Richard Biener <rguenther@suse.de>
5774 PR tree-optimization/82402
5775 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5776 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5778 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5780 * match.pd (0-ptr): New transformation.
5782 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5785 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5786 when skipping checking.
5788 2017-11-24 Richard Biener <rguenther@suse.de>
5790 PR tree-optimization/83128
5791 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5792 (vn_reference_lookup_3): Likewise.
5794 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5797 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5798 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5801 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5802 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5803 check for the latter with gimple_call_builtin_p. Do not handle
5804 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5806 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5808 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5811 2017-11-23 Julia Koval <julia.koval@intel.com>
5813 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5814 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5815 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5816 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5817 _mm512_maskz_expandloadu_epi16): New intrinsics.
5818 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5819 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5820 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5821 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5822 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5823 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5824 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5825 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5826 _mm256_maskz_expandloadu_epi8): New intrinsics.
5827 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5828 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5829 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5830 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5831 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5832 * config/i386/sse.md (VI248_VLBW): New iterator.
5833 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5835 2017-11-23 Julia Koval <julia.koval@intel.com>
5837 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5838 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5839 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5840 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5841 _mm512_mask_compressstoreu_epi16): New.
5842 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5843 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5844 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5845 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5846 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5847 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5848 _mm256_mask_compressstoreu_epi8): New.
5849 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5850 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5851 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5852 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5853 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5854 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5855 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5856 __builtin_ia32_compresshi128_mask,
5857 __builtin_ia32_compressstoreuqi512_mask,
5858 __builtin_ia32_compressstoreuhi512_mask,
5859 __builtin_ia32_compressstoreuqi256_mask,
5860 __builtin_ia32_compressstoreuqi128_mask,
5861 __builtin_ia32_compressstoreuhi256_mask,
5862 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5863 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5865 (ix86_expand_special_args_builtin): Handle new types.
5866 (s4fma_expand): Handle new builtin array.
5867 * config/i386/immintrin.h: Include new headers.
5868 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5869 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5871 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5873 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5874 fixes. Declare temp and g variables at the top in order to avoid
5875 {} in most of the cases.
5877 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5879 * match.pd (ptr-0): New transformation.
5881 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5883 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5884 (struct addr_mode_cost_table): New.
5885 (struct tune_params): Add field addr_mode_costs.
5886 * config/arm/arm.c (generic_addr_mode_costs): New.
5887 (arm_slowmul_tune): Initialise addr_mode_costs field.
5888 (arm_fastmul_tune): Likewise.
5889 (arm_strongarm_tune): Likewise.
5890 (arm_xscale_tune): Likewise.
5891 (arm_9e_tune): Likewise.
5892 (arm_marvell_pj4_tune): Likewise.
5893 (arm_v6t2_tune): Likewise.
5894 (arm_cortex_tune): Likewise.
5895 (arm_cortex_a8_tune): Likewise.
5896 (arm_cortex_a7_tune): Likewise.
5897 (arm_cortex_a15_tune): Likewise.
5898 (arm_cortex_a35_tune): Likewise.
5899 (arm_cortex_a53_tune): Likewise.
5900 (arm_cortex_a57_tune): Likewise.
5901 (arm_exynosm1_tune): Likewise.
5902 (arm_xgene1_tune): Likewise.
5903 (arm_cortex_a5_tune): Likewise.
5904 (arm_cortex_a9_tune): Likewise.
5905 (arm_cortex_a12_tune): Likewise.
5906 (arm_cortex_a73_tune): Likewise.
5907 (arm_v7m_tune): Likewise.
5908 (arm_cortex_m7_tune): Likewise.
5909 (arm_v6m_tune): Likewise.
5910 (arm_fa726te_tune): Likewise.
5911 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5913 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5915 * config/arm/arm.c (arm_mem_costs): New function.
5916 (arm_rtx_costs_internal): Use arm_mem_costs.
5918 2017-11-23 Mark Wielaard <mark@klomp.org>
5920 * dwarf2out.c (init_sections_and_labels): Use generation to create
5921 unique ranges_section_label and ranges_base_label. Return generation.
5922 (output_rnglists): Add generation argument. Use generation to create
5923 unique ranges labels.
5924 (dwarf2out_finish): Get generation from init_sections_and_labels
5925 and pass generation to output_rnglists.
5927 2017-11-23 Mike Stump <mikestump@comcast.net>
5928 Eric Botcazou <ebotcazou@adacore.com>
5930 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5931 * cfgloop.h (struct loop): Add unroll field.
5932 * function.h (struct function): Add has_unroll bitfield.
5933 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5934 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5935 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5936 (pass_rtl_unroll_loops::gate): Likewise.
5937 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5938 for which loop->unroll==1.
5939 (decide_unroll_constant_iterations): Use note for consistency and
5940 take loop->unroll into account. Return early if loop->unroll is set.
5941 Fix thinko in existing test.
5942 (decide_unroll_runtime_iterations): Use note for consistency and
5943 take loop->unroll into account.
5944 (decide_unroll_stupid): Likewise.
5945 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5946 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5947 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5949 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5950 (print_loop): Print loop->unroll if set.
5951 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5952 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5953 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5955 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5956 loop->unroll is set and smaller than the trip count. Otherwise bypass
5957 entirely the heuristics if loop->unroll is set. Remove dead note.
5958 Fix off-by-one bug in other note.
5959 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5960 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5962 (tree_unroll_loops_completely): Make static.
5963 (pass_complete_unroll::execute): Use correct type for variable.
5964 (pass_complete_unrolli::execute): Fix formatting.
5965 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5967 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5969 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5970 enable when TARGET_PREFER_AVX128 is set.
5972 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5974 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5976 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5977 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5979 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5981 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5982 frequencies. Fix estimation of aggregate parameters.
5984 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5986 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5989 2017-11-23 Tom de Vries <tom@codesourcery.com>
5991 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5993 2017-11-23 Tom de Vries <tom@codesourcery.com>
5995 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5998 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
6001 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
6002 sibcall_value_pcrel_fdpic): Use local variable instead of
6004 (calli_tbr_rel): Add missing operand 2.
6005 (call_valuei_tbr_rel): Add missing operand 3.
6007 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6010 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
6011 bitpos/bitsize covering the whole destination, use store_expr only if
6012 the complex mode is the same. Otherwise, use expand_normal and if
6013 it returns CONCAT, subreg each part separately instead of trying to
6014 subreg the whole result.
6016 2017-11-23 Richard Biener <rguenther@suse.de>
6018 PR tree-optimization/23094
6019 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
6020 come here from walking over backedges in the first iteration.
6021 (vn_reference_lookup_3): Skip clobbers that store the same value.
6023 2017-11-23 Richard Biener <rguenther@suse.de>
6025 PR tree-optimization/81403
6026 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
6027 a block we need a leader relative to.
6028 (phi_translate_1): For nary processing require a leader from
6029 get_representative_for given we run expression simplification
6030 using match-and-simplify. Remove previous fix.
6032 2017-11-22 Jeff Law <law@redhat.com>
6034 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
6035 Use new method allocate_value_range rather than accessing the
6036 vrp_value_range_pool data member directly.
6037 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
6038 to use extract_range_from_stmt method to avoid need for
6039 extract_range_from_assignment method.
6040 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
6041 method rather than setting values_propgated data member directly.
6042 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
6043 and values propagated data members and extract_range_from_assignment
6044 method. Reorder private data members to conform to standards.
6045 Add new methods set_lattice_propagation_complete and
6046 allocate_value_range.
6048 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
6050 PR rtl-optimization/83030
6051 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
6052 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
6053 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
6054 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
6055 CROSSING_JUMP_P flag is already set before setting it.
6056 * cfgrtl.c (fixup_partition_crossing): Likewise.
6057 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6060 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6062 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6063 Handle the case where both arguments are using gen_const_vec_series.
6065 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6068 * pretty-print.c (pp_format): Move quoting implementation to
6069 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6070 to pass address of "quote" local.
6071 (pp_begin_quote): New function.
6072 (pp_end_quote): New function.
6073 * pretty-print.h (printer_fn): Convert penultimate param from bool
6075 (pp_begin_quote): New decl.
6076 (pp_end_quote): New decl.
6077 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6078 param from bool to bool *.
6079 * tree-diagnostic.h (default_tree_printer): Likewise.
6081 2017-11-22 Jeff Law <law@redhat.com>
6083 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6084 Initialize vr_values.
6085 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6086 vr_values via vr_values class instance rather than delegators.
6087 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6088 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6089 (evrp_range_analyzer::push_value_range): Likewise.
6090 (evrp_range_analyzer::pop_value_range): Likewise.
6091 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6092 most delegators. Those remaining are exposed as public interfaces.
6093 Make vr_values a pointer and private.
6094 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6096 (evrp_range_analyzer::get_vr_value): New method.
6097 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6098 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6099 (class evrp_dom_walker): Attach evrp_folder class, initialize
6100 it in the ctor. Remove temporary delegators.
6101 (evrp_dom_walker::before_dom_children): Call methods in attached
6102 evrp_range_analyzer class via class instance pointer. Use
6103 free value_range_constant_singleton to remove need for
6104 op_with_constant_singleton_value delegator method. Do not
6105 create a vrp_prop class instance for every call! Narrow
6106 scope of a couple variables.
6107 (evrp_dom_walker::cleanup): Call methods in attached
6108 evrp_range_analyzer class via class instance pointer.
6109 * vr-values.h (class vr_values): Privatize many methods and
6112 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6114 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6115 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6116 * expr.c (expand_expr_real_2): Likewise.
6117 * fold-const.c (const_unop): Likewise.
6118 * optabs-tree.c (optab_for_tree_code): Likewise.
6119 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6120 * tree-inline.c (estimate_operator_cost): Likewise.
6121 * tree-pretty-print.c (dump_generic_node): Likewise.
6122 (op_code_prio): Likewise.
6123 (op_symbol_code): Likewise.
6124 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6125 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6126 * internal-fn.c (direct_internal_fn_optab): New function.
6127 (direct_internal_fn_array, direct_internal_fn_supported_p
6128 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6129 * fold-const-call.c (fold_const_reduction): New function.
6130 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6132 * tree-vect-loop.c: Include internal-fn.h.
6133 (reduction_code_for_scalar_code): Rename to...
6134 (reduction_fn_for_scalar_code): ...this and return an internal
6136 (vect_model_reduction_cost): Take an internal_fn rather than
6138 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6140 (vectorizable_reduction): Use internal functions rather than tree
6141 codes for the reduction operation. Update calls to the functions
6143 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6144 Use calls to internal functions rather than REDUC tree codes.
6145 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6147 2017-11-22 Olivier Hainque <hainque@adacore.com>
6149 * config/vxworks.c (vxworks_override_options): Pick default
6150 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6151 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6152 DWARF_GNAT_ENCODINGS_DEFAULT.
6153 * config/vxworksae.h: Likewise.
6155 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6157 PR tree-optimization/83104
6158 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6161 2017-11-22 Marek Polacek <polacek@redhat.com>
6162 H.J. Lu <hongjiu.lu@intel.com>
6163 Jason Merrill <jason@redhat.com>
6168 * calls.c (initialize_argument_information): Call
6169 warn_parameter_passing_abi target hook.
6170 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6171 argument onto stack.
6172 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6173 * common.opt: Update -fabi-version description.
6174 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6175 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6177 (ix86_is_empty_record): New function.
6178 (ix86_warn_parameter_passing_abi): New function.
6179 (TARGET_EMPTY_RECORD_P): Redefine.
6180 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6181 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6182 * doc/tm.texi: Regenerated.
6183 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6184 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6185 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6186 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6187 instead of int_size_in_bytes.
6188 * expr.c (copy_blkmode_to_reg): Likewise.
6189 * function.c (aggregate_value_p): Return 0 for empty types.
6190 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6192 (locate_and_pad_parm): Call arg size_in_bytes instead
6194 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6195 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6196 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6198 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6199 (std_gimplify_va_arg_expr): Skip empty records. Call
6200 arg_size_in_bytes instead size_in_bytes.
6201 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6202 * tree-core.h (tree_type_common): Add empty_flag.
6203 (tree_decl_common): Update comments.
6204 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6206 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6207 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6209 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6210 * tree.c (default_is_empty_type): New function.
6211 (default_is_empty_record): New function.
6212 (arg_int_size_in_bytes): New function.
6213 (arg_size_in_bytes): New function.
6214 (get_ultimate_context): New function.
6215 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6216 TRANSLATION_UNIT_WARN_EMPTY_P.
6217 (default_is_empty_record, arg_int_size_in_bytes,
6218 arg_size_in_bytes, get_ultimate_context): Declare.
6220 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6222 * config/arm/arm.c (cmse_clear_registers): New function.
6223 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6224 code by call to cmse_clear_registers.
6225 (cmse_nonsecure_entry_clear_before_return): Likewise.
6227 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6229 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6230 (vdot_s32, vdotq_s32): New.
6231 (vdot_lane_u32, vdotq_lane_u32): New.
6232 (vdot_lane_s32, vdotq_lane_s32): New.
6235 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6238 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6239 for unsigned values with fewer HWIs than the precision.
6240 (test_overflow): New function.
6241 (wide_int_cc_tests): Call it.
6243 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6245 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6246 has the same mode class as Pmode.
6248 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6250 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6251 padding_bits_to_clear_ptr.
6252 (cmse_nonsecure_entry_clear_before_return): Likewise.
6254 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6256 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6257 auto_sbitap instead of integer bitfield to control register needing
6260 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6262 PR tree-optimization/83044
6263 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6264 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6265 that need the upper bound. Subtract offset from
6266 get_addr_base_and_unit_offset only if positive and subtract it
6267 before division by eltsize rather than after it.
6270 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6271 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6272 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6276 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6277 expand_binop with *mul_widen_optab, make sure at least one of the
6278 operands doesn't have VOIDmode.
6281 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6283 PR rtl-optimization/82044
6284 PR tree-optimization/82042
6285 * dse.c (record_store): Check for overflow.
6286 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6287 add_wild_read instead of clear_rhs_from_active_local_stores on
6288 overflow and log it into dump_file.
6290 2017-11-22 Richard Biener <rguenther@suse.de>
6292 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6293 fake edges to exit when looking for a place to insert.
6294 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6296 (insert_into_preds_of_block): Commit edge insertion immediately,
6297 assert that doesn't require new BBs.
6298 (fini_pre): Release expressions.
6299 (pass_pre::execute): Shuffle things around a bit, if the fn
6300 is too large do not compute AVAIL either as this is really the
6303 2017-11-22 Richard Biener <rguenther@suse.de>
6305 PR tree-optimization/83089
6306 * tree-if-conv.c (pass_if_conversion::execute): If anything
6307 changed reset SCEV and free the number of iteration estimates.
6309 2017-11-21 Martin Sebor <msebor@redhat.com>
6311 PR tree-optimization/82945
6312 * calls.h (warn_nonstring_bound): Remove unused function.
6314 2017-11-21 Martin Sebor <msebor@redhat.com>
6316 PR tree-optimization/82945
6317 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6318 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6319 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6321 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6322 * calls.h (get_attr_nonstring_decl): Declare new function.
6323 * doc/extend.texi (attribute nonstring): Update.
6324 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6325 get_attr_nonstring_decl and handle it.
6326 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6327 detection of nul-termination.
6328 (strlen_to_stridx): Change to a pointer.
6329 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6330 (pass_strlen::execute): Same.
6332 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6334 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6335 for the new option -mprefer-vector-width=[none|128|256|512].
6336 * config/i386/i386.c (ix86_target_string): remove old style options
6337 -mprefer-avx256 and make -mprefer-avx128 as alias.
6338 (ix86_option_override_internal): Apply defaults for the
6339 -mprefer-vector-width=[128|256] option.
6340 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6341 Implement macros to work with -mprefer-vector-width=.
6342 * config/i386/i386.opt: Implemented option
6343 -mprefer-vector-width=[none|128|256|512].
6344 * doc/invoke.texi: Documentation for
6345 -mprefer-vector-width=[none|128|256|512].
6347 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6349 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6351 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6353 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6355 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6357 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6359 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6360 * expr.c (expand_expr_real_2): Likewise.
6361 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6362 fold_binary_loc): Likewise.
6363 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6364 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6365 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6366 MINUS_EXPR transformations.
6367 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6368 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6369 * tree-inline.c (estimate_operator_cost): Likewise.
6370 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6371 op_symbol_code): Likewise.
6372 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6373 * vr-values.c (extract_range_from_binary_expr): Likewise.
6374 * varasm.c (initializer_constant_valid_p_1): Likewise.
6375 * tree.def: New tree code POINTER_DIFF_EXPR.
6377 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6379 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6380 integer suffix to movbe mnemonic.
6381 (*bswaphi2_movbe): Ditto.
6382 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6384 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6387 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6388 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6389 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6391 2017-11-21 Martin Liska <mliska@suse.cz>
6393 * tree-inline.c (expand_call_inline): Remove not needed
6396 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6397 Jakub Jelinek <jakub@redhat.com>
6400 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6401 Remove static keyword from f variable.
6403 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6405 PR tree-optimization/83086
6406 * gimple-ssa-store-merging.c
6407 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6408 rather than n.base_addr.
6410 2017-11-21 Martin Liska <mliska@suse.cz>
6412 PR rtl-optimization/82044
6413 PR tree-optimization/82042
6414 * dse.c (check_mem_read_rtx): Check for overflow.
6416 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6418 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6421 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6423 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6424 and its corresponding call arg location note.
6426 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6427 Andrew Burgess <andrew.burgess@embecosm.com>
6429 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6431 (arc_return_slot_offset): Likewise.
6432 (arc_eh_return_address_location): New declaration.
6433 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6434 (MUST_SAVE_REGISTER): Add exception handler case.
6435 (MUST_SAVE_RETURN_ADDR): Likewise.
6436 (arc_frame_pointer_required): Likewise.
6437 (arc_frame_pointer_needed): New function.
6438 (arc_compute_frame_size): Changed.
6439 (arc_expand_prologue): Likewise.
6440 (arc_expand_epilogue): Likewise.
6441 (arc_initial_elimination_offset): Likewise.
6442 (arc_return_slot_offset): Delete.
6443 (arc_eh_return_address_location): New function.
6444 (arc_builtin_setjmp_frame_value): Likewise.
6445 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6446 (EH_RETURN_STACKADJ_RTX): Define.
6447 (EH_RETURN_HANDLER_RTX): Likewise.
6448 * config/arc/arc.md (eh_return): Delete.
6450 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6452 * print-tree.h (debug_vec_tree): Remove prototype.
6453 * gdbinit.in (pvt): Remove macro.
6455 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6457 PR tree-optimization/83047
6458 * gimple-ssa-store-merging.c
6459 (imm_store_chain_info::output_merged_store): If the loads with the
6460 same vuse are in different basic blocks, for load_gsi pick a load
6461 location that is dominated by the other loads.
6464 * config/i386/i386.c (ix86_memmodel_check): Start
6465 -Winvalid-memory-model diagnostics with lowercase letter.
6468 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6469 set in_cold_section_p to the partition containing loc_list->first.
6470 When seeing loc_list->last_before_switch node, update secname and
6471 perform range_across_switch second partition handling only after that.
6474 * run-rtl-passes.c: Include debug.h.
6475 (run_rtl_passes): Call debug_hooks->assembly_start.
6476 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6480 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6481 OPTAB_DIRECT in calls to expand_simple_binop.
6483 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6486 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6488 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6491 * spellcheck.h (best_match::blithely_get_best_candidate): New
6494 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6496 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6497 with lower case letter.
6499 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6501 * config/i386/i386.md (bswaphi2): New expander.
6502 (*bswaphi2_movbe): New insn pattern.
6503 (bswaphi -> rorhi pepehole2): New peephole pattern.
6505 2017-11-20 Jeff Law <law@redhat.com>
6507 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6508 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6509 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6510 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6511 gimple-ssa-evrp-analyze.h.
6513 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6514 set BB_VISITED here.
6515 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6517 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6519 * tree-predcom.c: Add general comment on Store-Store chains.
6520 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6522 (get_chain_last_ref_at): Rename into...
6523 (get_chain_last_write_at): ...this.
6524 (get_chain_last_write_before_load): New function.
6525 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6526 CT_STORE_STORE when write reference is added.
6527 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6528 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6529 (initialize_root_vars_store_elim_1): Ditto.
6530 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6531 definition is created.
6532 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6533 chain by replacing it with dominant stored value.
6535 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6537 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6539 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6541 * vr-values.c (extract_range_from_binary_expr): Use a full range
6544 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6546 * config/arm/arm.md (R4_REGNUM): Define constant.
6547 (nonsecure_call_internal): Remove r4 clobber.
6548 (nonsecure_call_value_internal): Likewise.
6549 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6550 clobber and resequence match_operands.
6551 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6552 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6553 (nonsecure_call_value_reg_thumb2): Likewise.
6555 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6557 PR tree-optimization/78821
6558 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6559 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6560 to the address of the base rather than the base itself.
6561 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6562 (find_bswap_or_nop_finalize): New function.
6563 (find_bswap_or_nop): Use it.
6564 (bswap_replace): Return a tree rather than bool, change first
6565 argument from gimple * to gimple_stmt_iterator, allow inserting
6566 into an empty sequence, allow ins_stmt to be NULL - then emit
6567 all stmts into gsi. Fix up MEM_REF address gimplification.
6568 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6569 (struct store_immediate_info): Add N and INS_STMT non-static
6571 (store_immediate_info::store_immediate_info): Initialize them
6572 from newly added ctor args.
6573 (merged_store_group::apply_stores): Formatting fixes. Sort by
6575 (stmts_may_clobber_ref_p): For stores call also
6576 refs_anti_dependent_p.
6577 (gather_bswap_load_refs): New function.
6578 (imm_store_chain_info::try_coalesce_bswap): New method.
6579 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6580 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6581 (imm_store_chain_info::output_merged_store): Fail if number of
6582 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6583 and NOP_EXPR rhs_code.
6584 (pass_store_merging::process_store): Compute n and ins_stmt, if
6585 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6586 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6588 (pass_store_merging::execute): Calculate dominators.
6590 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6591 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6592 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6593 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6594 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6595 class pass_optimize_bswap, bswap_replace,
6596 pass_optimize_bswap::execute): Moved to ...
6597 * gimple-ssa-store-merging.c: ... this file.
6598 Include optabs-tree.h.
6599 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6600 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6601 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6602 anonymous namespace, remove static keywords.
6603 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6604 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6606 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6609 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6611 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6613 * vec.h (debug_helper): New function.
6614 (DEFINE_DEBUG_VEC): New macro.
6615 * hash-set.h (debug_helper): New function.
6616 (DEFINE_DEBUG_HASH_SET): New macro.
6617 * cfg.c (debug_slim (edge)): New function.
6618 Call DEFINE_DEBUG_VEC for edges.
6619 Call DEFINE_DEBUG_HASH_SET for edges.
6620 * cfghooks.c (debug_slim (basic_block)): New function.
6621 Call DEFINE_DEBUG_VEC for basic blocks.
6622 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6623 * print-tree.c (debug_slim): New function to handle trees.
6624 Call DEFINE_DEBUG_VEC for trees.
6625 Call DEFINE_DEBUG_HASH_SET for trees.
6626 (debug (vec<tree, va_gc>) &): Remove.
6627 (debug (<vec<tree, va_gc>) *): Remove.
6628 * print-rtl.c (debug_slim): New function to handle const_rtx.
6629 Call DEFINE_DEBUG_VEC for rtx_def.
6630 Call DEFINE_DEBUG_VEC for rtx_insn.
6631 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6632 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6633 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6634 (debug (vec<rtx_insn *> *ptr): Remove.
6635 (debug_insn_vector): Remove.
6636 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6638 2017-11-20 Tom de Vries <tom@codesourcery.com>
6640 PR rtl-optimization/82020
6641 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6642 IF_THEN_ELSE condition.
6644 2017-11-19 Jeff Law <law@redhat.com>
6646 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6647 of degenerates resulting from ignoring an edge.
6649 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6652 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6654 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6657 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6658 for uninitialied counts.
6660 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6663 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6665 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6668 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6670 2017-11-19 Tom de Vries <tom@codesourcery.com>
6672 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6675 2017-11-19 Tom de Vries <tom@codesourcery.com>
6677 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6680 2017-11-19 Tom de Vries <tom@codesourcery.com>
6682 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6685 2017-11-19 Tom de Vries <tom@codesourcery.com>
6687 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6689 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6691 2017-11-19 Tom de Vries <tom@codesourcery.com>
6693 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6695 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6696 semicolon after MCORE_EXPORT_NAME call.
6698 2017-11-19 Tom de Vries <tom@codesourcery.com>
6701 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6702 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6704 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6706 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6707 (cgraph_edge::clone): Cleanup updating of profile.
6708 * ipa-cp.c (update_profiling_info): Likewise.
6709 * ipa-inline-transform.c (inline_transform): Likewise.
6710 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6711 * ipa-split.c (execute_split_functions): Do not split when function
6713 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6714 * profile-count.c (profile_count::dump): Add global0.
6715 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6717 (profile_count::to_sreal_scale): Likewise.
6718 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6719 (profile_count::combine_with_ipa_count): New function.
6720 * profile-count.h (profile_guessed_global0adjusted): New.
6721 (profile_count::adjusted_zero): New.
6722 (profile_count::global0adjusted): New.
6723 (profile_count::combine_with_ipa_count): New.
6724 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6725 correct profile of return block of split functions.
6726 (copy_cfg_body): Remove unused profile_count.
6727 (copy_body): Likewise.
6728 (expand_call_inline): Update.
6729 (tree_function_versioning): Update.
6731 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6733 * hash-set.h (hash_set::empty): New.
6734 * tree-ssa-threadbackward.h: Delete.
6735 * tree-ssa-threadbackward.c (class thread_jumps): New.
6736 Move max_threaded_paths into class.
6737 (fsm_find_thread_path): Remove arguments that are now in class.
6738 (profitable_jump_thread_path): Rename to...
6739 (thread_jumps::profitable_jump_thread_path): ...this.
6740 (convert_and_register_jump_thread_path): Rename to...
6741 (thread_jumps::convert_and_register_current_path): ...this.
6742 (check_subpath_and_update_thread_path): Rename to...
6743 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6744 (register_jump_thread_path_if_profitable): Rename to...
6745 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6746 (handle_phi): Rename to...
6747 (thread_jumps::handle_phi): ...this.
6748 (handle_assignment): Rename to...
6749 (thread_jumps::handle_assignment): ...this.
6750 (fsm_find_control_statement_thread_paths): Rename to...
6751 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6752 (find_jump_threads_backwards): Rename to...
6753 (thread_jumps::find_jump_threads_backwards): ...this.
6754 Initialize path local data.
6755 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6756 from within thread_jumps class.
6757 (pass_early_thread_jumps::execute): Same.
6759 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6761 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6763 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6765 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6766 to canonical location.
6768 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6771 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6773 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6775 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6777 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6778 rcrt1.o%s/grcrt1.o%s for -static-pie.
6780 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6782 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6783 ix86_shift_rotate_cost): Break out from ...
6784 (ix86_rtx_costs): ... here.
6785 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6788 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6790 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6791 when entry block was promoted unlikely.
6792 (estimate_bb_frequencies): Increase frequency scale.
6793 * profile-count.h (profile_count): Export precision info.
6795 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6797 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6798 disturbin profile of entry block.
6800 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6802 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6805 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6807 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6808 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6810 2017-11-17 Jeff Law <law@redhat.com>
6812 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6813 from evrp_dom_walker class. Various methods moved into new class.
6814 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6815 (evrp_range_analyzer::enter): New method.
6816 (evrp_range_analyzer::leave): New method.
6817 (evrp_dom_walker): Remove delegators no longer needed by this class.
6818 Replace vr_values data member with evrp_range_analyzer
6820 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6821 method extracted from evrp_dom_walker::before_dom_children.
6822 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6823 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6825 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6826 Add private copy constructor and move assignment operators.
6827 Privatize methods and class data where trivially possible.
6828 (evrp_dom_walker::cleanup): New function, extracted from
6829 execute_early_vrp. Simplify access to class data.
6831 * vr-values.h (get_output_for_vrp): Prototype.
6832 * vr-values.c (get_output_for_vrp): New function extracted from
6833 vrp_visit_assignment_or_call and extract_range_from_stmt.
6834 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6836 2017-11-17 Luis Machado <luis.machado@linaro.org>
6838 * config/aarch64/aarch64.c
6839 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6840 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6841 tune_params::AUTOPREFETCHER_WEAK.
6843 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6846 * config/arm/arm.c (arm_valid_target_attribute_rec):
6847 Parse "arch=" and "+<ext>".
6848 (arm_valid_target_attribute_tree): Re-init global options.
6849 (arm_option_override): Make non-static.
6850 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6851 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6852 (__ARM_FEATURE_CRC32): Support undef.
6853 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6854 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6856 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6858 * gdbinit.in (break-on-diagnostic): New command.
6860 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6862 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6863 return to indirect jump for EH return if control-flow
6864 protection is enabled. Change explicit 'false' argument in
6865 pro_epilogue_adjust_stack with a value of
6867 * config/i386/i386.md (simple_return_indirect_internal):
6868 Remove SImode restriction to support 64-bit.
6870 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6872 * combine.c (added_notes_insn): New.
6873 (try_combine): Handle added_notes_insn like added_links_insn.
6874 Rewrite return value code.
6875 (distribute_notes): Set added_notes_insn to the earliest insn we added
6878 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6880 PR rtl-optimization/82621
6881 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6882 dest of one of those SETs is unused.
6884 2017-11-17 Richard Biener <rguenther@suse.de>
6887 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6888 * tree-pretty-print.c (dump_generic_node): Handle
6889 annot_expr_parallel_kind.
6890 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6891 * gimplify.c (gimple_boolify): Likewise.
6893 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6895 * config.gcc (extra_headers): Add cet.h for x86 targets.
6896 * config/i386/cet.h: New file.
6897 * doc/install.texi: Add --enable-cet/--disable-cet.
6899 2017-11-17 Richard Biener <rguenther@suse.de>
6901 PR tree-optimization/83017
6902 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6903 (gen_parallel_loop): Properly count iterations.
6904 (parallelize_loops): Handle loop->can_be_parallel independent
6905 of flag_loop_parallelize_all. Make static profitability test match
6907 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6908 * invoke.texi (parloops-min-per-thread): Document.
6910 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6912 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6913 upstreaming review comments.
6915 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6917 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6918 with fast unaligned access.
6919 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6921 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6923 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6924 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6926 2017-11-17 Richard Biener <rguenther@suse.de>
6928 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6929 folding of references.
6931 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6934 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6935 of replacing call to strncmp with corresponding call to strcmp when
6938 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6940 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6941 option prefer-avx256 for skylake-avx512 configuration.
6942 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6943 (get_builtin_code_for_version): Ditto.
6945 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6946 Monk Chiang <sh.chiang04@gmail.com>
6948 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6949 (FIXED_REGISTERS): Reserve more register numbers.
6950 (CALL_USED_REGISTERS): Likewise.
6951 (REG_ALLOC_ORDER): Likewise.
6952 (REG_CLASS_CONTENTS): Likewise.
6953 (REGISTER_NAMES): Likewise.
6955 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6956 Kito Cheng <kito.cheng@gmail.com>
6958 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6960 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6962 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6964 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6965 parameter list for vec_splats.
6967 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6969 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6971 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6973 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6976 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6977 with the required version of automake 1.11.6.
6979 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6981 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6983 (DU_C2_3_power9): Correct reservation combinations.
6984 (FP_DIV_power9, VEC_DIV_power9): New.
6985 (power9-alu): Split out rotate/shift...
6986 (power9-rot): ...to here, correct dispatch resource.
6987 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6989 (power9-fp): Correct latency.
6990 (power9-sdiv): Add div/sqrt resource.
6991 (power9-ddiv): Correct latency, add div/sqrt resource.
6992 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6993 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6995 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6997 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6999 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
7000 switch statement mapping KF built-ins to TF built-ins if we don't
7001 have the proper ISA 3.0 assembler support.
7003 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7005 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
7006 (lower_emutls_function_body): Do not compute it.
7008 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7010 * ipa-split.c (split_bb_info): Turn time to sreal.
7011 (split_point): Likewise.
7012 (dump_split_point): Likewise.
7013 (fine_split_points): Likewise.
7014 (execute_split_functions): Only zero split_bbs; turn time to sreals.
7016 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7018 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
7021 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7023 * predict.c (combine_predictions_for_bb): Preserve zero predicted
7025 (expensive_function_p): Remove useless assert.
7026 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
7028 2017-11-16 Martin Sebor <msebor@redhat.com>
7030 PR tree-optimization/82588
7031 PR tree-optimization/82583
7032 * tree-vrp.c (check_array_ref): Handle flexible array members,
7033 string literals, and inner indices.
7034 (search_for_addr_array): Add detail to diagnostics.
7036 2017-11-16 Nathan Sidwell <nathan@acm.org>
7040 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
7041 (SET_DECL_ASSEMBLER_NAME): Forward to
7042 overwrite_decl_assembler_name.
7043 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
7044 (overwrite_decl_assembler_name): Declare.
7045 * tree.c (overwrite_decl_assembler_name): New.
7046 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
7047 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
7048 (LANG_HOOKS_INITIALIZER): Add it.
7049 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
7050 * langhooks.c (lhd_set_decl_assembler_name): Use
7051 SET_DECL_ASSEMBLER_NAME.
7052 (lhd_overwrite_decl_assembler_name): Default implementation.
7054 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7055 Jackson Woodruff <jackson.woodruff@arm.com>
7057 PR tree-optimization/71026
7058 * match.pd: Canonicalize constant multiplies in division.
7060 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7062 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7063 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7064 Remove OPT_fomit_frame_pointer entry.
7065 * common/config/arc/arc-common.c: Likewise.
7066 * common/config/arm/arm-common.c: Likewise.
7067 * common/config/avr/avr-common.c: Likewise.
7068 * common/config/c6x/c6x-common.c: Likewise.
7069 * common/config/cr16/cr16-common.c: Likewise.
7070 * common/config/cris/cris-common.c: Likewise.
7071 * common/config/epiphany/epiphany-common.c: Likewise.
7072 * common/config/fr30/fr30-common.c: Likewise.
7073 * common/config/frv/frv-common.c: Likewise.
7074 * common/config/ia64/ia64-common.c: Likewise.
7075 * common/config/iq2000/iq2000-common.c: Likewise.
7076 * common/config/lm32/lm32-common.c: Likewise.
7077 * common/config/m32r/m32r-common.c: Likewise.
7078 * common/config/mcore/mcore-common.c: Likewise.
7079 * common/config/microblaze/microblaze-common.c: Likewise.
7080 * common/config/mips/mips-common.c: Likewise.
7081 * common/config/mmix/mmix-common.c: Likewise.
7082 * common/config/mn10300/mn10300-common.c: Likewise.
7083 * common/config/nios2/nios2-common.c: Likewise.
7084 * common/config/pa/pa-common.c: Likewise.
7085 * common/config/pdp11/pdp11-common.c: Likewise.
7086 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7087 * common/config/riscv/riscv-common.c: Likewise.
7088 * common/config/rs6000/rs6000-common.c: Likewise.
7089 * common/config/rx/rx-common.c: Likewise.
7090 * common/config/s390/s390-common.c: Likewise.
7091 * common/config/sh/sh-common.c: Likewise.
7092 * common/config/sparc/sparc-common.c: Likewise.
7093 * common/config/tilegx/tilegx-common.c: Likewise.
7094 * common/config/tilepro/tilepro-common.c: Likewise.
7095 * common/config/v850/v850-common.c: Likewise.
7096 * common/config/visium/visium-common.c: Likewise.
7097 * common/config/xstormy16/xstormy16-common.c: Likewise.
7098 * common/config/xtensa/xtensa-common.c: Likewise.
7099 * invoke.texi (-fomit-frame-pointer): Update documentation.
7101 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7103 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7105 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7107 * cfg.c (scale_bbs_frequencies_int,
7108 cale_bbs_frequencies_gcov_type): Remove.
7109 * cfg.h (scale_bbs_frequencies_int,
7110 cale_bbs_frequencies_gcov_type): Remove.
7112 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7114 * tree-ssa-loop-manip.c
7115 (scale_dominated_blocks_in_loop): Update to profile counts.
7116 (tree_transform_and_unroll_loop): Likewise.
7118 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7120 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7121 scale_bbs_frequencies_int.
7123 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7125 * final.c (compute_alignments): Use counts rather than frequencies.
7127 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7129 * cfgloopanal.c: Include sreal.h
7130 (average_num_loop_insns): Use counts and sreal for accounting.
7132 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7134 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7137 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7138 Kito Cheng <kito.cheng@gmail.com>
7140 * config/nds32/constraints.md: Provide more constraints.
7141 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7142 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7143 support constraints usage.
7145 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7147 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7149 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7150 Kito Cheng <kito.cheng@gmail.com>
7152 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7153 * config/nds32/nds32.opt: Refine the layout.
7154 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7155 TARGET_EXT_STRING): Support new options.
7156 * config/nds32/nds32.h: Likewise.
7157 * config/nds32/nds32.md: Likewise.
7158 * config/nds32/nds32-predicates.c: Likewise.
7159 * config/nds32/constraints.md: Likewise.
7160 * common/config/nds32/nds32-common.c: Likewise.
7162 2017-11-16 Julia Koval <julia.koval@intel.com>
7165 * config/i386/gfniintrin.h: Add sse check.
7166 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7168 2017-11-16 Julia Koval <julia.koval@intel.com>
7170 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7171 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7172 (ix86_handle_option): Handle -mavx512vbmi2.
7173 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7174 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7175 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7176 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7177 (ix86_valid_target_attribute_inner_p): Ditto.
7178 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7179 * config/i386/i386.opt (mavx512vbmi2): New option.
7180 * doc/invoke.texi: Add new option.
7182 2017-11-16 Julia Koval <julia.koval@intel.com>
7184 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7185 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7186 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7187 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7188 _mm512_gf2p8mul_epi8): New intrinsics.
7189 * config/i386/i386-builtin-types.def
7190 (V64QI_FTYPE_V64QI_V64QI): New type.
7191 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7192 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7193 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7194 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7195 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7196 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7198 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7200 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7201 explicitly as a stream of bytes.
7203 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7205 * config/rs6000/altivec.h (vec_xst_be): New #define.
7206 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7207 and externalize from *altivec_vperm_<mode>_internal.
7208 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7210 (XL_BE_V8HI): Likewise.
7211 (XL_BE_V4SI): Likewise.
7212 (XL_BE_V4SI): Likewise.
7213 (XL_BE_V2DI): Likewise.
7214 (XL_BE_V4SF): Likewise.
7215 (XL_BE_V2DF): Likewise.
7216 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7217 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7218 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7219 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7220 VSX_BUILTIN_VEC_XST_BE.
7221 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7222 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7224 (altivec_init_builtins): Replace conditional calls to def_builtin
7225 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7226 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7227 with unconditional calls. Remove calls to def_builtin for
7228 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7229 __builtin_vec_xst_be.
7230 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7231 to define_expand, and add alternate RTL generation for P8.
7232 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7233 vsx_ld_elemrev_v8hi.
7234 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7235 add alternate RTL generation for P8.
7236 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7237 vsx_ld_elemrev_v16qi.
7238 (vsx_st_elemrev_v8hi): Convert define_insn
7239 to define_expand, and add alternate RTL generation for P8.
7240 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7241 vsx_st_elemrev_v8hi.
7242 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7243 add alternate RTL generation for P8.
7244 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7245 vsx_st_elemrev_v16qi.
7247 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7250 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7251 TARGET_AVX512ER check.
7252 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7253 neither -mzeroupper nor -mno-zeroupper is used and
7254 TARGET_EMIT_VZEROUPPER is set.
7255 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7256 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7258 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7260 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7261 folding of vector compares.
7262 (fold_build_vec_cmp): New helper function.
7263 (fold_compare_helper): New helper function.
7264 (builtin_function_type): Add compare builtins to the list of functions
7265 having unsigned arguments. Cosmetic updates to comment indentation.
7266 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7267 the not+eq combination.
7269 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7271 PR tree-optimization/82726
7272 PR tree-optimization/70754
7273 * tree-predcom.c (order_drefs_by_pos): New function.
7274 (combine_chains): Move code setting has_max_use_after to...
7275 (try_combine_chains): ...here. New parameter. Sort combined chains
7276 according to position information.
7277 (tree_predictive_commoning_loop): Update call to above function.
7278 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7280 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7282 PR tree-optimization/82726
7284 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7286 PR tree-optimization/70754
7287 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7288 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7289 combined stmt before it if not NULL.
7290 (combine_chains): Process refs reversely and compute dominance point
7294 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7296 PR tree-optimization/79663
7297 * tree-predcom.c (combine_chains): Process refs in reverse order
7298 only for ZERO length chains, and add explaining comment.
7300 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7302 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7303 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7304 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7306 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7308 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7310 2017-11-15 Martin Liska <mliska@suse.cz>
7312 * tree-cfg.c (pass_warn_function_return::execute):
7313 Compare warn_return_type for greater than zero.
7315 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7319 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7320 to return true on Xeon and not on Xeon Phi.
7321 (ix86_check_avx256_register): Changed to ...
7322 (ix86_check_avx_upper_register): ... this. Add extra check for
7323 VALID_AVX512F_REG_OR_XI_MODE.
7324 (ix86_avx_u128_mode_needed): Changed
7325 ix86_check_avx256_register to ix86_check_avx_upper_register.
7326 (ix86_check_avx256_stores): Changed to ...
7327 (ix86_check_avx_upper_stores): ... this. Changed
7328 ix86_check_avx256_register to ix86_check_avx_upper_register.
7329 (ix86_avx_u128_mode_after): Changed
7330 avx_reg256_found to avx_upper_reg_found. Changed
7331 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7332 (ix86_avx_u128_mode_entry): Changed
7333 ix86_check_avx256_register to ix86_check_avx_upper_register.
7334 (ix86_avx_u128_mode_exit): Ditto.
7335 * config/i386/i386.h: (host_detect_local_cpu): New define.
7337 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7339 * config/arm/xgene1.md (xgene1): Split into automatons
7340 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7341 (xgene1_f_load): Adjust reservations and/or types.
7342 (xgene1_f_store): Likewise.
7343 (xgene1_load_pair): Likewise.
7344 (xgene1_store_pair): Likewise.
7345 (xgene1_fp_load1): Likewise.
7346 (xgene1_load1): Likewise.
7347 (xgene1_store1): Likewise.
7348 (xgene1_move): Likewise.
7349 (xgene1_alu): Likewise.
7350 (xgene1_simd): Likewise.
7351 (xgene1_bfm): Likewise.
7352 (xgene1_neon_load1): Likewise.
7353 (xgene1_neon_store1): Likewise.
7354 (xgene1_neon_logic): Likewise.
7355 (xgene1_neon_st1): Likewise.
7356 (xgene1_neon_ld1r): Likewise.
7357 (xgene1_alu_cond): Added.
7358 (xgene1_shift_reg): Likwise.
7359 (xgene1_bfx): Likewise.
7360 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7362 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7365 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7367 (can_widen_mult_without_libcall): New function.
7368 (expand_mul_overflow): If only checking unsigned mul overflow,
7369 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7370 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7371 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7372 (expand_DIVMOD): Formatting fix.
7373 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7374 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7375 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7377 PR tree-optimization/82977
7378 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7379 constructed temporary to strlen_to_stridx.put.
7381 2017-11-15 Martin Liska <mliska@suse.cz>
7383 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7384 * configure: Regenerate.
7386 2017-11-15 Martin Liska <mliska@suse.cz>
7389 * config/sh/sh-mem.cc: Use proper probability for
7392 2017-11-14 Jeff Law <law@redhat.com>
7394 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7395 the red zone for stack_clash_protection_final_dynamic_probe targets
7396 when the total dynamic stack size is zero bytes.
7398 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7399 blocks is post order.
7401 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7403 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7404 * final.c (rest_of_clean_state): Set it for the
7405 -fcompare-debug dump.
7406 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7407 class when TDF_COMPARE_DEBUG is set.
7409 * dwarf2out.c (gen_producer_string): Discard
7412 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7415 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7416 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7417 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7418 (__TGMATH_REAL_2_3): Remove macros.
7419 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7420 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7422 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7424 (remquo): Define using __TGMATH_REAL_3.
7426 2017-11-14 Jeff Law <law@redhat.com>
7428 * vr-values.c: New file with contents extracted from tree-vrp.c.
7429 * Makefile.in (OBJS): Add vr-values.o
7430 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7431 (set_value_range, set_and_canonicalize_value_range): Likewise.
7432 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7433 (value_range_constant_singleton, symbolic_range_p): Likewise.
7434 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7435 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7436 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7437 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7438 (find_case_label_range, find_case_label_index): Likewise.
7439 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7440 (range_int_cst_singleton_p, value_inside_range): Likewise.
7441 (get_single_symbol): Likewise.
7442 (switch_update): Move structure definition here.
7443 (to_remove_edges, to_update_switch_stmts): Provide externs.
7444 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7445 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7446 (vrp_val_is_min, set_value_range): Likewise.
7447 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7448 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7449 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7450 (range_is_nonnull, range_int_cst_p): Likewwise.
7451 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7452 (get_single_symbol, operand_less_p): Likewise
7453 (compare_values_warnv, compare_values): Likewise.
7454 (value_inside_range, value_range_constant_singleton): Likewise.
7455 (zero_nonzero_bitgs_from_vr): Likewise.
7456 (extract_range_from_binary_expr_1): Likewise.
7457 (overflow_comparison_p): Likewise.
7458 (to_remove_edges, to_update_switch_stmts): Likewise.
7459 (find_case_label-index, find_case_label_range): Likewise.
7460 (switch_update, set_value_range_to_nonnegative): Remove.
7461 (set_value_range_to_truthvalue): Likewise.
7462 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7463 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7464 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7465 (find_case_label_ranges, test_for_singularity): Likewise.
7466 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7467 (x_vr_values): Move to its remaining use site.
7469 2017-11-10 Jeff Law <law@redhat.com>
7471 * vr-values.h (VR_INITIALIZER): Move #define here.
7472 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7473 * Makefile.in (OBJS): Add tree-evrp.o
7474 * tree-vrp.h (assert_info): Move structure definition here.
7475 (set_value_range_to_varying): Prototype.
7476 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7477 (infer_value_range, register_edge_assert_for): Likewise.
7478 (stmt_interesting_for_vrp): Likewise.
7479 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7480 (set_value_range_to_varying): No longer static.
7481 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7482 (infer_value_range, register_edge_assert_for): Likewise.
7484 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7486 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7487 generate the XXBRD instruction.
7489 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7490 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7491 as being compatible if -mabi=ieeelongdouble.
7492 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7493 to setup float128 built-ins with hardware support.
7494 (BU_FLOAT128_HW_2): Likewise.
7495 (BU_FLOAT128_HW_3): Likewise.
7496 (BU_FLOAT128_HW_VSX_1): Likewise.
7497 (BU_FLOAT128_HW_VSX_2): Likewise.
7498 (scalar_extract_expq): Change float128 built-in functions to
7499 accommodate having both KFmode and TFmode functions. Use the
7500 KFmode variant as the default.
7501 (scalar_extract_sigq): Likewise.
7502 (scalar_test_neg_qp): Likewise.
7503 (scalar_insert_exp_q): Likewise.
7504 (scalar_insert_exp_qp): Likewise.
7505 (scalar_test_data_class_qp): Likewise.
7506 (sqrtf128_round_to_odd): Delete processing the round to odd
7507 built-in functions as special built-in functions, and define them
7508 as float128 built-ins. Use the KFmode variant as the default.
7509 (truncf128_round_to_odd): Likewise.
7510 (addf128_round_to_odd): Likewise.
7511 (subf128_round_to_odd): Likewise.
7512 (mulf128_round_to_odd): Likewise.
7513 (divf128_round_to_odd): Likewise.
7514 (fmaf128_round_to_odd): Likewise.
7515 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7516 support for KFmode and TFmode xststdcqp calls.
7517 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7518 point, switch the built-in handlers for the get/set float128
7519 exponent, get float128 mantissa, float128 test built-ins, and the
7520 float128 round to odd built-in functions. Eliminate creating the
7521 float128 round to odd built-in functions as special built-ins.
7522 (rs6000_init_builtins): Eliminate special creation of the float128
7523 round to odd built-in functions.
7524 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7525 function insns to support both TFmode and KFmode variants.
7526 (xsxsigqp_<mode>): Likewise.
7527 (xsiexpqpf_<mode>): Likewise.
7528 (xsiexpqp_<mode>): Likewise.
7529 (xststdcqp_<mode>): Likewise.
7530 (xststdcnegqp_<mode>): Likewise.
7531 (xststdcqp_<mode>): Likewise.
7533 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7535 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7536 unused path_in_freq_ptr parameter.
7537 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7539 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7541 * ipa-inline.c (edge_badness): Dump sreal frequency.
7542 (compute_inlined_call_time): Match natural implementaiton ...
7543 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7544 forgotten division by CGRAPH_FREQ_BASE.
7546 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7548 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7549 Solaris 11. Update comment.
7550 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7552 * config/sol2.h (STARTFILE_SPEC): Likewise.
7553 * configure: Regenerate.
7555 2017-11-14 Carl Love <cel@us.ibm.com>
7557 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7558 le_ and be_ prefixes to swap* variables. Remove
7559 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7562 2017-11-14 Jason Merrill <jason@redhat.com>
7564 Support GTY((cache)) on hash_map.
7565 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7566 (ggc_cache_remove): Override it instead of ggc_mx.
7567 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7568 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7569 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7570 (simple_cache_map_traits): Override maybe_mx.
7571 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7572 (hash_map): Friend gt_cleare_cache.
7573 (gt_cleare_cache): New.
7574 * tree.h (tree_cache_traits): New hash_map traits class.
7575 (tree_cache_map): New typedef.
7577 2017-11-14 Richard Biener <rguenther@suse.de>
7579 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7580 paramter and handling.
7581 (cleanup_control_flow_bb): Likewise.
7582 (cleanup_control_flow_pre): New helper performing a DFS walk
7583 to call cleanup_control_flow_bb in PRE order.
7584 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7585 via cleanup_control_flow_pre.
7587 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7589 * config/aarch64/aarch64-simd.md
7590 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7591 (*aarch64_simd_bsl<mode>_alt): Likewise.
7592 (aarch64_simd_bsldi_internal): New.
7593 (aarch64_simd_bsldi_alt): Likewise.
7595 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7597 * tracer.c (better_p): Do not compare frequencies.
7598 * reg-stack.c (better_edge): Likewise.
7599 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7602 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7604 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7605 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7606 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7607 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7608 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7609 * predict.c (maybe_hot_frequency_p): Inline to ...
7610 (maybe_hot_count_p): ... here; rewrite to counts.
7611 (counts_to_freqs): Rename to ...
7612 (update_max_bb_count): ... this one.
7613 (expensive_function_p): Use counts.
7614 (estimate_bb_frequencies): Update.
7615 (rebuild_frequencies): Update.
7616 * predict.h (counts_to_freqs): Rename to ...
7617 (update_max_bb_count): ... this one.
7618 * profile.c (compute_branch_probabilities): Add debug info
7619 * tree-inline.c (expand_call_inline): Update debug info.
7620 (optimize_inline_calls): Use update_max_bb_count..
7621 (tree_function_versioning): Use update_max_bb_count..
7622 * value-prof.c (gimple_value_profile_transformations):
7623 Do not use update_max_bb_count.
7625 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7627 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7628 always use frequencies.
7630 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7632 * bb-reorder.c: Remove frequencies from comments.
7633 (better_edge_p): Use profile counts.
7634 (find_traces): Dump profile counts.
7635 (rotate_loop): Use profile counts.
7636 (find_traces_1_round): Likewise.
7637 (connect_better_edge_p): Use counts instead of probabilities for
7639 (copy_bb_p): Drop early check for non-0 frequency.
7640 (sanitize_hot_paths): Update comments.
7642 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7644 * ipa-split.c (struct split_point): Add count.
7645 (consider_split): Do not compute incoming frequency; compute incoming
7646 count and store it to split_point.
7647 (split_function): Set count of the call to split part correctly.
7649 2017-11-13 Carl Love <cel@us.ibm.com>
7651 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7653 2017-11-13 Tom Tromey <tom@tromey.com>
7655 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7657 2017-11-13 Carl Love <cel@us.ibm.com>
7659 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7660 Add support for builtins:
7661 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7662 vector {un,}signed {char,int,short},
7663 vector {un,}signed {char,int,short}) arguments.
7664 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7665 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7666 Add BU_P9V_AV_2 expansions for the builtins.
7667 * config/rs6000/altivec.h (vec_first_match_index,
7668 vec_first_mismatch_index, vec_first_match_or_eos_index,
7669 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7670 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7671 new extern declaration.
7672 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7673 * config/rs6000/vsx.md (first_match_index_<mode>,
7674 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7675 first_mismatch_or_eos_index_<mode>): Add define expand.
7676 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7677 * doc/extend.texi: Update the built-in documenation file for the new
7680 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7682 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7683 into the min/max operations for _Float<N> and _Float<N>X types.
7685 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7688 * dwarf2out.c (do_eh_frame): New static variable.
7689 (dwarf2out_begin_prologue): Set it.
7690 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7692 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7694 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7696 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7698 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7701 * debug.h (dwarf2out_do_eh_frame): Declare.
7702 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7703 (dwarf2out_do_frame): Use it.
7704 (dwarf2out_do_cfi_asm): Likewise.
7705 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7706 (dwarf2out_assembly_start): Likewise.
7707 (dwarf2out_begin_prologue): Fix comment.
7708 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7709 if the target needs either debug or unwind DWARF2 info.
7710 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7711 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7712 -ftrapping-math, -ftrapv and -fwrapv.
7714 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7716 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7717 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7718 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7719 (estimate_edge_size_and_time): Likewise.
7720 (ipa_merge_fn_summary_after_inlining): Likewise.
7721 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7722 (compute_uninlined_call_time): Use sreal_frequency.
7723 (compute_inlined_call_time): Likewise.
7724 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7725 * profile-count.c: Include sreal.h.
7726 (profile_count::to_sreal_scale): New.
7727 * profile-count.h: Forward declare sreal.
7728 (profile_count::to_sreal_scale): Declare.
7730 2017-11-13 Nathan Sidwell <nathan@acm.org>
7732 * diagnostic.c (maybe_line_and_column): New.
7733 (diagnostic_get_location_text): Use it.
7734 (diagnostic_report_current_module): Likewise.
7735 (test_diagnostic_get_location_text): Add tests.
7737 2017-11-13 Luis Machado <luis.machado@linaro.org>
7739 * doc/md.texi (Specifying processor pipeline description): Fix
7740 incorrect latency for the div instruction example.
7742 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7744 PR tree-optimization/78821
7745 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7746 that bit_not_p is the same.
7747 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7748 (split_group): Count precisely bit_not_p bits in each statement.
7749 (invert_op): New function.
7750 (imm_store_chain_info::output_merged_store): Use invert_op to
7751 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7752 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7754 2017-11-13 Martin Liska <mliska@suse.cz>
7756 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7757 (struct source_info): Likewise.
7758 (add_branch_counts): Likewise.
7759 (add_line_counts): Likewise.
7760 (function_summary): Likewise.
7761 (output_intermediate_line): Likewise.
7762 (generate_results): Likewise.
7764 2017-11-13 Martin Liska <mliska@suse.cz>
7766 * gcov.c (struct block_info): Remove typedef for block_t.
7767 (struct line_info): Likewise.
7768 (line_info::has_block): Likewise.
7769 (EXIT_BLOCK): Likewise.
7770 (unblock): Likewise.
7771 (circuit): Likewise.
7772 (get_cycles_count): Likewise.
7773 (process_file): Likewise.
7774 (read_graph_file): Likewise.
7775 (solve_flow_graph): Likewise.
7776 (find_exception_blocks): Likewise.
7777 (add_line_counts): Likewise.
7778 (accumulate_line_info): Likewise.
7779 (output_line_details): Likewise.
7781 2017-11-13 Martin Liska <mliska@suse.cz>
7783 * gcov.c (struct arc_info): Remove typedef for arc_t.
7784 (struct line_info): Likewise.
7785 (add_branch_counts): Likewise.
7786 (output_branch_count): Likewise.
7787 (function_info::~function_info): Likewise.
7788 (circuit): Likewise.
7789 (output_intermediate_line): Likewise.
7790 (read_graph_file): Likewise.
7791 (solve_flow_graph): Likewise.
7792 (find_exception_blocks): Likewise.
7793 (add_line_counts): Likewise.
7794 (accumulate_line_info): Likewise.
7795 (output_line_details): Likewise.
7796 (output_function_details): Likewise.
7798 2017-11-13 Martin Liska <mliska@suse.cz>
7800 * gcov.c (struct function_info): Remove typedef for function_t.
7801 (struct source_info): Likewise.
7802 (source_info::get_functions_at_location): Likewise.
7803 (solve_flow_graph): Likewise.
7804 (find_exception_blocks): Likewise.
7805 (add_line_counts): Likewise.
7806 (output_intermediate_file): Likewise.
7807 (process_file): Likewise.
7808 (generate_results): Likewise.
7809 (release_structures): Likewise.
7810 (read_graph_file): Likewise.
7811 (read_count_file): Likewise.
7812 (accumulate_line_counts): Likewise.
7813 (output_lines): Likewise.
7815 2017-11-13 Martin Liska <mliska@suse.cz>
7817 * gcov.c (function_info::function_info): Remove num_counts
7818 and add vector<gcov_type>.
7819 (function_info::~function_info): Use the vector.
7820 (process_file): Likewise.
7821 (read_graph_file): Likewise.
7822 (read_count_file): Likewise.
7823 (solve_flow_graph): Likewise.
7825 2017-11-13 Martin Liska <mliska@suse.cz>
7827 * gcov.c (function_info::is_artificial): New function.
7828 (process_file): Erase all artificial early.
7829 (generate_results): Skip as all artificial are already
7832 2017-11-13 Martin Liska <mliska@suse.cz>
7834 * gcov.c (read_graph_file): Store to global vector of functions.
7835 (read_count_file): Iterate the vector.
7836 (process_file): Likewise.
7837 (generate_results): Likewise.
7838 (release_structures): Likewise.
7840 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7842 PR tree-optimization/82954
7843 * gimple-ssa-store-merging.c
7844 (imm_store_chain_info::coalesce_immediate_stores): If
7845 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7847 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7849 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7850 Upddate call to ENDIAN_LANE_N.
7851 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7852 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7853 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7854 and use aarch64_endian_lane_rtx.
7855 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7857 2017-11-12 Tom de Vries <tom@codesourcery.com>
7859 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7861 2017-11-12 Tom de Vries <tom@codesourcery.com>
7863 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7865 2017-11-12 Tom de Vries <tom@codesourcery.com>
7867 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7869 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7870 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7871 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7873 2017-11-11 Martin Sebor <msebor@redhat.com>
7876 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7879 * prefic.c (translate_name): Replace strncpy with memcpy to
7880 avoid -Wstringop-truncation.
7882 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7884 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7886 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7888 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7890 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7892 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7895 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7897 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7898 enable generating XXBRH if the value is in a vector register.
7899 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7900 value is in a vector register.
7901 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7902 register to register bswap64's instead of doing the GPR sequence
7903 used on previous machines.
7904 (bswapdi2_xxbrd): New insn.
7905 (bswapdi2_reg): Disallow on ISA 3.0.
7906 (register to register bswap64 splitter): Do not split the insn on
7907 ISA 3.0 systems that use XXBRD.
7909 2017-11-10 Martin Sebor <msebor@redhat.com>
7912 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7914 (find_subframework_file): Same.
7916 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7918 * auto-profile.c (afdo_indirect_call): Drop frequency.
7919 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7920 (cgraph_node::create_edge): Drop frequency argument.
7921 (cgraph_node::create_indirect_edge): Drop frequency argument.
7922 (cgraph_edge::make_speculative): Drop frequency arguments.
7923 (cgraph_edge::resolve_speculation): Do not update frequencies
7924 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7925 (cgraph_node::dump): Check consistency in IPA mode.
7926 (cgraph_edge::maybe_hot_p): Use IPA counter.
7927 (cgraph_edge::verify_count_and_frequency): Rename to ...
7928 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7929 (cgraph_node::verify_node): Update.
7930 * cgraph.h (struct cgraph_edge): Drop frequency.
7931 (cgraph_edge::frequency): New function.
7932 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7934 (cgraph_edge::rebuild_edges): Likewise.
7935 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7936 (duplicate_thunk_for_node): Do not pass frequency.
7937 (cgraph_node::create_clone): Scale only counts.
7938 (cgraph_node::create_virtual_clone): Do not pass frequency.
7939 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7940 (cgraph_node::create_version_clone): Do not pass frequency.
7941 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7942 (cgraph_node::expand_thunk): Do not pass frequency.
7943 (cgraph_node::create_wrapper): Do not pass frequency.
7944 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7946 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7947 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7948 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7949 (gather_caller_stats): Use frequency function.
7950 (ipcp_cloning_candidate_p): Use frequency function.
7951 (ipcp_propagate_stage): Use frequency function.
7952 (get_info_about_necessary_edges): Use frequency function.
7953 (update_profiling_info): Update only IPA profile.
7954 (update_specialized_profile): Use frequency functoin.
7955 (perhaps_add_new_callers): Update only IPA profile.
7956 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7957 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7958 (dump_ipa_call_summary): Use frequency function.
7959 (estimate_edge_size_and_time): Use frequency function.
7960 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7961 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7962 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7963 (update_noncloned_counts): ... ths one; scale counts only.
7964 (clone_inlined_nodes): Do not scale frequency.
7965 (inline_call): Do not pass frequency.
7966 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7967 (compute_inlined_call_time): Use IPA profile.
7968 (want_inline_small_function_p): Use IPA profile.
7969 (want_inline_self_recursive_call_p): Use IPA profile.
7970 (edge_badness): Use IPA profile.
7971 (lookup_recursive_calls): Use IPA profile.
7972 (recursive_inlining): Do not pass frequency.
7973 (resolve_noninline_speculation): Do not update frequency.
7974 (inline_small_functions): Collect max of IPA profile.
7975 (dump_overall_stats): Dump IPA porfile.
7976 (dump_inline_stats): Dump IPA porfile.
7977 (ipa_inline): Collect IPA stats.
7978 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7979 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7980 (ipa_propagate_frequency): Use frequency function.
7981 (ipa_profile): Cleanup.
7982 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7983 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7984 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7985 (input_node): Do not stream frequency.
7986 (input_edge): Do not stream frequency.
7987 (merge_profile_summaries): Scale only IPA profiles.
7988 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7989 * predict.c (drop_profile): Do not recompute frequency.
7990 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7991 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7992 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7993 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7994 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7995 * tree-inline.c (copy_bb): Do not scale frequency.
7996 (expand_call_inline): Do not scale frequency.
7997 (tree_function_versioning): Do not scale frequency.
7998 * ubsan.c (ubsan_create_edge): Do not pass frequency.
8000 2017-11-10 Julia Koval <julia.koval@intel.com>
8002 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
8003 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
8004 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
8005 (_mm256_maskz_gf2p8affine_epi64_epi8)
8006 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
8007 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
8008 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
8009 (__builtin_ia32_vgf2p8affineqb_v32qi)
8010 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
8011 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
8013 2017-11-10 Tamar Christina <tamar.christina@arm.com>
8017 (arm_option_override): Refactor.
8018 (arm_option_reconfigure_globals): New.
8019 (arm_options_perform_arch_sanity_checks): New.
8020 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
8022 (arm_options_perform_arch_sanity_checks): Likewise
8024 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
8026 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
8027 (power9-qpmul): New.
8028 * rs6000/rs6000.md ("type" attr): Add qmul.
8029 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
8030 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
8031 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
8033 2017-11-10 Martin Sebor <msebor@redhat.com>
8036 * builtins.c (compute_objsize): Handle arrays that
8037 compute_builtin_object_size likes to fail for. Make extern.
8038 * builtins.h (compute_objsize): Declare.
8039 (check_strncpy_sizes): New function.
8040 (expand_builtin_strncpy): Call check_strncpy_sizes.
8041 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
8042 -Wstringop-truncation.
8043 (gimple_fold_builtin_strncat): Same.
8044 * gimple.c (gimple_build_call_from_tree): Set call location.
8045 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
8046 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
8047 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
8048 (handle_builtin_strlen): Use strlen_to_stridx.
8049 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
8051 Use strlen_to_stridx.
8052 (pass_strlen::execute): Release strlen_to_stridx.
8053 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
8054 (-Wstringop-truncation): Document new option.
8056 2017-11-10 Martin Liska <mliska@suse.cz>
8058 PR gcov-profile/82702
8059 * gcov.c (main): Handle intermediate files in a different
8061 (get_gcov_intermediate_filename): New function.
8062 (output_gcov_file): Remove support of intermediate files.
8063 (generate_results): Allocate intermediate file.
8064 (release_structures): Clean-up properly fn_end.
8065 (output_intermediate_file): Start iterating with line 1.
8067 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8069 PR tree-optimization/82929
8070 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8071 ops_swapped_p non-static data member.
8072 (store_immediate_info::store_immediate_info): Clear it.
8073 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8074 ops set ops_swapped_p.
8075 (count_multiple_uses): Handle ops_swapped_p.
8077 2017-11-10 Martin Liska <mliska@suse.cz>
8079 * coverage.c (coverage_init): Stream information about
8080 support of has_unexecuted_blocks.
8081 * doc/gcov.texi: Document that.
8082 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8083 * gcov.c (read_graph_file): Likewise.
8084 (output_line_beginning): Fix a small issue with
8087 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8089 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8090 reference of trivial component.
8092 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8095 * gimple-ssa-store-merging.c
8096 (pass_store_merging::terminate_all_aliasing_chains): For
8097 gimple_store_p stmts also call refs_output_dependent_p.
8099 PR rtl-optimization/82913
8100 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8103 2017-11-09 Jeff Law <law@redhat.com>
8105 * vr-values.h: New file with vr_values class.
8106 * tree-vrp.c: Include vr-values.h
8107 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8108 data objects into the vr_values class.
8109 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8110 (get_value_range): Make it a member function within vr_values class.
8111 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8112 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8113 (op_with_constant_singleton_value_range): Likewise.
8114 (extract_range_for_var_from_comparison_expr): Likewise.
8115 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8116 (extract_range_from_binary_expr): Likewise.
8117 (extract_range_from_unary_expr): Likewise.
8118 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8119 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8120 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8121 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8122 (compare_name_with_value, compare_names): Likewise.
8123 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8124 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8125 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8126 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8127 (extract_range_from_phi_node): Likewise.
8128 (simplify_truth_ops_using_ranges): Likewise.
8129 (simplify_div_or_mod_using_ranges): Likewise.
8130 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8131 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8132 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8133 (simplify_float_conversion_using_ranges): Likewise.
8134 (simplify_internal_call_using_ranges): Likewise.
8135 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8136 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8137 poitner via x_vr_values for calls into gimple folder.
8138 (vrp_initialize_lattice): Make this the vr_values ctor.
8139 (vrp_free_lattice): Make this the vr_values dtor.
8140 (set_vr_value): New function.
8141 (class vrp_prop): Add vr_values data member. Add various member
8142 functions as well as member functions that delegate to vr_values.
8143 (check_array_ref): Make a member function within vrp_prop class.
8144 (search_for_addr_array, vrp_initialize): Likewise.
8145 (vrp_finalize): Likewise. Revamp to avoid direct access to
8146 vr_value, values_propagated, etc.
8147 (check_array_bounds): Extract vrp_prop class instance pointer from
8148 walk info structure. Use it to call member functions.
8149 (check_all_array_refs): Make a member function within vrp_prop class.
8150 Smuggle class instance pointer via walk info structure.
8151 (x_vr_values): New local static.
8152 (vrp_valueize): Use x_vr_values to get class instance.
8153 (vr_valueize_1): Likewise.
8154 (class vrp_folder): Add vr_values data member. Add various member
8155 functions as well as member functions that delegate to vr_values.
8156 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8157 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8158 class instance from vr_values. Use it to call member functions.
8159 (vrp_dom_walker): Add vr_values data member.
8160 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8161 instance via x_vr_values.
8162 (identify_jump_threads): Accept vr_values as argument. Store
8163 it into the walker structure.
8164 (evrp_dom_walker): Add vr_values class data member. Add various
8166 (evrp_dom_walker::try_find_new_range): Use vr_values data
8167 member to access the memory allocator.
8168 (evrp_dom_walker::before_dom_children): Store vr_values class
8169 instance into the vrp_folder class.
8170 (evrp_dom_walker::push_value_range): Rework to avoid direct
8171 access to num_vr_values and vr_value.
8172 (evrp_dom_walker::pop_value_range): Likewise.
8173 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8174 Use vr_values to get to dump_all_value_ranges member function.
8175 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8176 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8177 Pass vr_values class instance down to identify_jump_threads.
8178 Remove call to vrp_free_lattice.
8179 (debug_all_value_ranges): Remove.
8181 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8182 (vrp_folder): Likewise.
8184 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8185 Get it from the existing bitmap instead.
8186 (vrp_intersect_ranges_1): Likewise.
8188 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8190 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8192 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8193 set bit_not_p to it.
8194 (imm_store_chain_info::coalesce_immediate_stores): Break group
8195 if bit_not_p is different.
8196 (count_multiple_uses, split_group,
8197 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8198 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8199 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8200 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8201 to store_immediate_info ctor.
8203 2017-11-09 Jim Wilson <jimw@sifive.com>
8205 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8206 (scan_prog_file): Likewise.
8208 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8210 * bb-reorder.c (max_entry_frequency): Remove.
8211 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8212 connect_traces, push_to_next_round_p): Remove prototypes.
8213 (find_traces_1_round): Use counts only.
8214 (push_to_next_round_p): Likewise.
8215 (find_traces): Likewise.
8216 (rotate_loop): Likewise.
8217 (find_traces_1_round): Likewise.
8218 (connect_traces): Likewise.
8219 (edge_order): Likewise.
8221 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8223 * config/arm/arm.c (output_return_instruction): Add comments to
8224 indicate requirement for cmse_nonsecure_entry return to account
8225 for the size of clearing instruction output here.
8226 (thumb_exit): Likewise.
8227 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8228 return in hardfloat mode.
8230 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8232 * config/rs6000/rs6000.c (machine_function): Add a bool,
8233 "toc_is_wrapped_separately".
8234 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8235 if it wasn't explicitly set or unset, we are optimizing for speed, and
8236 doing separate shrink-wrapping.
8237 (rs6000_get_separate_components): Enable the TOC component if
8238 saving the TOC register in the prologue.
8239 (rs6000_components_for_bb): Handle the TOC component.
8240 (rs6000_emit_prologue_components): Store the TOC register where needed.
8241 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8242 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8244 2017-11-09 Martin Jambor <mjambor@suse.cz>
8246 * ipa-param-manipulation.c: New file.
8247 * ipa-param-manipulation.h: Likewise.
8248 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8249 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8250 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8251 (ipa_parm_adjustment): Likewise.
8252 (ipa_parm_adjustment_vec): Likewise.
8253 (ipa_get_vector_of_formal_parms): Moved declaration to
8254 ipa-param-manipulation.h.
8255 (ipa_get_vector_of_formal_parm_types): Likewise.
8256 (ipa_modify_formal_parameters): Likewise.
8257 (ipa_modify_call_arguments): Likewise.
8258 (ipa_combine_adjustments): Likewise.
8259 (ipa_dump_param_adjustments): Likewise.
8260 (ipa_modify_expr): Likewise.
8261 (ipa_get_adjustment_candidate): Likewise.
8262 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8263 ipa-param-manipulation.c.
8264 (ipa_get_vector_of_formal_parm_types): Likewise.
8265 (ipa_modify_formal_parameters): Likewise.
8266 (ipa_modify_call_arguments): Likewise.
8267 (ipa_modify_expr): Likewise.
8268 (get_ssa_base_param): Likewise.
8269 (ipa_get_adjustment_candidate): Likewise.
8270 (index_in_adjustments_multiple_times_p): Likewise.
8271 (ipa_combine_adjustments): Likewise.
8272 (ipa_dump_param_adjustments): Likewise.
8273 * tree-sra.c: Also include ipa-param-manipulation.h
8274 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8277 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8278 Alan Hayward <alan.hayward@arm.com>
8279 David Sherwood <david.sherwood@arm.com>
8281 * doc/sourcebuild.texi (vect_masked_store): Document.
8283 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8284 Alan Hayward <alan.hayward@arm.com>
8285 David Sherwood <david.sherwood@arm.com>
8287 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8289 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8290 Alan Hayward <alan.hayward@arm.com>
8291 David Sherwood <david.sherwood@arm.com>
8293 * doc/sourcebuild.texi (vect_variable_length): Document.
8295 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8296 Alan Hayward <alan.hayward@arm.com>
8297 David Sherwood <david.sherwood@arm.com>
8299 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8301 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8302 Alan Hayward <alan.hayward@arm.com>
8303 David Sherwood <david.sherwood@arm.com>
8305 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8307 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8308 Alan Hayward <alan.hayward@arm.com>
8309 David Sherwood <david.sherwood@arm.com>
8311 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8312 previously undocumented selectors.
8313 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8315 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8317 * doc/rtl.texi (const_vector): Say that elements can be
8318 const_wide_ints too.
8319 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8320 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8321 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8322 * optabs.c (expand_vector_broadcast): Likewise.
8324 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8325 Alan Hayward <alan.hayward@arm.com>
8326 David Sherwood <david.sherwood@arm.com>
8328 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8329 scaled index even if the unscaled address was invalid.
8330 Don't increase the complexity of using a scale in that case.
8332 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8333 Alan Hayward <alan.hayward@arm.com>
8334 David Sherwood <david.sherwood@arm.com>
8336 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8337 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8338 UNITS_PER_WORD bytes.
8339 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8340 based on REGMODE_NATURAL_SIZE of the inner mode.
8341 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8342 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8343 * expmed.c (lowpart_bit_field_p): Divide the value up into
8344 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8345 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8346 whether something is likely to occupy more than one register.
8348 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8351 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8352 profile_count::adjust_for_ipa_scaling.
8353 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8354 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8356 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8358 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8360 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8361 (split_group): Add total_orig and total_new arguments, estimate the
8362 number of statements related to the store group without store merging
8363 and with store merging.
8364 (imm_store_chain_info::output_merged_store): Adjust split_group
8365 callers, punt if estimated number of statements with store merging
8366 is not smaller than estimated number of statements without it.
8368 (handled_load): Remove has_single_use checks.
8369 (pass_store_merging::process_store): Likewise.
8371 2017-11-09 Richard Biener <rguenther@suse.de>
8373 PR tree-optimization/82902
8374 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8376 2017-11-09 Martin Liska <mliska@suse.cz>
8379 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8382 2017-11-09 Martin Liska <mliska@suse.cz>
8384 PR tree-optimization/82669
8385 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8387 2017-11-09 Martin Liska <mliska@suse.cz>
8389 PR gcov-profile/48463
8390 * coverage.c (coverage_begin_function): Output also end locus
8391 of a function and information whether the function is
8393 * gcov-dump.c (tag_function): Parse and print the information.
8394 * gcov.c (INCLUDE_MAP): Add include.
8395 (INCLUDE_SET): Likewise.
8396 (struct line_info): Move earlier in the source file because
8397 of vector<line_info> in function_info structure.
8398 (line_info::line_info): Likewise.
8399 (line_info::has_block): Likewise.
8400 (struct source_info): Add new member index.
8401 (source_info::get_functions_at_location): New function.
8402 (function_info::group_line_p): New function.
8403 (output_intermediate_line): New function.
8404 (output_intermediate_file): Use the mentioned function.
8405 (struct function_start): New.
8406 (struct function_start_pair_hash): Likewise.
8407 (process_file): Add code that identifies group functions.
8408 Assign lines either to global or function scope.
8409 (generate_results): Skip artificial functions.
8410 (find_source): Assign index for each source file.
8411 (read_graph_file): Read new flag artificial and end_line.
8412 (add_line_counts): Assign it either to global of function scope.
8413 (accumulate_line_counts): Isolate core of the function to
8414 accumulate_line_info and call it for both function and global
8416 (accumulate_line_info): New function.
8417 (output_line_beginning): Fix GNU coding style.
8418 (print_source_line): New function.
8419 (output_line_details): Likewise.
8420 (output_function_details): Likewise.
8421 (output_lines): Iterate both source (global) scope and function
8423 (struct function_line_start_cmp): New class.
8424 * doc/gcov.texi: Reflect changes in documentation.
8426 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8429 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8430 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8431 and similarly for not instead of neg.
8433 2017-11-08 Andi Kleen <ak@linux.intel.com>
8435 * config/i386/i386.opt: Add -mforce-indirect-call.
8436 * config/i386/predicates.md: Check for flag_force_indirect_call.
8437 * doc/invoke.texi: Document -mforce-indirect-call
8439 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8441 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8443 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8444 config/riscv/riscv.c (predict.h): New include.
8445 (riscv_slow_unaligned_access_p): No longer static.
8446 (riscv_block_move_straight): Add require.
8447 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8449 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8452 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8453 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8454 nonimmediate_operand predicate for operand 1 instead of
8457 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8459 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8461 * config/aarch64/constraints.md (Uml): New constraint.
8462 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8465 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8467 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8468 of two vec_duplicates into a vec_concat.
8470 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8472 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8473 Simplify vec_merge of vec_duplicate and vec_concat.
8474 * config/aarch64/constraints.md (Utq): New constraint.
8475 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8478 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8480 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8481 Simplify vec_merge of vec_duplicate and const_vector.
8482 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8484 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8485 mode iterator. Update predicate on operand 1 to
8486 handle non-const_vec constants. Delete constraints.
8487 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8489 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8491 PR tree-optimization/78821
8492 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8494 (store_operand_info::store_operand_info): Initialize it to false.
8495 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8496 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8497 store in the group, and if chain_info is non-NULL, to ignore altogether
8499 (compatible_load_p): Fail if bit_not_p does not match.
8500 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8501 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8502 (pass_store_merging::process_store): Adjust
8503 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8504 call terminate_all_aliasing_chains newly when adding a store into
8505 a chain with non-NULL chain_info.
8507 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8509 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8511 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8513 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8515 (aarch64_layout_frame): Initialise emit_frame_chain.
8516 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8517 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8519 2017-11-08 Martin Liska <mliska@suse.cz>
8521 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8522 of gimple_call_internal_p.
8524 2017-11-07 Tom de Vries <tom@codesourcery.com>
8526 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8528 2017-11-07 Tom de Vries <tom@codesourcery.com>
8530 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8533 2017-11-08 Martin Liska <mliska@suse.cz>
8536 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8538 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8540 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8542 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8545 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8546 (*cmp<mode>_ccz_1): New insn with $k alternative.
8549 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8550 RTX_COMM_COMPARE as commutative as well.
8551 (ix86_binary_operator_ok): Formatting fix.
8552 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8553 *<code><mode>3<mask_name><round_saeonly_name>,
8554 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8555 *<s>mul<mode>3_highpart<mask_name>,
8556 *vec_widen_umult_even_v16si<mask_name>,
8557 *vec_widen_umult_even_v8si<mask_name>,
8558 *vec_widen_umult_even_v4si<mask_name>,
8559 *vec_widen_smult_even_v16si<mask_name>,
8560 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8561 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8562 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8563 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8564 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8565 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8566 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8567 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8568 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8569 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8570 ix86_binary_operator_ok. Formatting fixes.
8571 (*<plusminus_insn><mode>3<mask_name><round_name>,
8572 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8575 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8577 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8578 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8579 eq and ne if TARGET_ISEL.
8580 (cmp): New code_iterator.
8581 (UNS, UNSU_, UNSIK): New code_attrs.
8582 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8583 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8585 ("eq<mode>3"): ... this.
8586 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8588 ("ne<mode>3"): ... this.
8590 2017-11-07 Julia Koval <julia.koval@intel.com>
8593 * common/config/i386/i386-common.c
8594 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8595 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8596 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8597 * config/i386/i386.opt: Ditto.
8598 * config/i386/i386.c (ix86_target_string): Ditto.
8599 (ix86_option_override_internal): Ditto.
8600 (ix86_init_mpx_builtins): Move MPX to args2.
8601 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8602 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8603 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8604 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8605 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8606 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8607 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8609 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8612 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8613 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8614 (zero-extendsidi peephole2): Remove peephole.
8616 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8619 * stor-layout.c: Include attribs.h.
8620 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8621 explicit lookup of "aligned" attribute.
8623 2017-11-07 Andrew Waterman <andrew@sifive.com>
8625 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8626 (riscv_expand_block_move): Likewise.
8627 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8629 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8630 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8631 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8632 (riscv_adjust_block_mem): Likewise.
8633 (riscv_block_move_loop): Likewise.
8634 (riscv_expand_block_move): Likewise.
8635 * config/riscv/riscv.md (movmemsi): New pattern.
8637 2017-11-07 Michael Clark <michaeljclark@mac.com>
8639 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8640 (MUSL_DYNAMIC_LINKER): Likewise.
8642 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8644 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8645 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8648 2017-11-07 Richard Biener <rguenther@suse.de>
8650 * match.pd: Fix build.
8652 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8653 Jackson Woodruff <jackson.woodruff@arm.com>
8655 PR tree-optimization/71026
8656 * match.pd: Canonicalize negate in division.
8658 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8661 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8663 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8665 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8666 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8668 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8670 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8671 non-scalar integral types.
8672 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8673 (-(A-B), -(~A)): New transformations.
8675 2017-11-07 Tom de Vries <tom@codesourcery.com>
8677 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8678 semicolon after "do {} while (0)".
8679 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8680 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8681 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8682 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8683 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8685 2017-11-07 Tom de Vries <tom@codesourcery.com>
8687 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8688 after "do {} while (0)".
8689 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8690 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8691 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8692 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8693 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8695 2017-11-07 Tom de Vries <tom@codesourcery.com>
8698 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8700 (arm_rtx_costs_internal): Add missing semicolon after
8701 HANDLE_NARROW_SHIFT_ARITH call.
8703 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8705 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8706 disable isel if it was not set explicitly.
8708 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8710 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8711 (add_type_attribute) likewise.
8713 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8715 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8716 of struct ix86_frame.
8717 (ix86_initial_elimination_offset): Likewise.
8718 (ix86_expand_split_stack_prologue): Likewise.
8720 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8722 * tree-vrp.h (enum value_range_type): Update stale comment.
8724 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8725 Alan Hayward <alan.hayward@arm.com>
8726 David Sherwood <david.sherwood@arm.com>
8728 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8729 (aarch64_expand_vec_perm_const): Take the number of units too.
8730 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8731 (aarch64_expand_vec_perm_const): Likewise.
8732 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8733 (vec_perm<mode>): Update accordingly.
8735 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8736 Alan Hayward <alan.hayward@arm.com>
8737 David Sherwood <david.sherwood@arm.com>
8739 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8740 Take the number of units too.
8741 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8742 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8743 but check for a vector mode before rather than after the call.
8744 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8745 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8746 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8747 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8748 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8749 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8750 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8751 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8752 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8753 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8754 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8755 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8756 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8757 (aarch64_sqdmull2_n<mode>): Update accordingly.
8759 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8760 Alan Hayward <alan.hayward@arm.com>
8761 David Sherwood <david.sherwood@arm.com>
8763 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8764 the number of units too.
8765 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8766 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8767 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8768 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8769 (vec_store_lanesxi<mode>): Update accordingly.
8771 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8772 Alan Hayward <alan.hayward@arm.com>
8773 David Sherwood <david.sherwood@arm.com>
8775 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8776 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8777 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8778 of units rather than the mode.
8779 * config/aarch64/iterators.md (nunits): New mode attribute.
8780 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8781 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8782 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8783 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8784 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8785 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8786 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8787 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8788 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8789 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8790 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8791 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8792 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8793 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8794 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8795 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8796 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8797 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8798 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8799 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8800 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8801 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8802 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8803 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8804 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8805 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8806 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8807 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8808 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8809 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8810 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8811 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8812 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8813 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8814 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8815 (aarch64_simd_vec_setv2di): Likewise.
8817 2017-11-06 Carl Love <cel@us.ibm.com>
8819 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8821 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8822 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8824 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8826 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8827 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8828 Add power 8 macro expansions.
8829 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8830 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8831 power 8 instructions. (VSX_XXBR): Add iterator.
8833 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8835 * config/arm/arm.md (predicable_short_it): Change default to "no",
8836 improve documentation, remove uses that are identical to the default.
8837 (enabled_for_depr_it): Rename to enabled_for_short_it.
8838 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8839 Remove default uses.
8840 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8841 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8842 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8843 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8845 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8848 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8849 float128 helper macros, which are no longer used after deleting
8850 the old 'q' built-in functions, and moving the round to odd
8851 built-in functions to being special built-in functions.
8852 (BU_FLOAT128_2): Likewise.
8853 (BU_FLOAT128_1_HW): Likewise.
8854 (BU_FLOAT128_2_HW): Likewise.
8855 (BU_FLOAT128_3_HW): Likewise.
8856 (FABSQ): Delete old 'q' built-in functions.
8857 (COPYSIGNQ): Likewise.
8858 (SQRTF128_ODD): Move round to odd built-in functions to be
8859 special built-in functions, so that we can handle
8860 -mabi=ieeelongdouble.
8861 (TRUNCF128_ODD): Likewise.
8862 (ADDF128_ODD): Likewise.
8863 (SUBF128_ODD): Likewise.
8864 (MULF128_ODD): Likewise.
8865 (DIVF128_ODD): Likewise.
8866 (FMAF128_ODD): Likewise.
8867 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8868 built-in names to 'f128'.
8869 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8870 old 'q' built-in functions, as the machine independent code for
8871 'f128' built-in functions handles this.
8872 (rs6000_expand_builtin): Add expansion for float128 round to odd
8873 functions, keying off on -mabi=ieeelongdouble of whether to use
8874 the KFmode or TFmode variant.
8875 (rs6000_init_builtins): Initialize the _Float128 round to odd
8877 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8878 _Float128 'q' built-in functions are now mapped into the new
8879 'f128' built-in functions.
8881 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8883 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8884 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8886 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8889 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8890 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8891 * toplev.c: Include "ipa-fnsummary.h".
8892 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8894 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8896 PR tree-optimization/82838
8897 * gimple-ssa-store-merging.c
8898 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8899 on a separate gimple_seq which is then appended to seq.
8901 2017-11-06 Jeff Law <law@redhat.com>
8904 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8905 (get_probe_interval): New functions.
8906 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8907 (ix86_adjust_stack_and_probe): Likewise.
8908 (output_adjust_stack_and_probe): Likewise.
8909 (ix86_emit_probe_stack_range): Likewise.
8910 (ix86_expand_prologue): Likewise.
8912 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8914 PR tree-optimization/82816
8915 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8916 if the modes of the two types are the same.
8917 (convert_plusminus_to_widen): Likewise.
8919 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8921 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8923 (usadv16qi): New define_expand.
8924 (usadv8hi): New define_expand.
8926 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8929 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8931 (inline_transform): Likewise.
8932 * predict.c (counts_to_freqs): Remove useless conditional.
8933 * profile-count.h (profile_count::apply_scale): Move sanity check.
8934 * tree-inline.c (copy_bb): Always scale.
8935 (copy_cfg_body): Likewise.
8937 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8940 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8942 (*cmp_ite0): Add enabled_for_depr_it attribute.
8943 (*cmp_ite1): Likewise.
8945 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8947 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8950 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8952 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8953 return the value by pointer.
8954 (extract_range_from_multiplicative_op_1): Update accordingly.
8955 Return as soon as an operation fails.
8957 2017-11-05 Tom de Vries <tom@codesourcery.com>
8960 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8961 (DEF_SANITIZER_BUILTIN): ... here.
8962 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8963 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8965 2017-11-05 Tom de Vries <tom@codesourcery.com>
8968 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8970 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8971 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8972 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8974 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8975 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8976 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8978 2017-11-05 Tom de Vries <tom@codesourcery.com>
8981 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8984 2017-11-04 Michael Clark <michaeljclark@mac.com>
8986 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8987 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8989 (*addsi3_extended): Likewise.
8990 (*addsi3_extended2): Likewise.
8991 (<optab>si3): Likewise.
8992 (<optab>di3): Likewise.
8993 (<optab><mode>3): Likewise.
8994 (<*optabe>si3_internal): Likewise.
8995 (zero_extendqi<SUPERQI:mode>2): Likewise.
8996 (*add<mode>hi3): Likewise.
8997 (*xor<mode>hi3): Likewise.
8998 (<optab>di3): Likewise.
8999 (*<optab>si3_extend): Likewise.
9000 (*sge<u>_<X:mode><GPR:mode>): Likewise.
9001 (*slt<u>_<X:mode><GPR:mode>): Likewise.
9002 (*sle<u>_<X:mode><GPR:mode>): Likewise.
9004 2017-11-04 Andrew Waterman <andrew@sifive.com>
9006 * config/riscv/riscv.c (riscv_option_override): Conditionally set
9007 TARGET_STRICT_ALIGN based upon -mtune argument.
9009 2017-11-04 Andrew Waterman <andrew@sifive.com>
9011 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
9013 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
9015 * config/i386/i386.c (choose_basereg): Use optional scratch
9016 register and add assertion.
9017 (x86_emit_outlined_ms2sysv_save): Use scratch register when
9018 needed, and don't allocate stack.
9019 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
9020 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
9021 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
9023 2017-11-03 Jeff Law <law@redhat.com>
9025 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
9026 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
9027 to probe at the start of a noreturn function.
9029 2017-11-03 Jakub Jelinek <jakub@redhat.com>
9031 PR tree-optimization/78821
9032 * gimple-ssa-store-merging.c: Update the file comment.
9033 (MAX_STORE_ALIAS_CHECKS): Define.
9034 (struct store_operand_info): New type.
9035 (store_operand_info::store_operand_info): New constructor.
9036 (struct store_immediate_info): Add rhs_code and ops data members.
9037 (store_immediate_info::store_immediate_info): Add rhscode, op0r
9038 and op1r arguments to the ctor, initialize corresponding data members.
9039 (struct merged_store_group): Add load_align_base and load_align
9041 (merged_store_group::merged_store_group): Initialize them.
9042 (merged_store_group::do_merge): Update them.
9043 (merged_store_group::apply_stores): Pick the constant for
9044 encode_tree_to_bitpos from one of the two operands, or skip
9045 encode_tree_to_bitpos if neither operand is a constant.
9046 (class pass_store_merging): Add process_store method decl. Remove
9047 bool argument from terminate_all_aliasing_chains method decl.
9048 (pass_store_merging::terminate_all_aliasing_chains): Remove
9049 var_offset_p argument and corresponding handling.
9050 (stmts_may_clobber_ref_p): New function.
9051 (compatible_load_p): New function.
9052 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
9053 if there is overlap and rhs_code is not INTEGER_CST. For
9054 non-overlapping stores terminate group if rhs is not mergeable.
9055 (get_alias_type_for_stmts): Change first argument from
9056 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
9057 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
9058 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9060 (get_location_for_stmts): Change first argument from
9061 auto_vec<gimple *> & to vec<gimple *> &.
9062 (struct split_store): Remove orig_stmts data member, add orig_stores.
9063 (split_store::split_store): Create orig_stores rather than orig_stmts.
9064 (find_constituent_stmts): Renamed to ...
9065 (find_constituent_stores): ... this. Change second argument from
9066 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9067 to info structures rather than the statements.
9068 (split_group): Rename ALLOW_UNALIGNED argument to
9069 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9070 it. Adjust find_constituent_stores caller.
9071 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9072 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9073 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9074 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9075 (mem_valid_for_store_merging): New function.
9076 (handled_load): New function.
9077 (pass_store_merging::process_store): New method.
9078 (pass_store_merging::execute): Use process_store method. Adjust
9079 terminate_all_aliasing_chains caller.
9081 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9083 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9084 Return true for more constants, symbols and label references.
9085 (aarch64_valid_floating_const): Remove unused function.
9087 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9090 * config/aarch64/aarch64.c (aarch64_layout_frame):
9091 Undo forcing of LR at bottom of frame.
9093 2017-11-03 Jeff Law <law@redhat.com>
9096 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9097 for int_registers_saved.
9099 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9100 extracted from tree-ssa-dom.c.
9101 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9102 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9103 (record_equivalences_from_incoming_edge): Add additional argument
9104 to single_pred_edge_ignoring_loop_edges call.
9105 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9106 (uncprop_dom_walker::before_dom_children): Add additional argument
9107 to single_pred_edge_ignoring_loop_edges call.
9108 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9109 single_pred_edge_ignoring_loop_edges rather than open coding.
9110 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9112 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9114 * match.pd (-(-A)): Rewrite.
9116 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9118 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9119 (rs6000_emit_int_cmove): New declaration.
9120 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9121 (rs6000_emit_sISEL): Delete.
9122 (rs6000_emit_int_cmove): Make non-static.
9123 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9124 instead of rs6000_emit_sISEL.
9126 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9128 * asan.c (create_cond_insert_point): Maintain profile.
9129 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9131 * basic-block.h (struct basic_block_def): Remove frequency.
9132 (EDGE_FREQUENCY): Use to_frequency
9133 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9135 (find_traces): Update to use to_frequency.
9136 (find_traces_1_round): Likewise; use only IPA counts.
9137 (bb_to_key): Likewise.
9138 (connect_traces): Use IPA counts only.
9139 (copy_bb_p): Update to use to_frequency.
9140 (fix_up_crossing_landing_pad): Likewise.
9141 (sanitize_hot_paths): Likewise.
9142 * bt-load.c (basic_block_freq): Likewise.
9143 * cfg.c (init_flow): Set count_max to uninitialized.
9144 (check_bb_profile): Remove frequencies; check counts.
9145 (dump_bb_info): Do not dump frequencies.
9146 (update_bb_profile_for_threading): Update counts only.
9147 (scale_bbs_frequencies_int): Likewise.
9148 (MAX_SAFE_MULTIPLIER): Remove.
9149 (scale_bbs_frequencies_gcov_type): Update counts only.
9150 (scale_bbs_frequencies_profile_count): Update counts only.
9151 (scale_bbs_frequencies): Update counts only.
9152 * cfg.h (struct control_flow_graph): Add count-max.
9153 (update_bb_profile_for_threading): Update prototype.
9154 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9155 (find_many_sub_basic_blocks): Likewise.
9156 * cfgcleanup.c (try_forward_edges): Likewise.
9157 (try_crossjump_to_edge): Likewise.
9158 * cfgexpand.c (expand_gimple_cond): Likewise.
9159 (expand_gimple_tailcall): Likewise.
9160 (construct_init_block): Likewise.
9161 (construct_exit_block): Likewise.
9162 * cfghooks.c (verify_flow_info): Check consistency of counts.
9163 (dump_bb_for_graph): Do not dump frequencies.
9164 (split_block_1): Do not update frequencies.
9165 (split_edge): Do not update frequencies.
9166 (make_forwarder_block): Do not update frequencies.
9167 (duplicate_block): Do not update frequencies.
9168 (account_profile_record): Do not update frequencies.
9169 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9170 for global heuristics.
9171 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9172 (expected_loop_iterations_unbounded): Use counts only.
9173 * cfgloopmanip.c (scale_loop_profile): Simplify.
9174 (create_empty_loop_on_edge): Simplify
9176 (duplicate_loop_to_header_edge): Simplify
9177 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9178 (update_br_prob_note): Take care of removing note when profile
9180 (relink_block_chain): Do not dump frequency.
9181 (rtl_account_profile_record): Use to_frequency.
9182 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9183 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9184 (cgraph_update_edges_for_call_stmt_node): Likewise.
9185 (cgraph_edge::verify_count_and_frequency): Update.
9186 (cgraph_node::verify_node): Temporarily disable frequency verification.
9187 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9188 to_cgraph_frequency.
9189 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9190 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9192 (cgraph_node::expand_thunk): Update profile.
9193 * except.c (dw2_build_landing_pads): Do not update frequency.
9194 * final.c (compute_alignments): Use to_frequency.
9195 (dump_basic_block_info): Do not dump frequency.
9196 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9197 (dump_gimple_bb_header): Do not dump frequency.
9198 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9200 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9201 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9202 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9203 (init_before_recovery): Do not update frequency.
9204 (sched_create_recovery_edges): Do not update frequency.
9205 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9206 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9207 (ipa_cp_c_finalize): Set max_count to uninitialized.
9208 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9209 (param_change_prob): Use counts.
9210 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9212 * ipa-split.c (consider_split): Use to_frequency.
9213 (split_function): Use to_frequency.
9214 * ira-build.c (loop_compare_func): Likewise.
9215 (mark_loops_for_removal): Likewise.
9216 (mark_all_loops_for_removal): Likewise.
9217 * loop-doloop.c (doloop_modify): Do not update frequency.
9218 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9220 * lto-streamer-in.c (input_function): Update count_max.
9221 * omp-expand.c (expand_omp_taskreg): Update count_max.
9222 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9223 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9224 (maybe_hot_count_p): Use ipa counts only.
9225 (maybe_hot_bb_p): Simplify.
9226 (maybe_hot_edge_p): Simplify.
9227 (probably_never_executed): Do not take frequency argument.
9228 (probably_never_executed_bb_p): Do not pass frequency.
9229 (probably_never_executed_edge_p): Likewise.
9230 (combine_predictions_for_bb): Check that profile is nonzero.
9231 (propagate_freq): Do not set frequency.
9232 (drop_profile): Simplify.
9233 (counts_to_freqs): Simplify.
9234 (expensive_function_p): Use to_frequency.
9235 (propagate_unlikely_bbs_forward): Simplify.
9236 (determine_unlikely_bbs): Simplify.
9237 (estimate_bb_frequencies): Add hack to silence graphite issues.
9238 (compute_function_frequency): Use ipa counts.
9239 (pass_profile::execute): Update.
9240 (rebuild_frequencies): Use counts only.
9241 (force_edge_cold): Use counts only.
9242 * profile-count.c (profile_count::dump): Dump new count types.
9243 (profile_count::differs_from_p): Check compatiblity.
9244 (profile_count::to_frequency): New function.
9245 (profile_count::to_cgraph_frequency): New function.
9246 * profile-count.h (struct function): Declare.
9247 (enum profile_quality): Add profile_guessed_local and
9248 profile_guessed_global0.
9249 (class profile_proability): Decrease number of bits to 29;
9250 update from_reg_br_prob_note and to_reg_br_prob_note.
9251 (class profile_count: Update comment; decrease number of bits
9252 to 61. Check compatibility.
9253 (profile_count::compatible_p): New private member function.
9254 (profile_count::ipa_p): New member function.
9255 (profile_count::operator<): Handle global zero correctly.
9256 (profile_count::operator>): Handle global zero correctly.
9257 (profile_count::operator<=): Handle global zero correctly.
9258 (profile_count::operator>=): Handle global zero correctly.
9259 (profile_count::nonzero_p): New member function.
9260 (profile_count::force_nonzero): New member function.
9261 (profile_count::max): New member function.
9262 (profile_count::apply_scale): Handle IPA scalling.
9263 (profile_count::guessed_local): New member function.
9264 (profile_count::global0): New member function.
9265 (profile_count::ipa): New member function.
9266 (profile_count::to_frequency): Declare.
9267 (profile_count::to_cgraph_frequency): Declare.
9268 * profile.c (OVERLAP_BASE): Delete.
9269 (compute_frequency_overlap): Delete.
9270 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9271 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9272 * sched-ebb.c (rank): Use counts only.
9273 * shrink-wrap.c (handle_simple_exit): Use counts only.
9274 (try_shrink_wrapping): Use counts only.
9275 (place_prologue_for_one_component): Use counts only.
9276 * tracer.c (find_best_predecessor): Use to_frequency.
9277 (find_trace): Use to_frequency.
9278 (tail_duplicate): Use to_frequency.
9279 * trans-mem.c (expand_transaction): Do not update frequency.
9280 * tree-call-cdce.c: Do not update frequency.
9281 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9282 (gimple_merge_blocks): Likewise.
9283 (gimple_split_edge): Likewise.
9284 (gimple_duplicate_sese_region): Likewise.
9285 (gimple_duplicate_sese_tail): Likewise.
9286 (move_sese_region_to_fn): Likewise.
9287 (gimple_account_profile_record): Likewise.
9288 (insert_cond_bb): Likewise.
9289 * tree-complex.c (expand_complex_div_wide): Likewise.
9290 * tree-eh.c (lower_resx): Update profile.
9291 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9293 (initialize_cfun): Do not initialize frequencies
9294 (freqs_to_counts): Delete.
9295 (copy_cfg_body): Ignore count parameter.
9296 (copy_body): Update.
9297 (expand_call_inline): Update count_max.
9298 (optimize_inline_calls): Update count_max.
9299 (tree_function_versioning): Update count_max.
9300 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9301 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9303 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9304 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9305 (try_peel_loop): Likewise.
9306 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9308 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9309 (tree_transform_and_unroll_loop): Do not use frequencies
9310 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9311 Use reliable prediction only.
9312 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9313 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9314 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9315 probability scaling.
9316 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9317 not update frequency
9318 (any_remaining_duplicated_blocks): Likewise.
9319 (update_profile): Likewise.
9320 (estimated_freqs_path): Delete.
9321 (freqs_to_counts_path): Delete.
9322 (clear_counts_path): Delete.
9323 (ssa_fix_duplicate_block_edges): Likewise.
9324 (duplicate_thread_path): Likewise.
9325 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9326 * tree-tailcall.c (decrease_profile): Do not update frequency.
9327 (eliminate_tail_call): Likewise.
9328 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9329 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9330 (optimize_mask_stores): Likewise.
9331 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9332 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9333 (ubsan_expand_ptr_ifn): Update profile.
9334 * value-prof.c (gimple_ic): Simplify.
9335 * value-prof.h (gimple_ic): Update prototype.
9336 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9337 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9339 (want_inline_self_recursive_call_p): Likewise.
9340 (resolve_noninline_speculation): Only cummulate defined counts.
9341 (inline_small_functions): Use nonzero_p.
9342 (ipa_inline): Do not access freed node.
9344 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9346 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9347 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9349 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9351 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9352 non-legitimate address.
9354 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9356 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9357 (*lt0_<mode>di, *lt0_<mode>si): New.
9359 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9361 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9362 TARGET_PAIRED_FLOAT.
9364 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9365 Jim Wilson <jim.wilson@linaro.org>
9367 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9368 * config/aarch64/aarch64-tune.md: Regenerated.
9369 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9370 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9372 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9374 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9375 (arc_expand_prologue): Restore blink for millicode.
9376 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9378 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9381 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9382 gen_vec_duplicate after forcing the scalar into a register.
9384 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9386 * combine (try_combine): Print the insns input to try_combine to the
9389 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9392 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9393 Remove second argument from aarch64_process_target_attr call.
9394 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9396 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9398 (aarch64_handle_attr_arch): Remove second argument.
9399 (aarch64_handle_attr_cpu): Ditto.
9400 (aarch64_handle_attr_tune): Ditto.
9401 (aarch64_handle_attr_isa_flags): Ditto.
9402 (aarch64_process_one_target_attr): Ditto.
9403 (aarch64_process_target_attr): Ditto.
9404 (aarch64_option_valid_attribute_p): Remove second argument.
9405 on aarch64_process_target_attr call.
9407 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9409 * diagnostic.c: Include "selftest-diagnostic.h".
9410 (selftest::assert_location_text): New function.
9411 (selftest::test_diagnostic_get_location_text): New function.
9412 (selftest::diagnostic_c_tests): Call it.
9414 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9416 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9417 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9418 (class selftest::test_diagnostic_context): Move to...
9419 * selftest-diagnostic.c: New file.
9420 * selftest-diagnostic.h: New file.
9422 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9424 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9425 offset range for FT32B.
9426 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9427 * config/ft32/ft32.md: Add TARGET_NOPM.
9428 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9429 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9431 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9433 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9435 2017-11-02 Jeff Law <law@redhat.com>
9437 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9438 virtual keyword on FINAL OVERRIDE members.
9440 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9441 virtuals together. Add virtual destructor.
9442 (substitute_and_fold_engine): Similarly.
9444 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9446 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9448 2017-11-02 Richard Biener <rguenther@suse.de>
9450 PR tree-optimization/82795
9451 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9453 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9455 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9457 (gcc_GAS_CHECK_FEATURE): Remove.
9458 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9459 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9460 * configure: Regenerate.
9462 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9464 * config/arc/arc.c (hwloop_optimize): Account for empty
9467 2017-11-02 Richard Biener <rguenther@suse.de>
9470 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9471 Truncate ARRAY_REF index and element size.
9473 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9475 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9477 2017-11-01 Jeff Law <law@redhat.com>
9479 * tree-ssa-ccp.c (ccp_folder): New class derived from
9480 substitute_and_fold_engine.
9481 (ccp_folder::get_value): New member function.
9482 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9483 (ccp_fold_stmt): Remove prototype.
9484 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9485 * tree-ssa-copy.c (copy_folder): New class derived from
9486 substitute_and_fold_engine.
9487 (copy_folder::get_value): Renamed from get_value.
9488 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9489 * tree-vrp.c (vrp_folder): New class derived from
9490 substitute_and_fold_engine.
9491 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9492 (vrp_folder::get_value): New member function.
9493 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9494 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9495 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9496 provide a class interface to folder/substitute routines.
9497 (ssa_prop_fold_stmt_fn): Remove typedef.
9498 (ssa_prop_get_value_fn): Likewise.
9499 (subsitute_and_fold): Remove prototype.
9500 (replace_uses_in): Likewise.
9501 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9502 Renamed from replace_uses_in. Call the virtual member function
9503 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9504 (substitute_and_fold_dom_walker): Remove initialization of
9505 data member entries for calbacks. Add substitute_and_fold_engine
9506 member and initialize it.
9507 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9508 member functions for get_value, replace_phi_args_in c
9509 replace_uses_in, and fold_stmt calls.
9510 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9511 substitute_and_fold. Remove assert. Update ctor call.
9513 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9514 (ssa_prop_visit_phi_fn): Likewise.
9515 (class ssa_propagation_engine): New class to provide an interface
9517 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9519 (ssa_prop_visit_phi): Likewise.
9520 (ssa_propagation_engine::simulate_stmt): Moved into class.
9521 Call visit_phi/visit_stmt from the class rather than via
9522 file scoped static variables.
9523 (ssa_propagation_engine::simulate_block): Moved into class.
9524 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9525 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9526 set file scoped statics for the visit_stmt/visit_phi callbacks.
9527 * tree-complex.c (complex_propagate): New class derived from
9528 ssa_propagation_engine.
9529 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9530 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9531 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9533 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9534 ssa_propagation_engine.
9535 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9536 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9537 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9538 * tree-ssa-copy.c (copy_prop): New class derived from
9539 ssa_propagation_engine.
9540 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9541 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9542 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9543 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9544 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9545 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9546 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9548 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9550 PR rtl-optimization/82778
9551 PR rtl-optimization/82597
9552 * compare-elim.c (struct comparison): Add in_a_setter field.
9553 (find_comparison_dom_walker::before_dom_children): Remove killed
9554 bitmap and df_simulate_find_defs call, instead walk the defs.
9555 Compute last_setter and initialize in_a_setter. Merge definitions
9556 with first initialization for a few variables.
9557 (try_validate_parallel): Use insn_invalid_p instead of
9558 recog_memoized. Return insn rather than just the pattern.
9559 (try_merge_compare): Fix up comment. Don't uselessly test if
9560 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9562 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9565 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9566 Alan Hayward <alan.hayward@arm.com>
9567 David Sherwood <david.sherwood@arm.com>
9569 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9570 aarch64_hard_regno_nregs to get the number of registers
9573 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9574 Alan Hayward <alan.hayward@arm.com>
9575 David Sherwood <david.sherwood@arm.com>
9577 * config/aarch64/constraints.md (Upl): Rename to...
9579 * config/aarch64/aarch64.md
9580 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9583 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9584 Alan Hayward <alan.hayward@arm.com>
9585 David Sherwood <david.sherwood@arm.com>
9587 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9588 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9591 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9592 Alan Hayward <alan.hayward@arm.com>
9593 David Sherwood <david.sherwood@arm.com>
9595 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9596 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9597 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9599 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9600 of a single element.
9601 * config/aarch64/iterators.md: Add a comment above the permute
9602 unspecs to say that they are generated directly by
9603 aarch64_expand_vec_perm_const.
9604 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9606 2017-11-01 Nathan Sidwell <nathan@acm.org>
9608 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9610 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9612 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9613 and medany code models, and describe what they do.
9615 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9617 Revert accidental duplicate:
9619 * combine.c (can_change_dest_mode): Reject changes in
9620 REGMODE_NATURAL_SIZE.
9622 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9624 PR rtl-optimization/64682
9625 PR rtl-optimization/69567
9626 PR rtl-optimization/69737
9627 PR rtl-optimization/82683
9628 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9629 register mentioned in the note, drop the note, unless it came from I3,
9630 in which case it should go to I3 again.
9632 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9634 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9635 and return false if not.
9636 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9638 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9640 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9641 range is known to be empty.
9643 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9644 Alan Hayward <alan.hayward@arm.com>
9645 David Sherwood <david.sherwood@arm.com>
9647 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9648 and CONST_VECTOR_NUNITS instead of computing the number of units from
9649 the byte sizes of the vector and element.
9650 (simplify_binary_operation_1): Likewise.
9651 (simplify_const_binary_operation): Likewise.
9652 (simplify_ternary_operation): Likewise.
9654 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9655 Alan Hayward <alan.hayward@arm.com>
9656 David Sherwood <david.sherwood@arm.com>
9658 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9659 (int_mem_offset): ...this new function.
9660 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9661 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9662 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9665 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9666 Alan Hayward <alan.hayward@arm.com>
9667 David Sherwood <david.sherwood@arm.com>
9669 * lower-subreg.c (interesting_mode_p): New function.
9670 (compute_costs, find_decomposable_subregs, decompose_register)
9671 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9672 (resolve_clobber, dump_choices): Use it.
9674 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9675 Alan Hayward <alan.hayward@arm.com>
9676 David Sherwood <david.sherwood@arm.com>
9678 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9680 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9681 Alan Hayward <alan.hayward@arm.com>
9682 David Sherwood <david.sherwood@arm.com>
9684 * alias.c (find_base_value, find_base_term): Only process integer
9685 truncations. Check the precision rather than the size.
9687 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9688 Alan Hayward <alan.hayward@arm.com>
9689 David Sherwood <david.sherwood@arm.com>
9691 * machmode.h (is_narrower_int_mode): New function
9692 * optabs.c (expand_float, expand_fix): Use it.
9693 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9695 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9696 Alan Hayward <alan.hayward@arm.com>
9697 David Sherwood <david.sherwood@arm.com>
9699 * rtl.h (narrower_subreg_mode): New function.
9700 * ira-color.c (update_costs_from_allocno): Use it.
9702 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9703 Alan Hayward <alan.hayward@arm.com>
9704 David Sherwood <david.sherwood@arm.com>
9706 * optabs-query.h (convert_optab_p): New function, split out from...
9707 (convert_optab_handler): ...here.
9708 (widening_optab_handler): Delete.
9709 (find_widening_optab_handler): Remove permit_non_widening parameter.
9710 (find_widening_optab_handler_and_mode): Likewise. Provide an
9711 override that operates on mode class wrappers.
9712 * optabs-query.c (widening_optab_handler): Delete.
9713 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9714 parameter. Assert that the two modes are the same class and that
9715 the "from" mode is narrower than the "to" mode. Use
9716 convert_optab_handler instead of widening_optab_handler.
9717 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9718 instead of widening_optab_handler.
9719 * expr.c (expand_expr_real_2): Update calls to
9720 find_widening_optab_handler.
9721 * optabs.c (expand_widen_pattern_expr): Likewise.
9722 (expand_binop_directly): Take the insn_code as a parameter.
9723 (expand_binop): Only call find_widening_optab_handler for
9724 conversion optabs; use optab_handler otherwise. Update calls
9725 to find_widening_optab_handler and expand_binop_directly.
9726 Use convert_optab_handler instead of widening_optab_handler.
9727 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9728 find_widening_optab_handler and use scalar_mode rather than
9730 (convert_plusminus_to_widen): Likewise.
9732 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9733 Alan Hayward <alan.hayward@arm.com>
9734 David Sherwood <david.sherwood@arm.com>
9736 * machmode.h (fixed_size_mode): New class.
9737 * rtl.h (get_pool_mode): Return fixed_size_mode.
9738 * gengtype.c (main): Add fixed_size_mode.
9739 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9740 (get_raw_arg_mode): Likewise.
9741 * doc/tm.texi: Regenerate.
9742 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9743 * targhooks.c (default_get_reg_raw_mode): Likewise.
9744 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9745 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9746 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9747 (msp430_get_raw_result_mode): Likewise.
9748 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9749 * dbxout.c (dbxout_parms): Require fixed-size modes.
9750 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9751 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9752 * omp-low.c (lower_oacc_reductions): Likewise.
9753 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9754 (simplify_subreg): Update accordingly.
9755 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9756 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9757 that aren't fixed-size.
9758 (get_pool_mode): Return a fixed_size_mode.
9759 (output_constant_pool_2): Take a fixed_size_mode.
9761 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9762 Alan Hayward <alan.hayward@arm.com>
9763 David Sherwood <david.sherwood@arm.com>
9765 * doc/rtl.texi (vec_series): Document.
9766 (const): Say that the operand can be a vec_series.
9767 * rtl.def (VEC_SERIES): New rtx code.
9768 * rtl.h (const_vec_series_p_1): Declare.
9769 (const_vec_series_p): New function.
9770 * emit-rtl.h (gen_const_vec_series): Declare.
9771 (gen_vec_series): Likewise.
9772 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9773 (gen_vec_series): Likewise.
9774 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9775 * simplify-rtx.c (simplify_unary_operation): Handle negations
9777 (simplify_binary_operation_series): New function.
9778 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9779 (test_vector_ops_series): New function.
9780 (test_vector_ops): Call it.
9781 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9782 gen_const_vec_series.
9783 (altivec_lvsr): Likewise.
9784 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9786 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9787 Alan Hayward <alan.hayward@arm.com>
9788 David Sherwood <david.sherwood@arm.com>
9790 * doc/rtl.texi (const): Update description of address constants.
9791 Say that vector constants are allowed too.
9792 * common.md (E, F): Use CONSTANT_P instead of checking for
9794 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9795 checking for CONST_VECTOR.
9796 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9798 * expr.c (expand_expr_real_2): Check for vector modes instead
9799 of checking for CONST_VECTOR.
9800 * rtl.h (const_vec_p): New function.
9801 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9802 (unwrap_const_vec_duplicate): Handle them here too.
9804 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9805 David Malcolm <dmalcolm@redhat.com>
9806 Alan Hayward <alan.hayward@arm.com>
9807 David Sherwood <david.sherwood@arm.com>
9809 * rtl.h (vec_duplicate_p): New function.
9810 * selftest-rtl.c (assert_rtx_eq_at): New function.
9811 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9812 (assert_rtx_eq_at): Declare.
9813 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9814 * selftest-run-tests.c (selftest::run_tests): Call it.
9815 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9816 (simplify_unary_operation_1): Recursively handle vector duplicates.
9817 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9819 (simplify_subreg): Handle subregs of vector duplicates.
9820 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9821 (selftest::simplify_rtx_c_tests): New functions.
9823 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9824 Alan Hayward <alan.hayward@arm.com>
9825 David Sherwood <david.sherwood@arm.com>
9827 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9828 (gen_vec_duplicate): Likewise.
9829 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9831 (gen_const_vector): ...here.
9832 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9833 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9834 whose elements are all equal.
9835 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9836 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9837 (simplify_relational_operation): Likewise.
9838 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9840 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9841 (aarch64_expand_vector_init): Likewise.
9842 * config/arm/arm.c (neon_vdup_constant): Likewise.
9843 (neon_expand_vector_init): Likewise.
9844 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9845 (arm_block_set_unaligned_vect): Likewise.
9846 (arm_block_set_aligned_vect): Likewise.
9847 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9848 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9849 (expand_vec_perm_even_odd_pack): Likewise.
9850 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9851 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9852 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9853 gen_const_vec_duplicate.
9854 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9855 * config/mips/mips.c (mips_gen_const_int_vector): Use
9856 gen_const_vec_duplicate.
9857 (mips_expand_vector_init): Use CONST0_RTX.
9858 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9859 (define_split): Use gen_const_vec_duplicate.
9860 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9861 (define_split): Use gen_const_vec_duplicate.
9862 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9863 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9864 * config/spu/spu.c (spu_const): Likewise.
9866 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9867 Alan Hayward <alan.hayward@arm.com>
9868 David Sherwood <david.sherwood@arm.com>
9870 * combine.c (can_change_dest_mode): Reject changes in
9871 REGMODE_NATURAL_SIZE.
9873 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9875 * configure.ac (--enable-libssp): New.
9876 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9877 trying to determine target-specific default. Adjust indentation.
9878 * configure: Regenerated.
9879 * doc/install.texi (Configuration): Expand --disable-libssp
9882 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9884 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9887 2017-10-31 Martin Jambor <mjambor@suse.cz>
9890 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9892 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9894 * auto-profile.c (autofdo_source_profile::read): Use
9895 UNKNOWN_LOCATION rather than 0.
9896 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9897 (warning_at): ...this overload.
9898 (warning_at_rich_loc_n): Rename to...
9899 (warning_n): ...this overload.
9900 (error_at_rich_loc): Rename to...
9901 (error_at): ...this overload.
9902 (pedwarn_at_rich_loc): Rename to...
9903 (pedwarn): ...this overload.
9904 (permerror_at_rich_loc): Rename to...
9905 (permerror): ...this overload.
9906 (inform_at_rich_loc): Rename to...
9907 (inform): ...this overload.
9908 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9909 (diagnostic_n_impl_richloc): Rename to...
9910 (diagnostic_n_impl): ...this rich_location *-based decl.
9911 (inform_at_rich_loc): Rename to...
9912 (inform): ...this, and add an assertion.
9913 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9914 (warning_at_rich_loc): Rename to...
9915 (warning_at): ...this, and add an assertion.
9916 (warning_at_rich_loc_n): Rename to...
9917 (warning_n): ...this, and add an assertion.
9918 (warning_n): Update location_t-based implementation for removal of
9919 location_t-based diagnostic_n_impl.
9920 (pedwarn_at_rich_loc): Rename to...
9921 (pedwarn): ...this, and add an assertion.
9922 (permerror_at_rich_loc): Rename to...
9923 (permerror): ...this, and add an assertion.
9924 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9925 (error_at_rich_loc): Rename to...
9926 (error_at): ...this, and add an assertion.
9927 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9928 (driver::do_spec_on_infiles): Likewise.
9929 * substring-locations.c (format_warning_va): Update for renaming
9930 of inform_at_rich_loc.
9932 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9934 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9935 _Float<N>X built-in functions so that the variant without the
9936 "__builtin_" prefix is only enabled for the GNU C and Objective C
9937 languages when they are in non-strict ANSI/ISO mode.
9938 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9939 * target.def (floatn_builtin_p): Add a target hook to control
9940 whether _Float<N> and _Float<N>X built-in functions without the
9941 "__builtin_" prefix are enabled, and return true for C and
9942 Objective C in the default hook. Include langhooks.h in
9944 * targhooks.h (default_floatn_builtin_p): Likewise.
9945 * targhooks.c (default_floatn_builtin_p): Likewise.
9946 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9947 floatn_builtin_p target hook.
9948 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9950 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9951 Eric Botcazou <ebotcazou@adacore.com>
9953 PR rtl-optimization/81803
9954 * lra-constraints.c (curr_insn_transform): Also reload the whole
9955 register for a strict subreg no wider than a word if this is for
9956 a WORD_REGISTER_OPERATIONS target.
9958 2017-10-31 Jason Merrill <jason@redhat.com>
9960 * gdbinit.in: Skip over inlines from timevar.h.
9962 2017-10-31 Martin Liska <mliska@suse.cz>
9964 * doc/gcov.texi: Document new option.
9965 * gcov.c (print_usage): Likewise print it.
9966 (process_args): Support the argument.
9967 (format_count): New function.
9968 (format_gcov): Use the function.
9970 2017-10-31 Martin Liska <mliska@suse.cz>
9972 * gcov.c (struct name_map): do not use typedef.
9973 Define operator== and operator<.
9974 (name_search): Remove.
9975 (name_sort): Remove.
9976 (main): Do not allocate names.
9977 (process_file): Add vertical space.
9978 (generate_results): Use std::find.
9979 (release_structures): Do not release memory.
9980 (find_source): Use std::find.
9982 2017-10-31 Martin Liska <mliska@suse.cz>
9984 * gcov.c (struct line_info): Remove it's typedef.
9985 (line_info::line_info): Add proper ctor.
9986 (line_info::has_block): Do not use a typedef.
9987 (struct source_info): Do not use typedef.
9988 (circuit): Likewise.
9989 (get_cycles_count): Likewise.
9990 (output_intermediate_file): Iterate via vector iterator.
9991 (add_line_counts): Use std::vector methods.
9992 (accumulate_line_counts): Likewise.
9993 (output_lines): Likewise.
9995 2017-10-31 Martin Liska <mliska@suse.cz>
9997 * gcov.c (struct source_info): Remove typedef.
9998 (source_info::source_info): Add proper ctor.
9999 (accumulate_line_counts): Use struct, not it's typedef.
10000 (output_gcov_file): Likewise.
10001 (output_lines): Likewise.
10002 (main): Do not allocate an array.
10003 (output_intermediate_file): Use size of vector container.
10004 (process_file): Resize the vector.
10005 (generate_results): Do not preallocate, use newly added vector
10007 (release_structures): Do not release sources.
10008 (find_source): Use vector methods.
10009 (add_line_counts): Do not use typedef.
10011 2017-10-31 Martin Liska <mliska@suse.cz>
10013 * doc/gcov.texi: Document that.
10014 * gcov.c (add_line_counts): Mark lines with a non-executed
10016 (output_line_beginning): Handle such lines.
10017 (output_lines): Pass new argument.
10018 (output_intermediate_file): Print it in intermediate format.
10020 2017-10-31 Martin Liska <mliska@suse.cz>
10022 * color-macros.h: New file.
10023 * diagnostic-color.c: Factor out color related to macros to
10025 * doc/gcov.texi: Document -k option.
10026 * gcov.c (INCLUDE_STRING): Include string.h.
10027 (print_usage): Add -k option.
10028 (process_args): Parse it.
10029 (pad_count_string): New function.
10030 (output_line_beginning): Likewise.
10031 (DEFAULT_LINE_START): New macro.
10032 (output_lines): Support color output.
10034 2017-10-31 Martin Liska <mliska@suse.cz>
10036 PR gcov-profile/82633
10037 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
10038 their interaction with GCOV infrastructure.
10039 * configure.ac: Add -fkeep-{inline,static}-functions to
10041 * configure: Regenerate.
10043 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
10046 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
10048 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
10051 * config/rs6000/rs6000.md (allocate_stack): Force update interval
10052 into a register if it does not fit into an immediate offset field.
10054 2017-10-31 Olivier Hainque <hainque@adacore.com>
10056 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
10058 2017-10-31 Julia Koval <julia.koval@intel.com>
10060 * config.gcc: Add gfniintrin.h.
10061 * config/i386/gfniintrin.h: New.
10062 * config/i386/i386-builtin-types.def
10063 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10064 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10065 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10066 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10067 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10068 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10069 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10070 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10071 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10072 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10073 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10074 * config/i386/immintrin.h: Include gfniintrin.h.
10075 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10077 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10079 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10081 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10083 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10084 (arm_ashldi3_1bit): Remove pattern.
10085 (ashrdi3): Remove shift by 1 expansion.
10086 (arm_ashrdi3_1bit): Remove pattern.
10087 (lshrdi3): Remove shift by 1 expansion.
10088 (arm_lshrdi3_1bit): Remove pattern.
10089 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10090 cost of ashldi3 by 1.
10091 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10092 (<shift>di3_neon): Likewise.
10094 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10096 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10097 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10098 and (*aarch64_simd_mov<VQ:mode>).
10099 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10100 pattern alternative.
10101 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10102 attributes to match pattern alternative.
10104 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10106 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10107 * config/rs6000/emmintrin.h: New file.
10108 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10110 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10112 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10113 * (movdi_vfp_cortexa8): Remove pattern.
10115 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10117 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10119 (Specific, alpha*-dec-osf5.1): Remove.
10120 (Specific, mips-sgi-irix5): Remove.
10121 (Specific, mips-sgi-irix6): Remove.
10123 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10125 PR middle-end/22141
10126 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10127 arguments to clear_bit_region_be.
10129 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10131 * gimplify.c: Include memmodel.h.
10133 2017-10-30 Martin Jambor <mjambor@suse.cz>
10135 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10136 condition whether loop should be executed at all.
10138 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10140 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10141 gimple folding of vec_madd() intrinsics.
10142 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10143 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10144 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10146 2017-10-30 Richard Biener <rguenther@suse.de>
10148 PR tree-optimization/82762
10150 2017-10-23 Richard Biener <rguenther@suse.de>
10152 PR tree-optimization/82129
10154 2017-08-01 Richard Biener <rguenther@suse.de>
10156 PR tree-optimization/81181
10157 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10158 (compute_antic): ... end of iteration here.
10160 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10162 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10163 -std=iso9899:2017 and -std=gnu17.
10164 * doc/standards.texi (C Language): Document C17 support.
10165 * doc/cpp.texi (Overview): Mention -std=c17.
10166 (Standard Predefined Macros): Document C11 and C17 values of
10167 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10168 * doc/extend.texi (Inline): Do not list individual options for
10169 standards newer than C99.
10170 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10172 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10175 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10177 * asan.c (asan_finish_file): Align asan globals array by shadow
10180 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10182 PR middle-end/22141
10183 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10184 (struct store_immediate_info): Add bitregion_start and bitregion_end
10186 (store_immediate_info::store_immediate_info): Add brs and bre
10187 arguments and initialize bitregion_{start,end} from those.
10188 (struct merged_store_group): Add bitregion_start, bitregion_end,
10189 align_base and mask fields. Drop unnecessary struct keyword from
10190 struct store_immediate_info. Add do_merge method.
10191 (clear_bit_region_be): Use memset instead of loop storing zeros.
10192 (merged_store_group::do_merge): New method.
10193 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10194 stores as long as the surrounding bitregions have no gaps.
10195 (merged_store_group::merge_overlapping): Use do_merge.
10196 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10197 is byte aligned, rather than requiring that start and width are
10198 byte aligned. Drop unnecessary struct keyword from
10199 struct store_immediate_info. Allocate and populate also mask array.
10200 Make start of the arrays relative to bitregion_start rather than
10201 start and size them according to bitregion_{end,start} difference.
10202 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10203 struct store_immediate_info.
10204 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10205 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10206 struct keyword from struct store_immediate_info.
10207 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10208 between stores as long as the surrounding bitregions have no gaps.
10210 (struct split_store): Add orig non-static data member.
10211 (split_store::split_store): Initialize orig to false.
10212 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10213 if there is exactly a single original stmt. Change stmts argument
10214 to pointer from reference, if NULL, don't push anything to it. Add
10215 first argument, use it to optimize skipping over orig stmts that
10216 are known to be before bitpos already. Simplify.
10217 (split_group): Return unsigned int count how many stores are or
10218 would be needed rather than a bool. Add allow_unaligned argument.
10219 Change split_stores argument from reference to pointer, if NULL,
10220 only do a dry run computing how many stores would be produced.
10221 Rewritten algorithm to use both alignment and misalign if
10222 !allow_unaligned and handle bitfield stores with gaps.
10223 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10224 from bitregion_start instead of start. Compute allow_unaligned
10225 here, if true, do 2 split_group dry runs to compute which one
10226 produces fewer stores and prefer aligned if equal. Punt if
10227 new count is bigger or equal than original before emitting any
10228 statements, rather than during that. Remove no longer needed
10229 new_ssa_names tracking. Replace num_stmts with
10230 split_stores.length (). Use 32-bit stack allocated entries
10231 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10232 if possible. Handle bitfields with gaps.
10233 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10234 Compute bitregion_{start,end} for the stores and construct
10235 store_immediate_info with that. Formatting fixes.
10237 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10240 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10241 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10243 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10245 * gimplify.c: Include tm_p.h.
10247 * common.opt (gcoff): Re-add as ignored option.
10248 (gcoff1, gcoff2, gcoff3): Likewise.
10250 * Makefile.in (OBJS): Delete sdbout.o.
10251 (GTFILES): Delete $(srcdir)/sdbout.c.
10252 * debug.h: Delete sdb_debug_hooks.
10253 * final.c: Delete sdbout.h include.
10254 (final_scan_insn): Delete SDB_DEBUG check.
10255 (rest_of_clean_state): Likewise.
10256 * output.h: Delete sdb_begin_function_line.
10257 * sdbout.c: Delete.
10258 * sdbout.h: Delete.
10259 * toplev.c: Delete sdbout.h include.
10260 (process_options): Delete SDB_DEBUG check.
10261 * tree-core.h (tree_type_common): Delete pointer field of
10263 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10264 TYPE_SYMTAB_POINTER.
10265 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10266 (TYPE_SYMTAB_IS_POINTER): Delete.
10267 (TYPE_SYMTAB_IS_DIE): Renumber.
10268 * xcoffout.c: Refer to former sdbout.c file.
10269 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10271 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10272 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10273 (Debugging Options): Delete -gcoff.
10274 (-gstabs): Delete SDB reference.
10276 (-gcoff@var{level}): Delete.
10277 * doc/passes.texi (Debugging information output): Delete SDB and
10278 sdbout.c references.
10279 * doc/tm.texi: Regenerate.
10280 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10281 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10283 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10284 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10285 (SDB_DEBUGGING_INFO): Delete.
10286 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10287 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10288 * target.def (output_source_filename): Delete COFF reference.
10290 * common.opt (gcoff): Delete.
10291 (gxcoff+): Update Negative chain.
10292 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10294 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10295 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10296 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10297 and SDB references.
10298 (expand_function_start): Change sdb reference to past tense.
10299 (expand_function_end): Change sdb reference to past tense.
10300 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10301 * opts.c (debug_type_names): Delete coff entry.
10302 (common_handle_option): Delete OPT_gcoff case.
10303 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10305 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10306 * config/cris/cris.h: Delete SDB reference in comment.
10307 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10308 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10309 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10310 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10312 (ix86_expand_prologue): Likewise.
10313 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10314 * config/ia64/ia64.h: Likewise.
10315 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10316 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10318 * config/mmix/mmix.h: Likewise.
10319 * config/nds32/nds32.c: Likewise.
10320 * config/stormy/storym16.h: Likewise.
10321 * config/visium/visium.h: Likewise.
10322 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10324 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10326 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10327 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10328 FRAME_POINTER_REGNUM point at high end of local var area.
10330 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10332 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10333 Move comment around. Do not reset best_edge for a copiable
10334 destination if the copy would cause a partition change.
10335 (better_edge_p): Remove redundant check.
10337 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10339 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10341 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10343 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10344 for math functions that have _Float<N> and _Float<N>X variants.
10345 (mathfn_built_in_2): Add support for math functions that have
10346 _Float<N> and _Float<N>X variants.
10347 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10348 (expand_builtin_mathfn_ternary): Add support for fma with
10349 _Float<N> and _Float<N>X variants.
10350 (expand_builtin): Likewise.
10351 (fold_builtin_3): Likewise.
10352 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10353 create math function _Float<N> and _Float<N>X variants as external
10355 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10356 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10357 the __builtin_ prefix and if not strict ansi, without the prefix.
10358 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10359 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10360 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10361 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10362 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10363 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10364 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10365 function signatures for fma _Float<N> and _Float<N>X variants.
10366 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10367 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10368 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10369 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10370 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10371 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10372 * gencfn-macros.c (print_case_cfn): Add support for math functions
10373 that have _Float<N> and _Float<N>X variants.
10374 (print_define_operator_list): Likewise.
10375 (fltfn_suffixes): Likewise.
10377 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10378 for math functions that have _Float<N> and _Float<N>X variants.
10379 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10380 and _Float<N>X variants.
10381 (COPYSIGN): Likewise.
10384 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10385 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10387 (integer_valued_read_call_p): Likewise.
10388 * fold-const-call.c (fold_const_call_ss): Likewise.
10389 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10390 _Float<N> and _Float<N>X variants.
10391 (fold_const_call_ssss): Add support for fma _Float<N> and
10392 _Float<N>X variants.
10393 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10394 support for copysign and fma _Float<N> and _Float<N>X variants.
10395 (backprop::process_builtin_call_use): Likewise.
10396 * tree-call-cdce.c (can_test_argument_range); Add support for
10397 sqrt _Float<N> and _Float<N>X variants.
10398 (edom_only_function): Likewise.
10399 (get_no_error_domain): Likewise.
10400 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10401 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10402 copysign _Float<N> and _Float<N>X variants.
10403 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10404 handled by machine independent code.
10405 (FMAF128): Likewise.
10406 * doc/cpp.texi (Common Predefined Macros): Document defining
10407 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10408 fma _Float<N> and _Float<N>X variants.
10410 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10413 * config/i386/i386-modes.def (CCFPU): Remove definition.
10414 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10415 (ix86_cc_modes_compatible): Ditto.
10416 (ix86_expand_carry_flag_compare): Ditto.
10417 (ix86_expand_int_movcc): Ditto.
10418 (ix86_expand_int_addcc): Ditto.
10419 (ix86_reverse_condition): Ditto.
10420 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10421 Return true/false for unordered/ordered fp comparisons.
10422 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10423 (ix86_prepare_fp_compare_args): Update for rename.
10424 (ix86_expand_fp_compare): Update for rename. Generate unordered
10425 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10426 (ix86_expand_sse_compare_and_jump): Ditto.
10427 * config/i386/predicates.md (fcmov_comparison_operator):
10428 Remove CCFPU mode handling.
10429 (ix86_comparison_operator): Ditto.
10430 (ix86_carry_flag_operator): Ditto.
10431 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10432 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10433 (*cmpu<mode>_cc_i387): Ditto.
10434 (FPCMP): Remove mode iterator.
10435 (unord): Remove mode attribute.
10436 (unord_subst): New define_subst transformation
10437 (unord): New define_subst attribute.
10438 (unordered): Ditto.
10439 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10440 (*cmpi<unord>xf_i387): Ditto.
10441 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10442 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10443 using unord_subst transformation.
10444 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10445 (round_saeonly): Also handle CCFP mode.
10446 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10447 Remove UNSPEC_SAHF unspec handling.
10449 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10451 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10453 2017-10-27 Jeff Law <law@redhat.com>
10455 * gimple-ssa-sprintf.c: Include domwalk.h.
10456 (class sprintf_dom_walker): New class, derived from dom_walker.
10457 (sprintf_dom_walker::before_dom_children): New function.
10458 (struct call_info): Moved into sprintf_dom_walker class
10459 (compute_formath_length, handle_gimple_call): Likewise.
10460 (sprintf_length::execute): Call the dominator walker rather
10461 than walking the statements.
10463 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10464 gimple statement locations.
10465 (check_array_bounds): Corresponding changes. Get the statement's
10466 location directly from wi->stmt.
10468 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10471 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10473 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10475 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10476 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10478 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10481 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10482 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10483 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10484 maybe_get_pool_constant.
10485 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10488 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10490 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10491 2.26 caveat. Update gas and gld versions.
10492 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10495 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10497 * cgraph.h (set_malloc_flag): Declare.
10498 * cgraph.c (set_malloc_flag_1): New function.
10499 (set_malloc_flag): Likewise.
10500 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10501 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10503 (read_ipa_call_summary): Add support for reading is_return_callee.
10504 (write_ipa_call_summary): Stream is_return_callee.
10505 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10506 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10507 ipa-prop.h, ipa-fnsummary.h.
10508 (pure_const_names): Change to static.
10509 (malloc_state_e): Define.
10510 (malloc_state_names): Define.
10511 (funct_state_d): Add field malloc_state.
10512 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10513 (check_retval_uses): New function.
10514 (malloc_candidate_p): Likewise.
10515 (analyze_function): Add support for malloc attribute.
10516 (pure_const_write_summary): Stream malloc_state.
10517 (pure_const_read_summary): Add support for reading malloc_state.
10518 (dump_malloc_lattice): New function.
10519 (propagate_malloc): New function.
10520 (warn_function_malloc): New function.
10521 (ipa_pure_const::execute): Call propagate_malloc and
10522 ipa_free_fn_summary.
10523 (pass_local_pure_const::execute): Add support for malloc attribute.
10524 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10525 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10527 2017-10-27 Martin Liska <mliska@suse.cz>
10529 PR gcov-profile/82457
10530 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10531 for fork-like functions to be properly instrumented.
10533 2017-10-27 Richard Biener <rguenther@suse.de>
10535 PR middle-end/81659
10536 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10537 info when we redirected EH.
10539 2017-10-26 Michael Collison <michael.collison@arm.com>
10541 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10543 (<optab>_trunchf<GPI:mode>2: New pattern.
10544 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10545 * config/aarch64/iterators.md (wv): New mode attribute.
10546 (vf, VF): New mode attributes.
10547 (vgp, VGP): New mode attributes.
10548 (s): Update attribute with SImode and DImode prefixes.
10550 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10552 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10553 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10554 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10555 (nios2_option_overide): Initialize it. Don't allow R0-relative
10556 addressing with PIC.
10557 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10558 (nios2_symbolic_constant_p): Likewise.
10559 (nios2_legitimate_address_p): Likewise.
10560 (nios2_r0rel_section_name_p): New.
10561 (nios2_symbol_ref_in_r0rel_data_p): New.
10562 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10563 (r0rel_constant_p): New.
10564 (nios2_print_operand_address): Handle r0rel_constant_p.
10565 (nios2_cdx_narrow_form_p): Likewise.
10566 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10567 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10568 (Nios II Options): Document -mr0rel-sec.
10570 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10572 * config/nios2/nios2.c: Include xregex.h.
10573 (nios2_gprel_sec_regex): New.
10574 (nios2_option_overide): Initialize it. Don't allow GP-relative
10575 addressing with PIC.
10576 (nios2_small_section_name_p): Check for regex match.
10577 * config/nios2/nios2.opt (mgprel-sec=): New option.
10578 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10579 (Nios II Options): Document -mgprel-sec.
10581 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10583 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10585 2017-10-26 Tom de Vries <tom@codesourcery.com>
10587 PR tree-optimization/82707
10588 * gimple.c (gimple_copy): Fix unsharing of
10589 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10591 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10593 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10594 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10595 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10596 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10597 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10598 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10599 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10600 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10601 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10602 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10603 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10604 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10605 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10606 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10607 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10608 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10609 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10611 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10613 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10615 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10616 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10617 warning to rs6000.c. Remove the Undocumented flag, since it has
10619 (-mabi=ibmlongdouble): Likewise.
10620 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10621 already set, set the default format for long double.
10622 (rs6000_debug_reg_global): Print whether long double is IBM or
10624 (rs6000_option_override_internal): Rework setting long double
10625 format. Only warn if the user is changing the long double default
10626 and they did not use -Wno-psabi.
10627 * doc/invoke.texi (PowerPC options): Update the documentation for
10628 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10630 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10631 Alan Hayward <alan.hayward@arm.com>
10632 David Sherwood <david.sherwood@arm.com>
10634 * rtl.h (wider_subreg_mode): New function.
10635 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10636 rather than an unsigned int *.
10637 * ira-color.c (regno_max_ref_width): Replace with...
10638 (regno_max_ref_mode): ...this new variable.
10639 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10640 Use wider_subreg_mode.
10641 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10642 rather than an unsigned int *.
10643 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10644 (process_alt_operands): Likewise.
10645 (invariant_p): Likewise.
10646 * lra-spills.c (assign_mem_slot): Likewise.
10647 (add_pseudo_to_slot): Likewise.
10648 * lra.c (collect_non_operand_hard_regs): Likewise.
10649 (add_regs_to_insn_regno_info): Likewise.
10650 * reload1.c (regno_max_ref_width): Replace with...
10651 (regno_max_ref_mode): ...this new variable.
10652 (reload): Update accordingly. Update call to
10653 ira_sort_regnos_for_alter_reg.
10654 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10655 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10656 (scan_paradoxical_subregs): Likewise.
10658 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10660 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10661 (aarch64_frame): Add emit_frame_chain boolean.
10662 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10663 Move eh_return case to aarch64_layout_frame.
10664 (aarch64_layout_frame): Initialize emit_frame_chain.
10665 (aarch64_expand_prologue): Use emit_frame_chain.
10667 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10669 * config/aarch64/aarch64.c (aarch64_layout_frame):
10670 Ensure LR is always stored at the bottom of the callee-saves.
10671 Remove rarely used frame layout which saves callee-saves at top of
10672 frame, so the store of LR can be used as a valid probe in all cases.
10674 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10676 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10677 Improve unaligned TImode/TFmode base/offset split.
10679 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10680 Alan Hayward <alan.hayward@arm.com>
10681 David Sherwood <david.sherwood@arm.com>
10683 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10684 * combine.c (find_single_use_1): Likewise.
10685 (expand_field_assignment): Likewise.
10686 (move_deaths): Likewise.
10687 * lra-constraints.c (simplify_operand_subreg): Likewise.
10688 (curr_insn_transform): Likewise.
10689 * lra.c (collect_non_operand_hard_regs): Likewise.
10690 (add_regs_to_insn_regno_info): Likewise.
10691 * rtlanal.c (reg_referenced_p): Likewise.
10692 (covers_regno_no_parallel_p): Likewise.
10694 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10696 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10697 Don't print any bits outside the precision of the value.
10698 * wide-int.cc (test_printing): Add some new tests.
10700 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10702 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10703 supports -xbrace_comment option.
10704 * configure: Regenerate.
10705 * config.in: Regenerate.
10706 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10707 (ASM_CPU_SPEC): Use it.
10709 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10711 * target.def (static_rtx_alignment): New hook.
10712 * targhooks.h (default_static_rtx_alignment): Declare.
10713 * targhooks.c (default_static_rtx_alignment): New function.
10714 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10715 * doc/tm.texi: Regenerate.
10716 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10717 instead of targetm.constant_alignment. Remove call to
10718 set_mem_attributes.
10719 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10720 (cris_preferred_mininum_alignment): New function, split out from...
10721 (cris_constant_alignment): ...here.
10722 (cris_static_rtx_alignment): New function.
10723 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10725 (ix86_constant_alignment): ...here.
10726 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10727 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10728 (mmix_static_rtx_alignment): New function.
10729 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10730 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10732 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10735 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10736 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10738 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10740 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10741 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10742 mark as a sign-extending load.
10743 (local_pic_load_u): Define.
10745 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10747 PR middle-end/82062
10748 * fold-const.c (operand_equal_for_comparison_p): Also return true
10749 if ARG0 is a simple variant of ARG1 with narrower precision.
10750 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10752 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10754 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10756 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10757 scatter_static, scatter_per_elt.
10758 * x86-tune-costs.h: Add new cost entries.
10760 2017-10-25 Richard Biener <rguenther@suse.de>
10762 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10763 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10764 class pass_fre): Move to ...
10765 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10766 class pass_fre): ... here and adjust for statistics.
10768 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10771 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10772 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10773 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10774 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10777 2017-10-25 Richard Biener <rguenther@suse.de>
10779 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10780 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10781 eliminate_push_avail, eliminate_insert): Move inside...
10782 (class eliminate_dom_walker): ... this class in preparation
10784 (fini_eliminate): Remove by merging with ...
10785 (eliminate): ... this function. Adjust for class changes.
10786 (pass_pre::execute): Remove fini_eliminate call.
10787 (pass_fre::execute): Likewise.
10789 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10792 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10793 (VPERMI2, VPERMI2I): New mode iterators.
10794 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10795 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10797 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10798 mode iterator. Remove 3 old define_insn patterns.
10799 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10800 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10801 VPERMI2 mode iterator, remove the other two expanders.
10802 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10803 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10804 instructions, remove the other two patterns.
10805 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10806 mode iterator, remove the other two patterns.
10807 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10808 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10809 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10810 and adjust argument order accordingly.
10811 (ix86_expand_vec_perm): Adjust caller.
10812 (expand_vec_perm_1): Likewise.
10813 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10814 (expand_vec_perm_vpermt2_vpshub2): ... this.
10815 (ix86_expand_vec_perm_const_1): Adjust caller.
10816 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10819 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10820 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10821 (vec_shl_<mode>): Remove unused expander.
10822 (avx512bw_<shift_insn><mode>3): New define_insn.
10823 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10824 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10826 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10829 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10832 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10834 PR rtl-optimization/82396
10835 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10836 (autopref_multipass_init): Simplify initialization.
10837 (autopref_rank_data): Simplify sort order.
10838 * gcc/sched-int.h (autopref_multipass_data_): Remove
10839 multi_mem_insn_p, min_offset and max_offset.
10841 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10843 PR middle-end/60580
10844 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10845 Check special value of flag_omit_frame_pointer.
10846 (aarch64_can_eliminate): Likewise.
10847 (aarch64_override_options_after_change_1): Simplify handling of
10848 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10850 2017-10-24 Richard Biener <rguenther@suse.de>
10852 PR tree-optimization/82697
10853 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10854 zero for conditional load and unconditional store.
10856 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10858 * doc/install.texi: Document bootstrap-cet.
10860 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10863 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10864 ENDBR instruction at function entrance if function is only
10867 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10870 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10871 patterns to better describe from which operation the CF is computed.
10872 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10873 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10874 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10875 is 0, use _0 suffixed expanders instead of emitting a comparison
10878 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10880 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10881 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10883 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10885 PR middle-end/82569
10886 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10887 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10888 * loop-iv.c (iv_get_reaching_def): Likewise.
10889 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10890 variable is promoted and the partition contains undefined values.
10892 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10894 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10896 (nios2_address_cost): Define.
10897 (nios2_legitimize_address): Recognize (exp + constant) directly.
10898 (TARGET_ADDRESS_COST): Define.
10900 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10902 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10903 (nios2_symbolic_memory_operand_p): Declare.
10904 (nios2_split_large_constant): Declare.
10905 (nios2_split_symbolic_memory_operand): Declare.
10906 * config/nios2/nios2.c: Adjust includes.
10907 (nios2_symbolic_constant_allowed): New.
10908 (nios2_symbolic_constant_p): New.
10909 (nios2_plus_symbolic_constant_p): New.
10910 (nios2_valid_addr_expr_p): Recognize addresses involving
10911 symbolic constants.
10912 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10913 (nios2_symbolic_memory_operand_p): New.
10914 (nios2_large_constant_p): New.
10915 (nios2_split_large_constant): New.
10916 (nios2_split_plus_large_constant): New.
10917 (nios2_split_symbolic_memory_operand): New.
10918 (nios2_legitimize_address): Code refactoring. Handle addresses
10919 involving symbolic constants.
10920 (nios2_emit_move_sequence): Likewise.
10921 (nios2_print_operand): Improve error output.
10922 (nios2_print_operand_address): Handle LO_SUM.
10923 (nios2_cdx_narrow_form_p): Likewise.
10924 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10925 operands involving symbolic constants.
10926 (movhi_internal, movsi_internal): Likewise.
10927 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10928 (extendhisi2, extendqi<mode>2): Likewise.
10930 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10932 * tree-pass.h (PROP_rtl_split_insns): Define.
10933 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10935 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10937 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10939 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10942 * target.def (const_not_ok_for_debug_p): Default to
10943 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10944 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10945 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10946 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10947 which targetm.const_not_ok_for_debug_p returned true.
10948 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10950 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10952 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10953 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10954 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10956 (ix86_const_not_ok_for_debug_p): New function.
10957 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10958 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10960 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10963 * system.h: Conditionally include "unique-ptr.h" if
10964 INCLUDE_UNIQUE_PTR is defined.
10965 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10966 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10968 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10970 * config/rl78/rl78.md: New define_expand "subdi3".
10972 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10975 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10976 DF_REF_INSN if DF_REF_INSN_INFO is false.
10978 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10980 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10981 xmm_move instead of sse_move.
10982 (sse_store_index): New function.
10983 (ix86_register_move_cost): Be more sensible about mismatch stall;
10984 model AVX moves correctly; make difference between sse->integer and
10986 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10987 moves; make difference between SSE and AVX.
10988 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10989 and zmm_move. Increase size of sse load and store tables;
10990 add unaligned load and store tables; add ssemmx_to_integer.
10991 * x86-tune-costs.h: Update all entries according to real
10992 move latencies from Agner Fog's manual and chip documentation.
10994 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10997 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10998 * config/i386/constraints.md (Wf): New constraint.
10999 * config/i386/i386.md (UNSPEC_SBB): New unspec.
11000 (cmp<dwi>_doubleword): Removed.
11001 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
11002 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
11003 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
11004 expand with cmp<dwi>_doubleword. For LTU and GEU use
11005 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
11007 * common.opt (gcolumn-info): Enable by default.
11008 * doc/invoke.texi (gcolumn-info): Document new default.
11010 2017-10-23 Richard Biener <rguenther@suse.de>
11012 PR tree-optimization/82672
11013 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
11014 Fold the stmt if we propagated into it.
11016 2017-10-23 Richard Biener <rguenther@suse.de>
11018 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
11019 (bitmap_remove_expr_from_set): ... this. All callers call this
11020 for non-constant values.
11021 (bitmap_set_subtract): Rename to...
11022 (bitmap_set_subtract_expressions): ... this. Adjust and
11024 (bitmap_set_contains_value): Remove superfluous check.
11025 (bitmap_set_replace_value): Inline into single caller ...
11026 (bitmap_value_replace_in_set): ... here and simplify.
11027 (dependent_clean): Merge into ...
11028 (clean): ... this using an overload. Adjust.
11029 (prune_clobbered_mems): Adjust.
11030 (compute_antic_aux): Likewise.
11031 (compute_partial_antic_aux): Likewise.
11033 2017-10-23 Richard Biener <rguenther@suse.de>
11035 PR tree-optimization/82129
11037 2017-08-01 Richard Biener <rguenther@suse.de>
11039 PR tree-optimization/81181
11040 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11041 (compute_antic): ... end of iteration here.
11043 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11045 * target.def (starting_frame_offset): New hook.
11046 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
11047 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
11048 * doc/tm.texi.in: Regenerate.
11049 * hooks.h (hook_hwi_void_0): Declare.
11050 * hooks.c (hook_hwi_void_0): New function.
11051 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
11052 STARTING_FRAME_OFFSET.
11053 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
11054 * reload1.c (reload): Likewise.
11055 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
11056 instead of STARTING_FRAME_OFFSET.
11057 * function.c (try_fit_stack_local): Likewise.
11058 (assign_stack_local_1): Likewise
11059 (instantiate_virtual_regs): Likewise.
11060 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11061 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11062 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11063 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11064 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11065 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11066 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11067 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11068 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11069 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11070 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11071 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11072 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11073 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11074 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11075 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11076 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11077 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11078 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11079 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11080 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11081 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11082 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11083 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11084 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11085 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11086 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11087 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11088 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11089 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11090 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11091 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11092 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11093 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11094 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11095 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11096 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11097 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11098 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11099 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11100 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11101 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11102 return a HOST_WIDE_INT.
11103 (avr_builtin_setjmp_frame_value): Use it instead of
11104 STARTING_FRAME_OFFSET.
11105 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11106 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11107 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11109 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11110 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11111 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11112 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11113 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11114 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11115 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11116 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11117 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11118 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11119 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11120 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11122 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11123 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11124 * config/mips/mips.c (mips_compute_frame_info): Refer to
11125 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11126 (mips_starting_frame_offset): New function.
11127 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11128 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11129 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11130 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11131 and return a HOST_WIDE_INT.
11132 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11133 (mmix_initial_elimination_offset): Refer to
11134 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11135 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11136 * config/pa/pa.c (pa_starting_frame_offset): New function.
11137 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11138 (pa_expand_prologue): Likewise.
11139 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11140 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11141 !FRAME_GROWS_DOWNWARD handling to...
11142 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11143 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11144 !FRAME_GROWS_DOWNWARD handling to...
11145 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11146 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11147 !FRAME_GROWS_DOWNWARD handling to...
11148 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11149 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11151 (rs6000_starting_frame_offset): New function.
11152 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11153 !FRAME_GROWS_DOWNWARD handling to...
11154 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11155 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11156 !FRAME_GROWS_DOWNWARD handling to...
11157 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11158 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11159 !FRAME_GROWS_DOWNWARD handling to...
11160 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11161 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11162 (rs6000_starting_frame_offset): New function.
11163 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11164 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11165 * config/vax/vax.c (vax_starting_frame_offset): New function.
11166 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11167 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11168 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11169 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11170 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11171 * system.h (STARTING_FRAME_OFFSET): Poison.
11173 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11175 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11176 SCALAR_TYPE_MODE instead of TYPE_MODE.
11178 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11179 Alan Hayward <alan.hayward@arm.com>
11180 David Sherwood <david.sherwood@arm.com>
11182 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11184 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11185 Alan Hayward <alan.hayward@arm.com>
11186 David Sherwood <david.sherwood@arm.com>
11188 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11190 2017-10-23 Richard Biener <rguenther@suse.de>
11192 PR tree-optimization/82129
11193 * tree-ssa-pre.c (bitmap_set_and): Remove.
11194 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11195 canonicalizing expressions in the set to those with lowest
11196 ID rather than taking that from the first edge.
11198 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11200 * combine.c (rtx_equal_for_field_assignment_p): Use
11201 byte_lowpart_offset.
11203 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11204 Alan Hayward <alan.hayward@arm.com>
11205 David Sherwood <david.sherwood@arm.com>
11207 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11208 to a SUBREG_PROMOTED_VAR.
11210 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11211 Alan Hayward <alan.hayward@arm.com>
11212 David Sherwood <david.sherwood@arm.com>
11214 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11215 (expand_debug_source_expr): Likewise.
11216 * combine.c (combine_simplify_rtx): Likewise.
11217 * cse.c (fold_rtx): Likewise.
11218 * optabs.c (expand_float): Likewise.
11219 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11220 (simplify_binary_operation_1): Likewise.
11222 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11223 Alan Hayward <alan.hayward@arm.com>
11224 David Sherwood <david.sherwood@arm.com>
11226 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11227 (record_promoted_value): Likewise.
11228 * expr.c (expand_expr_real_2): Likewise.
11229 * ree.c (update_reg_equal_equiv_notes): Likewise.
11230 (combine_set_extension): Likewise.
11231 * rtlanal.c (low_bitmask_len): Likewise.
11232 * simplify-rtx.c (neg_const_int): Likewise.
11233 (simplify_binary_operation_1): Likewise.
11235 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11236 Alan Hayward <alan.hayward@arm.com>
11237 David Sherwood <david.sherwood@arm.com>
11239 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11240 * regcprop.c (maybe_mode_change): Likewise.
11241 * reload1.c (alter_reg): Likewise.
11243 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11245 * inchash.h (inchash::hash::add_wide_int): New function.
11246 * lto-streamer-out.c (hash_tree): Use it.
11248 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11250 * inchash.h (inchash::hash::add_wide_int): Rename to...
11251 (inchash::hash::add_hwi): ...this.
11252 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11253 (polymorphic_call_target_hasher::hash): Likewise.
11254 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11255 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11256 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11257 * lto-streamer-out.c (hash_tree): Likewise.
11258 * optc-save-gen.awk: Likewise.
11259 * tree.c (add_expr): Likewise.
11261 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11264 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11265 for ordered inequality comparisons even with TARGET_IEEE_FP.
11267 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11270 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11271 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11272 Expand with cmp<dwi>_doubleword.
11274 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11276 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11277 List CET intrinsics.
11278 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11279 specific to -fcf-protection option.
11281 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11283 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11284 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11285 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11286 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11287 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11288 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11289 (extra_objs): Add cet.o for Linux/x86 targets.
11290 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11291 * config/i386/cet.c: New file.
11292 * config/i386/cetintrin.h: Likewise.
11293 * config/i386/t-cet: Likewise.
11294 * config/i386/cpuid.h (bit_SHSTK): New.
11295 (bit_IBT): Likewise.
11296 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11297 pass IBT and SHSTK bits.
11298 * config/i386/i386-builtin-types.def
11299 (VOID_FTYPE_UNSIGNED_PVOID): New.
11300 (VOID_FTYPE_UINT64_PVOID): Likewise.
11301 * config/i386/i386-builtin.def: Add CET intrinsics.
11302 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11303 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11304 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11305 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11307 * config/i386/i386.c (rest_of_insert_endbranch): New.
11308 (pass_data_insert_endbranch): Likewise.
11309 (pass_insert_endbranch): Likewise.
11310 (make_pass_insert_endbranch): Likewise.
11311 (ix86_notrack_prefixed_insn_p): Likewise.
11312 (ix86_target_string): Add -mibt, -mshstk flags.
11313 (ix86_option_override_internal): Add flag_cf_protection
11315 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11316 (ix86_print_operand): Add 'notrack' prefix output.
11317 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11318 (ix86_expand_builtin): Expand CET intrinsics.
11319 (x86_output_mi_thunk): Add 'endbranch' instruction.
11320 * config/i386/i386.h (TARGET_IBT): New.
11321 (TARGET_IBT_P): Likewise.
11322 (TARGET_SHSTK): Likewise.
11323 (TARGET_SHSTK_P): Likewise.
11324 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11325 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11326 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11327 (builtin_setjmp_setup): New pattern.
11328 (builtin_longjmp): Likewise.
11329 (rdssp<mode>): Likewise.
11330 (incssp<mode>): Likewise.
11331 (saveprevssp): Likewise.
11332 (rstorssp): Likewise.
11333 (wrss<mode>): Likewise.
11334 (wruss<mode>): Likewise.
11335 (setssbsy): Likewise.
11336 (clrssbsy): Likewise.
11337 (nop_endbr): Likewise.
11338 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11340 * config/i386/immintrin.h: Include <cetintrin.h>.
11341 * config/i386/linux-common.h
11342 (file_end_indicate_exec_stack_and_cet): New prototype.
11343 (TARGET_ASM_FILE_END): New.
11345 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11347 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11348 latencies instead of having separate table; make difference between
11349 integer and float costs.
11350 * i386.h (processor_costs): Remove scalar_stmt_cost,
11351 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11352 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11354 * x86-tune-costs.h: Remove entries which has been removed in
11355 procesor_costs from all tables; make cond_taken_branch_cost
11356 and cond_not_taken_branch_cost COST_N_INSNS based.
11358 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11360 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11362 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11364 * config/i386/i386.md (isa): Remove fma_avx512f.
11365 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11366 <avx512>_fmadd_<mode>_mask3<round_name>,
11367 <avx512>_fmsub_<mode>_mask<round_name>,
11368 <avx512>_fmsub_<mode>_mask3<round_name>,
11369 <avx512>_fnmadd_<mode>_mask<round_name>,
11370 <avx512>_fnmadd_<mode>_mask3<round_name>,
11371 <avx512>_fnmsub_<mode>_mask<round_name>,
11372 <avx512>_fnmsub_<mode>_mask3<round_name>,
11373 <avx512>_fmaddsub_<mode>_mask<round_name>,
11374 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11375 <avx512>_fmsubadd_<mode>_mask<round_name>,
11376 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11377 (*vec_widen_umult_even_v16si<mask_name>,
11378 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11379 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11381 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11383 * extend.texi: Add 'nocf_check' documentation.
11384 * gimple.texi: Add second parameter to
11385 gimple_build_call_from_tree.
11386 * invoke.texi: Add -fcf-protection documentation.
11387 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11389 2017-10-20 Richard Biener <rguenther@suse.de>
11391 PR tree-optimization/82473
11392 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11393 the largest input type.
11395 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11397 * c-attribs.c (handle_nocf_check_attribute): New function.
11398 (c_common_attribute_table): Add 'nocf_check' handling.
11399 * gimple-parser.c: Add second argument NULL to
11400 gimple_build_call_from_tree.
11401 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11402 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11404 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11405 * common.opt: Add fcf-protection flag.
11406 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11407 * flag-types.h: Add enum cf_protection_level.
11408 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11409 Add 'nocf_check' attribute propagation to gimple call.
11410 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11411 (gimple_build_call_from_tree): Update prototype.
11412 (gimple_call_nocf_check_p): New function.
11413 (gimple_call_set_nocf_check): Likewise.
11414 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11415 * ipa-icf.c: Add nocf_check attribute in statement hash.
11416 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11417 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11418 * toplev.c (process_options): Add flag_cf_protection handling.
11420 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11422 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11424 2017-10-20 Richard Biener <rguenther@suse.de>
11426 PR tree-optimization/82603
11427 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11428 remove false predicated stores.
11430 2017-10-20 Richard Biener <rguenther@suse.de>
11432 * graphite-isl-ast-to-gimple.c
11433 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11434 Remove return value and simplify, dump copied stmt after lhs
11436 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11437 Reduce dump verbosity.
11438 (gsi_insert_earliest): Likewise.
11439 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11440 * graphite.c (print_global_statistics): Adjust dumping.
11441 (print_graphite_scop_statistics): Likewise.
11442 (print_graphite_statistics): Do not dump loops here.
11443 (graphite_transform_loops): But here.
11445 2017-10-20 Nicolas Roche <roche@adacore.com>
11447 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11448 * configure: Regenerate.
11450 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11453 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11454 functions when optimizing replace GIMPLE_RETURN stmts with
11455 calls to __builtin_unreachable ().
11458 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11459 for -fsanitize=thread link of executables.
11460 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11461 link of executables.
11464 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11465 New mode iterators.
11466 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11467 define_insns for logical vector shifts to use VI248_AVX512BW
11468 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11469 condition, useless isa and prefix attributes. Change the first
11470 2 of these define_insns to ...
11471 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11472 define_insn for avx512vl.
11473 (<shift_insn><mode>3): ... and this, new define_insn without
11474 masking for non-avx512vl.
11477 * config/i386/sse.md (*andnot<mode>3,
11478 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11479 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11480 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11481 not applied use empty suffix even for TARGET_AVX512VL.
11482 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11483 is applied, supply evex,evex or evex,evex,evex instead of just
11486 2017-10-20 Julia Koval <julia.koval@intel.com>
11488 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11489 (OPTION_MASK_ISA_GFNI_UNSET): New.
11490 (ix86_handle_option): Handle OPT_mgfni.
11491 * config/i386/cpuid.h (bit_GFNI): New.
11492 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11493 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11494 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11495 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11496 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11497 * config/i386/i386.opt: Add mgfni.
11499 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11501 * config/msp430/msp430.c (msp430_option_override): Disable
11502 -fdelete-null-pointer-checks.
11503 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11505 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11507 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11508 of x87 and SSE instructions.
11510 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11512 * asan.c (create_cond_insert_point): Do not update edge count.
11513 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11514 (afdo_propagate_circuit): Likewise.
11515 (afdo_calculate_branch_prob): Likewise.
11516 (afdo_annotate_cfg): Likewise.
11517 * basic-block.h (struct edge_def): Remove count.
11518 (edge_def::count): New accessor.
11519 * bb-reorder.c (rotate_loop): Update.
11520 (find_traces_1_round): Update.
11521 (connect_traces): Update.
11522 (sanitize_hot_paths): Update.
11523 * cfg.c (unchecked_make_edge): Update.
11524 (make_single_succ_edge): Update.
11525 (check_bb_profile): Update.
11526 (dump_edge_info): Update.
11527 (update_bb_profile_for_threading): Update.
11528 (scale_bbs_frequencies_int): Update.
11529 (scale_bbs_frequencies_gcov_type): Update.
11530 (scale_bbs_frequencies_profile_count): Update.
11531 (scale_bbs_frequencies): Update.
11532 * cfganal.c (connect_infinite_loops_to_exit): Update.
11533 * cfgbuild.c (compute_outgoing_frequencies): Update.
11534 (find_many_sub_basic_blocks): Update.
11535 * cfgcleanup.c (try_forward_edges): Update.
11536 (try_crossjump_to_edge): Update
11537 * cfgexpand.c (expand_gimple_cond): Update
11538 (expand_gimple_tailcall): Update
11539 (construct_exit_block): Update
11540 * cfghooks.c (verify_flow_info): Update
11541 (redirect_edge_succ_nodup): Update
11542 (split_edge): Update
11543 (make_forwarder_block): Update
11544 (duplicate_block): Update
11545 (account_profile_record): Update
11546 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11547 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11548 * cfgloopmanip.c (scale_loop_profile): Update.
11550 (lv_adjust_loop_entry_edge): Update.
11551 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11552 (force_nonfallthru_and_redirect): Update.
11553 (purge_dead_edges): Update.
11554 (rtl_flow_call_edges_add): Update.
11555 * cgraphunit.c (init_lowered_empty_function): Update.
11556 (cgraph_node::expand_thunk): Update.
11557 * gimple-pretty-print.c (dump_probability): Update.
11558 (dump_edge_probability): Update.
11559 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11560 * haifa-sched.c (sched_create_recovery_edges): Update.
11561 * hsa-gen.c (convert_switch_statements): Update.
11562 * ifcvt.c (dead_or_predicable): Update.
11563 * ipa-inline-transform.c (inline_transform): Update.
11564 * ipa-split.c (split_function): Update.
11565 * ipa-utils.c (ipa_merge_profiles): Update.
11566 * loop-doloop.c (add_test): Update.
11567 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11568 * lto-streamer-in.c (input_cfg): Update.
11569 (input_function): Update.
11570 * lto-streamer-out.c (output_cfg): Update.
11571 * modulo-sched.c (sms_schedule): Update.
11572 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11573 * predict.c (maybe_hot_edge_p): Update.
11574 (unlikely_executed_edge_p): Update.
11575 (probably_never_executed_edge_p): Update.
11576 (dump_prediction): Update.
11577 (drop_profile): Update.
11578 (propagate_unlikely_bbs_forward): Update.
11579 (determine_unlikely_bbs): Update.
11580 (force_edge_cold): Update.
11581 * profile.c (compute_branch_probabilities): Update.
11582 * reg-stack.c (better_edge): Update.
11583 * shrink-wrap.c (handle_simple_exit): Update.
11584 * tracer.c (better_p): Update.
11585 * trans-mem.c (expand_transaction): Update.
11586 (split_bb_make_tm_edge): Update.
11587 * tree-call-cdce.c: Update.
11588 * tree-cfg.c (gimple_find_sub_bbs): Update.
11589 (gimple_split_edge): Update.
11590 (gimple_duplicate_sese_region): Update.
11591 (gimple_duplicate_sese_tail): Update.
11592 (gimple_flow_call_edges_add): Update.
11593 (insert_cond_bb): Update.
11594 (execute_fixup_cfg): Update.
11595 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11596 * tree-complex.c (expand_complex_div_wide): Update.
11597 * tree-eh.c (lower_resx): Update.
11598 (unsplit_eh): Update.
11599 (cleanup_empty_eh_move_lp): Update.
11600 * tree-inline.c (copy_edges_for_bb): Update.
11601 (freqs_to_counts): Update.
11602 (copy_cfg_body): Update.
11603 * tree-ssa-dce.c (remove_dead_stmt): Update.
11604 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11605 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11606 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11607 (unloop_loops): Update.
11608 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11609 * tree-ssa-loop-split.c (connect_loops): Update.
11610 (split_loop): Update.
11611 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11612 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11613 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11614 * tree-ssa-reassoc.c (branch_fixup): Update.
11615 * tree-ssa-tail-merge.c (replace_block_by): Update.
11616 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11617 (compute_path_counts): Update.
11618 (update_profile): Update.
11619 (recompute_probabilities): Update.
11620 (update_joiner_offpath_counts): Update.
11621 (estimated_freqs_path): Update.
11622 (freqs_to_counts_path): Update.
11623 (clear_counts_path): Update.
11624 (ssa_fix_duplicate_block_edges): Update.
11625 (duplicate_thread_path): Update.
11626 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11627 (case_bit_test_cmp): Update.
11628 (collect_switch_conv_info): Update.
11629 (gen_inbound_check): Update.
11630 (do_jump_if_equal): Update.
11631 (emit_cmp_and_jump_insns): Update.
11632 * tree-tailcall.c (decrease_profile): Update.
11633 (eliminate_tail_call): Update.
11634 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11635 (vect_do_peeling): Update.
11636 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11637 * ubsan.c (ubsan_expand_null_ifn): Update.
11638 (ubsan_expand_ptr_ifn): Update.
11639 * value-prof.c (gimple_divmod_fixed_value): Update.
11640 (gimple_mod_pow2): Update.
11641 (gimple_mod_subtract): Update.
11642 (gimple_ic): Update.
11643 (gimple_stringop_fixed_value): Update.
11645 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11648 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11650 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11652 PR rtl-optimization/82395
11653 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11654 based on non_spilled_static_chain_regno_p.
11656 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11658 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11659 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11660 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11662 2017-10-19 Martin Sebor <msebor@redhat.com>
11664 PR tree-optimization/82596
11665 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11667 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11669 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11670 (get_mem_refs_of_builtin_call): Likewise.
11671 * builtins.c (expand_builtin_apply): Adjust call to
11672 allocate_dynamic_stack_space.
11673 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11674 the third argument to allocate_dynamic_stack_space, otherwise -1.
11675 (expand_builtin): Deal with all alloca variants.
11676 (is_inexpensive_builtin): Likewise.
11677 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11678 * calls.c (special_function_p): Deal with all alloca variants.
11679 (initialize_argument_information): Adjust call to
11680 allocate_dynamic_stack_space.
11681 (expand_call): Likewise.
11682 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11683 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11684 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11685 use it for the stack usage computation.
11686 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11687 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11688 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11689 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11690 (in_loop_p): Remove first argument and useless check.
11691 (pass_walloca::execute): Remove useless test and adjust call to above.
11692 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11693 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11694 (gimplify_call_expr): Deal with all alloca variants.
11695 * hsa-gen.c (gen_hsa_alloca): Likewise.
11696 (gen_hsa_insns_for_call): Likewise.
11697 * ipa-pure-const.c (special_builtin_state): Likewise.
11698 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11699 * tree-object-size.c (alloc_object_size): Likewise.
11700 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11701 (call_may_clobber_ref_p_1): Likewise.
11702 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11703 (ccp_fold_stmt): Likewise.
11704 (optimize_stack_restore): Likewise.
11705 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11706 (mark_all_reaching_defs_necessary_1): Likewise.
11707 (propagate_necessity): Likewise.
11708 (eliminate_unnecessary_stmts): Likewise.
11709 * tree.c (build_common_builtin_nodes): Build
11710 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11711 (build_alloca_call_expr): New function.
11712 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11713 (CASE_BUILT_IN_ALLOCA): Likewise.
11714 (build_alloca_call_expr): Declare.
11715 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11717 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11720 * dwarf2out.c (new_die_raw): New static inline function.
11721 (new_die): Use it to create the DIE.
11722 (add_AT_external_die_ref): Likewise.
11723 (clone_die): Likewise.
11724 (clone_as_declaration): Likewise.
11725 (dwarf2out_vms_debug_main_pointer): Likewise.
11726 (base_type_die): Likewise. Remove early return for corner cases.
11727 Do not call add_pubtype on the DIE here.
11728 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11729 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11730 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11731 native order exists for base types, attach the DIE manually and call
11732 add_pubtype on it. Do not equate a reverse order DIE to the type.
11734 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11736 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11737 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11739 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11741 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11743 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11744 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11745 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11746 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11747 * opts.c (sanitizer_opts): Add builtin.
11748 * ubsan.c (instrument_builtin): New function.
11749 (pass_ubsan::execute): Call it.
11750 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11751 * doc/invoke.texi: Document -fsanitize=builtin.
11753 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11754 builtins, store max (log2 (align), 0) into uchar field instead of
11755 align into uptr field.
11756 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11757 store uchar 0 field instead of uptr 0 field.
11758 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11759 instead of passing one address of struct with 2 locations pass
11760 two addresses of structs with 1 location each.
11761 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11762 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11763 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11764 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11765 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11766 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11767 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11768 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11770 2017-10-19 Martin Liska <mliska@suse.cz>
11773 * file-find.c (remove_prefix): Remove.
11774 * file-find.h (remove_prefix): Likewise.
11775 * gcc-ar.c: Remove smartness of lookup.
11777 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11779 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11780 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11781 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11783 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11786 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11787 (setcc + and to xor + setcc): New peephole2.
11789 2017-10-19 Tom de Vries <tom@codesourcery.com>
11791 * doc/sourcebuild.texi (Test Directives, Variants of
11792 dg-require-support): Add dg-require-stack-size.
11794 2017-10-19 Martin Liska <mliska@suse.cz>
11797 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11798 that have a large alignment.
11799 (gimplify_target_expr): Likewise.
11801 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11803 PR rtl-optimization/82602
11804 * ira.c (rtx_moveable_p): Return false for volatile asm.
11806 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11809 * config/i386/i386-modes.def (CCGZ): New CC mode.
11810 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11811 * config/i386/predicates.md (ix86_comparison_operator):
11813 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11814 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11815 with double-word subtraction.
11816 (put_condition_code): Handle CCGZmode.
11818 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11820 * wide-int.cc (debug (const wide_int &)): New.
11821 (debug (const wide_int *)): New.
11822 (debug (const widest_int &)): New.
11823 (debug (const widest_int *)): New.
11825 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11827 PR middle-end/82556
11828 * lra-constraints.c (curr_insn_transform): Use non-input operand
11829 instead of output one for matched reload.
11831 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11833 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11834 (tree-ssa-loop-ivopts.h): New header file.
11835 (struct builtin_info): New fields.
11836 (classify_builtin_1): Compute and record base and offset parts for
11837 memset builtin partition by calling strip_offset.
11838 (offset_cmp, fuse_memset_builtins): New functions.
11839 (finalize_partitions): Fuse adjacent memset partitions by calling
11841 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11842 Expose the interface.
11843 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11845 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11847 PR tree-optimization/82574
11848 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11849 that data reference must be executed exactly once per iteration
11850 against the outermost loop in nest.
11851 (classify_partition): Update call to above function.
11853 2017-10-18 Richard Biener <rguenther@suse.de>
11855 PR tree-optimization/82591
11856 * graphite.c (graphite_transform_loops): Move code gen message
11858 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11859 Here. Handle scop_to_isl_ast failing.
11860 (scop_to_isl_ast): Limit the number of ISL operations.
11862 2017-10-18 Richard Biener <rguenther@suse.de>
11864 * graphite-isl-ast-to-gimple.c
11865 (translate_isl_ast_to_gimple::set_rename): Simplify.
11866 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11867 (graphite_copy_stmts_from_block): ... here.
11868 (copy_bb_and_scalar_dependences): Simplify.
11869 (add_parameters_to_ivs_params): Canonicalize.
11870 (generate_entry_out_of_ssa_copies): Simplify.
11871 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11872 by passing in ISL dimension.
11873 (parameter_index_in_region_1): Rename to ...
11874 (parameter_index_in_region): ... this.
11875 (extract_affine): Adjust assert, pass down parameter index.
11876 (add_param_constraints): Use range-info when available.
11877 (build_scop_context): Adjust.
11878 * sese.c (new_sese_info): Adjust.
11879 (free_sese_info): Likewise.
11880 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11881 Remove unused typedefs.
11882 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11884 2017-10-18 Martin Liska <mliska@suse.cz>
11886 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11888 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11890 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11891 (s390_sched_init): Do not reset s390_sched_state if we entered the
11892 current basic block via a fallthru edge and all others are unlikely.
11894 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11896 * config/s390/s390.c (NUM_SIDES): New variable.
11897 (LONGRUNNING_THRESHOLD): New variable.
11898 (LATENCY_FACTOR): New variable.
11899 (s390_sched_score): Decrease score for long-running instructions on
11901 (s390_sched_variable_issue): Perform bookkeeping for long-running
11904 2017-10-18 Richard Biener <rguenther@suse.de>
11906 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11907 Simplify with removal of the parameter rename map.
11908 (set_rename): Likewise.
11909 (should_copy_to_new_region): Likewise.
11910 (graphite_copy_stmts_from_block): Likewise.
11911 (copy_bb_and_scalar_dependences): Remove initialization of
11912 unused copied_bb_map.
11913 (copy_def): Remove.
11914 (copy_internal_parameters): Likewise.
11915 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11916 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11917 Use INTEGRAL_TYPE_P.
11918 (parameter_index_in_region_1): Rename to ...
11919 (assign_parameter_index_in_region): ... this. Assert we have
11920 a parameter we handle.
11921 (scan_tree_for_params): Adjust.
11922 * sese.h (parameter_rename_map_t): Remove.
11923 (struct sese_info_t): Remove unused parameter_rename_map and
11924 copied_bb_map members.
11925 * sese.c (new_sese_info): Adjust.
11926 (free_sese_info): Likewise.
11928 2017-10-18 Martin Liska <mliska@suse.cz>
11931 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11932 on an abnormal edge.
11934 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11936 * doc/invoke.texi (ffunction-sections and fdata-sections):
11939 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11941 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11942 the use statement can throw internally.
11944 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11946 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11947 any RTX present on the RHS of a SET.
11948 * compare-elim.c (try_eliminate_compare): Restore comment.
11950 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11952 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11953 may be also called on tcc_type nodes.
11954 * langhooks.c (lhd_tree_size): Likewise.
11956 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11958 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11959 format_warning_at_substring.
11960 (maybe_warn): Convert source_range * param to a location_t. Pass
11961 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11962 (format_directive): Remove code to extract source_ranges and
11963 source_range * in favor of just a location_t.
11964 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11966 * substring-locations.c (format_warning_va): Convert
11967 source_range * param to a location_t.
11968 (format_warning_at_substring): Likewise.
11969 * substring-locations.h (format_warning_va): Likewise.
11970 (format_warning_at_substring): Likewise.
11972 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11974 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11976 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11977 and scatter/gather ops.
11979 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11980 vec_gather_load and vec_scatter_store.
11981 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11982 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11983 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11984 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11985 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11986 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11988 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11990 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11991 Detect FTST instruction and handle its register pops. Only pop
11992 second operand if can_pop_second_op is true.
11993 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11994 set can_pop_second_op to false in the compare_for_stack_reg call.
11996 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11997 output_fp_compare for stack register operands.
11998 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11999 instructions here. Do not emit stack register pops here. Assert
12000 that FCOMPP pops next to top stack register. Rewrite function.
12002 2017-10-17 Nathan Sidwell <nathan@acm.org>
12004 PR middle-end/82577
12005 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
12006 use DECL_ASSEMBLER_NAME_RAW.
12008 PR middle-end/82546
12009 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
12012 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
12013 Wilco Dijkstra <wilco.dijkstra@arm.com>
12015 * builtins.c (expand_builtin_update_setjmp_buf): Add a
12016 converstion to Pmode from the buf_addr.
12018 2017-10-17 Richard Biener <rguenther@suse.de>
12020 * graphite-dependences.c (scop_get_reads_and_writes): Change
12021 output parameters to references.
12023 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
12025 PR 71026/tree-optimization
12026 * fold-const.c (distribute_real_division): Removed.
12027 (fold_binary_loc): Remove calls to distribute_real_divison.
12029 2017-10-17 Richard Biener <rguenther@suse.de>
12031 * graphite-scop-detection.c
12032 (scop_detection::stmt_has_simple_data_refs_p): Always use
12033 the full nest as region.
12034 (try_generate_gimple_bb): Likewise.
12035 * sese.c (scalar_evolution_in_region): Simplify now that
12036 SCEV can handle instantiation in regions.
12037 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
12038 in the non-loop part of a function if requested.
12040 2017-10-17 Richard Biener <rguenther@suse.de>
12042 PR tree-optimization/82563
12043 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
12045 (graphite_regenerate_ast_isl): Call it.
12046 * graphite-scop-detection.c (build_scops): Remove entry edge split.
12048 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12050 PR tree-optimization/82549
12051 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
12052 Formatting fixes. Instead of calling make_bit_field_ref with negative
12055 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
12057 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
12058 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12059 _mm_maskz_reduce_ss): New.
12060 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12061 __builtin_ia32_reducess_mask): Ditto..
12062 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12063 * config/i386/sse.md (reduces<mode>): Renamed to ...
12064 (reduces<mode><mask_scalar_name>): ... this.
12065 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12067 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12068 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12070 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12072 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12073 * selftest-run-tests.c (selftest::run_tests): Call
12074 selftest::unique_ptr_tests_cc_tests.
12075 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12076 * unique-ptr-tests.cc: New file.
12078 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12081 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12083 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12084 (make_hard_regno_born, make_hard_regno_dead): Update
12085 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12087 2017-10-16 Jeff Law <law@redhat.com>
12089 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12091 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12093 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12095 2017-10-16 Olivier Hainque <hainque@adacore.com>
12097 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12098 with_cpu if we were configured for an e500v2 target cpu name.
12100 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12102 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12103 * doc/invoke.texi: Document +nodsp as a valid extension for
12106 2017-10-16 Martin Liska <mliska@suse.cz>
12108 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12109 (test_set_range): Likewise.
12110 (test_range_functions): Rename to ...
12111 (test_bit_in_range): ... this.
12112 (sbitmap_c_tests): Add new test.
12114 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12116 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12118 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12119 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12121 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12123 * config/aarch64/aarch64-builtins.c
12124 (aarch64_types_quadopu_lane_qualifiers): New.
12125 (TYPES_QUADOPU_LANE): New.
12126 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12127 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12128 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12129 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12130 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12131 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12132 (Vdottype, DOTPROD): New.
12133 (sur): Add SDOT and UDOT.
12135 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12137 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12138 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12139 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12140 Add TARGET_DOTPROD.
12141 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12142 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12143 Enable TARGET_DOTPROD.
12144 (cortex-a75.cortex-a55): Likewise.
12145 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12147 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12149 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12150 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12152 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12154 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12155 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12156 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12157 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12158 * config/arm/types.md (neon_dot, neon_dot_q): New.
12159 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12161 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12163 * config/arm/arm.h (TARGET_DOTPROD): New.
12164 * config/arm/arm.c (arm_arch_dotprod): New.
12165 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12166 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12167 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12168 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12169 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12170 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12171 * doc/invoke.texi (armv8.2-a): Document dotprod
12173 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12175 * i386.c (ix86_vec_cost): New function.
12176 (ix86_rtx_costs): Handle vector operations better.
12177 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12178 * x86-tune-costs.h: Add new costs to all tables.
12180 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12182 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12184 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12185 divsd, sqrtss and sqrtsd
12186 * x86-tune-costs.h: Add new entries to all costs.
12187 (znver1_cost): Fix to match real instruction latencies.
12189 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12190 Michael Collison <michael.collison@arm.com>
12192 * compare-elim.c: Include emit-rtl.h.
12193 (can_merge_compare_into_arith): New function.
12194 (try_validate_parallel): Likewise.
12195 (try_merge_compare): Likewise.
12196 (try_eliminate_compare): Call the above when no previous clobber
12198 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12201 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12203 PR middle-end/62263
12204 PR middle-end/82498
12205 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12206 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12208 PR middle-end/62263
12209 PR middle-end/82498
12210 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12211 to be any operand_equal_p operands. For & (B - 1) require
12212 B to be power of 2. Recognize
12213 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12215 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12218 * optabs.c (expand_memory_blockage): Fix call of
12219 targetm.have_memory_blockage.
12221 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12224 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12225 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12226 objects to extra_objs instead of overwriting it.
12228 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12230 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12231 Use any_fp_register_operand as operand[3] predicate. Simplify
12232 equality test for operands[2] and operands[4] memory location.
12233 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12234 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12235 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12236 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12237 any_fp_register_operand as operand[1] predicate. Simplify
12238 equality test for operands[0] and operands[3] memory location.
12239 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12240 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12241 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12243 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12245 * target-insns.def: Add memory_blockage.
12246 * optabs.c (expand_memory_blockage): New function.
12247 (expand_asm_memory_barrier): Rename ...
12248 (expand_asm_memory_blockage): ... to this.
12249 (expand_mem_thread_fence): Call expand_memory_blockage
12250 instead of expand_asm_memory_barrier.
12251 (expand_mem_singnal_fence): Ditto.
12252 (expand_atomic_load): Ditto.
12253 (expand_atomic_store): Ditto.
12254 * doc/md.texi (Standard Pattern Names For Generation):
12255 Document memory_blockage instruction pattern.
12257 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12259 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12260 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12261 * config/rl78/rl78.md: New define_expand "adddi3".
12263 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12265 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12268 2017-10-13 Jeff Law <law@redhat.com>
12270 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12272 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12275 * internal-fn.c (expand_mul_overflow): If both operands have
12276 the same highpart of -1 or 0 and the topmost bit of lowpart
12277 is different, overflow is if res <= 0 rather than res < 0.
12279 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12281 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12282 TARGET_P9_VECTOR code for unaligned_load case.
12284 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12286 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12288 2017-10-13 Nathan Sidwell <nathan@acm.org>
12290 * tree-core.h (tree_contains_struct): Make bool.
12291 * tree.c (tree_contains_struct): Likewise.
12292 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12293 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12294 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12295 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12296 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12298 2017-10-13 Richard Biener <rguenther@suse.de>
12300 * graphite-isl-ast-to-gimple.c
12301 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12302 parameters and dominance check.
12303 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12304 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12305 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12306 Do not update SSA form here or do intermediate IL verification.
12307 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12308 (graphite_initialize): Remove check on the number of loops in
12309 the function and inline into graphite_transform_loops.
12310 (graphite_finalize): Inline into graphite_transform_loops.
12311 (graphite_transform_loops): Perform SSA update and IL verification
12313 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12315 2017-10-13 Richard Biener <rguenther@suse.de>
12317 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12318 graphite_expression_type_precision): Avoid global constructor
12320 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12321 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12322 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12323 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12324 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12325 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12327 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12330 * config/i386/i386.h (ix86_red_zone_size): New.
12331 * config/i386/i386.md (push peephole2s): Replace
12332 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12334 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12335 Alan Hayward <alan.hayward@arm.com>
12336 David Sherwood <david.sherwood@arm.com>
12338 * combine.c (can_change_dest_mode): Reject changes in
12339 REGMODE_NATURAL_SIZE.
12341 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12342 Alan Hayward <alan.hayward@arm.com>
12343 David Sherwood <david.sherwood@arm.com>
12345 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12346 (expand_debug_source_expr): Likewise.
12347 * combine.c (combine_simplify_rtx): Likewise.
12348 * cse.c (fold_rtx): Likewise.
12349 * fwprop.c (canonicalize_address): Likewise.
12350 * targhooks.c (default_shift_truncation_mask): Likewise.
12352 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12353 Alan Hayward <alan.hayward@arm.com>
12354 David Sherwood <david.sherwood@arm.com>
12356 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12357 (widened_mode): Likewise.
12358 (expand_unop): Likewise.
12359 * ree.c (transform_ifelse): Likewise.
12360 (merge_def_and_ext): Likewise.
12361 (combine_reaching_defs): Likewise.
12362 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12364 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12365 Alan Hayward <alan.hayward@arm.com>
12366 David Sherwood <david.sherwood@arm.com>
12368 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12369 * combine.c (gen_lowpart_for_combine): Likewise.
12370 * dwarf2out.c (rtl_for_decl_location): Likewise.
12371 * final.c (alter_subreg): Likewise.
12372 * rtlhooks.c (gen_lowpart_general): Likewise.
12373 (gen_lowpart_if_possible): Likewise.
12375 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12376 Alan Hayward <alan.hayward@arm.com>
12377 David Sherwood <david.sherwood@arm.com>
12379 * calls.c (expand_call): Use subreg_lowpart_offset.
12380 * cse.c (cse_insn): Likewise.
12381 * regcprop.c (copy_value): Likewise.
12382 (copyprop_hardreg_forward_1): Likewise.
12384 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12387 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12388 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12389 =Q constraints to +Q and into insn condition add check
12390 that operands[0] and operands[1] are equal.
12391 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12392 =Q constraints to +Q and into insn condition add check
12393 that operands[0] is equal to either operands[1] or operands[2].
12396 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12397 instead of handling MINUS_EXPR twice (once for each argument),
12398 canonicalize operand order and handle just once, use rtype where
12399 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12402 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12403 any values of __C while still being pattern recognizable as a simple
12404 rotate instruction.
12406 2017-10-13 Richard Biener <rguenther@suse.de>
12408 PR tree-optimization/82451
12410 2017-10-02 Richard Biener <rguenther@suse.de>
12412 PR tree-optimization/82355
12413 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12414 a mapping for the enclosing loop but avoid generating one for
12415 the loop tree root.
12416 (copy_bb_and_scalar_dependences): Remove premature codegen
12417 error on PHIs in blocks duplicated into multiple places.
12418 * graphite-scop-detection.c
12419 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12420 in the region use it as loop and nest to analyze the DR in.
12421 (try_generate_gimple_bb): Likewise.
12422 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12423 (add_loop_constraints): For blocks in a loop not in the region
12424 create a dimension with a single iteration.
12425 * sese.h (gbb_loop_at_index): Remove assert.
12427 * cfgloop.c (loop_preheader_edge): For the loop tree root
12428 return the single successor of the entry block.
12429 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12430 Reset the SCEV hashtable and niters.
12431 * graphite-scop-detection.c
12432 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12433 assert that we only have POLYNOMIAL_CHREC that vary in loops
12434 contained in the region.
12435 (scop_detection::graphite_can_represent_expr): Adjust.
12436 (scop_detection::stmt_has_simple_data_refs_p): For loops
12437 not in the region set loop to NULL. The nest is now the
12438 entry edge to the region.
12439 (try_generate_gimple_bb): Likewise.
12440 * sese.c (scalar_evolution_in_region): Adjust for
12441 instantiate_scev change.
12442 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12443 Make nest parameter the edge into the region.
12444 (create_data_ref): Likewise.
12445 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12446 entry edge into a region and adjust instantiate_scev calls.
12447 (create_data_ref): Likewise.
12448 (graphite_find_data_references_in_stmt): Likewise.
12449 (find_data_references_in_stmt): Pass the loop preheader edge
12450 from the nest argument.
12451 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12452 parameter the edge into the region.
12453 (instantiate_parameters): Use the loop preheader edge as entry.
12454 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12456 (get_instantiated_value_entry): Make instantiate_below parameter
12457 the edge into the region.
12458 (instantiate_scev_name): Likewise. Adjust dominance checks,
12459 when we cannot use loop-based instantiation instantiate by
12460 walking use-def chains.
12461 (instantiate_scev_poly): Adjust.
12462 (instantiate_scev_binary): Likewise.
12463 (instantiate_scev_convert): Likewise.
12464 (instantiate_scev_not): Likewise.
12465 (instantiate_array_ref): Remove.
12466 (instantiate_scev_3): Likewise.
12467 (instantiate_scev_2): Likewise.
12468 (instantiate_scev_1): Likewise.
12469 (instantiate_scev_r): Do not blindly handle N-operand trees.
12470 Do not instantiate array-refs. Handle all constants and invariants.
12471 (instantiate_scev): Make instantiate_below parameter
12472 the edge into the region.
12473 (resolve_mixers): Use the loop preheader edge for the region
12474 parameter to instantiate_scev_r.
12475 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12477 2017-10-13 Richard Biener <rguenther@suse.de>
12479 PR tree-optimization/82525
12480 * graphite-isl-ast-to-gimple.c
12481 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12483 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12484 Fail code generation when we cannot represent the isl integer.
12485 (binary_op_to_tree): Elide modulo operations that are no-ops
12486 in the type we code generate. Remove now superfluous code
12489 2017-10-13 Richard Biener <rguenther@suse.de>
12491 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12492 (scop_detection::harmful_loop_in_region): Remove premature
12493 IV type restriction.
12494 (scop_detection::graphite_can_represent_scev): We can handle
12495 pointer IVs just fine.
12497 2017-10-13 Alan Modra <amodra@gmail.com>
12499 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12500 "Clobbers and Scratch Registers". Add paragraph on
12501 alternative to clobbers for scratch registers and OpenBLAS
12504 2017-10-13 Alan Modra <amodra@gmail.com>
12506 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12507 example of a memory input for a string of known length. Move
12508 commentary out of table. Add a number of new examples
12509 covering array memory inputs.
12511 2017-10-12 Martin Liska <mliska@suse.cz>
12513 PR tree-optimization/82493
12514 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12515 (test_range_functions): New function.
12516 (sbitmap_c_tests): Likewise.
12517 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12518 * selftest.h (sbitmap_c_tests): New function.
12520 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12522 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12524 * config/rs6000/amo.h: Fix spacing issue.
12526 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12529 * config/i386/i386.md (*ashl<mode>3_mask_1,
12530 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12531 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12534 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12536 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12537 (profile_probability): Set max_probability
12538 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12540 * profile-count.c (profile_probability::differs_from_p): Do not
12541 rely on max_probaiblity == 10000
12543 2017-10-12 Jeff Law <law@redhat.com>
12545 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12548 2017-10-12 Martin Sebor <msebor@redhat.com>
12552 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12553 (handle_alias_pairs): Call it.
12554 * common.opt (-Wattribute-alias): New option.
12555 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12556 * doc/invoke.texi (-Wattribute-alias): Document.
12558 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12561 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12563 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12565 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12566 (make_hard_regno_born, make_hard_regno_dead): Update
12567 bb_killed_pseudos and bb_gen_pseudos.
12569 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12571 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12573 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12575 * config/alpha/alpha.c (alpha_split_conditional_move):
12576 Use std::swap instead of manually swapping.
12577 (alpha_stdarg_optimize_hook): Ditto.
12578 (alpha_canonicalize_comparison): Ditto.
12580 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12582 * tree-loop-distribution.c (struct builtin_info): New struct.
12583 (struct partition): Refactor fields into struct builtin_info.
12584 (partition_free): Free struct builtin_info.
12585 (build_size_arg_loc, build_addr_arg_loc): Delete.
12586 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12587 information from struct builtin_info.
12588 (find_single_drs): New function refactored from classify_partition.
12589 Also moved builtin validity checks to this function.
12590 (compute_access_range, alloc_builtin): New functions.
12591 (classify_builtin_st, classify_builtin_ldst): New functions.
12592 (classify_partition): Refactor code into functions find_single_drs,
12593 classify_builtin_st and classify_builtin_ldst.
12594 (distribute_loop): Don't do runtime alias check when distributing
12596 (find_seed_stmts_for_distribution): New function.
12597 (pass_loop_distribution::execute): Refactor code finding seed
12598 stmts into above function. Support distribution for the innermost
12599 two-level loop nest. Adjust dump information.
12601 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12603 * tree-loop-distribution.c: Adjust the general comment.
12604 (NUM_PARTITION_THRESHOLD): New macro.
12605 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12606 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12607 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12608 in call to build_partition_graph.
12609 (finalize_partitions): New parameter. Make loop distribution more
12610 conservative by fusing more partitions.
12611 (distribute_loop): Don't do runtime alias check in case of loop nest
12613 (find_seed_stmts_for_distribution): New function.
12614 (prepare_perfect_loop_nest): New function.
12615 (pass_loop_distribution::execute): Refactor code finding seed stmts
12616 and loop nest into above functions. Support loop nest distribution.
12617 Adjust dump information accordingly.
12619 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12621 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12622 and set PTYPE_SEQUENTIAL for merged partition.
12624 2017-10-12 Richard Biener <rguenther@suse.de>
12626 PR tree-optimization/69728
12628 2017-09-19 Richard Biener <rguenther@suse.de>
12630 PR tree-optimization/69728
12631 * graphite-sese-to-poly.c (schedule_error): New global.
12632 (add_loop_schedule): Handle empty domain by failing the
12634 (build_original_schedule): Handle schedule_error.
12636 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12637 domain by returning an unchanged schedule.
12639 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12641 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12642 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12644 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12646 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12649 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12652 * expr.c (store_field): Don't optimize away bitsize == 0 store
12653 from CALL_EXPR with addressable return type.
12655 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12657 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12658 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12659 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12660 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12661 TARGET_ISEL instead of TARGET_ISEL<sel>.
12663 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12665 * config/rs6000/rs6000.c
12666 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12668 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12670 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12672 (reg_or_cint_operand): Fix comment.
12673 (reg_or_zero_operand): New predicate.
12674 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12675 * config/rs6000/rs6000.c (output_isel): Delete.
12676 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12677 instead of reg_or_cint_operand. Output instruction directly (not via
12679 (isel_unsigned_<mode>): Ditto.
12680 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12681 gpc_reg_operand. Add an instruction alternative for this. Output
12682 instruction directly.
12683 (*isel_reversed_unsigned_<mode>): Ditto.
12685 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12687 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12688 (TARGET_CANONICALIZE_COMPARISON): Define.
12690 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12693 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12694 Check whether the dest is REG before adding REG_EQUIV note.
12696 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12699 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12701 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12702 (make_hard_regno_born, make_hard_regno_dead): Update
12703 bb_killed_pseudos and bb_gen_pseudos.
12705 2017-10-11 Nathan Sidwell <nathan@acm.org>
12707 * incpath.h (enum incpath_kind): Name enum, prefix values.
12708 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12709 * incpath.c (heads, tails): Use INC_MAX.
12710 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12711 (merge_include_chains, split_quote_chain,
12712 register_include_chains): Update incpath_kind names.
12713 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12714 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12716 (add_framework_path, darwin_register_objc_includes): Likewise.
12717 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12719 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12721 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12722 Do not use float_operator operator predicate.
12723 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12724 * config/i386/predicates.md (float_operator): Remove predicate.
12726 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12728 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12729 (*jccxf_i387): Ditto.
12730 (*jcc<mode>_i387): Ditto.
12731 (*jccu<mode>_i387): Ditto.
12732 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12733 (*jcc_*_i387 splitters): Remove.
12734 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12735 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12736 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12739 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12741 * profile-count.h (slow_safe_scale_64bit): New function.
12742 (safe_scale_64bit): New inline.
12743 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12744 * profile-count.c: Include wide-int.h
12745 (slow_safe_scale_64bit): New.
12747 2017-10-11 Nathan Sidwell <nathan@acm.org>
12749 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12750 HAS_DECL_ASSEMBLER_NAME_P.
12751 * gimple-expr.c (gimple_decl_printable_name: Check
12752 HAS_DECL_ASSEMBLER_NAME_P too.
12753 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12754 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12755 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12756 * passes.c (rest_of_decl_compilation): Check
12757 HAS_DECL_ASSEMBLER_NAME_P too.
12758 * recog.c (verify_changes): Likewise.
12759 * tree-pretty-print.c (dump_decl_name): Likewise.
12760 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12762 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12763 (SET_DECL_ASSEMBLER_NAME): Use it.
12764 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12765 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12766 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12768 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12770 * config.gcc (i386, x86_64): Add extra objects.
12771 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12772 (ix86_min_insn_size): Declare.
12773 (ix86_issue_rate): Declare.
12774 (ix86_adjust_cost): Declare.
12775 (ia32_multipass_dfa_lookahead): Declare.
12776 (ix86_macro_fusion_p): Declare.
12777 (ix86_macro_fusion_pair_p): Declare.
12778 (ix86_bd_has_dispatch): Declare.
12779 (ix86_bd_do_dispatch): Declare.
12780 (ix86_core2i7_init_hooks): Declare.
12781 (ix86_atom_sched_reorder): Declare.
12782 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12783 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12784 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12785 (rip_relative_addr_p): Rename to ...
12786 (ix86_rip_relative_addr_p): ... this one; export.
12787 (memory_address_length): Update.
12788 (ix86_issue_rate): Move to x86-tune-sched.c.
12789 (ix86_flags_dependent): Move to x86-tune-sched.c.
12790 (ix86_agi_dependent): Move to x86-tune-sched.c.
12791 (exact_dependency_1): Move to x86-tune-sched.c.
12792 (exact_store_load_dependency): Move to x86-tune-sched.c.
12793 (ix86_adjust_cost): Move to x86-tune-sched.c.
12794 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12795 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12796 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12797 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12798 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12799 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12800 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12801 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12802 (min_insn_size): Rename to ...
12803 (ix86_min_insn_size): ... this one; export.
12804 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12805 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12806 (core2i7_first_cycle_multipass_backtrack): Move to
12807 x86-tune-sched-core.c.
12808 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12809 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12810 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12811 (ix86_avoid_jump_mispredicts): Update.
12812 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12813 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12814 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12815 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12816 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12817 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12818 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12819 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12820 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12821 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12822 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12823 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12824 (BIG): Move to ix86-tune-sched-bd.c.
12825 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12826 (enum insn_path): Move to ix86-tune-sched-bd.c.
12827 (get_mem_group): Move to ix86-tune-sched-bd.c.
12828 (is_cmp): Move to ix86-tune-sched-bd.c.
12829 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12830 (is_branch): Move to ix86-tune-sched-bd.c.
12831 (is_prefetch): Move to ix86-tune-sched-bd.c.
12832 (init_window): Move to ix86-tune-sched-bd.c.
12833 (allocate_window): Move to ix86-tune-sched-bd.c.
12834 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12835 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12836 (process_end_window): Move to ix86-tune-sched-bd.c.
12837 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12838 (find_constant): Move to ix86-tune-sched-bd.c.
12839 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12840 (has_immediate): Move to ix86-tune-sched-bd.c.
12841 (get_insn_path): Move to ix86-tune-sched-bd.c.
12842 (get_insn_group): Move to ix86-tune-sched-bd.c.
12843 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12844 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12845 (add_insn_window): Move to ix86-tune-sched-bd.c.
12846 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12847 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12848 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12849 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12850 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12851 (do_dispatch): Move to ix86-tune-sched-bd.c.
12852 (has_dispatch): Move to ix86-tune-sched-bd.c.
12853 * config/i386/t-i386: Add new object files.
12854 * config/i386/x86-tune-costs.h: New file.
12855 * config/i386/x86-tune-sched-atom.c: New file.
12856 * config/i386/x86-tune-sched-bd.c: New file.
12857 * config/i386/x86-tune-sched-core.c: New file.
12858 * config/i386/x86-tune-sched.c: New file.
12860 2017-10-11 Liu Hao <lh_mouse@126.com>
12862 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12863 the generic version below instead.
12864 (should_colorize): Recognize Windows consoles as terminals
12866 * pretty-print.c [__MINGW32__] (write_all): New function.
12867 [__MINGW32__] (find_esc_head): Likewise.
12868 [__MINGW32__] (find_esc_terminator): Likewise.
12869 [__MINGW32__] (eat_esc_sequence): Likewise.
12870 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12872 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12875 2017-10-11 Richard Biener <rguenther@suse.de>
12877 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12878 Properly call analyze_scalar_evolution with the loop of the stmt.
12880 2017-10-11 Richard Biener <rguenther@suse.de>
12882 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12883 * tree-core.h (tree_base): Add chrec_var union member.
12884 * tree.h (CHREC_VAR): Remove.
12885 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12886 * tree-chrec.h (build_polynomial_chrec): Adjust.
12887 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12888 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12890 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12892 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12893 * match.pd: ... here.
12894 ((T) X == (T) Y): Relax condition.
12896 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12898 PR tree-optimization/82472
12899 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12901 (break_alias_scc_partitions): Update postorder number.
12903 2017-10-11 Martin Liska <mliska@suse.cz>
12906 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12908 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12910 2017-10-11 Martin Liska <mliska@suse.cz>
12912 * print-rtl.c (print_insn): Move declaration of idbuf
12913 to same scope as name.
12915 2017-10-11 Martin Liska <mliska@suse.cz>
12920 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12921 volatile arguments.
12923 2017-10-11 Martin Liska <mliska@suse.cz>
12926 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12927 volatile arguments.
12929 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12931 * config.gcc (default_gnu_indirect_function): Default to yes for
12932 arm*-*-linux* with glibc.
12934 2017-10-11 Richard Biener <rguenther@suse.de>
12936 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12937 default-defs and types we do not want to analyze.
12938 (interpret_loop_phi): Replace unreachable code with an assert.
12939 (compute_scalar_evolution_in_loop): Remove and inline ...
12940 (analyze_scalar_evolution_1): ... here, replacing condition with
12941 what makes the intent clearer. Remove handling of cases
12942 get_scalar_evolution now handles.
12944 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12946 PR rtl-optimization/81434
12947 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12948 comment for main loop. In sched_group_found if, also add checks for
12949 pass and min_cost_group.
12951 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12953 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12954 (rs6000_insn_cost): New function.
12955 * config/rs6000/rs6000.md (cost): New attribute.
12957 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12958 H.J. Lu <hongjiu.lu@intel.com>
12962 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12963 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12964 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12965 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12966 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12967 OPTION_MASK_ISA_MMX and some other bit set require both
12968 mmx and the ISAs without the mmx bit.
12969 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12970 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12971 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12972 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12973 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12974 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12975 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12976 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12977 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12978 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12979 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12980 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12981 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12982 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12983 Add OPTION_MASK_ISA_MMX.
12985 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12987 * config.gcc (armv7*-*-freebsd*): New target.
12988 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12990 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12992 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12993 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12996 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12998 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12999 a template parameter.
13000 (WIDE_INT_REF_FOR): Update accordingly.
13001 * tree.h (wi::int_traits <const_tree>): Delete.
13002 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
13003 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
13004 (wi::tree_to_wide_ref): New typedef.
13005 (wi::to_wide): New function.
13006 * calls.c (get_size_range): Use wi::to_wide when operating on
13007 trees as wide_ints.
13008 * cgraph.c (cgraph_node::create_thunk): Likewise.
13009 * config/i386/i386.c (ix86_data_alignment): Likewise.
13010 (ix86_local_alignment): Likewise.
13011 * dbxout.c (stabstr_O): Likewise.
13012 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
13013 * expr.c (const_vector_from_tree): Likewise.
13014 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
13015 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
13016 (fold_negate_expr_1, int_const_binop_1, const_binop)
13017 (fold_convert_const_int_from_real, optimize_bit_field_compare)
13018 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
13019 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
13020 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
13021 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
13022 (fold_not_const, round_up_loc): Likewise.
13023 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
13024 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
13025 (alloca_call_type): Likewise.
13026 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
13027 * godump.c (go_output_typedef): Likewise.
13028 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
13029 * internal-fn.c (get_min_precision): Likewise.
13030 * ipa-cp.c (ipcp_store_vr_results): Likewise.
13031 * ipa-polymorphic-call.c
13032 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
13033 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
13034 (ipa_modify_call_arguments): Likewise.
13035 * match.pd: Likewise.
13036 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
13037 * print-tree.c (print_node_brief, print_node): Likewise.
13038 * stmt.c (expand_case): Likewise.
13039 * stor-layout.c (layout_type): Likewise.
13040 * tree-affine.c (tree_to_aff_combination): Likewise.
13041 * tree-cfg.c (group_case_labels_stmt): Likewise.
13042 * tree-data-ref.c (dr_analyze_indices): Likewise.
13043 (prune_runtime_alias_test_list): Likewise.
13044 * tree-dump.c (dequeue_and_dump): Likewise.
13045 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
13046 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
13047 * tree-pretty-print.c (dump_generic_node): Likewise.
13048 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
13049 (simple_iv_with_niters): Likewise.
13050 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
13051 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
13052 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
13053 * tree-ssa-loop-niter.c (split_to_var_and_offset)
13054 (refine_value_range_using_guard, number_of_iterations_ne_max)
13055 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
13056 (get_cst_init_from_scev, record_nonwrapping_iv)
13057 (scev_var_range_cant_overflow): Likewise.
13058 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13059 * tree-ssa-pre.c (compute_avail): Likewise.
13060 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13061 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13062 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13063 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13064 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13065 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13066 (dump_case_nodes, try_switch_expansion): Likewise.
13067 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13068 (vect_do_peeling): Likewise.
13069 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13070 * tree-vect-stmts.c (vectorizable_load): Likewise.
13071 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13072 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13073 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13074 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13075 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13076 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13077 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13078 (evrp_dom_walker::before_dom_children): Likewise.
13079 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13080 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13081 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13082 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13083 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13084 (get_range_pos_neg): Likewise.
13085 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13086 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13087 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13088 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13089 * config/avr/avr.c (avr_fold_builtin): Likewise.
13090 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13091 * config/msp430/msp430.c (msp430_attr): Likewise.
13092 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13093 * config/powerpcspe/powerpcspe-c.c
13094 (altivec_resolve_overloaded_builtin): Likewise.
13095 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13096 (rs6000_expand_ternop_builtin): Likewise.
13097 * config/rs6000/rs6000-c.c
13098 (altivec_resolve_overloaded_builtin): Likewise.
13099 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13100 (rs6000_expand_ternop_builtin): Likewise.
13101 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13103 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13105 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13106 when copying loop nest with only one inner loop.
13108 2017-10-10 Richard Biener <rguenther@suse.de>
13110 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13111 blocks if SCEV is active.
13112 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13114 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13115 (scev_initialize): Assert we are not yet initialized.
13117 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13119 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13120 inner loop's exit stmt by making it always exit the loop, otherwise
13121 we would generate an infinite empty loop.
13123 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13125 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13126 renaming variables in new preheader if it's deleted.
13128 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13130 * tree-loop-distribution.c (struct partition): Remove unused field
13131 loops of the structure.
13132 (partition_alloc, partition_free): Ditto.
13133 (build_rdg_partition_for_vertex): Ditto.
13135 2017-10-09 Jeff Law <law@redhat.com>
13137 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13138 return type to match prototype and documentation.
13140 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13142 * config/rs6000/rs6000.c (processor_costs): Move to ...
13143 * config/rs6000/rs6000.h: ... here.
13144 (rs6000_cost): Declare.
13146 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13148 * except.c (setjmp_fn): New global variable.
13149 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13150 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13151 if DONT_USE_BUILTIN_SETJMP is defined.
13153 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13155 * target.def (insn_cost): New hook.
13156 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13157 * doc/tm.texi: Regenerate.
13158 * rtlanal.c (insn_cost): Use the new hook.
13160 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13162 * combine.c (combine_validate_cost): Compute the new insn_cost,
13163 not just pattern_cost.
13164 (try_combine): Adjust comment.
13166 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13168 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13170 * combine.c (uid_insn_cost): Adjust comment.
13171 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13173 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13174 * dse.c (find_shift_sequence): Ditto.
13175 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13176 (bb_valid_for_noce_process_p): Use pattern_cost.
13177 * rtl.h (insn_rtx_cost): Delete.
13178 (pattern_cost): New prototype.
13179 (insn_cost): New prototype.
13180 * rtlanal.c (insn_rtx_cost): Rename to...
13181 (pattern_cost): ... this.
13184 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13186 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13187 (*jcc<mode>_0_r_i387): Ditto.
13188 (*jccxf_r_i387): Ditto.
13189 (*jcc<mode>_r_i387): Ditto.
13190 (*jccu<mode>_r_i387): Ditto.
13191 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13192 (*jcc): Rename from *jcc_1.
13194 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13196 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13197 deferred rescans after the lvx/stvx recombination pre-pass.
13199 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13201 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13202 memory operation instruction support.
13203 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13204 (rs6000-ibm-aix[789]*): Likewise.
13205 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13206 Document new functions.
13208 2017-10-09 Richard Biener <rguenther@suse.de>
13210 PR tree-optimization/82397
13211 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13212 equality only for semantically equal trees.
13214 2017-10-09 Richard Biener <rguenther@suse.de>
13216 PR tree-optimization/82449
13217 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13218 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13219 allow constant addresses.
13220 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13223 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13225 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13228 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13231 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13234 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13237 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13239 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13242 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13243 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13245 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13247 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13248 (WI_BINARY_PREDICATE_RESULT): Likewise.
13249 (wi::binary_traits::operator_result): New type.
13250 (wi::binary_traits::predicate_result): Likewise.
13251 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13252 (generic_wide_int::operator==, generic_wide_int::operator!=)
13253 (generic_wide_int::operator&, generic_wide_int::and_not)
13254 (generic_wide_int::operator|, generic_wide_int::or_not)
13255 (generic_wide_int::operator^, generic_wide_int::operator+
13256 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13258 (operator~, unary operator-, operator==, operator!=, operator&)
13259 (operator|, operator^, operator+, binary operator-, operator*): New
13261 * expr.c (get_inner_reference): Use wi::bit_and_not.
13262 * fold-const.c (fold_binary_loc): Likewise.
13263 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13264 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13265 (bit_value_binop): Likewise.
13266 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13267 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13268 (extract_range_from_binary_expr_1): Likewise.
13269 (masked_increment): Likewise.
13270 (simplify_bit_ops_using_ranges): Likewise.
13272 2017-10-09 Martin Jambor <mjambor@suse.cz>
13275 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13276 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13277 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13279 (hsa_fixup_mov_insn_type): New function.
13280 (hsa_op_with_type::get_in_type): Use it.
13281 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13282 immediates in an assert.
13283 (hsa_op_with_type::extend_int_to_32bit): New method.
13284 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13285 types. Convert to dest type if necessary.
13286 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13287 (reg_for_gimple_ssa): Pass false as min32int to
13288 hsa_type_for_scalar_tree_type.
13289 (gen_hsa_addr): Fixup type when creating addresable temporary.
13290 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13291 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13292 necessary. Call hsa_fixup_mov_insn_type.
13293 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13294 extend operands and convert to dest type if necessary.
13295 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13296 to dest type if necessary.
13297 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13298 if conversion nt necessary and size matches.
13299 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13300 to dest type if necessary.
13301 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13302 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13304 (gen_hsa_clrsb): Likewise.
13305 (gen_hsa_ffs): Likewise.
13306 (gen_hsa_divmod): Extend operands and convert to dest type if
13308 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13310 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13312 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13313 Remove empty default arguments. Use a brace block as output
13315 (conditional return): Ditto.
13317 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13318 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13320 (group_ending_nop): Ditto.
13321 (doloop_end): Ditto.
13322 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13323 (splitters for those): Ditto.
13325 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13327 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13328 a conditional jump (and the compare for it) so that pc_rtx is the
13330 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13331 for the deleted and renamed ctr<mode>_internal[234] patterns.
13332 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13333 Delete second conditional return pattern.
13334 (ctr<mode>_internal2): Delete this second bdnz pattern.
13335 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13336 (ctr<mode>_internal4): Delete this second bdz pattern.
13338 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13340 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13341 (always_initialized_rtx_for_ssa_name_p): New predicate.
13342 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13343 (finish_out_of_ssa): Free new field of SA.
13344 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13345 * tree-ssa-coalesce.c: Include tree-ssa.h.
13346 (get_parm_default_def_partitions): Remove extern keyword.
13347 (get_undefined_value_partitions): New function.
13348 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13349 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13350 uninitialized bits.
13351 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13353 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13355 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13357 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13359 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13360 for targets that preffer 128bit.
13362 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13364 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13366 2017-10-08 Olivier Hainque <hainque@adacore.com>
13368 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13369 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13371 (thumb_set_return_address): Likewise.
13373 2017-10-08 Olivier Hainque <hainque@adacore.com>
13375 * common/config/arm/arm-common.c (arm_except_unwind_info):
13376 Handle DWARF2_UNWIND_INFO.
13378 2017-10-07 Michael Collison <michael.collison@arm.com>
13380 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13383 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13385 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13386 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13387 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13388 defined, force the creation of a new block for a dispatch label.
13390 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13392 * invoke.texi (Wsuggest-attribute=cold): Document.
13393 * common.opt (Wsuggest-attribute=cold): New
13394 * ipa-pure-const.c (warn_function_cold): New function.
13395 * predict.c (compute_function_frequency): Use it.
13396 * predict.h (warn_function_cold): Declare.
13398 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13400 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13403 2017-10-06 Martin Liska <mliska@suse.cz>
13405 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13406 keyword for member functions.
13407 (struct sanopt_tree_couple): New struct.
13408 (struct sanopt_tree_couple_hash): New function.
13409 (struct sanopt_ctx): Add new hash_map.
13410 (has_dominating_ubsan_ptr_check): New function.
13411 (record_ubsan_ptr_check_stmt): Likewise.
13412 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13413 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13414 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13416 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13419 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13420 aarch64_simd_valid_immediate on CONST_VECTORs.
13421 (aarch64_reg_or_bic_imm): Likewise.
13423 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13425 PR rtl-optimization/82396
13426 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13428 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13430 * graphite-dependences.c (scop_get_reads): Move code to...
13431 (scop_get_must_writes): Move code to...
13432 (scop_get_may_writes): Move code to...
13433 (scop_get_reads_and_writes): ... here.
13434 (scop_get_dependences): Call scop_get_reads_and_writes.
13436 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13438 PR tree-optimization/82434
13439 * fold-const.h (can_native_encode_type_p,
13440 can_native_encode_string_p): Remove.
13441 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13442 don't encode anything, just return what would be otherwise returned.
13443 (native_encode_fixed, native_encode_complex, native_encode_vector):
13445 (native_encode_string): Likewise. Inline by hand
13446 can_native_encode_string_p.
13447 (can_native_encode_type_p): Remove.
13448 (can_native_encode_string_p): Remove.
13449 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13450 STRING_CSTs using can_native_encode_string_p, test all
13451 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13452 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13453 argument from native_encode_expr.
13454 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13455 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13458 2017-10-06 Richard Biener <rguenther@suse.de>
13460 PR tree-optimization/82397
13461 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13462 operand_equal_p but rely on data_ref_compare_tree for detecting
13464 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13465 to match up with dr_group_sort_cmp.
13467 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13470 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13472 * config/s390/s390-builtin-types.def: Regenerate.
13474 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13477 * config/s390/s390-builtin-types.def: Regenerate.
13478 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13479 Change flag from B_VXE to B_VX.
13480 (s390_vec_min_dbl): Remove B_VXE flag.
13482 2017-10-06 Richard Biener <rguenther@suse.de>
13484 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13485 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13486 translate_isl_ast_to_gimple::is_valid_rename,
13487 translate_isl_ast_to_gimple::get_rename,
13488 translate_isl_ast_to_gimple::get_def_bb_for_const,
13489 translate_isl_ast_to_gimple::get_new_name,
13490 translate_isl_ast_to_gimple::collect_all_ssa_names,
13491 translate_isl_ast_to_gimple::copy_loop_phi_args,
13492 translate_isl_ast_to_gimple::collect_all_ssa_names,
13493 translate_isl_ast_to_gimple::copy_loop_phi_args,
13494 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13495 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13496 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13497 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13498 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13499 translate_isl_ast_to_gimple::copy_cond_phi_args,
13500 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13501 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13502 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13503 translate_isl_ast_to_gimple::rename_uses,
13504 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13505 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13506 (set_rename_for_each_def): Likewise.
13507 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13508 here. Handle rewriting SCEV analyzable uses here.
13509 (copy_bb_and_scalar_dependences): Generate code for PHI
13511 (graphite_regenerate_ast_isl): Adjust.
13512 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13513 (add_write, add_read): New functions.
13514 (build_cross_bb_scalars_def): Use it and simplify.
13515 (build_cross_bb_scalars_use): Likewise.
13516 (graphite_find_cross_bb_scalar_vars): Inline into...
13517 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13518 simulating out-of-SSA. Compute liveout and add dependencies.
13519 (build_scops): Force an empty entry block.
13520 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13522 (sese_build_liveouts): Declare.
13523 (sese_trivially_empty_bb_p): Likewise.
13524 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13525 compute liveout and debug_liveout.
13526 (sese_bad_liveouts_use): Remove.
13527 (sese_reset_debug_liveouts_bb): Likewise.
13528 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13529 (sese_build_liveouts): Build liveout and debug_liveout and store
13531 (new_sese_info): Adjust.
13532 (free_sese_info): Likewise.
13533 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13534 do not build liveout here.
13535 (move_sese_in_condition): Adjust region entry.
13536 (scev_analyzable_p): Match up with chrec_apply requirements.
13537 (sese_trivially_empty_bb_p): New.
13538 * tree-into-ssa.c (get_reaching_def): Properly support generating
13539 default-defs for incremental rewrite of anonymous names.
13541 2017-10-06 Richard Biener <rguenther@suse.de>
13543 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13544 precision do not perform modulo reduction.
13546 2017-10-06 Richard Biener <rguenther@suse.de>
13548 PR tree-optimization/82436
13549 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13550 conservatively choose the vectorization factor when checking
13551 whether we can perform the required load permutation.
13552 (vect_transform_slp_perm_load): Assert when we may not fail.
13554 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13556 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13557 message for incompatible -msdata=* and -mcall-* options.
13559 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13561 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13562 rate for post-reload scheduling.
13564 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13566 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13568 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13570 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13571 to improve monte carlo in scimark.
13573 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13575 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13576 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13577 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13578 pentium4_cost, nocona_cost): Set reassociation width to 1.
13579 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13580 width to 2 for fp operations and 1 otherwise.
13581 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13583 (atom_cost): Set reassociation width to 2.
13584 (slm_cost, generic_cost): Set fp reassociation width
13585 to 2 and 1 otherwise.
13586 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13587 (core_cost): Set fp reassociation width to 4 and vector to 2.
13588 (ix86_reassociation_width): Rewrite using cost table; special case
13589 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13590 and TARGET_AVX128_OPTIMAL.
13591 * config/i386/i386.h (processor_costs): Add
13592 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13593 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13594 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13595 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13596 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13597 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13599 2017-10-05 Nathan Sidwell <nathan@acm.org>
13601 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13603 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13605 * config/arm/arm.c (arm_test_fpu_data): New.
13606 (arm_run_selftests): Call arm_test_fpu_data.
13608 2017-10-04 Nathan Sidwell <nathan@acm.org>
13610 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13612 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13614 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13616 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13617 check type for aarch64_simd_valid_immediate.
13618 (aarch64_output_simd_mov_immediate): Update prototype.
13619 (aarch64_simd_valid_immediate): Update prototype.
13620 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13621 support for ORR-immediate.
13622 (and<mode>3): modified pattern to add support for BIC-immediate.
13623 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13624 now checks for valid immediate for BIC and ORR based on new enum
13626 (aarch64_output_simd_mov_immediate): Function now used to output
13627 BIC/ORR imm as well based on new enum argument.
13628 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13630 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13631 (aarch64_reg_or_bic_imm): Likewise.
13633 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13635 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13636 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13638 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13642 PR rtl-optimization/82396
13643 * haifa-sched.c (autopref_multipass_init): Simplify
13645 (autopref_rank_data): Simplify sort order.
13646 * sched-int.h (autopref_multipass_data_): Remove
13647 multi_mem_insn_p, min_offset and max_offset.
13649 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13651 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13653 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13655 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13656 vect_doubleint_cvt.
13658 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13660 * doc/sourcebuild.texi: Document vect_long_mult.
13662 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13664 PR tree-optimization/82413
13665 * fold-const.c (build_range_check): Use widest_int when comparing
13666 the maximum ETYPE value with HIGH.
13668 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13670 PR rtl-optimization/82396
13671 * haifa-sched.c (autopref_multipass_init): Simplify
13673 (autopref_rank_data): Simplify sort order.
13674 * sched-int.h (autopref_multipass_data_): Remove
13675 multi_mem_insn_p, min_offset and max_offset.
13677 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13679 PR tree-optimization/82381
13680 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13681 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13684 PR tree-optimization/82374
13685 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13686 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13687 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13688 current_function_decl to the new decl.
13690 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13692 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13693 helper macro for IEEE float128 hardware built-in functions.
13694 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13696 (TRUNCF128_ODD): Likewise.
13697 (ADDF128_ODD): Likewise.
13698 (SUBF128_ODD): Likewise.
13699 (MULF128_ODD): Likewise.
13700 (DIVF128_ODD): Likewise.
13701 (FMAF128_ODD): Likewise.
13702 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13703 UNSPEC_TRUNC_ROUND_TO_ODD.
13704 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13705 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13706 floating point round to odd instructions.
13707 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13708 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13709 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13710 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13711 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13712 (trunc<mode>sf2_hw): Change the truncate with round to odd
13713 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13714 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13715 to odd hardware instructions.
13716 (sub<mode>3_odd): Likewise.
13717 (mul<mode>3_odd): Likewise.
13718 (div<mode>3_odd): Likewise.
13719 (sqrt<mode>2_odd): Likewise.
13720 (fma<mode>4_odd): Likewise.
13721 (fms<mode>4_odd): Likewise.
13722 (nfma<mode>4_odd): Likewise.
13723 (nfms<mode>4_odd): Likewise.
13724 (trunc<mode>df2_odd): Change the truncate with round to odd
13725 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13727 * doc/extend.texi (PowerPC built-in functions): Update documentation
13728 for existing IEEE float128-bit built-in functions. Add built-in
13729 functions that generate the IEEE 128-bit floating point round to
13732 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13734 PR rtl-optimization/77729
13735 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13736 to (X&(C1&~C2))|C2 transformations.
13738 2017-10-03 Martin Jambor <mjambor@suse.cz>
13740 PR tree-optimization/82363
13741 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13742 mismatch, mark lacc written regardless of racc.
13744 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13746 PR tree-optimization/82381
13747 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13748 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13749 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13750 fallthrough into reassoc_stmt_dominates_stmt_p.
13753 * combine.c (combine_instructions): Don't combine in unreachable
13756 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13759 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13760 function to not use the have_cpu variable. Do not set cpu_index,
13761 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13762 or the default cpu.
13763 (rs6000_valid_attribute_p): Remove duplicate initializations of
13764 old_optimize and func_optimize.
13765 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13766 (rs6000_activate_target_options): Make global.
13767 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13770 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13772 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13773 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13774 Set *poffset to 0 and *psize and *pmax_size to -1 if
13775 *poffset + *psize overflows in HOST_WIDE_INT.
13777 PR tree-optimization/82387
13778 PR tree-optimization/82388
13779 PR tree-optimization/82389
13780 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13781 instead of live_bytes non-NULL.
13783 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13786 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13789 2017-10-02 Richard Biener <rguenther@suse.de>
13791 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13792 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13793 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13795 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13797 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13798 requested precision matches the type's.
13799 * calls.c (alloc_max_size): Calculate the new candidate size as
13800 a widest_int and use wi::to_widest when comparing it with the
13801 current candidate size.
13802 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13803 zero rather than integer_zero_node.
13804 * match.pd: Check for a no-op conversion before using wi::add
13805 rather than after. Use tree_to_uhwi when summing small shift
13806 counts into an unsigned int.
13808 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13809 Alan Hayward <alan.hayward@arm.com>
13810 David Sherwood <david.sherwood@arm.com>
13813 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13814 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13815 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13816 POINTER_AND_FP_REGS.
13818 2017-10-02 Richard Biener <rguenther@suse.de>
13820 PR tree-optimization/82355
13821 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13822 a mapping for the enclosing loop but avoid generating one for
13823 the loop tree root.
13824 (copy_bb_and_scalar_dependences): Remove premature codegen
13825 error on PHIs in blocks duplicated into multiple places.
13826 * graphite-scop-detection.c
13827 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13828 in the region use it as loop and nest to analyze the DR in.
13829 (try_generate_gimple_bb): Likewise.
13830 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13831 (add_loop_constraints): For blocks in a loop not in the region
13832 create a dimension with a single iteration.
13833 * sese.h (gbb_loop_at_index): Remove assert.
13835 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13837 * omp-expand.c (adjust_context_scope): New function.
13838 (expand_parallel_call): Call adjust_context_scope.
13840 2017-10-01 Jeff Law <law@redhat.com>
13842 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13843 dom_opt_dom_walker class with direct access to private members.
13844 Add comments. Call test_for_singularity.
13845 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13846 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13847 m_dummy_cond anymore.
13848 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13850 (pass_dominator:execute): Build the dummy_cond here and pass it
13851 to the dom_opt_dom_walker ctor.
13852 (test_for_singularity): New function.
13854 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13855 Maya Rashish <coypu@sdf.org>
13857 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13858 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13859 (alpha*-*-netbsd*) Use nbsd_tm_file.
13860 (arm*-*-netbsdelf*) Likewise.
13861 (i[34567]86-*-netbsdelf*) Likewise.
13862 (x86_64-*-netbsd*) Likewise.
13863 (mips*-*-netbsd*) Likewise.
13864 (powerpc-*-netbsd*) Likewise.
13865 (sh*-*-netbsd*) Likewise.
13866 (sparc-*-netbsdelf*) Likewise.
13867 (sparc64-*-netbsd*) Likewise.
13868 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13870 (vax-*-netbsdelf*) Likewise.
13871 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13872 (UINT_FAST8_TYPE) Likewise.
13873 (INT_FAST16_TYPE) Check CHAR_FAST16.
13874 (UINT_FAST16_TYPE) Likewise.
13876 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13879 * config/i386/i386.md
13880 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13881 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13882 *divmodsi4_zext_2): New define_insn_and_split.
13883 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13884 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13885 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13886 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13887 New define_insn_and_split.
13888 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13889 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13890 operands[1] having DImode when mode is SImode.
13892 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13893 always SImode for DIV and MOD in REG_EQUAL notes.
13895 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13897 PR middle-end/82319
13898 * match.pd: Fix handling of NaNs in pattern.
13900 2017-09-29 Jeff Law <law@redhat.com>
13902 * sbitmap.c (bitmap_bit_in_range_p): New function.
13903 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13904 * tree-ssa-dse.c (live_bytes_read): New function.
13905 (dse_classify_store): Ignore reads of dead bytes.
13907 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13908 typos and whitespace errors.
13909 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13910 * config/s390/s390.c (s390_emit_prologue): Likewise.
13912 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13915 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13916 with a symbol for LRA.
13918 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13920 PR rtl-optimization/82338
13921 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13923 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13925 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13926 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13927 (qsort_chk): Declare.
13928 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13929 (qsort_chk): New function.
13931 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13933 PR tree-optimization/82337
13934 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13935 phi definition if the PHI result appears in an abnormal PHI.
13936 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13937 basis appears in an abnormal PHI.
13939 2017-09-29 Richard Biener <rguenther@suse.de>
13941 * graphite-isl-ast-to-gimple.c
13942 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13943 (binary_op_to_tree): Use it.
13944 (get_rename_from_scev): Likewise.
13945 (copy_loop_phi_nodes): Likewise.
13946 (copy_bb_and_scalar_dependences): Likewise.
13947 (translate_pending_phi_nodes): Likewise.
13949 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13952 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13953 for movabsq $(i32 << shift), r64.
13955 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13957 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13958 index when encoding %esp as %rsp to avoid 0x67 prefix.
13960 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13962 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13963 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13965 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13967 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13968 Extensions with more than 16 double VFP registers.
13969 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13970 to_clear_mask and all code related to it. Replace the remaining
13971 entry by a sbitmap and adapt code accordingly.
13973 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13975 * brig-builtins.def: Change pure attributes to const.
13977 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13979 * config.gcc (default_gnu_indirect_function): Default to yes for
13980 sparc*-*-linux* with glibc.
13982 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13984 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13985 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13986 when creating .init_array and .fini_array sections with priority
13989 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13992 * config/aarch64/aarch64.c
13993 (aarch64_builtin_support_vector_misalignment): Always return false
13994 when misalignment is unknown.
13996 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13998 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13999 this function to return false if the definition used by the swap
14000 instruction is artificial, or if the memory address from which the
14001 constant value is loaded is not represented by a base address held
14002 in a register or if the base address register is a frame or stack
14003 pointer. Additionally, return false if the base address of the
14004 loaded constant is a SYMBOL_REF but is not considered to be a
14006 (replace_swapped_load_constant): New function.
14007 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
14008 loaded constant vector with a load of a swapped constant vector.
14010 2017-09-27 Carl Love <cel@us.ibm.com>
14012 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14013 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14014 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14017 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
14019 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
14020 first, always call autopref_rank_data otherwise.
14022 2017-09-27 Richard Biener <rguenther@suse.de>
14024 * graphite-scop-detection.c (find_scop_parameters): Move
14025 loop bound handling ...
14026 (gather_bbs::before_dom_children): ... here, avoiding the need
14027 to build scop_info->loop_nest.
14028 (record_loop_in_sese): Remove.
14029 * sese.h (sese_info_t::loop_nest): Remove.
14030 * sese.c (new_sese_info): Do not allocate loop_nest.
14031 (free_sese_info): Do not free loop_nest.
14033 2017-09-27 Jakub Jelinek <jakub@redhat.com>
14036 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
14037 lhs from calls if the lhs has addressable type.
14039 2017-09-27 Richard Biener <rguenther@suse.de>
14041 * graphite.h (scop::max_alias_set): New member.
14042 * graphite-scop-detection.c: Remove references to non-existing
14043 --param in comments.
14044 (build_alias_sets): Record the maximum alias set used for drs.
14045 (build_scops): Support zero as unlimited for
14046 --param graphite-max-arrays-per-scop.
14047 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
14048 and inline into ...
14049 (build_poly_sr_1): ... here. Compute alias set based on the
14050 maximum alias set used for drs rather than
14051 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
14053 2017-09-27 Richard Biener <rguenther@suse.de>
14055 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
14056 --param loop-block-tile-size=0 to disable tiling.
14058 2017-09-27 Richard Biener <rguenther@suse.de>
14060 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14061 (graphite-max-nb-scop-params): Document special value zero.
14062 * domwalk.h (dom_walker::STOP): New symbolical constant.
14063 (dom_walker::dom_walker): Add optional parameter for bb to
14065 (dom_walker::~dom_walker): Declare.
14066 (dom_walker::before_dom_children): Document STOP return value.
14067 (dom_walker::m_user_bb_to_rpo): New member.
14068 (dom_walker::m_bb_to_rpo): Likewise.
14069 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14070 mapping here if not provided by the user.
14071 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14072 provided by the user.
14073 (dom_walker::STOP): Define.
14074 (dom_walker::walk): Do not compute bb to RPO mapping here.
14075 Support STOP return value from before_dom_children to stop
14077 * graphite-optimize-isl.c (optimize_isl): If the schedule
14078 is the same still generate code if -fgraphite-identity
14079 or -floop-parallelize-all are given.
14080 * graphite-scop-detection.c: Include cfganal.h.
14081 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14083 (gather_bbs::before_dom_children): Return STOP for BBs
14085 (build_scops): Compute bb to RPO mapping and pass it to
14086 the domwalk. Treat --param graphite-max-nb-scop-params=0
14087 as not limiting the number of params.
14088 * graphite.c (graphite_initialize): Remove limit on the
14089 number of basic-blocks in a function.
14090 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14091 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14092 default value of 10.
14094 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14096 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14097 Adjust code to eliminate needing to do the shift right 32-bits
14098 operation after XSCVDPSPN.
14100 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14102 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14103 ((X / Y) != 0 -> X >= Y): Likewise.
14105 2017-09-26 Carl Love <cel@us.ibm.com>
14107 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14108 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14109 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14110 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14111 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14112 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14113 definitions and overloading.
14114 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14115 statement for P9V_BUILTIN_XST_LEN_R.
14116 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14117 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14118 define_expand and define_insn for the instructions and builtins.
14119 * doc/extend.texi: Update the built-in documentation file for the new
14120 built-in functions.
14121 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14122 define_insn for the instructions
14124 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14127 * gcc/config/netbsd-protos.h: New file.
14128 * gcc/config/netbsd.c: New file.
14129 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14130 * gcc/config/t-netbsd: New file.
14131 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14132 (tmake_file) Add t-netbsd.
14133 (extra_objs) Add netbsd.o.
14135 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14139 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14141 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14143 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14144 sign extension from a vector register to a GPR by doing a 32-bit
14145 direct move and then an EXTSW.
14146 (extendsi<mode>2 splitter): Likewise.
14147 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14148 right or vector extract after doing XSCVDPSPN. Use
14149 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14151 (movdi_from_sf_zero_ext): Likewise.
14152 (reload_gpr_from_vsxsf): Likewise.
14153 (p8_mfvsrd_4_disf): Delete, no longer used.
14154 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14155 then needing to move the SFmode to a GPR to use the XSCVDPSP
14156 instruction instead of FRSP and XSCVDPSPN.
14157 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14158 it is adjacent to the other XSCVSPDP insns.
14159 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14160 SFmode to be in traditional Altivec registers.
14161 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14162 (vsx_xscvspdpn): Likewise.
14163 (vsx_xscvspdpn_scalar): Likewise.
14165 2017-09-26 Martin Jambor <mjambor@suse.cz>
14167 * tree-sra.c (compare_access_positions): Put integral types first,
14168 stabilize sorting of integral types, remove conditions putting
14169 non-full-precision integers last.
14170 (sort_and_splice_var_accesses): Disable scalarization if a
14171 non-integert would be represented by a non-full-precision integer.
14173 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14175 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14176 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14177 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14178 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14179 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14180 conditionals inside the function instead of around it. Call
14181 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14182 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14184 2017-09-26 Richard Biener <rguenther@suse.de>
14186 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14188 (scop_detection::build_scop_breadth): ... this. Removed.
14189 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14190 (scop_detection::harmful_stmt_in_bb): Likewise.
14191 (scop_detection::graphite_can_represent_stmt): Likewise.
14192 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14193 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14194 (scop_detection::can_represent_loop_1): ... this. Removed.
14195 (scop_detection::harmful_loop_in_region): Simplify after inlining
14196 the above and remove more quadraticness.
14197 (build_scops): Adjust.
14198 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14201 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14204 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14205 REGNO (base) == SP_REG if base is a REG.
14207 PR middle-end/35691
14208 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14209 if it is different SSA_NAME.
14210 (optimize_range_tests_cmp_bitwise): New function.
14211 (optimize_range_tests): Call it.
14213 2017-09-26 Richard Biener <rguenther@suse.de>
14215 PR tree-optimization/82321
14216 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14217 for the def being inside the loop.
14219 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14221 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14223 * config/s390/s390-builtins.def: Fix constraint on op4.
14225 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14227 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14228 independent expanders.
14229 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14230 ("vec_ordered", "vec_unordered"): New expanders.
14232 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14234 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14237 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14239 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14240 vec_unpacks_lo_v16qi.
14241 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14243 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14245 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14246 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14247 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14249 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14251 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14253 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14255 ("*vec_slb<mode>"): New insn pattern.
14256 ("vec_shr_<mode>"): New expander.
14257 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14258 and force the shift count operand to V16QImode.
14259 ("vec_srb<mode>"): Set shift count mode to V16QI.
14261 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14263 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14264 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14265 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14267 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14270 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14272 2017-09-26 Richard Biener <rguenther@suse.de>
14274 PR tree-optimization/82320
14275 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14278 2017-09-25 Jeff Law <law@redhat.com>
14280 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14281 prototype for new argument.
14282 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14283 mostly extracted from rs6000_emit_allocate_stack.
14284 (rs6000_emit_probe_stack_range_stack_clash): New function.
14285 (rs6000_emit_allocate_stack): Call
14286 rs6000_emit_probe_stack_range_stack_clash as needed.
14287 (rs6000_emit_probe_stack_range): Add additional argument
14288 to call to gen_probe_stack_range{si,di}.
14289 (output_probe_stack_range): New.
14290 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14291 (output_probe_stack_range_stack_clash): New.
14292 (rs6000_emit_prologue): Emit notes into dump file as requested.
14293 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14294 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14295 Add additional operand and pass it to output_probe_stack_range.
14297 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14299 PR tree-optimization/82163
14300 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14301 (checking_verify_loop_closed_ssa): New parameter.
14302 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14303 (check_loop_closed_ssa_stmt): Delete.
14304 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14305 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14306 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14309 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14311 * brig-builtins.def: Treat HSAIL barrier builtins as
14312 setjmp/longjump style functions.
14314 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14316 * target.def (constant_alignment): New hook.
14317 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14318 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14319 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14320 * doc/tm.texi: Regenerate.
14321 * targhooks.h (default_constant_alignment): Declare.
14322 (constant_alignment_word_strings): Likewise.
14323 * targhooks.c (default_constant_alignment): New function.
14324 (constant_alignment_word_strings): Likewise.
14325 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14326 instead of CONSTANT_ALIGNMENT.
14327 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14328 (force_const_mem): Likewise.
14329 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14330 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14331 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14332 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14333 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14335 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14336 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14337 constant_alignment_word_strings.
14338 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14339 (CONSTANT_ALIGNMENT): Likewise.
14340 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14341 (arm_constant_alignment): New function.
14342 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14343 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14344 constant_alignment_word_strings.
14345 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14346 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14347 constant_alignment_word_strings.
14348 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14349 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14350 (cris_constant_alignment): New function.
14351 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14352 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14353 (epiphany_constant_alignment): New function.
14354 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14355 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14356 constant_alignment_word_strings.
14357 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14358 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14359 constant_alignment_word_strings.
14360 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14361 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14362 constant_alignment_word_strings.
14363 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14364 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14365 * config/i386/i386.c (ix86_constant_alignment): Make static.
14366 Use the same interface as the target hook.
14367 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14368 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14369 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14370 constant_alignment_word_strings.
14371 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14372 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14373 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14374 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14375 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14376 constant_alignment_word_strings.
14377 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14378 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14379 constant_alignment_word_strings.
14380 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14381 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14382 constant_alignment_word_strings.
14383 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14384 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14386 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14387 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14388 * config/mips/mips.c (mips_constant_alignment): New function.
14389 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14390 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14391 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14392 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14393 (mmix_constant_alignment): Make static. Use the same interface
14394 as the target hook.
14395 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14396 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14397 constant_alignment_word_strings.
14398 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14399 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14400 constant_alignment_word_strings.
14401 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14402 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14403 constant_alignment_word_strings.
14404 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14405 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14406 (rs6000_constant_alignment): New function.
14407 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14408 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14409 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14410 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14411 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14412 (rs6000_constant_alignment): New function.
14413 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14414 * config/s390/s390.c (s390_constant_alignment): New function.
14415 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14416 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14417 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14418 constant_alignment_word_strings.
14419 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14420 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14421 (sparc_constant_alignment): New function.
14422 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14423 * config/spu/spu.c (spu_constant_alignment): New function.
14424 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14425 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14426 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14427 constant_alignment_word_strings.
14428 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14429 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14430 constant_alignment_word_strings.
14431 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14432 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14433 constant_alignment_word_strings.
14434 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14435 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14436 (visium_constant_alignment): New function.
14437 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14438 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14439 (xtensa_constant_alignment): New function.
14440 * system.h (CONSTANT_ALIGNMENT): Poison.
14442 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14444 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14445 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14446 (rs6000_builtin_valid_without_lhs): New helper function.
14447 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14448 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14450 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14452 * target.h (vec_perm_indices): Use unsigned short rather than
14454 (auto_vec_perm_indices): Likewise.
14455 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14456 Use unsigned int rather than unsigned char.
14457 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14459 2017-09-25 Richard Biener <rguenther@suse.de>
14461 * cfgloop.h (sort_sibling_loops): Declare.
14462 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14463 (sort_sibling_loops): New function sorting the sibling loop list
14465 * graphite.c (graphite_transform_loops): Sort sibling loops.
14467 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14469 * target.def (vec_perm_const_ok): Change sel parameter to
14471 * optabs-query.c (can_vec_perm_p): Update accordingly.
14472 * doc/tm.texi: Regenerate.
14473 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14474 auto_vec_perm_indices and remove separate nelt field.
14475 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14476 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14477 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14478 (aarch64_expand_vec_perm_const): Update accordingly.
14479 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14480 to vec_perm_indices.
14481 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14482 auto_vec_perm_indices and remove separate nelt field.
14483 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14484 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14485 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14487 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14488 to vec_perm_indices.
14489 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14490 sel to vec_perm_indices.
14491 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14492 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14493 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14495 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14498 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14501 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14502 on the FUNCTION_DECL function context if it has a DIE that is a
14505 2017-09-25 Richard Biener <rguenther@suse.de>
14507 PR tree-optimization/82285
14508 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14511 2017-09-25 Tom de Vries <tom@codesourcery.com>
14515 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14518 2017-09-25 Richard Biener <rguenther@suse.de>
14520 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14521 ISL errors other than isl_error_quota happen. Dump if the
14522 schedule is the same.
14523 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14524 errors instead of aborting inside ISL.
14526 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14529 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14530 of libgcc_eh for m64.
14531 * config/i386/darwin64.h: Likewise.
14533 2017-09-25 Richard Biener <rguenther@suse.de>
14535 PR middle-end/82144
14536 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14537 attribute for incomplete types nor twice for complete ones.
14539 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14542 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14543 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14546 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14549 * config/i386/i386.opt (mprefer-avx256): Use
14550 ix86_target_flags variable.
14551 * config/i386/i386.c (ix86_target_string): Move
14552 -mprefer-avx256 to flag2_opts.
14554 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14556 PR middle-end/35691
14557 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14558 and x != -1 | y != -1 into (x & y) != -1.
14560 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14562 * config.gcc: Add new case statement to set
14563 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14564 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14565 s390x-*-linux* case statements. Added aarch64 to the list of
14566 supported architectures.
14568 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14570 PR tree-optimization/82289
14571 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14572 STMT_VINFO_RELEVANT_P.
14574 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14575 Alan Hayward <alan.hayward@arm.com>
14576 David Sherwood <david.sherwood@arm.com>
14578 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14579 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14580 (extract_range_from_binary_expr_1): Don't call
14581 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14583 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14584 Alan Hayward <alan.hayward@arm.com>
14585 David Sherwood <david.sherwood@arm.com>
14587 * target.def (preferred_vector_alignment): New hook.
14588 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14590 * doc/tm.texi: Regenerate.
14591 * targhooks.h (default_preferred_vector_alignment): Declare.
14592 * targhooks.c (default_preferred_vector_alignment): New function.
14593 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14595 (DR_TARGET_ALIGNMENT): New macro.
14596 (aligned_access_p): Update commentary.
14597 (vect_known_alignment_in_bytes): New function.
14598 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14600 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14601 Calculate the misalignment based on the target alignment rather than
14603 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14604 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14605 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14606 the target alignment, rather than masking the element misalignment
14607 with the number of elements in a vector. Also use the target
14608 alignment when calculating the maximum number of peels.
14609 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14610 instead of TYPE_ALIGN_UNIT.
14611 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14612 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14613 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14614 (vect_create_data_ref_ptr): Likewise.
14615 (vect_setup_realignment): Realign by ANDing with
14616 -DR_TARGET_MISALIGNMENT.
14617 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14618 the number of peels based on DR_TARGET_ALIGNMENT.
14619 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14620 with the guaranteed alignment boundary when deciding whether
14622 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14623 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14624 (ensure_base_align): Remove stmt_info parameter. Get the
14625 target base alignment from DR_TARGET_ALIGNMENT.
14626 (vectorizable_store): Update call accordingly. Interpret
14627 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14629 (vectorizable_load): Likewise.
14631 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14632 Alan Hayward <alan.hayward@arm.com>
14633 David Sherwood <david.sherwood@arm.com>
14635 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14636 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14637 (vect_enhance_data_refs_alignment): Likewise.
14639 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14641 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14642 error. Only quit immediately if parsing is complete.
14643 (BEGIN): Initialize fatal_err and parse_done.
14644 (begin fpu, end fpu): Check number of arguments.
14645 (begin arch, end arch): Likewise.
14646 (begin cpu, end cpu): Likewise.
14647 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14648 (optalias): Likewise.
14650 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14652 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14653 * config/arm/arm-isa.h: Delete. Move definitions to ...
14654 * arm-cpus.in: ... here. Use new feature and fgroup values.
14655 * config/arm/arm.c (arm_option_override): Use lower case for feature
14657 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14658 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14659 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14661 (print_isa_bits_for): New function.
14662 (gen_isa): New function.
14663 (gen_comm_data): Use print_isa_bits_for.
14664 (define feature): New keyword.
14665 (define fgroup): New keyword.
14666 * config/arm/t-arm (TM_H): Remove.
14667 (GTM_H): Add arm-isa.h.
14668 (arm-isa.h): Add rule to generate file.
14669 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14670 case for feature bit names.
14672 2017-09-22 Richard Biener <rguenther@suse.de>
14674 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14676 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14677 print of no dependency loops ...
14678 * graphite.c (graphite_transform_loops): ... here.
14679 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14681 (same_close_phi_node, remove_duplicate_close_phi,
14682 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14683 (canonicalize_loop_closed_ssa): ... here and simplify.
14684 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14685 (optimize_isl): Use dump_printf_loc to tell when we stopped
14686 optimizing because of an ISL timeout.
14688 2017-09-22 Richard Biener <rguenther@suse.de>
14690 PR tree-optimization/82291
14691 * tree-if-conv.c (predicate_mem_writes): Make sure to
14692 remove writes in blocks predicated with false.
14694 2017-09-22 Richard Biener <rguenther@suse.de>
14696 * sese.c: Include cfganal.h.
14697 (if_region_set_false_region): Remove.
14698 (create_if_region_on_edge): Likewise.
14699 (move_sese_in_condition): Re-implement without destroying
14702 2017-09-22 Richard Biener <rguenther@suse.de>
14704 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14705 Verify both BBs contain loop PHI nodes before dispatching to
14706 copy_loop_phi_args.
14707 (graphite_regenerate_ast_isl): Do not recompute dominators,
14708 do not verify three times. Restructure for clarity.
14709 * graphite-scop-detection.c (same_close_phi_node,
14710 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14711 defined_in_loop_p, canonicalize_loop_closed_ssa,
14712 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14713 checking and SSA rewrite, move to ...
14714 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14715 (graphite_initialize): Do not pass in ctx, do not reset the
14716 SCEV cache, compute only dominators.
14717 (graphite_transform_loops): Allocate ISL ctx after
14718 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14719 Maintain post-dominators only around build_scops.
14720 * sese.c (if_region_set_false_region): Make static. Free
14721 and recompute dominators.
14722 (move_sese_in_condition): Assert we don't get called with
14723 post-dominators computed.
14724 * sese.h (if_region_set_false_region): Remove.
14726 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14728 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14729 mode attribute for TARGET_AVX512VL.
14731 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14733 * config/i386/i386.opt (mprefer-avx256): New option.
14734 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14736 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14737 for TARGET_PREFER_AVX256.
14738 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14740 2017-09-21 Jeff Law <law@redhat.com>
14742 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14743 Fix dump output if the only stack space is for pushed registers.
14745 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14747 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14750 2017-09-21 Martin Sebor <msebor@redhat.com>
14753 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14754 code (in C++) or code that triggers warnings.
14756 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14758 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14760 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14762 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14763 * sched-rgn.c: Ditto.
14764 * sel-sched-ir.c: Ditto.
14766 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14768 * toplev.h (set_random_seed): Adjust return type.
14769 * toplev.c (init_local_tick): Move eager initialization of random_seed
14770 to get_random_seed. Adjust comment.
14771 (init_random_seed): Inline to get_random_seed, delete.
14772 (get_random_seed): Initialize random_seed lazily.
14773 (set_random_seed): Do not return previous value.
14774 (print_switch_value): Do not call get_random_seed.
14776 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14778 * cgraph.c (delete_function_version): New, broken out from...
14779 (cgraph_node::delete_function_version): ...here. Rename to
14780 cgraph_node::delete_function_version_by_decl. Update all uses.
14781 (cgraph_node::remove): Call delete_function_version.
14783 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14786 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14787 VAR_DECLs to which addressable non-volatile parameters are mapped
14788 and for id->retvar after the return value assignment. Clear
14789 id->retval and id->retbnd after inlining.
14791 2017-09-21 Richard Biener <rguenther@suse.de>
14793 PR tree-optimization/82276
14794 PR tree-optimization/82244
14795 * tree-vrp.c (build_assert_expr_for): Set
14796 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14798 (remove_range_assertions): Revert earlier change.
14800 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14803 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14805 2017-09-21 Richard Biener <rguenther@suse.de>
14807 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14808 Restore valid IL after code generation errors.
14809 * graphite.c (graphite_transform_loops): Diagnose code
14810 generation issues as MSG_MISSED_OPTIMIZATION and continue
14811 with processing SCOPs.
14813 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14814 Alan Hayward <alan.hayward@arm.com>
14815 David Sherwood <david.sherwood@arm.com>
14817 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14818 rather than choosing between plus_constant and gen_rtx_<CODE>.
14819 * expr.c (emit_push_insn): Likewise.
14820 (expand_expr_real_2): Likewise.
14822 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14823 Alan Hayward <alan.hayward@arm.com>
14824 David Sherwood <david.sherwood@arm.com>
14826 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14828 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14829 Alan Hayward <alan.hayward@arm.com>
14830 David Sherwood <david.sherwood@arm.com>
14832 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14833 calling tree_to_uhwi.
14835 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14836 Alan Hayward <alan.hayward@arm.com>
14837 David Sherwood <david.sherwood@arm.com>
14839 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14840 INTEGER_CST rather than a negative test for ADDR_EXPR.
14842 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14843 Alan Hayward <alan.hayward@arm.com>
14844 David Sherwood <david.sherwood@arm.com>
14846 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14847 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14849 2017-09-21 Richard Biener <rguenther@suse.de>
14851 PR tree-optimization/71351
14852 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14853 graphite_create_new_loop_guard): Remove, fold remaining parts
14855 (translate_isl_ast_node_for): ... here and simplify.
14857 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14860 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14861 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14862 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14863 alternative always use QI mode, for -Os imov (=R,R) alternative
14864 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14867 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14868 Jeff Law <law@redhat.com>
14870 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14871 (allocate_stack_space): New function, partially extracted from
14872 s390_emit_prologue.
14873 (s390_emit_prologue): Track offset to most recent stack probe.
14874 Code to allocate space moved into allocate_stack_space.
14875 Dump actions when no stack is allocated.
14876 (s390_prologue_plus_offset): New function.
14877 (s390_emit_stack_probe): Likewise.
14879 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14881 * common.opt (Wa, Wl, Wp, g, gz=): Add
14883 (gno-column-info): Remove.
14884 (gcolumn-info): Drop RejectNegative.
14885 (gno-): New prefix.
14886 (gno-record-gcc-switches): Remove.
14887 (grecord-gcc-switches): Drop RejectNegative.
14888 (gno-split-dwarf): Remove.
14889 (gsplit-dwarf): Drop RejectNegative.
14890 (gno-strict-dwarf): Remove.
14891 (gstrict-dwarf): Drop RejectNegative.
14892 * config/darwin.opt (gfull, gused): Add RejectNegative.
14893 * dwarf2out.c (gen_producer_string): Drop
14894 gno-record-gcc-switches handler.
14895 * optc-gen.awk: Add g to prefixes with negative forms.
14896 * opts-common.c (remapping_prefix_p): New.
14897 (find_opt): Check it.
14898 (generate_canonical_option): Test g prefix.
14899 (option_map): Add -gno- mapping.
14900 (add_misspelling_candidates): Check remapping_prefix_p.
14902 2017-09-20 Jeff Law <law@redhat.com>
14904 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14905 thinko in stack clash protection support.
14907 * explow.c (compute_stack_clash_protection_loop_data): Use
14908 CONST_INT_P instead of explicit test. Verify object is a
14909 CONST_INT_P before looking at INTVAL.
14910 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14911 instead of explicit test.
14913 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14916 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14917 address instead of to r1 and r11.
14919 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14921 * config.gcc: Support "knm".
14922 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14923 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14925 * config/i386/i386.c (m_KNM): Define.
14926 (processor_target_table): Add "knm".
14928 (ix86_option_override_internal): Add "knm".
14929 (ix86_issue_rate): Add PROCESSOR_KNM.
14930 (ix86_adjust_cost): Ditto.
14931 (ia32_multipass_dfa_lookahead): Ditto.
14932 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14933 (fold_builtin_cpu): Add M_INTEL_KNM.
14934 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14935 (processor_type): Add PROCESSOR_KNM.
14936 * config/i386/x86-tune.def: Add m_KNM.
14937 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14939 2017-09-20 Richard Biener <rguenther@suse.de>
14941 PR tree-optimization/80213
14942 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14943 are allowed in empty BBs as well.
14944 (canonicalize_loop_closed_ssa): Also look for other complex
14946 (scop_detection::get_sese): Include the loop-closed PHI block
14948 (scop_detection::merge_sese): Remove code adding extra blocks.
14949 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14950 (build_scops): Assert the final returned scop is invalid.
14952 2017-09-20 Richard Biener <rguenther@suse.de>
14954 PR tree-optimization/82264
14955 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14957 (vn_phi_lookup): Likewise.
14958 (vn_phi_insert): Likewise.
14960 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14962 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14963 that fits into uhwi or shwi, add DW_AT_const_value regardless
14964 of early_dwarf without going through RTL, using add_AT_unsigned
14967 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14968 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14969 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14970 (DEBUG_MACRO_SECTION): Likewise.
14971 (DEBUG_LTO_MACRO_SECTION): Likewise.
14972 (DEBUG_STR_DWO_SECTION): Likewise.
14973 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14974 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14975 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14976 (DEBUG_LTO_LINE_STR_SECTION): Define.
14977 (init_sections_and_labels): Initialize debug_line_str_section
14978 variable. Initialize debug_loc_section for -gdwarf-5 to
14979 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14981 2017-09-20 Richard Biener <rguenther@suse.de>
14983 * graphite-sese-to-poly.c (extract_affine): Properly handle
14984 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14986 2017-09-20 Richard Biener <rguenther@suse.de>
14988 PR tree-optimization/81373
14989 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14990 Force SESE live-out defs to be handled even if they are
14993 2017-09-19 Jeff Law <law@redhat.com>
14995 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14996 nothing for stack adjustments with REG_STACK_CHECK.
14997 * sched-deps.c (parse_add_or_inc): Reject insns with
14998 REG_STACK_CHECK from dependency breaking.
14999 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
15000 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
15001 * reg-notes.def (STACK_CHECK): New note.
15003 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
15004 (ix86_expand_prologue): Dump stack clash info as needed.
15005 Call ix86_adjust_stack_and_probe_stack_clash as needed.
15007 * function.c (dump_stack_clash_frame_info): New function.
15008 * function.h (dump_stack_clash_frame_info): Prototype.
15009 (enum stack_clash_probes): New enum.
15011 * config/alpha/alpha.c (alpha_expand_prologue): Also check
15012 flag_stack_clash_protection.
15013 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
15014 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
15015 (arm_frame_pointer_required): Likewise.
15016 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
15017 (ia64_expand_prologue): Likewise.
15018 * config/mips/mips.c (mips_expand_prologue): Likewise.
15019 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
15020 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15021 (sparc_flat_expand_prologue): Likewise.
15022 * config/spu/spu.c (spu_expand_prologue): Likewise.
15024 * explow.c: Include "params.h".
15025 (anti_adjust_stack_and_probe_stack_clash): New function.
15026 (get_stack_check_protect): Likewise.
15027 (compute_stack_clash_protection_loop_data): Likewise.
15028 (emit_stack_clash_protection_loop_start): Likewise.
15029 (emit_stack_clash_protection_loop_end): Likewise.
15030 (allocate_dynamic_stack_space): Use get_stack_check_protect.
15031 Use anti_adjust_stack_and_probe_stack_clash.
15032 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
15033 (emit_stack_clash_protection_loop_start): Likewise.
15034 (emit_stack_clash_protection_loop_end): Likewise.
15035 * rtl.h (get_stack_check_protect): Prototype.
15036 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
15037 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
15038 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
15040 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
15042 * doc/tm.texi: Rebuilt.
15043 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
15044 get_stack_check_protect.
15045 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
15046 * config/arm/arm.c (arm_expand_prologue): Likewise.
15047 (arm_frame_pointer_required): Likewise.
15048 * config/i386/i386.c (ix86_expand_prologue): Likewise.
15049 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
15050 * config/mips/mips.c (mips_expand_prologue): Likewise.
15051 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
15052 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
15053 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15054 (sparc_flat_expand_prologue): Likewise.
15056 * common.opt (-fstack-clash-protection): New option.
15057 * flag-types.h (enum stack_check_type): Note difference between
15058 -fstack-check= and -fstack-clash-protection.
15059 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15060 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15061 * toplev.c (process_options): Issue warnings/errors for cases
15062 not handled with -fstack-clash-protection.
15063 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15064 (-fstack-check): Note additional problem with -fstack-check=generic.
15065 Note that -fstack-check is primarily for Ada and refer users
15066 to -fstack-clash-protection for stack-clash-protection.
15067 Document new params for stack clash protection.
15069 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15071 * config/i386/i386.c (ix86_split_long_move): Do not handle
15072 address used for LEA in a special way.
15074 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15076 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15078 2017-09-19 Martin Sebor <msebor@redhat.com>
15081 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15082 of incompatible types.
15084 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15086 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15087 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15088 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15089 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15091 2017-09-19 Richard Biener <rguenther@suse.de>
15093 PR tree-optimization/82244
15094 * tree-vrp.c (remove_range_assertions): Do not propagate
15095 a constant to abnormals but replace the assert with a copy.
15097 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15099 PR rtl-optimization/57878
15100 PR rtl-optimization/68988
15101 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15102 avoidance test involving non_reload_pseudos. Move frequency test
15103 below the general fragmentation avoidance test.
15105 2017-09-19 Richard Biener <rguenther@suse.de>
15107 PR tree-optimization/69728
15108 * graphite-sese-to-poly.c (schedule_error): New global.
15109 (add_loop_schedule): Handle empty domain by failing the
15111 (build_original_schedule): Handle schedule_error.
15113 2017-09-19 Richard Biener <rguenther@suse.de>
15115 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15116 Do not iterate to sibling loops but only to siblings of inner
15119 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15122 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15123 that the registers are different.
15125 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15127 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15128 to processor_model and "amdfam17h" to arch_names_table.
15129 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15131 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15134 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15136 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15137 Alan Hayward <alan.hayward@arm.com>
15138 David Sherwood <david.sherwood@arm.com>
15140 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15142 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15144 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15145 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15146 vect_schedule_slp_instance.
15147 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15148 Update call to vect_slp_analyze_node_operations. Simplify return
15150 (vect_slp_analyze_bb_1): Update call accordingly.
15151 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15152 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15153 (vect_schedule_slp): Update call accordingly.
15155 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15156 Alan Hayward <alan.hayward@arm.com>
15157 David Sherwood <david.sherwood@arm.com>
15159 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15160 with types that aren't in fact scalar.
15162 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15164 * tree-vect-slp.c (vect_record_max_nunits): New function,
15166 (vect_build_slp_tree_1): ...here.
15167 (vect_build_slp_tree_2): Call it for phis too.
15169 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15171 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15172 to vect_get_vec_def_for_operand when getting the mask operand.
15174 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15175 Alan Hayward <alan.hayward@arm.com>
15176 David Sherwood <david.sherwood@arm.com>
15178 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15181 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15182 Alan Hayward <alan.hayward@arm.com>
15183 David Sherwood <david.sherwood@arm.com>
15185 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15186 calculation for vector booleans.
15188 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15189 Alan Hayward <alan.hayward@arm.com>
15190 David Sherwood <david.sherwood@arm.com>
15192 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15194 (vect_transform_stmt): ...here.
15195 (vect_analyze_stmt): Use it instead of calling
15196 vectorizable_live_operation directly.
15198 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15200 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15201 non-SIMT targets in acc vector loops.
15203 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15205 * configure.ac: Add arc and check if assembler supports gdwarf2.
15206 * configure: Regenerate.
15208 2017-09-18 Richard Biener <rguenther@suse.de>
15210 PR tree-optimization/82220
15211 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15212 epilogue niters from the min_profitable_iters compute.
15214 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15217 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15218 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15219 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15221 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15224 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15225 switching to a new text section.
15227 2017-09-18 Richard Biener <rguenther@suse.de>
15229 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15231 (build_alias_set): Reject aliases with no access function.
15233 2017-09-18 Richard Biener <rguenther@suse.de>
15235 PR tree-optimization/79622
15236 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15238 (build_cross_bb_scalars_use): Likewise.
15240 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15242 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15244 2017-09-18 Alan Modra <amodra@gmail.com>
15247 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15248 stack_pointer_rtx for count 0. Update comments. Break up
15249 large rtl expression.
15251 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15253 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15254 Increase to 20 bytes.
15255 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15256 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15257 or avx version of the stub.
15259 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15262 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15263 compute the minimum stack alignment. Also update preferred stack
15264 boundary for leaf functions.
15266 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15268 PR tree-optimization/82228
15269 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15272 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15274 * common/config/nds32/nds32-common.c
15275 (nds32_option_optimization_table): Refine formatting.
15276 (nds32_option_optimization_table): Use -fsched-pressure and
15277 -fomit-frame-pointer for specific optimization level.
15279 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15281 * config/nds32/nds32.c: Refine formatting and comments.
15282 * config/nds32/nds32.h: Likewise.
15283 * config/nds32/nds32.md: Likewise.
15284 * config/nds32/nds32-cost.c: Likewise.
15285 * config/nds32/nds32-isr.c: Likewise.
15286 * config/nds32/nds32-md-auxiliary.c: Likewise.
15287 * config/nds32/nds32-multiple.md: Likewise.
15288 * config/nds32/nds32-predicates.c: Likewise.
15290 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15291 Jakub Jelinek <jakub@redhat.com>
15293 Add support for -std=c++2a.
15294 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15296 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15298 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15301 * doc/extend.texi (Common Function Attributes): Add
15302 references to ARM, AArch64, and S/390 specific attributes.
15303 (Function Specific Option Pragmas): Add AArch64 and S/390
15304 to list of back ends that support the target pragma.
15306 2017-09-15 Nathan Sidwell <nathan@acm.org>
15308 * doc/standards.texi: Fix C++17 description. Update URLs for
15311 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15313 * common.opt (Wcast-align=strict): New warning option.
15314 * doc/invoke.texi: Document -Wcast-align=strict.
15316 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15318 * cgraph.h (cgraph_thunk_info): Add comments.
15319 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15320 assert for VIRTUAL_* arguments stricter.
15322 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15324 PR tree-optimization/71026
15325 * match.pd: Move RDIV patterns from fold-const.c
15326 * fold-const.c (distribute_real_division): Removed.
15327 (fold_binary_loc): Remove calls to distribute_real_divison.
15329 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15331 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15332 c++1z and gnu++1z as deprecated. Change other references to
15333 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15334 Change -Wc++1z-compat to -Wc++17-compat.
15335 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15336 * dwarf2out.c (highest_c_language): Handle C++17.
15337 (gen_compile_unit_die): Likewise.
15339 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15341 PR rtl-optimization/82192
15342 * combine.c (make_extraction): Don't look through non-paradoxical
15343 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15346 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15347 Alan Hayward <alan.hayward@arm.com>
15348 David Sherwood <david.sherwood@arm.com>
15350 * target.def (function_arg_offset): New hook.
15351 * targhooks.h (default_function_arg_offset): Declare.
15352 * targhooks.c (default_function_arg_offset): New function.
15353 * function.c (locate_and_pad_parm): Use
15354 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15355 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15356 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15357 * doc/tm.texi: Regenerate.
15358 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15359 * config/spu/spu.c (spu_function_arg_offset): New function.
15360 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15361 * system.h (FUNCTION_ARG_OFFSET): Poison.
15363 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15364 Alan Hayard <alan.hayward@arm.com>
15365 David Sherwood <david.sherwood@arm.com>
15367 * target.def (truly_noop_truncation): New hook.
15368 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15369 than TRULY_NOOP_TRUNCATION.
15370 * hooks.h (hook_bool_uint_uint_true): Declare.
15371 * hooks.c (hook_bool_uint_uint_true): New function.
15372 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15373 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15374 * doc/tm.texi: Regenerate.
15375 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15376 rather than TRULY_NOOP_TRUNCATION in comments.
15377 (simplify_comparison): Likewise.
15378 (record_truncated_value): Likewise.
15379 * expmed.c (extract_bit_field_1): Likewise.
15380 (extract_split_bit_field): Likewise.
15381 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15382 instead of TRULY_NOOP_TRUNCATION.
15383 * function.c (assign_parm_setup_block): Likewise.
15384 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15385 * rtlhooks.c: Include target.h.
15386 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15387 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15388 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15389 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15390 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15391 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15392 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15393 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15394 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15395 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15396 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15397 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15398 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15399 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15400 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15401 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15402 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15403 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15404 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15405 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15406 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15407 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15408 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15409 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15410 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15411 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15412 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15413 rather than TRULY_NOOP_TRUNCATION in comments.
15414 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15415 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15416 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15417 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15418 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15419 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15420 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15421 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15422 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15423 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15424 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15425 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15426 rather than TRULY_NOOP_TRUNCATION in comments.
15427 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15428 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15429 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15430 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15431 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15432 TRULY_NOOP_TRUNCATION condition.
15433 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15434 (TRULY_NOOP_TRUNCATION): Delete.
15435 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15436 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15437 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15438 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15439 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15440 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15441 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15442 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15443 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15444 rather than TRULY_NOOP_TRUNCATION in comments.
15445 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15446 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15447 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15448 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15449 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15450 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15452 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15455 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15456 (*cmp_ior): Likewise.
15457 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15458 (*ior_scc_scc_cmp): Likewise.
15459 (*and_scc_scc): Likewise.
15460 (*and_scc_scc_cmp): Likewise.
15462 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15463 Alan Hayard <alan.hayward@arm.com>
15464 David Sherwood <david.sherwood@arm.com>
15466 * target.def (can_change_mode_class): New hook.
15467 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15468 (hard_regno_nregs): Likewise.
15469 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15470 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15471 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15472 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15473 (LOAD_EXTEND_OP): Update accordingly.
15474 * doc/tm.texi: Regenerate.
15475 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15476 CANNOT_CHANGE_MODE_CLASS.
15477 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15478 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15479 * combine.c (simplify_set): Update accordingly.
15480 * emit-rtl.c (validate_subreg): Likewise.
15481 * recog.c (general_operand): Likewise.
15482 * regcprop.c (mode_change_ok): Likewise.
15483 * reload1.c (choose_reload_regs): Likewise.
15484 (inherit_piecemeal_p): Likewise.
15485 * rtlanal.c (simplify_subreg_regno): Likewise.
15486 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15487 instead of CANNOT_CHANGE_MODE_CLASS.
15488 (reload_cse_simplify_operands): Likewise.
15489 * reload.c (push_reload): Use targetm.can_change_mode_class
15490 instead of CANNOT_CHANGE_MODE_CLASS.
15491 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15492 REG_CANNOT_CHANGE_MODE_P.
15493 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15494 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15495 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15496 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15497 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15498 (arm_can_change_mode_class): New function.
15499 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15500 than CANNOT_CHANGE_MODE_CLASS in comments.
15501 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15502 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15503 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15504 (ix86_can_change_mode_class): ...this new function, inverting the
15505 sense of the return value.
15506 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15507 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15508 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15509 (ia64_can_change_mode_class): New function.
15510 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15511 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15512 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15513 (m32c_can_change_mode_class): ...this new function, inverting the
15514 sense of the return value.
15515 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15516 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15517 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15518 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15519 (mips_can_change_mode_class): ...this new function, inverting the
15520 sense of the return value.
15521 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15522 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15523 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15524 (msp430_can_change_mode_class): New function.
15525 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15526 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15527 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15528 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15529 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15530 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15531 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15532 (pa_cannot_change_mode_class): Replace with...
15533 (pa_can_change_mode_class): ...this new function, inverting the
15534 sense of the return value.
15535 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15536 than CANNOT_CHANGE_MODE_CLASS in comments.
15537 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15538 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15539 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15540 (pdp11_cannot_change_mode_class): Replace with...
15541 (pdp11_can_change_mode_class): ...this new function, inverting the
15542 sense of the return value.
15543 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15544 * config/powerpcspe/powerpcspe-protos.h
15545 (rs6000_cannot_change_mode_class_ptr): Delete.
15546 * config/powerpcspe/powerpcspe.c
15547 (rs6000_cannot_change_mode_class_ptr): Delete.
15548 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15549 (rs6000_option_override_internal): Assign to
15550 targetm.can_change_mode_class instead of
15551 rs6000_cannot_change_mode_class_ptr.
15552 (rs6000_cannot_change_mode_class): Replace with...
15553 (rs6000_can_change_mode_class): ...this new function, inverting the
15554 sense of the return value.
15555 (rs6000_debug_cannot_change_mode_class): Replace with...
15556 (rs6000_debug_can_change_mode_class): ...this new function.
15557 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15558 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15559 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15560 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15561 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15563 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15564 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15565 (rs6000_option_override_internal): Assign to
15566 targetm.can_change_mode_class instead of
15567 rs6000_cannot_change_mode_class_ptr.
15568 (rs6000_cannot_change_mode_class): Replace with...
15569 (rs6000_can_change_mode_class): ...this new function, inverting the
15570 sense of the return value.
15571 (rs6000_debug_cannot_change_mode_class): Replace with...
15572 (rs6000_debug_can_change_mode_class): ...this new function.
15573 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15574 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15575 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15576 (s390_can_change_mode_class): ...this new function, inverting the
15577 sense of the return value.
15578 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15579 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15580 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15581 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15582 (sh_cannot_change_mode_class): Replace with...
15583 (sh_can_change_mode_class): ...this new function, inverting the
15584 sense of the return value.
15585 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15586 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15587 (sparc_can_change_mode_class): New function.
15588 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15589 * config/spu/spu.c (spu_can_change_mode_class): New function.
15590 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15591 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15592 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15593 (visium_can_change_mode_class): New function.
15594 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15596 2017-09-15 Richard Biener <rguenther@suse.de>
15598 PR tree-optimization/82217
15599 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15600 but not undefined case.
15602 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15605 * postreload.c (reload_cse_simplify_operands): Skip
15606 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15608 2017-09-15 Richard Biener <rguenther@suse.de>
15610 PR tree-optimization/68823
15611 * graphite-scop-detection.c (build_alias_set): If we have a
15612 possible dependence check whether we can handle them by just
15613 looking at the DRs DR_ACCESS_FNs.
15614 (build_scops): If build_alias_set fails, fail the SCOP.
15616 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15618 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15619 to support float128 built-in functions that require the ISA 3.0
15621 (BU_FLOAT128_3_HW): Likewise.
15622 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15623 built-in functions.
15624 (FMAF128): Likewise.
15626 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15627 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15628 floating point instructions.
15629 (rs6000_invalid_builtin): Likewise.
15630 (rs6000_builtin_mask_names): Likewise.
15631 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15632 (RS6000_BTM_FLOAT128_HW): Likewise.
15633 (RS6000_BTM_COMMON): Likewise.
15634 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15636 * doc/extend.texi (RS/6000 built-in functions): Document the
15637 IEEE 128-bit floating point square root and fused multiply-add
15638 built-in functions.
15640 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15642 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15643 reg (r2) isn't in the set of registers defined in the prologue.
15645 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15646 Alan Hayward <alan.hayward@arm.com>
15647 David Sherwood <david.sherwood@arm.com>
15649 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15650 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15651 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15652 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15653 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15655 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15656 max_vectorization_factor.
15657 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15659 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15660 Alan Hayward <alan.hayward@arm.com>
15661 David Sherwood <david.sherwood@arm.com>
15663 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15664 (vect_worthwhile_without_simd_p): Declare.
15665 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15666 (vectorizable_reduction): Use it.
15667 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15668 (vectorizable_operation): Likewise.
15670 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15671 Alan Hayward <alan.hayward@arm.com>
15672 David Sherwood <david.sherwood@arm.com>
15674 * tree-vectorizer.h (vect_get_num_copies): New function.
15675 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15676 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15677 (vectorizable_induction): Likewise.
15678 (vectorizable_live_operation): Likewise.
15679 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15680 (vectorizable_bswap): Likewise.
15681 (vectorizable_call): Likewise.
15682 (vectorizable_conversion): Likewise.
15683 (vectorizable_assignment): Likewise.
15684 (vectorizable_shift): Likewise.
15685 (vectorizable_operation): Likewise.
15686 (vectorizable_store): Likewise.
15687 (vectorizable_load): Likewise.
15688 (vectorizable_condition): Likewise.
15689 (vectorizable_comparison): Likewise.
15690 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15692 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15693 Alan Hayward <alan.hayward@arm.com>
15694 David Sherwood <david.sherwood@arm.com>
15696 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15697 of vect_init_vector.
15699 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15700 Alan Hayward <alan.hayward@arm.com>
15701 David Sherwood <david.sherwood@arm.com>
15703 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15704 an inline wrapper that provides a location.
15705 (gimple_build_vector): Likewise.
15706 * gimple-fold.c (gimple_build_vector_from_val): New function.
15707 (gimple_build_vector): Likewise.
15708 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15709 functions to build the initial value. Always return a gimple value.
15710 (get_initial_defs_for_reduction): Likewise. Only compute
15712 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15713 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15714 (vectorizable_induction): Use gimple_build_vector rather than
15717 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15718 Alan Hayward <alan.hayward@arm.com>
15719 David Sherwood <david.sherwood@arm.com>
15721 * target.h (vec_perm_indices): New typedef.
15722 (auto_vec_perm_indices): Likewise.
15723 * optabs-query.h: Include target.h
15724 (can_vec_perm_p): Take a vec_perm_indices *.
15725 * optabs-query.c (can_vec_perm_p): Likewise.
15726 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15727 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15728 * tree-vect-generic.c (lower_vec_perm): Likewise.
15729 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15730 (vect_grouped_load_supported): Likewise.
15731 (vect_shift_permute_load_chain): Likewise.
15732 (vect_permute_store_chain): Use auto_vec_perm_indices.
15733 (vect_permute_load_chain): Likewise.
15734 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15735 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15736 Update uses of can_vec_perm_p.
15737 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15738 mode with a number of elements. Take a vec_perm_indices *.
15739 (vect_create_epilog_for_reduction): Update accordingly.
15740 Use auto_vec_perm_indices.
15741 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15742 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15743 (vect_transform_slp_perm_load): Likewise.
15744 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15745 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15746 (vect_gen_perm_mask_checked): Likewise.
15747 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15748 (vect_gen_perm_mask_checked): Likewise.
15749 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15750 (vectorizable_store): Likewise.
15751 (vectorizable_load): Likewise.
15752 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15753 (vectorizable_bswap): Likewise.
15755 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15756 Alan Hayward <alan.hayward@arm.com>
15757 David Sherwood <david.sherwood@arm.com>
15759 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15760 * tree.c (build_vector): Likewise.
15761 (build_vector_from_ctor): Update accordingly.
15762 (build_vector_from_val): Likewise.
15763 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15764 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15765 * tree-vect-generic.c (add_rshift): Likewise.
15766 (expand_vector_divmod): Likewise.
15767 (optimize_vector_constructor): Likewise.
15768 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15769 (vect_transform_slp_perm_load): Likewise.
15770 (vect_schedule_slp_instance): Likewise.
15771 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15772 (vectorizable_call): Likewise.
15773 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15774 * expmed.c (make_tree): Likewise.
15775 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15776 a vector passed to build_vector.
15777 (fold_convert_const): Likewise.
15778 (exact_inverse): Likewise.
15779 (fold_ternary_loc): Likewise.
15780 (fold_relational_const): Likewise.
15781 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15782 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15783 (const_unop): Likewise. Store the reduction accumulator in a
15784 variable rather than an array.
15785 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15786 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15787 the new vector, rather than constructing it after the input arrays.
15788 (native_interpret_vector): Use auto_vec<tree> when building
15789 a vector passed to build_vector. Add elements in order.
15790 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15791 auto_vec<tree> when building a vector passed to build_vector.
15792 (vect_create_epilog_for_reduction): Likewise.
15793 (vectorizable_induction): Likewise.
15794 (get_initial_def_for_reduction): Likewise. Fix indentation of
15796 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15798 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15799 passed to build_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 * tree-core.h (tree_base::u): Add an "nelts" field.
15806 (tree_vector): Use VECTOR_CST_NELTS as the length.
15807 * tree.c (tree_size): Likewise.
15808 (make_vector): Initialize VECTOR_CST_NELTS.
15809 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15810 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15811 TYPE_VECTOR_SUBPARTS.
15812 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15813 as the number of units, setting it from VECTOR_CST_NELTS.
15814 (const_vector_from_tree): Likewise.
15815 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15816 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15817 (fold_negate_expr_1): Likewise.
15818 (fold_convert_const): Likewise.
15819 (const_binop): Likewise. Differentiate the number of output and
15821 (const_unop): Likewise.
15822 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15823 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15824 in cases that did the opposite.
15826 2017-09-14 Richard Biener <rguenther@suse.de>
15828 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15831 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15833 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15835 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15838 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15839 if and where to split a bb, except for splitting before debug insn
15840 sequences followed by non-label real insn. Delete debug insns
15841 in between basic blocks.
15843 * combine.c (make_compound_operation_int): Formatting fixes.
15845 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15846 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15847 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15848 * config/sol2.h (LINK_EH_SPEC): Likewise.
15849 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15850 * config/s390/linux.h (LINK_SPEC): Likewise.
15851 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15852 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15853 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15854 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15855 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15856 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15857 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15858 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15860 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15862 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15864 (ENDFILE_LINUX_SPEC): Likewise.
15865 (LINK_EH_SPEC): Likewise.
15866 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15867 (LINK_OS_LINUX_SPEC32): Likewise.
15868 (LINK_OS_LINUX_SPEC64): Likewise.
15869 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15870 (LINK_OS_LINUX_SPEC): Likewise.
15872 2017-09-13 Martin Liska <mliska@suse.cz>
15874 PR middle-end/82154
15875 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15876 CASE_HIGH is NULL_TREE.
15878 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15879 Alan Hayward <alan.hayward@arm.com>
15880 David Sherwood <david.sherwood@arm.com>
15882 * target.def (secondary_memory_needed): New hook.
15883 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15884 instead of SECONDARY_MEMORY_NEEDED.
15885 (secondary_memory_needed_mode): Likewise.
15886 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15887 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15888 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15889 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15890 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15891 * doc/tm.texi: Regenerate.
15892 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15893 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15894 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15895 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15896 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15897 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15898 mode argument first and change the reg_class arguments to reg_class_t.
15899 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15900 Make static. Update the call to inline_secondary_memory_needed.
15901 (ix86_register_move_cost): Update the call to
15902 inline_secondary_memory_needed.
15903 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15904 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15906 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15907 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15909 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15910 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15911 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15912 and match hook interface. Add comment from mips.h.
15913 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15914 * config/mmix/mmix.md (truncdfsf2): Refer to
15915 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15917 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15918 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15919 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15920 (pa_secondary_memory_needed): New function.
15921 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15922 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15923 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15924 (pdp11_secondary_memory_needed): Make static and match hook interface.
15925 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15926 * config/powerpcspe/powerpcspe-protos.h
15927 (rs6000_secondary_memory_needed_ptr): Delete.
15928 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15930 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15931 (rs6000_option_override_internal): Assign to
15932 targetm.secondary_memory_needed rather than
15933 rs6000_secondary_memory_needed_ptr.
15934 (rs6000_secondary_memory_needed): Match hook interface.
15935 (rs6000_debug_secondary_memory_needed): Likewise.
15936 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15937 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15938 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15939 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15940 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15941 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15943 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15944 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15945 (rs6000_option_override_internal): Assign to
15946 targetm.secondary_memory_needed rather than
15947 rs6000_secondary_memory_needed_ptr.
15948 (rs6000_secondary_memory_needed): Match hook interface.
15949 (rs6000_debug_secondary_memory_needed): Likewise.
15950 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15951 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15952 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15953 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15954 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15955 (sparc_secondary_memory_needed): New function.
15956 * lra-constraints.c (check_and_process_move): Refer to
15957 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15959 (curr_insn_transform): Likewise.
15960 (process_alt_operands): Use targetm.secondary_memory_needed
15961 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15962 (check_secondary_memory_needed_p): Likewise.
15963 (choose_split_class): Likewise.
15964 * reload.c: Unconditionally include code that was previously
15965 conditional on SECONDARY_MEMORY_NEEDED.
15966 (push_secondary_reload): Use targetm.secondary_memory_needed
15967 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15968 (push_reload): Likewise.
15969 * reload1.c: Unconditionally include code that was previously
15970 conditional on SECONDARY_MEMORY_NEEDED.
15971 (choose_reload_regs): Use targetm.secondary_memory_needed
15972 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15973 (gen_reload): Likewise.
15974 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15976 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15977 Alan Hayward <alan.hayward@arm.com>
15978 David Sherwood <david.sherwood@arm.com>
15980 * target.def (secondary_memory_needed_mode): New hook:
15981 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15982 * targhooks.h (default_secondary_memory_needed_mode): New function.
15983 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15984 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15985 * doc/tm.texi: Regenerate.
15986 * lra-constraints.c (check_and_process_move): Use
15987 targetm.secondary_memory_needed_mode instead of
15988 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15989 (curr_insn_transform): Likewise.
15990 * reload.c (get_secondary_mem): Likewise.
15991 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15992 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15994 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15995 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15996 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15997 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15998 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
16000 * config/powerpcspe/powerpcspe-protos.h
16001 (rs6000_secondary_memory_needed_mode): Delete.
16002 * config/powerpcspe/powerpcspe.c
16003 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16004 (rs6000_secondary_memory_needed_mode): Make static.
16005 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16006 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
16008 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16010 (rs6000_secondary_memory_needed_mode): Make static.
16011 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16012 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
16013 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16014 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16015 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16017 (sparc_secondary_memory_needed_mode): New function.
16018 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
16020 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
16022 * config/aarch64/constraints.md (Umq): New constraint.
16023 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
16025 (mov<mode>): Update condition.
16027 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16029 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
16030 when bitposition is the same.
16032 2017-09-13 Richard Biener <rguenther@suse.de>
16034 * dwarf2out.c (output_die_symbol): Remove.
16035 (output_die): Do not output a DIEs symbol.
16037 2017-09-13 Richard Biener <rguenther@suse.de>
16039 PR middle-end/82128
16040 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
16041 default-def to avoid breaking iterator update with the weird
16042 interaction with cgraph_update_edges_for_call_stmt_node.
16044 2017-09-13 Richard Biener <rguenther@suse.de>
16046 * tree-cfg.c (verify_gimple_assign_binary): Add verification
16047 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
16048 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
16049 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
16051 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
16053 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
16054 Disable pc relative literal load irrespective of
16055 TARGET_FIX_ERR_A53_84341 for default.
16057 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16059 * config/sparc/sparc.c (output_return): Output the source location of
16060 the insn in the delay slot, if any.
16061 (output_sibcall): Likewise.
16063 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16066 * common.opt (-static-pie): New alias.
16067 (shared): Negate static-pie.
16068 (-no-pie): Update help text.
16070 (static-pie): New option.
16071 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16072 -static-pie support.
16073 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16074 (LINK_EH_SPEC): Likewise.
16075 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16076 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16077 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16078 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16079 (init_gcc_specs): Likewise.
16080 (init_spec): Likewise.
16081 (display_help): Update help message for -pie.
16082 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16085 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16087 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16088 (movdi_aarch64): Likewise.
16089 (movti_aarch64): Likewise.
16091 2017-09-12 Simon Wright <simon@pushface.org>
16094 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16095 calculation of the minor version, always output as 0.
16097 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16100 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16101 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16102 on it early, rather than manual conversion late. For
16103 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16104 instead of performing manual conversion.
16106 2017-09-12 Carl Love <cel@us.ibm.com>
16108 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16109 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16111 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16112 VMULOSW): Add definitions.
16113 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16114 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16115 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16116 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16117 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16119 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16121 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16123 (movti_aarch64): Likewise.
16124 (movdf_aarch64): Likewise.
16125 (movtf_aarch64): Likewise.
16126 (load_pairdi): Likewise.
16127 (store_pairdi): Likewise.
16128 (load_pairdf): Likewise.
16129 (store_pairdf): Likewise.
16130 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16131 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16132 (ldr_got_small_<mode>): Likewise.
16133 (ldr_got_small_28k_<mode>): Likewise.
16134 (ldr_got_tiny): Likewise.
16135 * config/aarch64/iterators.md (ldst_sz): New.
16136 (ldpstp_sz): Likewise.
16137 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16139 (thunderx_load): Split load_8 to load_16.
16140 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16142 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16143 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16144 (xgene1_store_pair): Split store_8 to store_16.
16145 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16146 (falkor_st_0_st_sd): Split store_8 to store_16.
16148 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16150 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16151 and store1/2/3/4 to store_4/8/12/16.
16152 * config/aarch64/aarch64.md: Update for rename.
16153 * config/arm/arm.md: Likewise.: Likewise.
16154 * config/arm/arm.c: Likewise.
16155 * config/arm/thumb1.md: Likewise.
16156 * config/arm/thumb2.md: Likewise.
16157 * config/arm/vfp.md: Likewise.
16158 * config/arm/arm-generic.md: Likewise.
16159 * config/arm/arm1020e.md: Likewise.
16160 * config/arm/arm1026ejs.md: Likewise.
16161 * config/arm/arm1136jfs.md: Likewise.
16162 * config/arm/arm926ejs.md: Likewise.
16163 * config/arm/cortex-a15.md: Likewise.
16164 * config/arm/cortex-a17.md: Likewise.
16165 * config/arm/cortex-a5.md: Likewise.
16166 * config/arm/cortex-a53.md: Likewise.
16167 * config/arm/cortex-a57.md: Likewise.
16168 * config/arm/cortex-a7.md: Likewise.
16169 * config/arm/cortex-a8.md: Likewise.
16170 * config/arm/cortex-a9.md: Likewise.
16171 * config/arm/cortex-m4.md: Likewise.
16172 * config/arm/cortex-m7.md: Likewise.
16173 * config/arm/cortex-r4.md: Likewise.
16174 * config/arm/exynos-m1.md: Likewise.
16175 * config/arm/fa526.md: Likewise.
16176 * config/arm/fa606te.md: Likewise.
16177 * config/arm/fa626te.md: Likewise.
16178 * config/arm/fa726te.md: Likewise.
16179 * config/arm/fmp626.md: Likewise.
16180 * config/arm/iwmmxt.md: Likewise.
16181 * config/arm/ldmstm.md: Likewise.
16182 * config/arm/marvell-pj4.md: Likewise.
16183 * config/arm/xgene1.md: Likewise.
16184 * config/aarch64/thunderx.md: Likewise.
16185 * config/aarch64/thunderx2t99.md: Likewise.
16186 * config/aarch64/falkor.md: Likewise.
16188 2017-09-12 Martin Liska <mliska@suse.cz>
16190 * attribs.c (private_lookup_attribute): New function.
16191 * attribs.h (private_lookup_attribute): Declared here.
16192 (lookup_attribute): Called from this place.
16194 2017-09-12 Richard Biener <rguenther@suse.de>
16196 PR tree-optimization/82157
16197 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16198 stmts with side-effects.
16200 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16201 Alan Hayward <alan.hayward@arm.com>
16202 David Sherwood <david.sherwood@arm.com>
16204 * target.def (hard_regno_nregs): New hook.
16205 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16206 * targhooks.h (default_hard_regno_nregs): Declare.
16207 * targhooks.c (default_hard_regno_nregs): New function.
16208 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16209 (TARGET_HARD_REGNO_NREGS): ...this hook.
16210 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16211 (CLASS_MAX_NREGS): Likewise.
16212 * doc/tm.texi: Regenerate.
16213 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16214 instead of HARD_REGNO_NREGS.
16215 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16216 HARD_REGNO_NREGS in the comment.
16217 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16218 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16219 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16220 Return an unsigned int.
16221 (TARGET_HARD_REGNO_NREGS): Redefine.
16222 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16223 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16224 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16225 (arc_hard_regno_nregs): New function.
16226 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16227 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16228 (arm_hard_regno_nregs): New function.
16229 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16230 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16231 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16232 (TARGET_HARD_REGNO_NREGS): Redefine.
16233 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16234 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16235 (HARD_REGNO_NREGS): Delete.
16236 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16237 (cr16_hard_regno_nregs): New function.
16238 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16239 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16240 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16241 (cris_hard_regno_nregs): New function.
16242 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16243 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16244 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16245 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16246 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16247 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16248 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16249 (frv_hard_regno_nregs): Make static. Take and return an
16251 (frv_class_max_nregs): Remove outdated copy of documentation.
16252 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16253 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16254 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16255 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16256 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16257 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16258 (TARGET_HARD_REGNO_NREGS): Redefine.
16259 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16260 (CLASS_MAX_NREGS): Update comment.
16261 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16262 (ia64_hard_regno_nregs): New function.
16263 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16264 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16265 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16266 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16267 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16269 (m32c_hard_regno_nregs): Likewise. Make static.
16270 (TARGET_HARD_REGNO_NREGS): Redefine.
16271 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16272 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16273 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16274 (m68k_hard_regno_nregs): New function.
16275 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16276 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16277 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16278 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16279 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16280 Take and return an unsigned int.
16281 (TARGET_HARD_REGNO_NREGS): Redefine.
16282 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16283 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16284 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16285 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16286 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16287 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16288 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16289 (msp430_hard_regno_nregs): Make static. Take and return an
16291 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16292 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16293 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16294 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16295 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16296 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16297 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16298 (TARGET_HARD_REGNO_NREGS): Redefine.
16299 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16300 (PA_HARD_REGNO_NREGS): ...this.
16301 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16302 (PA_HARD_REGNO_NREGS): ...this.
16303 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16304 (pa_hard_regno_nregs): New function.
16305 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16306 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16307 (pdp11_hard_regno_nregs): New function.
16308 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16309 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16310 (rs6000_hard_regno_nregs_hook): New function.
16311 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16312 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16313 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16314 Take and return an unsigned int. Move earlier in file.
16315 (TARGET_HARD_REGNO_NREGS): Redefine.
16316 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16317 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16318 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16319 (rl78_hard_regno_nregs): Make static. Take and return an
16321 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16322 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16323 (rs6000_hard_regno_nregs_hook): New function.
16324 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16325 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16326 (TARGET_HARD_REGNO_NREGS): Redefine.
16327 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16328 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16329 instead of HARD_REGNO_NREGS.
16330 (s390_hard_regno_nregs): New function.
16331 (s390_hard_regno_mode_ok): Add comment from s390.h.
16332 (TARGET_HARD_REGNO_NREGS): Redefine.
16333 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16334 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16335 (sh_hard_regno_nregs): New function.
16336 (sh_pass_in_reg_p): Use it.
16337 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16338 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16339 (sparc_hard_regno_nregs): New function.
16340 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16341 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16342 (spu_function_arg_advance): Use it, supplying a valid register number.
16343 (TARGET_HARD_REGNO_NREGS): Redefine.
16344 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16345 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16346 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16347 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16348 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16349 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16350 (CLASS_MAX_NREGS): Remove copy of old documentation.
16351 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16352 (visium_hard_regno_nregs): New function.
16353 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16354 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16355 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16356 xtensa_hard_regno_nregs): New function.
16357 * system.h (HARD_REGNO_NREGS): Poison.
16359 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16361 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16362 hard_regno_nregs instead of HARD_REGNO_NREGS.
16363 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16364 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16365 (c6x_expand_epilogue): Likewise.
16366 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16367 (frv_read_iacc_argument): Likewise.
16368 * config/sh/sh.c: Include regs.h.
16369 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16370 (regs_used): Likewise.
16371 (output_stack_adjust): Likewise.
16372 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16373 * expmed.c: Include regs.h.
16374 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16375 * ree.c: Include regs.h.
16376 (combine_reaching_defs): Use hard_regno_nregs instead of
16378 (add_removable_extension): Likewise.
16380 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16382 * regs.h (hard_regno_nregs): Turn into a function.
16383 (end_hard_regno): Update accordingly.
16384 * caller-save.c (setup_save_areas): Likewise.
16385 (save_call_clobbered_regs): Likewise.
16386 (replace_reg_with_saved_mem): Likewise.
16387 (insert_restore): Likewise.
16388 (insert_save): Likewise.
16389 * combine.c (can_change_dest_mode): Likewise.
16390 (move_deaths): Likewise.
16391 (distribute_notes): Likewise.
16392 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16393 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16394 (rs6000_split_multireg_move): Likewise.
16395 (rs6000_register_move_cost): Likewise.
16396 (rs6000_memory_move_cost): Likewise.
16397 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16398 (rs6000_split_multireg_move): Likewise.
16399 (rs6000_register_move_cost): Likewise.
16400 (rs6000_memory_move_cost): Likewise.
16401 * cselib.c (cselib_reset_table): Likewise.
16402 (cselib_lookup_1): Likewise.
16403 * emit-rtl.c (set_mode_and_regno): Likewise.
16404 * function.c (aggregate_value_p): Likewise.
16405 * ira-color.c (setup_profitable_hard_regs): Likewise.
16406 (check_hard_reg_p): Likewise.
16407 (calculate_saved_nregs): Likewise.
16408 (assign_hard_reg): Likewise.
16409 (improve_allocation): Likewise.
16410 (calculate_spill_cost): Likewise.
16411 * ira-emit.c (modify_move_list): Likewise.
16412 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16413 (ira_hard_reg_in_set_p): Likewise.
16414 * ira.c (setup_reg_mode_hard_regset): Likewise.
16415 (clarify_prohibited_class_mode_regs): Likewise.
16416 (check_allocation): Likewise.
16417 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16418 (lra_setup_reg_renumber): Likewise.
16419 (setup_try_hard_regno_pseudos): Likewise.
16420 (spill_for): Likewise.
16421 (assign_hard_regno): Likewise.
16422 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16423 * lra-constraints.c (in_class_p): Likewise.
16424 (lra_constraint_offset): Likewise.
16425 (simplify_operand_subreg): Likewise.
16426 (lra_constraints): Likewise.
16427 (split_reg): Likewise.
16428 (split_if_necessary): Likewise.
16429 (invariant_p): Likewise.
16430 (inherit_in_ebb): Likewise.
16431 * lra-lives.c (process_bb_lives): Likewise.
16432 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16433 (get_hard_regs): Likewise.
16434 (do_remat): Likewise.
16435 * lra-spills.c (assign_spill_hard_regs): Likewise.
16436 * mode-switching.c (create_pre_exit): Likewise.
16437 * postreload.c (reload_combine_recognize_pattern): Likewise.
16438 * recog.c (peep2_find_free_register): Likewise.
16439 * regcprop.c (kill_value_regno): Likewise.
16440 (set_value_regno): Likewise.
16441 (copy_value): Likewise.
16442 (maybe_mode_change): Likewise.
16443 (find_oldest_value_reg): Likewise.
16444 (copyprop_hardreg_forward_1): Likewise.
16445 * regrename.c (check_new_reg_p): Likewise.
16446 (regrename_do_replace): Likewise.
16447 * reload.c (push_reload): Likewise.
16448 (combine_reloads): Likewise.
16449 (find_dummy_reload): Likewise.
16450 (operands_match_p): Likewise.
16451 (find_reloads): Likewise.
16452 (find_equiv_reg): Likewise.
16453 (reload_adjust_reg_for_mode): Likewise.
16454 * reload1.c (count_pseudo): Likewise.
16455 (count_spilled_pseudo): Likewise.
16456 (find_reg): Likewise.
16457 (clear_reload_reg_in_use): Likewise.
16458 (free_for_value_p): Likewise.
16459 (allocate_reload_reg): Likewise.
16460 (choose_reload_regs): Likewise.
16461 (reload_adjust_reg_for_temp): Likewise.
16462 (emit_reload_insns): Likewise.
16463 (delete_output_reload): Likewise.
16464 * rtlanal.c (subreg_get_info): Likewise.
16465 * sched-deps.c (sched_analyze_reg): Likewise.
16466 * sel-sched.c (init_regs_for_mode): Likewise.
16467 (mark_unavailable_hard_regs): Likewise.
16468 (choose_best_reg_1): Likewise.
16469 (verify_target_availability): Likewise.
16470 * valtrack.c (dead_debug_insert_temp): Likewise.
16471 * var-tracking.c (track_loc_p): Likewise.
16472 (emit_note_insn_var_location): Likewise.
16473 * varasm.c (make_decl_rtl): Likewise.
16474 * reginfo.c (choose_hard_reg_mode): Likewise.
16475 (init_reg_modes_target): Refer directly to
16476 this_target_regs->x_hard_regno_nregs.
16478 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16480 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16481 instead of hard_regno_nregs.
16483 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16485 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16486 end_hard_regno instead of hard_regno_nregs.
16487 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16488 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16489 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16490 * ira-color.c (improve_allocation): Likewise.
16491 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16492 * lra-lives.c (mark_regno_live): Likewise.
16493 (mark_regno_dead): Likewise.
16494 * lra-remat.c (operand_to_remat): Likewise.
16495 * lra.c (collect_non_operand_hard_regs): Likewise.
16496 * postreload.c (reload_combine_note_store): Likewise.
16497 (move2add_valid_value_p): Likewise.
16498 * reload.c (regno_clobbered_p): Likewise.
16500 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16502 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16504 * config/v850/v850.c (v850_reorg): Likewise.
16505 * reload.c (refers_to_regno_for_reload_p): Likewise.
16506 (find_equiv_reg): Likewise.
16507 * reload1.c (reload_reg_reaches_end_p): Likewise.
16509 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16511 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16513 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16514 * config/arm/arm.c (output_move_neon): Likewise.
16515 (arm_attr_length_move_neon): Likewise.
16516 (neon_split_vcombine): Likewise.
16517 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16518 (c6x_mark_reg_written): Likewise.
16519 (c6x_dwarf_register_span): Likewise.
16520 * config/i386/i386.c (ix86_save_reg): Likewise.
16521 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16522 (rws_access_reg): Likewise.
16523 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16524 * mode-switching.c (create_pre_exit): Likewise.
16525 * ree.c (combine_reaching_defs): Likewise.
16526 (add_removable_extension): Likewise.
16527 * regcprop.c (find_oldest_value_reg): Likewise.
16528 (copyprop_hardreg_forward_1): Likewise.
16529 * reload.c (reload_inner_reg_of_subreg): Likewise.
16530 (push_reload): Likewise.
16531 (combine_reloads): Likewise.
16532 (find_dummy_reload): Likewise.
16533 (reload_adjust_reg_for_mode): Likewise.
16534 * reload1.c (find_reload_regs): Likewise.
16535 (forget_old_reloads_1): Likewise.
16536 (reload_reg_free_for_value_p): Likewise.
16537 (reload_adjust_reg_for_temp): Likewise.
16538 (emit_reload_insns): Likewise.
16539 (delete_output_reload): Likewise.
16540 * sel-sched.c (choose_best_reg_1): Likewise.
16541 (choose_best_pseudo_reg): Likewise.
16543 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16544 Alan Hayward <alan.hayward@arm.com>
16545 David Sherwood <david.sherwood@arm.com>
16547 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16548 * target.def (slow_unaligned_access): New hook.
16549 * targhooks.h (default_slow_unaligned_access): Declare.
16550 * targhooks.c (default_slow_unaligned_access): New function.
16551 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16552 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16553 * doc/tm.texi: Regenerate.
16554 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16555 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16556 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16558 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16559 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16561 (rs6000_slow_unaligned_access): New function.
16562 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16563 (expand_block_compare): Likewise.
16564 (expand_strn_compare): Likewise.
16565 (rs6000_rtx_costs): Likewise.
16566 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16567 (riscv_slow_unaligned_access): Likewise.
16568 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16569 (riscv_slow_unaligned_access_p): ...this and make static.
16570 (riscv_option_override): Update accordingly.
16571 (riscv_slow_unaligned_access): New function.
16572 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16573 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16574 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16575 (rs6000_slow_unaligned_access): New function.
16576 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16577 (rs6000_rtx_costs): Likewise.
16578 * config/rs6000/rs6000-string.c (expand_block_compare)
16579 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16580 of SLOW_UNALIGNED_ACCESS.
16581 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16582 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16583 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16584 of SLOW_UNALIGNED_ACCESS.
16585 * expmed.c (simple_mem_bitfield_p): Likewise.
16586 * expr.c (alignment_for_piecewise_move): Likewise.
16587 (emit_group_load_1): Likewise.
16588 (emit_group_store): Likewise.
16589 (copy_blkmode_from_reg): Likewise.
16590 (emit_push_insn): Likewise.
16591 (expand_assignment): Likewise.
16592 (store_field): Likewise.
16593 (expand_expr_real_1): Likewise.
16594 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16595 * lra-constraints.c (simplify_operand_subreg): Likewise.
16596 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16597 * gimple-ssa-store-merging.c: Likewise in block comment at start
16599 * tree-ssa-strlen.c: Include target.h.
16600 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16601 of SLOW_UNALIGNED_ACCESS.
16602 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16604 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16606 PR rtl-optimization/82185
16607 * expmed.c (emit_store_flag_int): Only test tem if it has been
16610 2017-09-12 Richard Biener <rguenther@suse.de>
16612 PR middle-end/82149
16613 * match.pd ((FTYPE) N CMP CST): Fix typo.
16615 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16617 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16619 (mips_near_type_p): Add 'short_call' attribute as a synonym
16621 * doc/extend.texi (short_call): Document new function attribute.
16623 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16626 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16627 assertion check that in the condition.
16628 (get_atomic_generic_size): Likewise. Before testing if parameter
16629 has pointer type, if it has array type, call for C++
16630 default_conversion to perform array-to-pointer conversion.
16632 2017-09-12 Richard Biener <rguenther@suse.de>
16634 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16635 for operations we cannot scalarize.
16637 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16639 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16640 vectors heap vectors. Clean up comments.
16641 Make visited_bbs a reference.
16642 (profitable_jump_thread_path): Make GC
16643 vectors heap vectors. Clean up comments.
16645 (convert_and_register_jump_thread_path): Make GC vectors heap
16647 (check_subpath_and_update_thread_path): Same. Clean up comments.
16648 Make visited_bbs a reference.
16649 (handle_phi): Abstract common code to to
16650 register_jump_thread_path_if_profitable.
16651 Rename VAR_BB to DEF_BB.
16653 Make GC vectors heap vectors.
16654 Make visited_bbs a reference.
16655 (handle_assignment): Same.
16656 (register_jump_thread_path_if_profitable): New.
16657 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16659 Make GC vectors heap vectors. Clean up comments.
16660 Make visited_bbs a reference.
16661 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16662 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16665 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16668 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16669 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16671 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16673 Revert r251800 and r251799.
16675 2017-09-11 Martin Jambor <mjambor@suse.cz>
16678 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16679 arguments in advance.
16680 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16681 use it to find predecessor edges.
16682 (naive_outof_ssa): Collect vector of predecessors.
16684 2017-09-08 Jason Merrill <jason@redhat.com>
16686 PR c++/70029 - ICE with ref-qualifier and -flto
16687 * langhooks.h (struct lang_hooks_for_types): Add
16688 copy_lang_qualifiers.
16689 * attribs.c (build_type_attribute_qual_variant): Use it.
16690 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16692 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16693 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16695 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16698 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16699 (*mulsi3_sp64): New instruction.
16700 (mulsi3): New expander.
16702 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16704 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16706 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16708 * sancov.c: Include memmodel.h.
16710 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16713 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16716 2017-09-07 Carl Love <cel@us.ibm.com>
16718 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16719 the sldi instruction.
16721 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16723 * sancov.c: Include tm_p.h.
16725 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16728 * output.h (switch_to_other_text_partition): New declaration.
16729 * varasm.c (switch_to_other_text_partition): New function.
16730 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16731 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16732 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16733 to the other text partition before emitting LCL label and switch back
16734 after emitting the word after it.
16736 2017-09-07 Richard Biener <rguenther@suse.de>
16738 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16739 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16740 critical edges here, after loop init.
16741 (pass_data_pre): Remove PROP_no_crit_edges flags.
16742 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16743 for valueization of call args to avoid leaking VN_TOP.
16744 (visit_use): Assert we do not visit default defs.
16745 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16746 Use error_mark_node to more easily detect leaking VN_TOP.
16747 All default-defs are varying, not VN_TOP. Mark them visited.
16748 (run_scc_vn): Make code match comment.
16750 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16752 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16753 OPTION_MASK_FLOAT128_KEYWORD.
16754 (POWERPC_MASKS): Likewise.
16755 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16756 support for the -mfloat128-type option, and make -mfloat128
16757 default on PowerPC Linux systems. Define or undefine
16758 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16759 Define __float128 to be __ieee128 if IEEE 128-bit support is
16760 enabled, or undefine it.
16761 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16762 Delete defining __FLOAT128_TYPE__.
16763 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16764 -mfloat128-type option and make -mfloat128 default on PowerPC
16766 (TARGET_FLOAT128_TYPE): Likewise.
16767 (-mfloat128-type): Likewise.
16768 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16769 Delete the -mfloat128-type option and make -mfloat128 default on
16770 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16771 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16772 128-bit floating point is enabled. Change tests from using
16773 -mfloat128-type to -mfloat128.
16774 (rs6000_mangle_type): Use the correct mangling for the __float128
16775 type even if normal long double is restricted to 64-bits.
16776 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16778 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16779 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16780 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16782 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16783 documentation for -mfloat128.
16785 2017-09-06 Olivier Hainque <hainque@adacore.com>
16787 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16789 2017-09-06 Wish Wu <wishwu007@gmail.com>
16790 Jakub Jelinek <jakub@redhat.com>
16792 * asan.c (initialize_sanitizer_builtins): Add
16793 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16794 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16795 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16796 BT_FN_VOID_UINT64_PTR variables.
16797 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16798 (BT_FN_VOID_UINT16_UINT16): Likewise.
16799 (BT_FN_VOID_UINT32_UINT32): Likewise.
16800 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16801 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16802 (BT_FN_VOID_UINT64_PTR): Likewise.
16803 * common.opt (flag_sanitize_coverage): New variable.
16804 (fsanitize-coverage=trace-pc): Remove.
16805 (fsanitize-coverage=): Add.
16806 * flag-types.h (enum sanitize_coverage_code): New enum.
16807 * fold-const.c (fold_range_test): Disable non-short-circuit
16808 optimization if flag_sanitize_coverage.
16809 (fold_truth_andor): Likewise.
16810 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16811 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16812 (coverage_sanitizer_opts): New array.
16813 (get_closest_sanitizer_option): Add OPTS argument, handle also
16814 OPT_fsanitize_coverage_.
16815 (parse_sanitizer_options): Adjusted to also handle
16816 OPT_fsanitize_coverage_.
16817 (common_handle_option): Add OPT_fsanitize_coverage_.
16818 * sancov.c (instrument_comparison, instrument_switch): New function.
16819 (sancov_pass): Add trace-cmp support.
16820 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16821 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16822 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16823 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16824 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16825 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16826 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16827 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16828 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16829 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16831 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16833 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16834 error. Only quit immediately if parsing is complete.
16835 (BEGIN): Initialize fatal_err and parse_done.
16836 (begin fpu, end fpu): Check number of arguments.
16837 (begin arch, end arch): Likewise.
16838 (begin cpu, end cpu): Likewise.
16839 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16840 (optalias): Likewise.
16842 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16844 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16845 * config/arm/arm-isa.h: Delete. Move definitions to ...
16846 * arm-cpus.in: ... here. Use new feature and fgroup values.
16847 * config/arm/arm.c (arm_option_override): Use lower case for feature
16849 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16850 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16851 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16853 (print_isa_bits_for): New function.
16854 (gen_isa): New function.
16855 (gen_comm_data): Use print_isa_bits_for.
16856 (define feature): New keyword.
16857 (define fgroup): New keyword.
16858 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16859 (arm-isa.h): Add rule to generate file.
16860 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16861 case for feature bit names.
16863 2017-09-06 Richard Biener <rguenther@suse.de>
16865 * tree-ssa-pre.c (NECESSARY): Remove.
16866 (create_expression_by_pieces): Do not touch pass-local flags.
16867 (insert_into_preds_of_block): Likewise.
16868 (do_pre_regular_insertion): Likewise.
16869 (eliminate_insert): Likewise.
16870 (eliminate_dom_walker::before_dom_children): Likewise.
16871 (fini_eliminate): Do not look at inserted_exprs.
16872 (mark_operand_necessary): Remove.
16873 (remove_dead_inserted_code): Replace with simple work-list
16874 algorithm based on inserted_exprs and SSA uses.
16875 (pass_pre::execute): Re-order fini_eliminate and
16876 remove_dead_inserted_code.
16878 2017-09-06 Olivier Hainque <hainque@adacore.com>
16880 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16881 for VxWorks 7. Adjust surrounding comments.
16883 2017-09-06 Richard Biener <rguenther@suse.de>
16885 * gimple-ssa-strength-reduction.c
16886 (find_candidates_dom_walker::before_dom_children): Also allow
16889 2017-09-06 Richard Biener <rguenther@suse.de>
16891 PR tree-optimization/82108
16892 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16893 for gap in the non-permutation SLP case.
16895 2017-09-06 Martin Jambor <mjambor@suse.cz>
16897 PR tree-optimization/82078
16898 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16899 add_access_to_work_queue...
16900 (build_accesses_from_assign): ...here.
16901 (propagate_all_subaccesses): Make sure racc is the group
16902 representative, if there is one.
16904 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16906 PR middle-end/82095
16907 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16910 2017-09-06 Richard Biener <rguenther@suse.de>
16912 * gimple-ssa-strength-reduction.c
16913 (find_candidates_dom_walker::before_doom_children): Use a
16914 type and not a mode check.
16916 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16919 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16921 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16923 2017-09-05 Jeff Law <law@redhat.com>
16925 PR tree-optimization/64910
16926 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16927 cases where we have 3 or more operands.
16929 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16931 PR middle-end/81768
16932 * omp-low.c (lower_omp_for): Recompute tree invariant if
16933 gimple_omp_for_initial/final is ADDR_EXPR.
16935 PR middle-end/81768
16936 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16937 into gimple val before gimplification fo the COND_EXPR.
16939 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16941 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16942 REGION_COPY argument.
16943 (thread_through_all_blocks): Remove unused argument to
16944 duplicate_thread_path.
16946 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16947 Alan Hayward <alan.hayward@arm.com>
16948 David Sherwood <david.sherwood@arm.com>
16950 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16951 Take a scalar_mode rather than a machine_mode.
16952 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16953 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16954 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16955 (aarch64_gen_adjusted_ldpstp): Likewise.
16956 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16958 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16959 Alan Hayward <alan.hayward@arm.com>
16960 David Sherwood <david.sherwood@arm.com>
16962 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16963 Take a scalar_int_mode instead of a machine_mode.
16964 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16965 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16966 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16967 (aarch64_simd_attr_length_rglist): Delete.
16968 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16969 a scalar_int_mode instead of a machine_mode.
16970 (aarch64_add_offset): Likewise.
16971 (aarch64_internal_mov_immediate): Likewise
16972 (aarch64_add_constant_internal): Likewise.
16973 (aarch64_add_constant): Likewise.
16974 (aarch64_movw_imm): Likewise.
16975 (aarch64_rtx_arith_op_extract_p): Likewise.
16976 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16977 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16978 Remove assert that the mode isn't a vector.
16979 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16980 (aarch64_expand_mov_immediate): Update calls after above changes.
16981 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16982 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16983 (aarch64_move_imm): Likewise.
16984 (aarch64_can_const_movi_rtx_p): Likewise.
16985 (aarch64_strip_extend): Likewise.
16986 (aarch64_extr_rtx_p): Likewise.
16987 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16988 a CONST_INT when the mode parameter is VOIDmode.
16989 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16991 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16993 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16994 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16995 (bitwise_type_for_mode): Update accordingly.
16997 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16999 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
17000 * stor-layout.c (mode_for_size_tree): Likewise.
17001 (mode_for_array): Update accordingly.
17002 (layout_decl): Likewise.
17003 (compute_record_mode): Likewise. Only set the mode once.
17005 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17007 * target.def (get_mask_mode): Change return type to opt_mode.
17009 * doc/tm.texi: Regenerate.
17010 * targhooks.h (default_get_mask_mode): Return an opt_mode.
17011 * targhooks.c (default_get_mask_mode): Likewise.
17012 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17013 * optabs-query.c (can_vec_mask_load_store_p): Update use of
17014 targetm.get_mask_mode.
17015 * tree.c (build_truth_vector_type): Likewise.
17017 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17019 * machmode.h (mode_for_vector): Return an opt_mode.
17020 * stor-layout.c (mode_for_vector): Likewise.
17021 (mode_for_int_vector): Update accordingly.
17022 (layout_type): Likewise.
17023 * config/i386/i386.c (emit_memmov): Likewise.
17024 (ix86_expand_set_or_movmem): Likewise.
17025 (ix86_expand_vector_init): Likewise.
17026 (ix86_get_mask_mode): Likewise.
17027 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
17029 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
17030 * expmed.c (extract_bit_field_1): Likewise.
17031 * expr.c (expand_expr_real_2): Likewise.
17032 * optabs-query.c (can_vec_perm_p): Likewise.
17033 (can_vec_mask_load_store_p): Likewise.
17034 * optabs.c (expand_vec_perm): Likewise.
17035 * targhooks.c (default_get_mask_mode): Likewise.
17036 * tree-vect-stmts.c (vectorizable_store): Likewise.
17037 (vectorizable_load): Likewise.
17038 (get_vectype_for_scalar_type_and_size): Likewise.
17040 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17042 * machmode.h (mode_for_int_vector): New function.
17043 * stor-layout.c (mode_for_int_vector): Likewise.
17044 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
17045 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
17046 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17047 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17048 (s390_expand_vcond): Likewise.
17050 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17052 * machmode.h (opt_machine_mode): New type.
17053 (opt_mode<T>): Allow construction from anything that can be
17055 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17056 (mode_for_size): Return an opt_machine_mode.
17057 * stor-layout.c (mode_for_size): Likewise.
17058 (mode_for_size_tree): Update call accordingly.
17059 (bitwise_mode_for_mode): Likewise.
17060 (make_fract_type): Likewise.
17061 (make_accum_type): Likewise.
17062 * caller-save.c (replace_reg_with_saved_mem): Update call
17064 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17065 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17066 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17067 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17068 * expmed.c (extract_bit_field_1): Likewise.
17069 * reload.c (get_secondary_mem): Likewise.
17070 * varasm.c (assemble_integer): Likewise.
17071 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17074 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17076 * machmode.h (decimal_float_mode_for_size): New function.
17077 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17078 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17079 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17080 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17081 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17082 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17084 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17086 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17087 (get_builtin_sync_mode): Likewise.
17088 (expand_ifn_atomic_compare_exchange): Likewise.
17089 (expand_builtin_atomic_clear): Likewise.
17090 (expand_builtin_atomic_test_and_set): Likewise.
17091 (fold_builtin_atomic_always_lock_free): Likewise.
17092 * calls.c (compute_argument_addresses): Likewise.
17093 (emit_library_call_value_1): Likewise.
17094 (store_one_arg): Likewise.
17095 * combine.c (combine_instructions): Likewise.
17096 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17097 * config/arm/arm.c (arm_function_value): Likewise.
17098 (aapcs_allocate_return_reg): Likewise.
17099 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17100 * config/i386/i386.c (construct_container): Likewise.
17101 (ix86_gimplify_va_arg): Likewise.
17102 (ix86_expand_sse_cmp): Likewise.
17103 (emit_memmov): Likewise.
17104 (emit_memset): Likewise.
17105 (expand_small_movmem_or_setmem): Likewise.
17106 (ix86_expand_pextr): Likewise.
17107 (ix86_expand_pinsr): Likewise.
17108 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17109 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17111 * config/mips/mips.c (mips_function_value_1) Likewise.
17112 (mips_block_move_straight): Likewise.
17113 (mips_expand_ins_as_unaligned_store): Likewise.
17114 * config/powerpcspe/powerpcspe.c
17115 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17116 (rs6000_darwin64_record_arg_flush): Likewise.
17117 * config/rs6000/rs6000.c
17118 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17119 (rs6000_darwin64_record_arg_flush): Likewise.
17120 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17121 (sparc_function_value_1): Likewise.
17122 * config/spu/spu.c (adjust_operand): Likewise.
17123 (spu_emit_branch_or_set): Likewise.
17124 (arith_immediate_p): Likewise.
17125 * emit-rtl.c (gen_lowpart_common): Likewise.
17126 * expr.c (expand_expr_real_1): Likewise.
17127 * function.c (assign_parm_setup_block): Likewise.
17128 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17129 * reload1.c (alter_reg): Likewise.
17130 * stor-layout.c (mode_for_vector): Likewise.
17131 (layout_type): Likewise.
17133 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17135 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17136 (spu_convert_move): Likewise.
17137 * lower-subreg.c (resolve_simple_move): Likewise.
17139 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17142 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17143 define_insn to a define_expand.
17144 (altivec_vsum2sws_direct): New define_insn.
17145 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17147 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17149 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17152 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17157 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17158 realign stack if stack alignment needed is less than incoming
17161 2017-09-05 Marek Polacek <polacek@redhat.com>
17164 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17167 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17169 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17171 2017-09-05 Richard Biener <rguenther@suse.de>
17173 PR tree-optimization/82084
17174 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17176 2017-09-05 Richard Biener <rguenther@suse.de>
17178 PR tree-optimization/82102
17179 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17181 2017-09-05 Martin Liska <mliska@suse.cz>
17183 PR tree-optimization/82032
17184 * tree-cfg.c (generate_range_test): New function.
17185 * tree-cfg.h (generate_range_test): Declared here.
17186 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17187 (cleanup_control_expr_graph): Use it.
17188 * tree-switch-conversion.c (try_switch_expansion): Remove
17190 (emit_case_nodes): Use generate_range_test.
17192 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17195 * config/i386/i386.md (*<btsc><mode>_mask): Add
17196 TARGET_USE_BT to insn constraint.
17197 (*btr<mode>_mask): Ditto.
17199 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17201 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17202 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17204 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17207 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17208 TARGET_NEON and TARGET_IWMMXT.
17209 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17210 TARGET_NEON and TARGET_IWMMXT.
17211 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17213 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17215 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17216 (ix86_rewrite_tls_address): Ditto.
17217 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17218 (ix86_rewrite_tls_address_1): Ditto.
17219 (ix86_rewrite_tls_address): Ditto.
17220 * config/i386/predicates.md (tls_address_pattern): New predicate.
17221 * config/i386/i386.md (TLS address splitter): New splitter.
17223 2017-09-04 Richard Biener <rguenther@suse.de>
17225 PR tree-optimization/82084
17226 * fold-const.h (can_native_encode_string_p): Declare.
17227 * fold-const.c (can_native_encode_string_p): Factor out from ...
17228 (native_encode_string): ... here.
17229 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17230 vectorizing stores from constants we later cannot handle.
17232 2017-09-04 Marek Polacek <polacek@redhat.com>
17235 * doc/invoke.texi: Update -Wtautological-compare documentation.
17237 2017-09-04 Jeff Law <law@redhat.com>
17239 PR tree-optimization/64910
17240 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17241 swap the first and last operand if the last is a constant.
17243 2017-09-04 Marek Polacek <polacek@redhat.com>
17246 * convert.c (do_narrow): When sanitizing signed integer overflows,
17247 bail out for signed types.
17248 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17250 2017-09-04 Richard Biener <rguenther@suse.de>
17252 PR tree-optimization/82060
17253 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17254 Move devirtualization after stmt folding and before EH/AB/noreturn
17255 cleanup to get the stmt refs canonicalized. Use a bool instead
17256 of gimple_modified_p since that doesn't work for NOPs. Schedule
17257 NOPs generated by folding for removal.
17259 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17260 Alan Hayward <alan.hayward@arm.com>
17261 David Sherwood <david.sherwood@arm.com>
17263 * coretypes.h (pad_direction): New enum.
17264 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17265 (FUNCTION_ARG_PADDING): Likewise.
17266 * target.def (function_arg_padding): New hook.
17267 * targhooks.h (default_function_arg_padding): Declare.
17268 * targhooks.c (default_function_arg_padding): New function.
17269 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17270 (TARGET_FUNCTION_ARG_PADDING): ...this.
17271 * doc/tm.texi: Regenerate.
17272 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17273 instead of direction.
17274 (compute_argument_addresses): Likewise.
17275 (load_register_parameters): Likewise.
17276 (emit_library_call_value_1): Likewise.
17277 (store_one_arg): Use targetm.calls.function_arg_padding instead
17278 of FUNCTION_ARG_PADDING.
17279 (must_pass_in_stack_var_size_or_pad): Likewise.
17280 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17281 (emit_group_store): Likewise.
17282 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17283 instead of FUNCTION_ARG_PADDING.
17284 (emit_push_insn): Likewise, and propagate enum change throughout
17286 * function.h (direction): Delete.
17287 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17289 * function.c (assign_parm_find_stack_rtl): Likewise.
17290 (assign_parm_setup_block_p): Likewise.
17291 (assign_parm_setup_block): Likewise.
17292 (gimplify_parameters): Likewise.
17293 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17294 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17296 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17297 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17298 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17299 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17300 (aarch64_function_arg_padding): ...this new function.
17301 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17302 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17303 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17304 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17305 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17306 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17307 (arm_pad_arg_upward): Replace with...
17308 (arm_function_arg_padding): ...this new function.
17309 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17311 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17312 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17313 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17314 (ia64_hpux_function_arg_padding): Replace with...
17315 (ia64_function_arg_padding): ...this new function. Use pad_direction
17316 instead of direction. Check for TARGET_HPUX.
17317 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17318 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17319 (iq2000_function_arg_padding): New function.
17320 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17321 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17322 (mips_function_arg_padding): ...this new function.
17323 (mips_pad_reg_upward): Update accordingly.
17324 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17325 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17326 targetm.calls.function_arg_padding.
17327 (FUNCTION_ARG_PADDING): Delete.
17328 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17329 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17330 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17331 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17332 (nios2_block_reg_padding): Return pad_direction instead of direction.
17333 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17334 instead of direction.
17335 (nios2_function_arg_padding): Likewise. Make static.
17336 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17337 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17338 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17339 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17340 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17341 (pa_function_arg_padding): Make static. Return pad_direction instead
17343 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17344 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17345 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17346 instead of direction. Use targetm.calls.function_arg_padding.
17347 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17348 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17349 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17350 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17351 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17353 (function_arg_padding): Rename to...
17354 (rs6000_function_arg_padding): ...this. Make static. Return
17355 pad_direction instead of direction.
17356 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17357 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17358 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17359 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17360 instead of direction. Use targetm.calls.function_arg_padding.
17361 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17362 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17363 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17364 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17365 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17366 (function_arg_padding): Rename to...
17367 (rs6000_function_arg_padding): ...this. Make static. Return
17368 pad_direction instead of direction.
17369 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17370 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17371 * config/s390/s390.c (s390_function_arg_padding): New function.
17372 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17373 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17374 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17375 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17376 (function_arg_padding): Rename to...
17377 (sparc_function_arg_padding): ...this. Make static. Return
17378 pad_direction instead of direction.
17379 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17380 * config/spu/spu.c (spu_function_arg_padding): New function.
17381 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17382 * system.h (FUNCTION_ARG_PADDING): Poison.
17384 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17385 Alan Hayward <alan.hayward@arm.com>
17386 David Sherwood <david.sherwood@arm.com>
17388 * target.def (modes_tieable_p): New hook.
17389 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17390 (TARGET_MODES_TIEABLE_P): ...this.
17391 * doc/tm.texi.in: Regenerate.
17392 * hooks.h (hook_bool_mode_mode_true): Declare.
17393 * hooks.c (hook_bool_mode_mode_true): New function.
17394 * combine.c (subst): Use targetm.modes_tieable_p instead of
17396 * dse.c (find_shift_sequence): Likewise.
17397 * expmed.c (extract_low_bits): Likewise.
17398 * lower-subreg.c: Include target.h.
17399 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17401 * rtlanal.c (rtx_cost): Likewise.
17402 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17403 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17404 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17405 (TARGET_MODES_TIEABLE_P): Redefine.
17406 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17407 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17408 (TARGET_MODES_TIEABLE_P): Redefine.
17409 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17410 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17411 (arc_modes_tieable_p): New function.
17412 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17413 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17414 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17415 (arm_modes_tieable_p): Make static.
17416 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17417 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17418 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17419 (TARGET_MODES_TIEABLE_P): Redefine.
17420 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17421 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17422 (TARGET_MODES_TIEABLE_P): Redefine.
17423 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17424 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17425 (cr16_modes_tieable_p): New function.
17426 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17427 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17428 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17429 (TRULY_NOOP_TRUNCATION): Update comment.
17430 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17431 (TRULY_NOOP_TRUNCATION): Update comment.
17432 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17433 (frv_modes_tieable_p): New function.
17434 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17435 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17436 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17437 (TARGET_MODES_TIEABLE_P): Redefine.
17438 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17439 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17440 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17441 (TARGET_MODES_TIEABLE_P): Redefine.
17442 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17443 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17444 (ia64_modes_tieable_p): New function.
17445 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17446 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17447 (iq2000_modes_tieable_p): New function.
17448 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17449 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17450 (lm32_modes_tieable_p): New function.
17451 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17452 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17453 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17454 (TARGET_MODES_TIEABLE_P): Redefine.
17455 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17456 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17457 (m32r_modes_tieable_p): New function.
17458 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17459 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17460 (m68k_modes_tieable_p): New function.
17461 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17462 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17463 (mcore_modes_tieable_p): New function.
17464 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17465 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17467 (TARGET_MODES_TIEABLE_P): Redefine.
17468 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17469 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17470 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17471 (TARGET_MODES_TIEABLE_P): Redefine.
17472 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17473 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17474 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17475 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17476 (mn10300_modes_tieable_p): ...this and make static.
17477 (TARGET_MODES_TIEABLE_P): Redefine.
17478 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17479 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17480 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17481 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17482 (msp430_modes_tieable_p): Make static.
17483 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17484 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17485 (TARGET_MODES_TIEABLE_P): Redefine.
17486 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17487 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17488 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17489 (TARGET_MODES_TIEABLE_P): Redefine.
17490 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17491 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17492 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17493 (TARGET_MODES_TIEABLE_P): Redefine.
17494 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17495 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17496 (pdp11_modes_tieable_p): New function.
17497 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17498 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17499 (rs6000_modes_tieable_p): New function.
17500 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17501 * config/powerpcspe/powerpcspe.md: Update comment.
17502 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17503 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17504 (TARGET_MODES_TIEABLE_P): Redefine.
17505 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17506 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17507 (rl78_modes_tieable_p): New function.
17508 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17509 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17510 (rs6000_modes_tieable_p): New function.
17511 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17512 * config/rs6000/rs6000.md: Update comment.
17513 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17514 * config/rx/rx.c (rx_modes_tieable_p): New function.
17515 (TARGET_MODES_TIEABLE_P): Redefine.
17516 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17517 * config/s390/s390.c (s390_modes_tieable_p): New function.
17518 (TARGET_MODES_TIEABLE_P): Redefine.
17519 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17520 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17521 (sh_modes_tieable_p): New function.
17522 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17523 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17524 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17525 (sparc_modes_tieable_p): Make static.
17526 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17527 * config/spu/spu.c (spu_modes_tieable_p): New function.
17528 (TARGET_MODES_TIEABLE_P): Redefine.
17529 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17530 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17531 (TARGET_MODES_TIEABLE_P): Redefine.
17532 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17533 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17534 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17535 * config/v850/v850.c (v850_modes_tieable_p): New function.
17536 (TARGET_MODES_TIEABLE_P): Redefine.
17537 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17538 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17539 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17540 (visium_modes_tieable_p): New function.
17541 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17542 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17543 (xtensa_modes_tieable_p): New function.
17544 * system.h (MODES_TIEABLE_P): Poison.
17546 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17547 Alan Hayward <alan.hayward@arm.com>
17548 David Sherwood <david.sherwood@arm.com>
17550 * target.def (hard_regno_mode_ok): New hook.
17551 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17552 (TARGET_HARD_REGNO_MODE_OK): ...this.
17553 * doc/tm.texi.in: Regenerate.
17554 * hooks.h (hook_bool_uint_mode_true): Declare.
17555 * hooks.c (hook_bool_uint_mode_true): New function.
17556 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17557 HARD_REGNO_MODE_OK.
17558 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17559 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17560 instead of HARD_REGNO_MODE_OK.
17561 * caller-save.c: Include target.h.
17562 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17563 HARD_REGNO_MODE_OK.
17564 * combine.c (can_combine_p): Likewise.
17565 (combinable_i3pat): Likewise.
17566 (can_change_dest_mode): Likewise.
17567 * expr.c (init_expr_target): Likewise.
17568 (convert_move): Likewise.
17569 (convert_modes): Likewise.
17570 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17571 (setup_prohibited_mode_move_regs): Likewise.
17572 * ira.h (target_ira): Likewise.
17573 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17574 * lra-constraints.c (process_alt_operands): Likewise.
17575 (split_reg): Likewise.
17576 * recog.c (peep2_find_free_register): Likewise.
17577 * ree.c (combine_reaching_defs): Likewise.
17578 * regcprop.c (maybe_mode_change): Likewise.
17579 * reginfo.c (init_reg_sets_1): Likewise.
17580 (choose_hard_reg_mode): Likewise.
17581 (simplifiable_subregs): Likewise.
17582 * regrename.c (check_new_reg_p): Likewise.
17583 * reload.c (find_valid_class): Likewise.
17584 (find_valid_class_1): Likewise.
17585 (reload_inner_reg_of_subreg): Likewise.
17586 (push_reload): Likewise.
17587 (combine_reloads): Likewise.
17588 (find_dummy_reload): Likewise.
17589 (find_reloads): Likewise.
17590 * reload1.c (find_reg): Likewise.
17591 (set_reload_reg): Likewise.
17592 (allocate_reload_reg): Likewise.
17593 (choose_reload_regs): Likewise.
17594 (reload_adjust_reg_for_temp): Likewise.
17595 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17596 (simplify_subreg_regno): Likewise.
17597 * sel-sched.c (init_regs_for_mode): Likewise.
17598 * varasm.c (make_decl_rtl): Likewise.
17599 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17600 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17601 HARD_REGNO_MODE_OK.
17602 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17603 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17604 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17605 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17606 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17607 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17608 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17609 (arc_mode_class): Delete.
17610 (HARD_REGNO_MODE_OK): Delete.
17611 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17612 (arc_hard_regno_mode_ok): Rename old array to...
17613 (arc_hard_regno_mode_ok_modes): ...this.
17614 (arc_conditional_register_usage): Update accordingly.
17615 (arc_mode_class): Make static.
17616 (arc_hard_regno_mode_ok): New function.
17617 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17618 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17619 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17620 (arm_hard_regno_mode_ok): Make static.
17621 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17622 HARD_REGNO_MODE_OK.
17623 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17624 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17625 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17627 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17628 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17629 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17630 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17631 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17632 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17633 * config/bfin/predicates.md (valid_reg_operand): Use
17634 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17635 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17636 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17637 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17638 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17639 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17640 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17641 (cr16_hard_regno_mode_ok): Make static and return a bool.
17642 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17643 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17644 (cris_hard_regno_mode_ok): New function.
17645 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17646 (epiphany_mode_class): Delete.
17647 (HARD_REGNO_MODE_OK): Delete.
17648 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17649 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17650 (hard_regno_mode_ok): Rename to...
17651 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17652 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17653 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17654 HARD_REGNO_MODE_OK.
17655 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17656 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17657 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17658 (frv_hard_regno_mode_ok): Make static and return a bool.
17659 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17660 HARD_REGNO_MODE_OK.
17661 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17662 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17663 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17664 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17666 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17667 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17668 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17669 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17671 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17672 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17673 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17674 (ia64_hard_regno_mode_ok): New function.
17675 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17676 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17677 (iq2000_hard_regno_mode_ok): New function.
17678 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17679 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17680 (lm32_hard_regno_mode_ok): New function.
17681 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17682 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17683 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17684 instead of HARD_REGNO_MODE_OK.
17685 (m32c_hard_regno_ok): Rename to...
17686 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17687 (m32c_cannot_change_mode_class): Update accordingly.
17688 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17689 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17690 (m32r_mode_class): Delete.
17691 (HARD_REGNO_MODE_OK): Delete.
17692 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17693 (m32r_hard_regno_mode_ok): Rename to...
17694 (m32r_hard_regno_modes): ...this.
17695 (m32r_mode_class): Make static.
17696 (m32r_hard_regno_mode_ok): New function.
17697 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17698 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17699 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17700 (m68k_hard_regno_mode_ok): Make static.
17701 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17702 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17703 (mcore_hard_regno_mode_ok): New function.
17704 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17705 (HARD_REGNO_MODE_OK): Delete.
17706 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17708 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17709 (microblaze_hard_regno_mode_ok): New function.
17710 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17711 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17712 (mips_hard_regno_mode_ok): Delete.
17713 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17714 (mips_hard_regno_mode_ok_p): ...this and make static.
17715 (mips_hard_regno_mode_ok_p): Rename to...
17716 (mips_hard_regno_mode_ok_uncached): ...this.
17717 (mips_hard_regno_mode_ok): New function.
17718 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17719 of HARD_REGNO_MODE_OK.
17720 (mips_option_override): Update after above name changes.
17721 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17722 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17723 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17724 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17725 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17726 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17727 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17728 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17729 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17730 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17731 (msp430_hard_regno_mode_ok): Make static and return a bool.
17732 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17733 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17734 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17736 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17737 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17738 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17739 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17740 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17741 (PA_HARD_REGNO_MODE_OK): ...this
17742 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17743 (PA_HARD_REGNO_MODE_OK): ...this.
17744 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17745 (pa_hard_regno_mode_ok): New function.
17746 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17747 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17748 (pdp11_hard_regno_mode_ok): New function.
17749 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17750 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17752 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17754 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17755 (rs6000_hard_regno_mode_ok): Rename to...
17756 (rs6000_hard_regno_mode_ok_uncached): ...this.
17757 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17758 (rs6000_hard_regno_mode_ok): New function.
17759 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17760 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17761 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17762 (riscv_hard_regno_mode_ok): ...this and make static.
17763 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17764 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17765 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17766 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17767 (rl78_hard_regno_mode_ok): Make static and return bool.
17768 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17769 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17771 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17772 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17773 (rs6000_hard_regno_mode_ok): Rename to...
17774 (rs6000_hard_regno_mode_ok_uncached): ...this.
17775 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17776 (rs6000_hard_regno_mode_ok): New function.
17777 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17778 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17779 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17780 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17781 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17782 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17783 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17784 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17785 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17786 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17787 (sh_hard_regno_mode_ok): Make static.
17788 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17789 instead of HARD_REGNO_MODE_OK.
17790 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17791 (sparc_mode_class): Delete.
17792 (HARD_REGNO_MODE_OK): Delete.
17793 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17794 (hard_regno_mode_classes): Make static.
17795 (sparc_mode_class): Likewise.
17796 (sparc_hard_regno_mode_ok): New function.
17797 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17798 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17799 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17801 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17802 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17803 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17804 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17805 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17806 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17807 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17808 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17809 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17810 (visium_hard_regno_mode_ok): New function.
17811 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17812 instead of HARD_REGNO_MODE_OK.
17813 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17814 (HARD_REGNO_MODE_OK): Delete.
17815 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17816 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17817 (xtensa_option_override): Update accordingly.
17818 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17819 (xtensa_hard_regno_mode_ok): New function.
17820 * system.h (HARD_REGNO_MODE_OK): Poison.
17822 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17823 Alan Hayward <alan.hayward@arm.com>
17824 David Sherwood <david.sherwood@arm.com>
17826 * target.def (hard_regno_call_part_clobbered): New hook.
17827 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17828 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17829 * doc/tm.texi: Regenerate.
17830 * hooks.h (hook_bool_uint_mode_false): Declare.
17831 * hooks.c (hook_bool_uint_mode_false): New function.
17832 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17833 * cselib.c (cselib_process_insn): Use
17834 targetm.hard_regno_call_part_clobbered instead of
17835 HARD_REGNO_CALL_PART_CLOBBERED.
17836 * ira-conflicts.c (ira_build_conflicts): Likewise.
17837 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17838 * lra-constraints.c (need_for_call_save_p): Likewise.
17839 * lra-lives.c: Include target.h.
17840 (check_pseudos_live_through_calls): Use
17841 targetm.hard_regno_call_part_clobbered instead of
17842 HARD_REGNO_CALL_PART_CLOBBERED.
17843 * regcprop.c: Include target.h.
17844 (copyprop_hardreg_forward_1): Use
17845 targetm.hard_regno_call_part_clobbered instead of
17846 HARD_REGNO_CALL_PART_CLOBBERED.
17847 * reginfo.c (choose_hard_reg_mode): Likewise.
17848 * regrename.c (check_new_reg_p): Likewise.
17849 * reload.c (find_equiv_reg): Likewise.
17850 * reload1.c (emit_reload_insns): Likewise.
17851 * sched-deps.c (deps_analyze_insn): Likewise.
17852 * sel-sched.c (init_regs_for_mode): Likewise.
17853 (mark_unavailable_hard_regs): Likewise.
17854 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17855 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17856 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17858 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17859 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17860 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17862 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17864 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17865 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17866 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17868 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17869 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17870 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17872 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17873 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17875 * config/powerpcspe/powerpcspe.c
17876 (rs6000_hard_regno_call_part_clobbered): New function.
17877 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17878 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17879 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17881 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17882 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17883 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17885 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17886 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17887 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17889 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17890 Alan Hayward <alan.hayward@arm.com>
17891 David Sherwood <david.sherwood@arm.com>
17893 * rtl.h (subreg_memory_offset): Declare.
17894 * emit-rtl.c (subreg_memory_offset): New function.
17895 * expmed.c (store_bit_field_1): Use it.
17896 * expr.c (undefined_operand_subword_p): Likewise.
17897 * simplify-rtx.c (simplify_subreg): Likewise.
17899 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17901 PR rtl-optimization/57448
17904 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17905 using atomic_load pattern.
17906 (expand_atomic_store): Likewise.
17908 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17911 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17912 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17913 handling. Use replace_call_with_value with NULL instead of
17914 gsi_replace, unlink_stmt_vdef and release_defs.
17916 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17919 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17921 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17924 * rtl.h (emit_library_call_value_1): Declare.
17925 (emit_library_call): Replace declaration with a series of overloads.
17926 Remove the parameter count argument.
17927 (emit_library_call_value): Likewise.
17928 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17929 with an "rtx_mode_t *".
17930 (emit_library_call_value): Delete.
17931 (emit_library_call): Likewise.
17932 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17933 (asan_emit_allocas_unpoison): Likewise.
17934 * builtins.c (expand_builtin_powi): Likewise.
17935 (expand_asan_emit_allocas_unpoison): Likewise.
17936 * cfgexpand.c (expand_main_function): Likewise.
17937 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17938 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17939 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17940 * config/arm/arm.c (arm_trampoline_init): Likewise.
17941 (arm_call_tls_get_addr): Likewise.
17942 (arm_expand_divmod_libfunc): Likewise.
17943 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17944 (smulsi3_highpart): Likewise.
17945 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17946 (c6x_expand_compare): Likewise.
17947 (c6x_expand_movmem): Likewise.
17948 * config/frv/frv.c (frv_trampoline_init): Likewise.
17949 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17950 (ix86_expand_divmod_libfunc): Likewise.
17951 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17952 (ia64_expand_compare): Likewise.
17953 (ia64_profile_hook): Likewise.
17954 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17955 (nonlocal_goto): Likewise.
17956 (restore_stack_nonlocal): Likewise.
17957 * config/m32r/m32r.c (block_move_call): Likewise.
17958 (m32r_trampoline_init): Likewise.
17959 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17960 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17961 (m68k_call_m68k_read_tp): Likewise.
17962 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17963 (microblaze_expand_divide): Likewise.
17964 * config/mips/mips.h (mips_args): Likewise.
17965 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17966 (MIPS_ICACHE_SYNC): Likewise.
17967 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17968 (nios2_trampoline_init): Likewise.
17969 * config/pa/pa.c (hppa_tls_call): Likewise.
17970 (pa_trampoline_init): Likewise.
17971 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17972 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17973 (expand_strn_compare): Likewise.
17974 (rs6000_generate_compare): Likewise.
17975 (rs6000_expand_float128_convert): Likewise.
17976 (output_profile_hook): Likewise.
17977 (rs6000_trampoline_init): Likewise.
17978 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17979 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17980 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17981 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17982 (rs6000_generate_compare): Likewise.
17983 (rs6000_expand_float128_convert): Likewise.
17984 (output_profile_hook): Likewise.
17985 (rs6000_trampoline_init): Likewise.
17986 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17987 * config/sh/sh.c (sh_trampoline_init): Likewise.
17988 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17989 (sparc_emit_float_lib_cmp): Likewise.
17990 (sparc32_initialize_trampoline): Likewise.
17991 (sparc64_initialize_trampoline): Likewise.
17992 (sparc_profile_hook): Likewise.
17993 * config/spu/spu.c (ea_load_store): Likewise.
17994 * config/spu/spu.md (floatunssidf2): Likewise.
17995 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17996 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17997 * config/visium/visium.c (expand_block_move_4): Likewise.
17998 (expand_block_move_2): Likewise.
17999 (expand_block_move_1): Likewise.
18000 (expand_block_set_4): Likewise.
18001 (expand_block_set_2): Likewise.
18002 (expand_block_set_1): Likewise.
18003 (visium_trampoline_init): Likewise.
18004 (visium_profile_hook): Likewise.
18005 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
18006 (xtensa_setup_frame_addresses): Likewise.
18007 (xtensa_trampoline_init): Likewise.
18008 * except.c (sjlj_emit_function_enter): Likewise.
18009 (sjlj_emit_function_exit): Likewise.
18010 * explow.c (allocate_dynamic_stack_space): Likewise.
18011 (probe_stack_range): Likewise.
18012 * expr.c (convert_mode_scalar): Likewise.
18013 * optabs.c (expand_binop): Likewise.
18014 (expand_twoval_binop_libfunc): Likewise.
18015 (expand_unop): Likewise.
18016 (prepare_cmp_insn): Likewise.
18017 (prepare_float_lib_cmp): Likewise.
18018 (expand_float): Likewise.
18019 (expand_fix): Likewise.
18020 (expand_fixed_convert): Likewise.
18021 (maybe_emit_sync_lock_test_and_set): Likewise.
18022 (expand_atomic_compare_and_swap): Likewise.
18023 (expand_mem_thread_fence): Likewise.
18024 (expand_atomic_fetch_op): Likewise.
18026 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
18028 * doc/generic.texi (OpenACC): Adjust URL.
18029 * doc/invoke.texi (C Dialect Options): Ditto.
18031 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
18033 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
18034 predicate for operand 1. Add (m,<S>) constraint.
18035 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
18036 Prevent memory operand 1 with register operand 2.
18038 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
18040 PR rtl-optimization/82024
18041 * combine.c (try_combine): If the combination result is a PARALLEL,
18042 and we only need to retain the SET in there that would be placed
18043 at I2, check that we can place that at I3 instead, before doing so.
18045 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18048 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
18050 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
18051 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
18054 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
18055 Jeff Law <law@redhat.com>
18057 * varasm.c (bss_initializer_p): Do not put constants into .bss
18058 (categorize_decl_for_section): Handle bss_initializer_p returning
18059 false when DECL_INITIAL is NULL.
18061 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18064 * config/s390/s390.c (s390_can_inline_p): New function.
18066 2017-09-01 Jeff Law <law@redhat.com>
18068 PR tree-optimization/82052
18069 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18070 Always initialize the returned slot after a hash table miss
18071 when INSERT is true.
18073 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18075 * config/s390/s390.md (mem_signal_fence): Remove.
18076 * doc/md.texi (mem_signal_fence): Remove.
18077 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18079 * target-insns.def (mem_signal_fence): Remove.
18081 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18084 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18087 * asan.c (create_odr_indicator): Strip name encoding from assembler
18088 name before appending it after __odr_asan_.
18090 2017-09-01 Martin Liska <mliska@suse.cz>
18092 PR tree-optimization/82059
18093 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18094 frequency only when an edge is redirected.
18096 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18098 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18099 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18100 (arc_conditional_register_usage): Remove ARC600 lp_count
18102 (arc_file_start): Emit Tag_ARC_CPU_variation.
18103 (arc_can_use_doloop_p): New conditions to use ZOLs.
18104 (hwloop_fail): New function.
18105 (hwloop_optimize): Likewise.
18106 (hwloop_pattern_reg): Likewise.
18107 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18108 (arc_reorg_loops): New function, calls reorg_loops.
18109 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18110 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18112 (arc_loop_hazard): Remove function, functionality moved into
18114 (arc_hazard): Remove arc_loop_hazard call.
18115 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18116 into hwloop_optimize.
18117 (arc_label_align): Remove ZOL handling.
18118 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18119 * config/arc/arc.md (doloop_begin): Remove pattern.
18120 (doloop_begin_i): Likewise.
18121 (doloop_end_i): Likewise.
18122 (doloop_fallback): Likewise.
18123 (doloop_fallback_m): Likewise.
18124 (doloop_end): Reimplement expand.
18125 (arc_lp): New pattern for LP instruction.
18126 (loop_end): New pattern.
18127 (loop_fail): Likewise.
18128 (decrement_and_branch_until_zero): Likewise.
18129 * config/arc/arc.opt (mlpc-width): New option.
18130 * doc/invoke.texi (mlpc-width): Document option.
18132 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18134 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18135 (arc_ccfsm_advance): Fix checking for delay slots.
18136 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18138 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18140 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18142 (movhi_insn): Update store instruction constraint which are saving
18143 6-bit short immediates.
18144 (movsi_insn): Consider also short scaled load operations.
18145 (zero_extendhisi2_i): Use Usd constraint instead of T.
18146 (extendhisi2_i): Add q constraint.
18147 (arc_clzsi2): Add type and length attributes.
18148 (arc_ctzsi2): Likewise.
18149 * config/arc/constraints.md (Usc): Update constraint, the
18150 assembler can parse two relocations for a single instruction.
18152 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18154 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18155 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18157 2017-08-31 Olivier Hainque <hainque@adacore.com>
18159 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18160 match as powerpc-wrs-vxworks*.
18162 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18164 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18165 register constraint for by-element operand.
18166 (aarch64_mls_elt_merge<mode>): Likewise.
18168 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18170 * config/arc/arc.c (arc_can_follow_jump): Check for short
18173 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18175 * config.gcc: Use g.opt for arc.
18176 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18177 functionality moved to ...
18178 (legitimate_scaled_address_p): New function, ...here.
18179 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18180 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18181 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18183 (arc_override_options): Handle G option.
18184 (arc_output_pic_addr_const): Correct function definition.
18185 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18186 (arc_decl_anon_ns_mem_p): Delete.
18187 (arc_in_small_data_p): Overhaul this function to take into
18188 consideration the value given via G option.
18189 (arc_rewrite_small_data_1): Renamed and corrected old
18190 arc_rewrite_small_data function.
18191 (arc_rewrite_small_data): New function.
18192 (small_data_pattern): Don't use pic_offset_table_rtx.
18193 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18194 * config/arc/simdext.md (movmisalignv2hi): Use
18195 prepare_move_operands function.
18197 (movmisalign*): Likewise.
18198 * doc/invoke.texi (ARC options): Document -G option.
18200 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18202 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18204 * config/arc/arc.c (arc_print_operand): Output scalled address for
18205 sdata whenever is possible.
18206 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18207 load/stores are available.
18208 (compact_sda_memory_operand): Check for the alignment required by
18209 code density instructions.
18210 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18212 * config/arc/constraints.md (Usd): Update constraint.
18213 (Us0): New constraint.
18214 (Usc): Update constraint.
18216 2017-08-31 Richard Biener <rguenther@suse.de>
18218 PR middle-end/82054
18219 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18220 function only once.
18222 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18224 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18225 Resize type_signature.
18227 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18228 Alan Hayward <alan.hayward@arm.com>
18229 David Sherwood <david.sherwood@arm.com>
18231 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18232 subregs whose inner modes can be stored in GPRs.
18233 (aarch64_classify_index): Likewise.
18235 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18236 Alan Hayward <alan.hayward@arm.com>
18237 David Sherwood <david.sherwood@arm.com>
18239 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18240 (V_INT_EQUIV): ...this.
18241 (v_cmp_result): Rename to...
18242 (v_int_equiv): ...this.
18243 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18244 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18245 (copysign<mode>3): Likewise.
18246 (aarch64_simd_bsl<mode>_internal): Likewise.
18247 (aarch64_simd_bsl<mode>): Likewise.
18248 (vec_cmp<mode><mode>): Likewise.
18249 (vcond<mode><mode>): Likewise.
18250 (vcond<v_cmp_mixed><mode>): Likewise.
18251 (vcondu<mode><v_cmp_mixed>): Likewise.
18252 (aarch64_cm<optab><mode>): Likewise.
18253 (aarch64_cmtst<mode>): Likewise.
18254 (aarch64_fac<optab><mode>): Likewise.
18255 (vec_perm_const<mode>): Likewise.
18256 (vcond_mask_<mode><v_cmp_result>): Rename to...
18257 (vcond_mask_<mode><v_int_equiv>): ...this.
18258 (vec_cmp<mode><v_cmp_result>): Rename to...
18259 (vec_cmp<mode><v_int_equiv>): ...this.
18261 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18262 Alan Hayward <alan.hayward@arm.com>
18263 David Sherwood <david.sherwood@arm.com>
18265 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18267 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18268 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18269 (UNSPEC_LD4_DREG): New unspecs.
18270 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18271 (aarch64_ld2<mode>_dreg_be): Replace with...
18272 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18274 (aarch64_ld3<mode>_dreg_le)
18275 (aarch64_ld3<mode>_dreg_be): Replace with...
18276 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18278 (aarch64_ld4<mode>_dreg_le)
18279 (aarch64_ld4<mode>_dreg_be): Replace with...
18280 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18283 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18285 PR tree-optimization/81987
18286 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18287 insert an initializer in a location not dominated by the stride
18290 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18292 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18293 on the entire header of the finally block in the fallthru case.
18295 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18297 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18299 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18301 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18302 rs6000_emit_move_from_cr and call renamed function.
18303 (rs6000_emit_prologue): Call renamed functions.
18304 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18305 movesi_from_cr, remove volatile CRs.
18307 2017-08-30 Jon Beniston <jon@beniston.com>
18308 Richard Biener <rguenther@suse.de>
18310 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18311 of VECTOR_MODE_P check.
18312 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18313 element vector types.
18315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18317 * df.h (df_read_modify_subreg_p): Remove in favor of...
18318 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18319 const_rtx instead of an rtx.
18320 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18321 * df-problems.c (df_word_lr_mark_ref): Likewise.
18322 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18323 (mark_pseudo_reg_dead): Likewise.
18324 (mark_ref_dead): Likewise.
18325 * reginfo.c (init_subregs_of_mode): Likewise.
18326 * sched-deps.c (sched_analyze_1): Likewise.
18327 * df-scan.c (df_def_record_1): Likewise.
18328 (df_uses_record): Likewise.
18329 (df_read_modify_subreg_p): Remove in favor of...
18330 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18331 const_rtx instead of an rtx.
18333 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18334 Alan Hayward <alan.hayward@arm.com>
18335 David Sherwood <david.sherwood@arm.com>
18337 * rtl.h (partial_subreg_p): New function.
18338 * caller-save.c (save_call_clobbered_regs): Use it.
18339 * calls.c (expand_call): Likewise.
18340 * combine.c (combinable_i3pat): Likewise.
18341 (simplify_set): Likewise.
18342 (make_extraction): Likewise.
18343 (make_compound_operation_int): Likewise.
18344 (gen_lowpart_or_truncate): Likewise.
18345 (force_to_mode): Likewise.
18346 (make_field_assignment): Likewise.
18347 (reg_truncated_to_mode): Likewise.
18348 (record_truncated_value): Likewise.
18349 (move_deaths): Likewise.
18350 * cse.c (record_jump_cond): Likewise.
18351 (cse_insn): Likewise.
18352 * cselib.c (cselib_lookup_1): Likewise.
18353 * expmed.c (extract_bit_field_using_extv): Likewise.
18354 * function.c (assign_parm_setup_reg): Likewise.
18355 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18356 * ira-build.c (create_insn_allocnos): Likewise.
18357 * lra-coalesce.c (merge_pseudos): Likewise.
18358 * lra-constraints.c (match_reload): Likewise.
18359 (simplify_operand_subreg): Likewise.
18360 (curr_insn_transform): Likewise.
18361 * lra-lives.c (process_bb_lives): Likewise.
18362 * lra.c (new_insn_reg): Likewise.
18363 (lra_substitute_pseudo): Likewise.
18364 * regcprop.c (mode_change_ok): Likewise.
18365 (maybe_mode_change): Likewise.
18366 (copyprop_hardreg_forward_1): Likewise.
18367 * reload.c (push_reload): Likewise.
18368 (find_reloads): Likewise.
18369 (find_reloads_subreg_address): Likewise.
18370 * reload1.c (alter_reg): Likewise.
18371 (eliminate_regs_1): Likewise.
18372 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18374 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18376 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18377 back to if statements, including unpack.
18379 2017-08-30 Martin Liska <mliska@suse.cz>
18381 PR inline-asm/82001
18382 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18384 (func_checker::compare_asm_inputs_outputs): ... this function.
18385 (func_checker::compare_gimple_asm): Use the function to compare
18386 also ASM constrains.
18387 * ipa-icf-gimple.h: Rename the function.
18389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18390 Alan Hayward <alan.hayward@arm.com>
18391 David Sherwood <david.sherwood@arm.com>
18393 * coretypes.h (complex_mode): New type.
18394 * gdbhooks.py (build_pretty_printer): Handle it.
18395 * machmode.h (complex_mode): New class.
18396 (complex_mode::includes_p): New function.
18397 (is_complex_int_mode): Likewise.
18398 (is_complex_float_mode): Likewise.
18399 * genmodes.c (get_mode_class): Handle complex mode classes.
18400 * function.c (expand_function_end): Use is_complex_int_mode.
18402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18403 Alan Hayward <alan.hayward@arm.com>
18404 David Sherwood <david.sherwood@arm.com>
18406 * coretypes.h (scalar_mode_pod): New typedef.
18407 * gdbhooks.py (build_pretty_printer): Handle it.
18408 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18409 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18410 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18411 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18412 as_a <scalar_mode>.
18414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18415 Alan Hayward <alan.hayward@arm.com>
18416 David Sherwood <david.sherwood@arm.com>
18418 * machmode.h (mode_for_vector): Take a scalar_mode instead
18420 * stor-layout.c (mode_for_vector): Likewise.
18421 * explow.c (promote_mode): Use as_a <scalar_mode>.
18422 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18425 Alan Hayward <alan.hayward@arm.com>
18426 David Sherwood <david.sherwood@arm.com>
18428 * target.def (preferred_simd_mode): Take a scalar_mode
18429 instead of a machine_mode.
18430 * targhooks.h (default_preferred_simd_mode): Likewise.
18431 * targhooks.c (default_preferred_simd_mode): Likewise.
18432 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18433 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18434 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18435 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18436 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18437 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18438 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18439 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18441 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18442 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18443 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18444 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18445 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18446 * doc/tm.texi: Regenerate.
18447 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18451 Alan Hayward <alan.hayward@arm.com>
18452 David Sherwood <david.sherwood@arm.com>
18454 * target.def (scalar_mode_supported_p): Take a scalar_mode
18455 instead of a machine_mode.
18456 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18457 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18458 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18459 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18460 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18461 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18462 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18463 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18464 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18465 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18466 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18467 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18468 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18469 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18471 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18472 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18473 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18474 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18475 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18477 * doc/tm.texi: Regenerate.
18479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18480 Alan Hayward <alan.hayward@arm.com>
18481 David Sherwood <david.sherwood@arm.com>
18483 * coretypes.h (opt_scalar_mode): New typedef.
18484 * gdbhooks.py (build_pretty_printers): Handle it.
18485 * machmode.h (mode_iterator::get_2xwider): Add overload for
18487 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18489 * expr.c (convert_mode_scalar): Likewise.
18490 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18491 * optabs.c (expand_float): Likewise.
18492 (expand_fix): Likewise.
18493 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18496 Alan Hayward <alan.hayward@arm.com>
18497 David Sherwood <david.sherwood@arm.com>
18499 * optabs.c (expand_float): Explicitly check for scalars before
18500 using a branching expansion.
18501 (expand_fix): Likewise.
18503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18504 Alan Hayward <alan.hayward@arm.com>
18505 David Sherwood <david.sherwood@arm.com>
18507 * expr.c (convert_mode): Split scalar handling out into...
18508 (convert_mode_scalar): ...this new function. Treat the modes
18511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18512 Alan Hayward <alan.hayward@arm.com>
18513 David Sherwood <david.sherwood@arm.com>
18515 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18517 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18520 Alan Hayward <alan.hayward@arm.com>
18521 David Sherwood <david.sherwood@arm.com>
18523 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18524 rather than a machine_mode.
18525 (fixed_from_string): Likewise.
18526 (fixed_convert): Likewise.
18527 (fixed_convert_from_int): Likewise.
18528 (fixed_convert_from_real): Likewise.
18529 (real_convert_from_fixed): Likewise.
18530 * fixed-value.c (fixed_from_double_int): Likewise.
18531 (fixed_from_string): Likewise.
18532 (fixed_convert): Likewise.
18533 (fixed_convert_from_int): Likewise.
18534 (fixed_convert_from_real): Likewise.
18535 (real_convert_from_fixed): Likewise.
18536 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18539 Alan Hayward <alan.hayward@arm.com>
18540 David Sherwood <david.sherwood@arm.com>
18542 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18543 of separate mode class checks. Do not allow vector modes here.
18544 (immed_wide_int_const): Use as_a <scalar_mode>.
18545 * explow.c (trunc_int_for_mode): Likewise.
18546 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18547 (wi::shwi): Likewise.
18548 (wi::min_value): Likewise.
18549 (wi::max_value): Likewise.
18550 * dwarf2out.c (loc_descriptor): Likewise.
18551 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18552 for CONST_WIDE_INT.
18554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18555 Alan Hayward <alan.hayward@arm.com>
18556 David Sherwood <david.sherwood@arm.com>
18558 * tree.h (SCALAR_TYPE_MODE): New macro.
18559 * expr.c (expand_expr_addr_expr_1): Use it.
18560 (expand_expr_real_2): Likewise.
18561 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18562 (fold_convert_const_fixed_from_int): Likewise.
18563 (fold_convert_const_fixed_from_real): Likewise.
18564 (native_encode_fixed): Likewise
18565 (native_encode_complex): Likewise
18566 (native_encode_vector): Likewise.
18567 (native_interpret_fixed): Likewise.
18568 (native_interpret_real): Likewise.
18569 (native_interpret_complex): Likewise.
18570 (native_interpret_vector): Likewise.
18571 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18572 (simd_clone_adjust_argument_types): Likewise.
18573 (simd_clone_init_simd_arrays): Likewise.
18574 (simd_clone_adjust): Likewise.
18575 * stor-layout.c (layout_type): Likewise.
18576 * tree.c (build_minus_one_cst): Likewise.
18577 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18578 * tree-inline.c (estimate_move_cost): Likewise.
18579 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18580 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18581 (vectorizable_reduction): Likewise.
18582 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18583 (vect_recog_mixed_size_cond_pattern): Likewise.
18584 (check_bool_pattern): Likewise.
18585 (adjust_bool_pattern): Likewise.
18586 (search_type_for_mask_1): Likewise.
18587 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18588 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18589 (vectorizable_load): Likewise.
18590 (vectorizable_store): Likewise.
18591 * ubsan.c (ubsan_encode_value): Likewise.
18592 * varasm.c (output_constant): Likewise.
18594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18595 Alan Hayward <alan.hayward@arm.com>
18596 David Sherwood <david.sherwood@arm.com>
18598 * coretypes.h (scalar_mode): New class.
18599 * machmode.h (scalar_mode): Likewise.
18600 (scalar_mode::includes_p): New function.
18601 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18602 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18603 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18604 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18605 * expmed.c (store_bit_field_1): Likewise.
18606 (extract_bit_field_1): Likewise.
18607 * expr.c (write_complex_part): Likewise.
18608 (read_complex_part): Likewise.
18609 (emit_move_complex_push): Likewise.
18610 (expand_expr_real_2): Likewise.
18611 * function.c (assign_parm_setup_reg): Likewise.
18612 (assign_parms_unsplit_complex): Likewise.
18613 * optabs.c (expand_binop): Likewise.
18614 * rtlanal.c (subreg_get_info): Likewise.
18615 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18616 * varasm.c (output_constant_pool_2): Likewise.
18618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18619 Alan Hayward <alan.hayward@arm.com>
18620 David Sherwood <david.sherwood@arm.com>
18622 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18624 (expmed_mult_highpart_optab): Likewise.
18625 (expmed_mult_highpart): Likewise.
18627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18628 Alan Hayward <alan.hayward@arm.com>
18629 David Sherwood <david.sherwood@arm.com>
18631 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18632 instead of a machine_mode.
18633 (builtin_memset_read_str): Likewise.
18634 * builtins.c (c_readstr): Likewise.
18635 (builtin_memcpy_read_str): Likewise.
18636 (builtin_strncpy_read_str): Likewise.
18637 (builtin_memset_read_str): Likewise.
18638 (builtin_memset_gen_str): Likewise.
18639 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18640 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18641 instead of a machine_mode.
18642 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18644 (make_extraction): Likewise.
18645 (try_widen_shift_mode): Take and return scalar_int_modes instead
18647 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18648 a scalar_int_mode instead of a machine_mode.
18649 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18650 (avr_addr_space_pointer_mode): Likewise.
18651 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18652 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18653 (msp430_unwind_word_mode): Likewise.
18654 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18655 (spu_addr_space_pointer_mode): Likewise.
18656 (spu_addr_space_address_mode): Likewise.
18657 (spu_libgcc_cmp_return_mode): Likewise.
18658 (spu_libgcc_shift_count_mode): Likewise.
18659 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18660 (rl78_addr_space_pointer_mode): Likewise.
18661 (fl78_unwind_word_mode): Likewise.
18662 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18664 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18665 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18666 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18667 (mips_valid_pointer_mode): Likewise.
18668 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18669 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18670 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18672 (ft32_addr_space_address_mode): Likewise.
18673 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18674 scalar_int_mode instead of a machine_mode.
18675 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18677 (m32c_addr_space_address_mode): Likewise.
18678 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18679 (rs6000_eh_return_filter_mode): Likewise.
18680 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18681 (rs6000_eh_return_filter_mode): Likewise.
18682 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18683 (s390_libgcc_shift_count_mode): Likewise.
18684 (s390_unwind_word_mode): Likewise.
18685 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18687 * target.def (mode_rep_extended): Likewise.
18688 (valid_pointer_mode): Likewise.
18689 (addr_space.valid_pointer_mode): Likewise.
18690 (eh_return_filter_mode): Return a scalar_int_mode rather than
18692 (libgcc_cmp_return_mode): Likewise.
18693 (libgcc_shift_count_mode): Likewise.
18694 (unwind_word_mode): Likewise.
18695 (addr_space.pointer_mode): Likewise.
18696 (addr_space.address_mode): Likewise.
18697 * doc/tm.texi: Regenerate.
18698 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18700 (do_jump): Use scalar_int_mode for local variables.
18701 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18702 rather than a machine_mode.
18703 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18704 (scompare_loc_descriptor_wide): Likewise.
18705 (scompare_loc_descriptor_narrow): Likewise.
18706 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18708 * except.c (sjlj_emit_dispatch_table): Likewise.
18709 (expand_builtin_eh_copy_values): Likewise.
18710 * explow.c (convert_memory_address_addr_space_1): Likewise.
18711 Take a scalar_int_mode rather than a machine_mode.
18712 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18713 than a machine_mode.
18714 (memory_address_addr_space): Use scalar_int_mode for local variables.
18715 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18716 rather than a machine_mode.
18717 * expmed.c (mask_rtx): Likewise.
18718 (init_expmed_one_conv): Likewise.
18719 (expand_mult_highpart_adjust): Likewise.
18720 (extract_high_half): Likewise.
18721 (expmed_mult_highpart_optab): Likewise.
18722 (expmed_mult_highpart): Likewise.
18723 (expand_smod_pow2): Likewise.
18724 (expand_sdiv_pow2): Likewise.
18725 (emit_store_flag_int): Likewise.
18726 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18728 (extract_low_bits): Likewise.
18729 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18731 * expr.c (pieces_addr::adjust): Likewise.
18732 (can_store_by_pieces): Likewise.
18733 (store_by_pieces): Likewise.
18734 (clear_by_pieces_1): Likewise.
18735 (expand_expr_addr_expr_1): Likewise.
18736 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18737 (expand_expr_real_1): Likewise.
18738 (try_casesi): Likewise.
18739 * final.c (shorten_branches): Likewise.
18740 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18741 type of "mode" to machine_mode.
18742 * internal-fn.c (expand_arith_overflow_result_store): Take a
18743 scalar_int_mode rather than a machine_mode.
18744 (expand_mul_overflow): Use scalar_int_mode for local variables.
18745 * loop-doloop.c (doloop_modify): Likewise.
18746 (doloop_optimize): Likewise.
18747 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18748 than a machine_mode.
18749 (expand_doubleword_shift_condmove): Likewise.
18750 (expand_doubleword_shift): Likewise.
18751 (expand_doubleword_clz): Likewise.
18752 (expand_doubleword_popcount): Likewise.
18753 (expand_doubleword_parity): Likewise.
18754 (expand_absneg_bit): Use scalar_int_mode for local variables.
18755 (prepare_float_lib_cmp): Likewise.
18756 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18757 rather than a machine_mode.
18758 (convert_memory_address_addr_space): Likewise.
18759 (get_mode_bounds): Likewise.
18760 (get_address_mode): Return a scalar_int_mode rather than a
18762 * rtlanal.c (get_address_mode): Likewise.
18763 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18764 than a machine_mode.
18765 * targhooks.c (default_mode_rep_extended): Likewise.
18766 (default_valid_pointer_mode): Likewise.
18767 (default_addr_space_valid_pointer_mode): Likewise.
18768 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18769 than a machine_mode.
18770 (default_libgcc_cmp_return_mode): Likewise.
18771 (default_libgcc_shift_count_mode): Likewise.
18772 (default_unwind_word_mode): Likewise.
18773 (default_addr_space_pointer_mode): Likewise.
18774 (default_addr_space_address_mode): Likewise.
18775 * targhooks.h (default_eh_return_filter_mode): Likewise.
18776 (default_libgcc_cmp_return_mode): Likewise.
18777 (default_libgcc_shift_count_mode): Likewise.
18778 (default_unwind_word_mode): Likewise.
18779 (default_addr_space_pointer_mode): Likewise.
18780 (default_addr_space_address_mode): Likewise.
18781 (default_mode_rep_extended): Take a scalar_int_mode rather than
18783 (default_valid_pointer_mode): Likewise.
18784 (default_addr_space_valid_pointer_mode): Likewise.
18785 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18787 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18788 rather than a machine_mode.
18789 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18790 for local variables.
18791 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18792 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18793 than a machine_mode.
18795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18796 Alan Hayward <alan.hayward@arm.com>
18797 David Sherwood <david.sherwood@arm.com>
18799 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18800 the mode argument to scalar_int_mode.
18801 (do_jump_by_parts_zero_rtx): Likewise.
18802 (do_jump_by_parts_equality_rtx): Likewise.
18803 (do_jump_by_parts_greater): Take a mode argument.
18804 (do_jump_by_parts_equality): Likewise.
18805 (do_jump_1): Update calls accordingly.
18807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18808 Alan Hayward <alan.hayward@arm.com>
18809 David Sherwood <david.sherwood@arm.com>
18811 * is-a.h (safe_dyn_cast): New function.
18812 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18813 (jump_table_for_label): Likewise.
18814 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18815 instead of an rtx_insn *.
18816 (shorten_branches): Use dyn_cast instead of LABEL_P and
18817 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18818 rtx_jump_table_data::get_data_mode.
18819 (final_scan_insn): Likewise.
18821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18822 Alan Hayward <alan.hayward@arm.com>
18823 David Sherwood <david.sherwood@arm.com>
18825 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18826 trying to combine a full-register integer set with a subreg
18829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18830 Alan Hayward <alan.hayward@arm.com>
18831 David Sherwood <david.sherwood@arm.com>
18833 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18834 that is always either address_mode or pointer_mode.
18836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18837 Alan Hayward <alan.hayward@arm.com>
18838 David Sherwood <david.sherwood@arm.com>
18840 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18841 when the two are known to be equal.
18843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18844 Alan Hayward <alan.hayward@arm.com>
18845 David Sherwood <david.sherwood@arm.com>
18847 * simplify-rtx.c (simplify_const_unary_operation): Use
18848 is_a <scalar_int_mode> instead of checking for a nonzero
18849 precision. Forcibly convert op_mode to a scalar_int_mode
18850 in that case. More clearly differentiate the operand and
18851 result modes and use the former when deciding what the value
18852 of a count-bits operation should be. Use is_int_mode instead
18853 of checking for a MODE_INT. Remove redundant check for whether
18854 this mode has a zero precision.
18856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18857 Alan Hayward <alan.hayward@arm.com>
18858 David Sherwood <david.sherwood@arm.com>
18860 * optabs.c (widen_leading): Change the type of the mode argument
18861 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18862 (widen_bswap): Likewise.
18863 (expand_parity): Likewise.
18864 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18865 (expand_ffs): Likewise.
18866 (epand_unop): Check for scalar integer modes before calling the
18869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18870 Alan Hayward <alan.hayward@arm.com>
18871 David Sherwood <david.sherwood@arm.com>
18873 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18875 (expand_expr_real_1): Update call accordingly.
18877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18878 Alan Hayward <alan.hayward@arm.com>
18879 David Sherwood <david.sherwood@arm.com>
18881 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18882 value_mode arguments. Use scalar_int_mode internally.
18883 (store_bit_field_1): Rename the new integer mode from imode
18884 to op0_mode and use it instead of GET_MODE (op0). Update calls
18885 to store_split_bit_field, store_bit_field_using_insv and
18886 store_fixed_bit_field.
18887 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18888 Use scalar_int_mode internally. Use a bit count rather than a mode
18889 when calculating the largest bit size for get_best_mode.
18890 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18891 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18892 Remove assertion that OP0 has a scalar integer mode.
18893 (store_split_bit_field): Add op0_mode and value_mode arguments.
18894 Update calls to extract_fixed_bit_field.
18895 (extract_bit_field_using_extv): Add an op0_mode argument.
18896 Use scalar_int_mode internally.
18897 (extract_bit_field_1): Rename the new integer mode from imode to
18898 op0_mode and use it instead of GET_MODE (op0). Update calls to
18899 extract_split_bit_field, extract_bit_field_using_extv and
18900 extract_fixed_bit_field.
18901 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18902 to extract_split_bit_field and extract_fixed_bit_field_1.
18903 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18904 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18905 on the target mode.
18906 (extract_split_bit_field): Add an op0_mode argument. Update call
18907 to extract_fixed_bit_field.
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 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18914 * explow.c (hard_function_value): Likewise.
18915 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18916 convert_to_mode call outside the loop.
18917 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18918 for the mode iterator. Require the mode specified by max_pieces
18920 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18922 (copy_blkmode_to_reg): Likewise.
18923 (set_storage_via_setmem): Likewise.
18924 * optabs.c (prepare_cmp_insn): Likewise.
18925 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18926 * stor-layout.c (finish_bitfield_representative): Likewise.
18928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18929 Alan Hayward <alan.hayward@arm.com>
18930 David Sherwood <david.sherwood@arm.com>
18932 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18933 * expr.c (convert_move): Use them.
18934 (convert_modes): Likewise.
18935 (store_expr_with_bounds): Likewise.
18937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18938 Alan Hayward <alan.hayward@arm.com>
18939 David Sherwood <david.sherwood@arm.com>
18941 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18942 parameter for the mode of "x". Remove the "known_x", "known_mode"
18943 and "known_ret" arguments. Change the type of the mode argument
18944 to scalar_int_mode.
18945 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18946 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18947 (reg_num_sign_bit_copies_for_combine): Likewise.
18948 * rtlanal.c (nonzero_bits1): Likewise.
18949 (num_sign_bit_copies1): Likewise.
18950 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18951 (reg_num_sign_bit_copies_general): Likewise.
18952 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18953 (reg_nonzero_bits_general): Likewise.
18955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18956 Alan Hayward <alan.hayward@arm.com>
18957 David Sherwood <david.sherwood@arm.com>
18959 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18960 than in subroutines. Return 1 for non-integer modes.
18961 (cached_num_sign_bit_copies): Change the type of the mode parameter
18962 to scalar_int_mode.
18963 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18964 classes. Handle CONST_INT_P first and then check whether X also
18965 has a scalar integer mode. Check the same thing for inner registers
18966 of a SUBREG and for values that are being extended or truncated.
18968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18969 Alan Hayward <alan.hayward@arm.com>
18970 David Sherwood <david.sherwood@arm.com>
18972 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18973 in subroutines. Return the mode mask for non-integer modes.
18974 (cached_nonzero_bits): Change the type of the mode parameter
18975 to scalar_int_mode.
18976 (nonzero_bits1): Likewise. Remove early exit for other mode
18977 classes. Handle CONST_INT_P first and then check whether X
18978 also has a scalar integer mode.
18980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18981 Alan Hayward <alan.hayward@arm.com>
18982 David Sherwood <david.sherwood@arm.com>
18984 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18985 Return a scalar_int_mode and assert that the size is greater than
18987 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18988 (op_by_pieces_d::op_by_pieces_d): Likewise.
18989 (op_by_pieces_d::run): Likewise.
18990 (can_store_by_pieces): Likewise.
18992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18993 Alan Hayward <alan.hayward@arm.com>
18994 David Sherwood <david.sherwood@arm.com>
18996 * combine.c (extract_left_shift): Add a mode argument and update
18998 (make_compound_operation_int): Change the type of the mode parameter
18999 to scalar_int_mode and update the call to extract_left_shift.
19001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19002 Alan Hayward <alan.hayward@arm.com>
19003 David Sherwood <david.sherwood@arm.com>
19005 * combine.c (simplify_and_const_int): Change the type of the mode
19006 parameter to scalar_int_mode.
19007 (simplify_and_const_int_1): Likewise. Update recursive call.
19009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19010 Alan Hayward <alan.hayward@arm.com>
19011 David Sherwood <david.sherwood@arm.com>
19013 * combine.c (simplify_compare_const): Check that the mode is a
19014 scalar_int_mode (rather than VOIDmode) before testing its
19016 (simplify_comparison): Move COMPARISON_P handling out of the
19017 loop and restrict the latter part of the loop to scalar_int_modes.
19018 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
19019 and when considering SUBREG_REGs. Use is_int_mode instead of
19020 checking GET_MODE_CLASS against MODE_INT.
19022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19023 Alan Hayward <alan.hayward@arm.com>
19024 David Sherwood <david.sherwood@arm.com>
19026 * combine.c (try_widen_shift_mode): Move check for equal modes to...
19027 (simplify_shift_const_1): ...here. Use scalar_int_mode for
19028 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
19035 (force_to_mode): ...here. Keep xmode up-to-date and use it
19036 instead of GET_MODE (x).
19038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19039 Alan Hayward <alan.hayward@arm.com>
19040 David Sherwood <david.sherwood@arm.com>
19042 * optabs-query.h (extraction_insn::struct_mode): Change type to
19043 opt_scalar_int_mode and update comment.
19044 (extraction_insn::field_mode): Change type to scalar_int_mode.
19045 (extraction_insn::pos_mode): Likewise.
19046 * combine.c (make_extraction): Update accordingly.
19047 * optabs-query.c (get_traditional_extraction_insn): Likewise.
19048 (get_optab_extraction_insn): Likewise.
19049 * recog.c (simplify_while_replacing): Likewise.
19050 * expmed.c (narrow_bit_field_mem): Change the type of the mode
19051 parameter to opt_scalar_int_mode.
19053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19054 Alan Hayward <alan.hayward@arm.com>
19055 David Sherwood <david.sherwood@arm.com>
19057 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
19058 to a scalar_int_mode instead of a machine_mode.
19059 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19060 (get_best_mode): Return a boolean and use a pointer argument to store
19061 the selected mode. Replace the limit mode parameter with a bit limit.
19062 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19063 for the values returned by bit_field_mode_iterator::next_mode.
19064 (store_bit_field): Update call to get_best_mode.
19065 (store_fixed_bit_field): Likewise.
19066 (extract_fixed_bit_field): Likewise.
19067 * expr.c (optimize_bitfield_assignment_op): Likewise.
19068 * fold-const.c (optimize_bit_field_compare): Likewise.
19069 (fold_truth_andor_1): Likewise.
19070 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19071 Update for new type of m_mode.
19072 (get_best_mode): As above.
19074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19075 Alan Hayward <alan.hayward@arm.com>
19076 David Sherwood <david.sherwood@arm.com>
19078 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19079 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19080 (store_bit_field): Check is_a <scalar_int_mode> before calling
19081 strict_volatile_bitfield_p.
19082 (extract_bit_field): Likewise.
19084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19085 Alan Hayward <alan.hayward@arm.com>
19086 David Sherwood <david.sherwood@arm.com>
19088 * target.def (cstore_mode): Return a scalar_int_mode.
19089 * doc/tm.texi: Regenerate.
19090 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19091 * targhooks.h (default_cstore_mode): Likewise.
19092 * targhooks.c (default_cstore_mode): Likewise, using a forced
19094 * expmed.c (emit_cstore): Expect the target of the cstore to be
19097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19098 Alan Hayward <alan.hayward@arm.com>
19099 David Sherwood <david.sherwood@arm.com>
19101 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19103 (niter_desc): Likewise mode.
19104 (iv_analyze): Add a mode parameter.
19106 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19107 and change its type to scalar_int_mode.
19108 * loop-iv.c: Update commentary at head of file.
19109 (iv_constant): Pass the mode paraeter before the rtx it describes
19110 and change its type to scalar_int_mode. Remove VOIDmode handling.
19111 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19112 (iv_extend): Likewise.
19113 (shorten_into_mode): Likewise.
19114 (iv_add): Use scalar_int_mode.
19115 (iv_mult): Likewise.
19116 (iv_shift): Likewise.
19117 (canonicalize_iv_subregs): Likewise.
19118 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19119 it describes and change its mode to scalar_int_mode. Also change
19120 the type of the returned inner_mode to scalar_int_mode.
19121 (get_biv_step): Likewise, turning outer_mode from a pointer
19122 into a direct parameter. Update call to get_biv_step_1.
19123 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19124 iv_constant and get_biv_step.
19125 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19126 and change its type to scalar_int_mode. Don't initialise iv->mode
19127 to VOIDmode and remove later checks for its still being VOIDmode.
19128 Update calls to iv_analyze_op and iv_analyze_expr. Check
19129 is_a <scalar_int_mode> when changing the mode under consideration.
19130 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19131 Update call to iv_analyze_expr.
19132 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19133 inner register is not also a scalar_int_mode. Update call to
19135 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19136 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19137 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19138 separate mode class checks. Update calls to iv_analyze. Remove
19139 fix-up of VOIDmodes after iv_analyze_biv.
19140 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19141 don't have a scalar_int_mode. Update call to biv_p.
19143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19144 Alan Hayward <alan.hayward@arm.com>
19145 David Sherwood <david.sherwood@arm.com>
19147 * cfgexpand.c (convert_debug_memory_address): Use
19148 as_a <scalar_int_mode>.
19149 * combine.c (expand_compound_operation): Likewise.
19150 (make_extraction): Likewise.
19151 (change_zero_ext): Likewise.
19152 (simplify_comparison): Likewise.
19153 * cse.c (cse_insn): Likewise.
19154 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19155 (mem_loc_descriptor): Likewise.
19156 (loc_descriptor): Likewise.
19157 * expmed.c (init_expmed_one_mode): Likewise.
19158 (synth_mult): Likewise.
19159 (emit_store_flag_1): Likewise.
19160 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19161 of a comparison with size.
19162 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19163 (reduce_to_bit_field_precision): Likewise.
19164 * function.c (expand_function_end): Likewise.
19165 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19166 * loop-doloop.c (doloop_modify): Likewise.
19167 * optabs.c (expand_binop): Likewise.
19168 (expand_unop): Likewise.
19169 (expand_copysign_absneg): Likewise.
19170 (prepare_cmp_insn): Likewise.
19171 (maybe_legitimize_operand): Likewise.
19172 * recog.c (const_scalar_int_operand): Likewise.
19173 * rtlanal.c (get_address_mode): Likewise.
19174 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19175 (simplify_cond_clz_ctz): Likewise.
19176 * tree-nested.c (get_nl_goto_field): Likewise.
19177 * tree.c (build_vector_type_for_mode): Likewise.
19178 * var-tracking.c (use_narrower_mode): Likewise.
19180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19181 Alan Hayward <alan.hayward@arm.com>
19182 David Sherwood <david.sherwood@arm.com>
19184 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19185 * builtins.c (expand_builtin_signbit): Use it.
19186 * cfgexpand.c (expand_debug_expr): Likewise.
19187 * dojump.c (do_jump): Likewise.
19188 (do_compare_and_jump): Likewise.
19189 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19190 * expmed.c (make_tree): Likewise.
19191 * expr.c (expand_expr_real_2): Likewise.
19192 (expand_expr_real_1): Likewise.
19193 (try_casesi): Likewise.
19194 * fold-const-call.c (fold_const_call_ss): Likewise.
19195 * fold-const.c (unextend): Likewise.
19196 (extract_muldiv_1): Likewise.
19197 (fold_single_bit_test): Likewise.
19198 (native_encode_int): Likewise.
19199 (native_encode_string): Likewise.
19200 (native_interpret_int): Likewise.
19201 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19202 * internal-fn.c (expand_addsub_overflow): Likewise.
19203 (expand_neg_overflow): Likewise.
19204 (expand_mul_overflow): Likewise.
19205 (expand_arith_overflow): Likewise.
19206 * match.pd: Likewise.
19207 * stor-layout.c (layout_type): Likewise.
19208 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19209 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19210 * tree-ssanames.c (get_range_info): Likewise.
19211 * tree-switch-conversion.c (array_value_type) Likewise.
19212 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19213 (vect_recog_divmod_pattern): Likewise.
19214 (vect_recog_mixed_size_cond_pattern): Likewise.
19215 * tree-vrp.c (extract_range_basic): Likewise.
19216 (simplify_float_conversion_using_ranges): Likewise.
19217 * tree.c (int_fits_type_p): Likewise.
19218 * ubsan.c (instrument_bool_enum_load): Likewise.
19219 * varasm.c (mergeable_string_section): Likewise.
19220 (narrowing_initializer_constant_valid_p): Likewise.
19221 (output_constant): Likewise.
19223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19224 Alan Hayward <alan.hayward@arm.com>
19225 David Sherwood <david.sherwood@arm.com>
19227 * machmode.h (NARROWEST_INT_MODE): New macro.
19228 * expr.c (alignment_for_piecewise_move): Use it instead of
19229 GET_CLASS_NARROWEST_MODE (MODE_INT).
19230 (push_block): Likewise.
19231 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19233 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19236 Alan Hayward <alan.hayward@arm.com>
19237 David Sherwood <david.sherwood@arm.com>
19239 * postreload.c (move2add_valid_value_p): Change the type of the
19240 mode parameter to scalar_int_mode.
19241 (move2add_use_add2_insn): Add a mode parameter and use it instead
19243 (move2add_use_add3_insn): Likewise.
19244 (reload_cse_move2add): Update accordingly.
19246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19247 Alan Hayward <alan.hayward@arm.com>
19248 David Sherwood <david.sherwood@arm.com>
19250 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19252 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19253 * optabs.c (expand_unop): Likewise.
19255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19256 Alan Hayward <alan.hayward@arm.com>
19257 David Sherwood <david.sherwood@arm.com>
19259 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19260 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19261 (popcount_loc_descriptor): Likewise.
19262 (bswap_loc_descriptor): Likewise.
19263 (rotate_loc_descriptor): Likewise.
19264 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19265 calling the functions above.
19267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19268 Alan Hayward <alan.hayward@arm.com>
19269 David Sherwood <david.sherwood@arm.com>
19271 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19273 (try_combine): Likewise.
19274 (simplify_if_then_else): Likewise.
19275 * cse.c (cse_insn): Likewise.
19276 * dwarf2out.c (mem_loc_descriptor): Likewise.
19277 * emit-rtl.c (gen_lowpart_common): Likewise.
19278 * simplify-rtx.c (simplify_truncation): Likewise.
19279 (simplify_binary_operation_1): Likewise.
19280 (simplify_const_relational_operation): Likewise.
19281 (simplify_ternary_operation): Likewise.
19282 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19285 Alan Hayward <alan.hayward@arm.com>
19286 David Sherwood <david.sherwood@arm.com>
19288 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19289 * reload.c (push_reload): Likewise.
19290 (find_reloads): Likewise.
19292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19293 Alan Hayward <alan.hayward@arm.com>
19294 David Sherwood <david.sherwood@arm.com>
19296 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19297 (make_compound_operation_int): Likewise.
19298 (change_zero_ext): Likewise.
19299 * expr.c (convert_move): Likewise.
19300 (convert_modes): Likewise.
19301 * fwprop.c (forward_propagate_subreg): Likewise.
19302 * loop-iv.c (get_biv_step_1): Likewise.
19303 * optabs.c (widen_operand): Likewise.
19304 * postreload.c (move2add_valid_value_p): Likewise.
19305 * recog.c (simplify_while_replacing): Likewise.
19306 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19307 (simplify_binary_operation_1): Likewise. Remove redundant
19308 mode equality check.
19310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19311 Alan Hayward <alan.hayward@arm.com>
19312 David Sherwood <david.sherwood@arm.com>
19314 * combine.c (combine_simplify_rtx): Add checks for
19315 is_a <scalar_int_mode>.
19316 (simplify_if_then_else): Likewise.
19317 (make_field_assignment): Likewise.
19318 (simplify_comparison): Likewise.
19319 * ifcvt.c (noce_try_bitop): Likewise.
19320 * loop-invariant.c (canonicalize_address_mult): Likewise.
19321 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19324 Alan Hayward <alan.hayward@arm.com>
19325 David Sherwood <david.sherwood@arm.com>
19327 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19328 is_a <scalar_int_mode> instead of != BLKmode.
19330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19331 Alan Hayward <alan.hayward@arm.com>
19332 David Sherwood <david.sherwood@arm.com>
19334 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19335 instead of != VOIDmode.
19336 * combine.c (if_then_else_cond): Likewise.
19337 (change_zero_ext): Likewise.
19338 * dwarf2out.c (mem_loc_descriptor): Likewise.
19339 (loc_descriptor): Likewise.
19340 * rtlanal.c (canonicalize_condition): Likewise.
19341 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19344 Alan Hayward <alan.hayward@arm.com>
19345 David Sherwood <david.sherwood@arm.com>
19347 * simplify-rtx.c (simplify_binary_operation_1): Use
19348 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19350 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19351 Alan Hayward <alan.hayward@arm.com>
19352 David Sherwood <david.sherwood@arm.com>
19354 * wide-int.h (int_traits<unsigned char>) New class.
19355 (int_traits<unsigned short>) Likewise.
19356 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19357 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19359 * combine.c (set_nonzero_bits_and_sign_copies): Use
19360 is_a <scalar_int_mode>.
19361 (find_split_point): Likewise.
19362 (combine_simplify_rtx): Likewise.
19363 (simplify_logical): Likewise.
19364 (expand_compound_operation): Likewise.
19365 (expand_field_assignment): Likewise.
19366 (make_compound_operation): Likewise.
19367 (extended_count): Likewise.
19368 (change_zero_ext): Likewise.
19369 (simplify_comparison): Likewise.
19370 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19371 (ucompare_loc_descriptor): Likewise.
19372 (minmax_loc_descriptor): Likewise.
19373 (mem_loc_descriptor): Likewise.
19374 (loc_descriptor): Likewise.
19375 * expmed.c (init_expmed_one_mode): Likewise.
19376 * lra-constraints.c (lra_constraint_offset): Likewise.
19377 * optabs.c (prepare_libcall_arg): Likewise.
19378 * postreload.c (move2add_note_store): Likewise.
19379 * reload.c (operands_match_p): Likewise.
19380 * rtl.h (load_extend_op): Likewise.
19381 * rtlhooks.c (gen_lowpart_general): Likewise.
19382 * simplify-rtx.c (simplify_truncation): Likewise.
19383 (simplify_unary_operation_1): Likewise.
19384 (simplify_binary_operation_1): Likewise.
19385 (simplify_const_binary_operation): Likewise.
19386 (simplify_const_relational_operation): Likewise.
19387 (simplify_subreg): Likewise.
19388 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19389 * var-tracking.c (adjust_mems): Likewise.
19390 (prepare_call_arguments): Likewise.
19392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19393 Alan Hayward <alan.hayward@arm.com>
19394 David Sherwood <david.sherwood@arm.com>
19396 * machmode.h (is_int_mode): New fuction.
19397 * combine.c (find_split_point): Use it.
19398 (combine_simplify_rtx): Likewise.
19399 (simplify_if_then_else): Likewise.
19400 (simplify_set): Likewise.
19401 (simplify_shift_const_1): Likewise.
19402 (simplify_comparison): Likewise.
19403 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19404 * cse.c (notreg_cost): Likewise.
19405 (cse_insn): Likewise.
19406 * cselib.c (cselib_lookup_1): Likewise.
19407 * dojump.c (do_jump_1): Likewise.
19408 (do_compare_rtx_and_jump): Likewise.
19409 * dse.c (get_call_args): Likewise.
19410 * dwarf2out.c (rtl_for_decl_init): Likewise.
19411 (native_encode_initializer): Likewise.
19412 * expmed.c (emit_store_flag_1): Likewise.
19413 (emit_store_flag): Likewise.
19414 * expr.c (convert_modes): Likewise.
19415 (store_field): Likewise.
19416 (expand_expr_real_1): Likewise.
19417 * fold-const.c (fold_read_from_constant_string): Likewise.
19418 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19419 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19420 * optabs.c (expand_binop): Likewise.
19421 (expand_unop): Likewise.
19422 (expand_abs_nojump): Likewise.
19423 (expand_one_cmpl_abs_nojump): Likewise.
19424 * simplify-rtx.c (mode_signbit_p): Likewise.
19425 (val_signbit_p): Likewise.
19426 (val_signbit_known_set_p): Likewise.
19427 (val_signbit_known_clear_p): Likewise.
19428 (simplify_relational_operation_1): Likewise.
19429 * tree.c (vector_type_mode): Likewise.
19431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19432 Alan Hayward <alan.hayward@arm.com>
19433 David Sherwood <david.sherwood@arm.com>
19435 * machmode.h (smallest_mode_for_size): Fix formatting.
19436 (smallest_int_mode_for_size): New function.
19437 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19438 instead of smallest_mode_for_size.
19439 * combine.c (make_extraction): Likewise.
19440 * config/arc/arc.c (arc_expand_movmem): Likewise.
19441 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19442 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19443 * config/s390/s390.c (s390_expand_insv): Likewise.
19444 * config/sparc/sparc.c (assign_int_registers): Likewise.
19445 * config/spu/spu.c (spu_function_value): Likewise.
19446 (spu_function_arg): Likewise.
19447 * coverage.c (get_gcov_type): Likewise.
19448 (get_gcov_unsigned_t): Likewise.
19449 * dse.c (find_shift_sequence): Likewise.
19450 * expmed.c (store_bit_field_1): Likewise.
19451 * expr.c (convert_move): Likewise.
19452 (store_field): Likewise.
19453 * internal-fn.c (expand_arith_overflow): Likewise.
19454 * optabs-query.c (get_best_extraction_insn): Likewise.
19455 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19456 * stor-layout.c (layout_type): Likewise.
19457 (initialize_sizetypes): Likewise.
19458 * targhooks.c (default_get_mask_mode): Likewise.
19459 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19462 Alan Hayward <alan.hayward@arm.com>
19463 David Sherwood <david.sherwood@arm.com>
19465 * machmode.h (opt_mode::else_blk): New function.
19466 (int_mode_for_mode): Declare.
19467 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19468 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19470 * cfgexpand.c (expand_debug_expr): Likewise.
19471 * combine.c (gen_lowpart_or_truncate): Likewise.
19472 (gen_lowpart_for_combine): Likewise.
19473 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19474 * config/avr/avr.c (avr_to_int_mode): Likewise.
19475 (avr_out_plus_1): Likewise.
19476 (avr_out_plus): Likewise.
19477 (avr_out_round): Likewise.
19478 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19479 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19480 (s390_expand_vcond): Likewise.
19481 * config/spu/spu.c (spu_split_immediate): Likewise.
19482 (spu_expand_mov): Likewise.
19483 * dse.c (get_stored_val): Likewise.
19484 * expmed.c (store_bit_field_1): Likewise.
19485 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19487 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19488 (extract_low_bits): Likewise.
19489 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19490 handling rather than repeating the check.
19491 (emit_group_store): Likewise.
19492 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19493 * optabs.c (expand_absneg_bit): Likewise.
19494 (expand_copysign_absneg): Likewise.
19495 (expand_copysign_bit): Likewise.
19496 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19497 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19498 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19499 * var-tracking.c (prepare_call_arguments): Likewise.
19500 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19501 int_mode_for_mode instead of mode_for_size.
19502 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19505 Alan Hayward <alan.hayward@arm.com>
19506 David Sherwood <david.sherwood@arm.com>
19508 * machmode.h (int_mode_for_size): New function.
19509 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19510 instead of mode_for_size.
19511 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19513 * combine.c (expand_field_assignment): Use int_mode_for_size
19514 instead of mode_for_size.
19515 (make_extraction): Likewise.
19516 (simplify_shift_const_1): Likewise.
19517 (simplify_comparison): Likewise.
19518 * dojump.c (do_jump): Likewise.
19519 * dwarf2out.c (mem_loc_descriptor): Likewise.
19520 * emit-rtl.c (init_derived_machine_modes): Likewise.
19521 * expmed.c (flip_storage_order): Likewise.
19522 (convert_extracted_bit_field): Likewise.
19523 * expr.c (copy_blkmode_from_reg): Likewise.
19524 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19525 * internal-fn.c (expand_mul_overflow): Likewise.
19526 * lower-subreg.c (simple_move): Likewise.
19527 * optabs-libfuncs.c (init_optabs): Likewise.
19528 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19529 * tree.c (vector_type_mode): Likewise.
19530 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19531 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19532 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19533 * tree-vect-stmts.c (vectorizable_load): Likewise.
19534 (vectorizable_store): Likewise.
19536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19537 Alan Hayward <alan.hayward@arm.com>
19538 David Sherwood <david.sherwood@arm.com>
19540 * coretypes.h (pod_mode): New type.
19541 (scalar_int_mode_pod): New typedef.
19542 * machmode.h (pod_mode): New class.
19543 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19544 * genmodes.c (emit_mode_int_n): Update accordingly.
19545 * lower-subreg.h (target_lower_subreg): Change type to
19546 scalar_int_mode_pod.
19547 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19548 scalar_int_mode_pod.
19550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19551 Alan Hayward <alan.hayward@arm.com>
19552 David Sherwood <david.sherwood@arm.com>
19554 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19555 machine_mode to scalar_int_mode.
19556 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19557 (rs6000_option_override_internal): Remove cast to int.
19558 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19559 machine_mode to scalar_int_mode.
19560 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19561 (rs6000_option_override_internal): Remove cast to int.
19562 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19563 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19565 * config/s390/s390.c (s390_expand_builtin): Likewise.
19566 * coretypes.h (scalar_int_mode): New type.
19567 (opt_scalar_int_mode): New typedef.
19568 * machmode.h (scalar_int_mode): New class.
19569 (scalar_int_mode::includes_p): New function.
19570 (byte_mode): Change type to scalar_int_mode.
19571 (word_mode): Likewise.
19572 (ptr_mode): Likewise.
19573 * emit-rtl.c (byte_mode): Likewise.
19574 (word_mode): Likewise.
19575 (ptr_mode): Likewise.
19576 (init_derived_machine_modes): Update accordingly.
19577 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19578 and MODE_PARTIAL_INT.
19579 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19580 opt_scalar_int_mode.
19582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19583 Alan Hayward <alan.hayward@arm.com>
19584 David Sherwood <david.sherwood@arm.com>
19586 * target.def (libgcc_floating_mode_supported_p): Take a
19588 * doc/tm.texi: Regenerate.
19589 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19591 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19592 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19596 Alan Hayward <alan.hayward@arm.com>
19597 David Sherwood <david.sherwood@arm.com>
19599 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19600 * doc/tm.texi: Regenerate.
19601 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19602 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19603 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19604 * targhooks.h (default_floatn_mode): Likewise.
19605 * targhooks.c (default_floatn_mode): Likewise.
19606 * tree.c (build_common_tree_nodes): Update accordingly.
19608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19609 Alan Hayward <alan.hayward@arm.com>
19610 David Sherwood <david.sherwood@arm.com>
19612 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19613 (mode_iterator::iterate_p): Likewise.
19614 (mode_iterator::get_wider): Likewise.
19615 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19618 Alan Hayward <alan.hayward@arm.com>
19619 David Sherwood <david.sherwood@arm.com>
19621 * coretypes.h (opt_scalar_float_mode): New typedef.
19622 * machmode.h (float_mode_for_size): New function.
19623 * emit-rtl.c (double_mode): Delete.
19624 (init_emit_once): Use float_mode_for_size.
19625 * stor-layout.c (layout_type): Likewise.
19626 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19629 Alan Hayward <alan.hayward@arm.com>
19630 David Sherwood <david.sherwood@arm.com>
19632 * output.h (assemble_real): Take a scalar_float_mode.
19633 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19634 * config/arm/arm.md (consttable_4): Likewise.
19635 (consttable_8): Likewise.
19636 (consttable_16): Likewise.
19637 * config/mips/mips.md (consttable_float): Likewise.
19638 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19639 * varasm.c (assemble_real): Take a scalar_float_mode.
19640 (output_constant_pool_2): Update accordingly.
19641 (output_constant): Likewise.
19643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19644 Alan Hayward <alan.hayward@arm.com>
19645 David Sherwood <david.sherwood@arm.com>
19647 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19648 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19649 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19650 (native_encode_real): Likewise.
19651 (native_interpret_real): Likewise.
19652 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19653 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19656 Alan Hayward <alan.hayward@arm.com>
19657 David Sherwood <david.sherwood@arm.com>
19659 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19660 <scalar_float_mode>. Simplify.
19661 (gen_extend_conv_libfunc): Likewise.
19663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19664 Alan Hayward <alan.hayward@arm.com>
19665 David Sherwood <david.sherwood@arm.com>
19667 * coretypes.h (scalar_float_mode): New type.
19668 * machmode.h (mode_traits::from_int): Use machine_mode if
19669 USE_ENUM_MODES is defined.
19670 (is_a): New function.
19672 (dyn_cast): Likewise.
19673 (scalar_float_mode): New class.
19674 (scalar_float_mode::includes_p): New function.
19675 (is_float_mode): Likewise.
19676 * gdbhooks.py (MachineModePrinter): New class.
19677 (build_pretty_printer): Use it for scalar_float_mode.
19678 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19679 (format_helper::format_helper): Turn into a template.
19680 * genmodes.c (get_mode_class): New function.
19681 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19682 or machine_mode if none.
19683 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19684 as_a <scalar_float_mode>.
19685 * dwarf2out.c (mem_loc_descriptor): Likewise.
19686 (insert_float): Likewise.
19687 (add_const_value_attribute): Likewise.
19688 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19689 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19690 (expand_unop): Update accordingly.
19691 (expand_abs_nojump): Likewise.
19692 (expand_copysign_absneg): Take a scalar_float_mode.
19693 (expand_copysign_bit): Likewise.
19694 (expand_copysign): Update accordingly.
19696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19697 Alan Hayward <alan.hayward@arm.com>
19698 David Sherwood <david.sherwood@arm.com>
19700 * coretypes.h (opt_mode): New class.
19701 * machmode.h (opt_mode): Likewise.
19702 (opt_mode::else_void): New function.
19703 (opt_mode::require): Likewise.
19704 (opt_mode::exists): Likewise.
19705 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19706 (GET_MODE_2XWIDER_MODE): Likewise.
19707 (mode_iterator::get_wider): Update accordingly.
19708 (mode_iterator::get_2xwider): Likewise.
19709 (mode_iterator::get_known_wider): Likewise, turning into a template.
19710 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19711 forcing a wider mode to exist.
19712 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19713 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19714 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19715 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19716 * lower-subreg.c (init_lower_subreg): Likewise.
19717 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19718 on the final iteration.
19719 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19720 a wider mode exists before asking for a move pattern.
19721 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19722 forcing a wider mode to exist.
19723 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19724 returning false if no such mode exists.
19725 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19726 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19727 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19728 Avoid checking for a MODE_INT if we already know the mode is not a
19730 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19731 forcing a wider mode to exist.
19732 (expmed_mult_highpart_optab): Likewise.
19733 (expmed_mult_highpart): Likewise.
19734 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19736 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19737 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19738 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19739 * internal-fn.c (expand_mul_overflow): Update use of
19740 GET_MODE_2XWIDER_MODE.
19741 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19742 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19743 GET_MODE_WIDER_MODE.
19744 (convert_plusminus_to_widen): Likewise.
19745 * tree-switch-conversion.c (array_value_type): Likewise.
19746 * var-tracking.c (emit_note_insn_var_location): Likewise.
19747 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19748 Return false inside rather than outside the loop if no wider mode
19750 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19751 and GET_MODE_2XWIDER_MODE
19752 (can_compare_p): Use else_void.
19753 * gdbhooks.py (OptMachineModePrinter): New class.
19754 (build_pretty_printer): Use it for opt_mode.
19756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19757 Alan Hayward <alan.hayward@arm.com>
19758 David Sherwood <david.sherwood@arm.com>
19760 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19761 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19764 Alan Hayward <alan.hayward@arm.com>
19765 David Sherwood <david.sherwood@arm.com>
19767 * machmode.h (mode_traits): New structure.
19768 (get_narrowest_mode): New function.
19769 (mode_iterator::start): Likewise.
19770 (mode_iterator::iterate_p): Likewise.
19771 (mode_iterator::get_wider): Likewise.
19772 (mode_iterator::get_known_wider): Likewise.
19773 (mode_iterator::get_2xwider): Likewise.
19774 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19775 (FOR_EACH_MODE): Likewise.
19776 (FOR_EACH_MODE_FROM): Likewise.
19777 (FOR_EACH_MODE_UNTIL): Likewise.
19778 (FOR_EACH_WIDER_MODE): Likewise.
19779 (FOR_EACH_2XWIDER_MODE): Likewise.
19780 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19781 * combine.c (simplify_comparison): Likewise
19782 * config/i386/i386.c (type_natural_mode): Likewise.
19783 * cse.c (cse_insn): Likewise.
19784 * dse.c (find_shift_sequence): Likewise.
19785 * emit-rtl.c (init_derived_machine_modes): Likewise.
19786 (init_emit_once): Likewise.
19787 * explow.c (hard_function_value): Likewise.
19788 * expmed.c (extract_fixed_bit_field_1): Likewise.
19789 (extract_bit_field_1): Likewise.
19790 (expand_divmod): Likewise.
19791 (emit_store_flag_1): Likewise.
19792 * expr.c (init_expr_target): Likewise.
19793 (convert_move): Likewise.
19794 (alignment_for_piecewise_move): Likewise.
19795 (widest_int_mode_for_size): Likewise.
19796 (emit_block_move_via_movmem): Likewise.
19797 (copy_blkmode_to_reg): Likewise.
19798 (set_storage_via_setmem): Likewise.
19799 (compress_float_constant): Likewise.
19800 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19801 * optabs-query.c (get_best_extraction_insn): Likewise.
19802 * optabs.c (expand_binop): Likewise.
19803 (expand_twoval_unop): Likewise.
19804 (expand_twoval_binop): Likewise.
19805 (widen_leading): Likewise.
19806 (widen_bswap): Likewise.
19807 (expand_parity): Likewise.
19808 (expand_unop): Likewise.
19809 (prepare_cmp_insn): Likewise.
19810 (prepare_float_lib_cmp): Likewise.
19811 (expand_float): Likewise.
19812 (expand_fix): Likewise.
19813 (expand_sfix_optab): Likewise.
19814 * postreload.c (move2add_use_add2_insn): Likewise.
19815 * reg-stack.c (reg_to_stack): Likewise.
19816 * reginfo.c (choose_hard_reg_mode): Likewise.
19817 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19818 * stor-layout.c (mode_for_size): Likewise.
19819 (smallest_mode_for_size): Likewise.
19820 (mode_for_vector): Likewise.
19821 (finish_bitfield_representative): Likewise.
19822 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19823 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19824 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19825 * var-tracking.c (prepare_call_arguments): Likewise.
19827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19828 Alan Hayward <alan.hayward@arm.com>
19829 David Sherwood <david.sherwood@arm.com>
19831 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19832 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19833 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19834 * machmode.h (mode_size): Move earlier in file.
19835 (mode_precision): Likewise.
19836 (mode_inner): Likewise.
19837 (mode_nunits): Likewise.
19838 (mode_unit_size): Likewise.
19839 (unit_unit_precision): Likewise.
19840 (mode_wider): Likewise.
19841 (mode_2xwider): Likewise.
19842 (machine_mode): New class.
19843 (mode_to_bytes): New function.
19844 (mode_to_bits): Likewise.
19845 (mode_to_precision): Likewise.
19846 (mode_to_inner): Likewise.
19847 (mode_to_unit_size): Likewise.
19848 (mode_to_unit_precision): Likewise.
19849 (mode_to_nunits): Likewise.
19850 (GET_MODE_SIZE): Use mode_to_bytes.
19851 (GET_MODE_BITSIZE): Use mode_to_bits.
19852 (GET_MODE_PRECISION): Use mode_to_precision.
19853 (GET_MODE_INNER): Use mode_to_inner.
19854 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19855 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19856 (GET_MODE_NUNITS): Use mode_to_nunits.
19857 * system.h (ALWAYS_INLINE): New macro.
19858 * config/powerpcspe/powerpcspe-c.c
19859 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19860 int for arg1_mode and arg2_mode.
19862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19863 Alan Hayward <alan.hayward@arm.com>
19864 David Sherwood <david.sherwood@arm.com>
19866 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19867 Prefix mode names with E_ in case statements.
19868 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19869 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19870 (aarch64_split_simd_move): Likewise.
19871 (aarch64_gen_storewb_pair): Likewise.
19872 (aarch64_gen_loadwb_pair): Likewise.
19873 (aarch64_gen_store_pair): Likewise.
19874 (aarch64_gen_load_pair): Likewise.
19875 (aarch64_get_condition_code_1): Likewise.
19876 (aarch64_constant_pool_reload_icode): Likewise.
19877 (get_rsqrte_type): Likewise.
19878 (get_rsqrts_type): Likewise.
19879 (get_recpe_type): Likewise.
19880 (get_recps_type): Likewise.
19881 (aarch64_gimplify_va_arg_expr): Likewise.
19882 (aarch64_simd_container_mode): Likewise.
19883 (aarch64_emit_load_exclusive): Likewise.
19884 (aarch64_emit_store_exclusive): Likewise.
19885 (aarch64_expand_compare_and_swap): Likewise.
19886 (aarch64_gen_atomic_cas): Likewise.
19887 (aarch64_emit_bic): Likewise.
19888 (aarch64_emit_atomic_swap): Likewise.
19889 (aarch64_emit_atomic_load_op): Likewise.
19890 (aarch64_evpc_trn): Likewise.
19891 (aarch64_evpc_uzp): Likewise.
19892 (aarch64_evpc_zip): Likewise.
19893 (aarch64_evpc_ext): Likewise.
19894 (aarch64_evpc_rev): Likewise.
19895 (aarch64_evpc_dup): Likewise.
19896 (aarch64_gen_ccmp_first): Likewise.
19897 (aarch64_gen_ccmp_next): Likewise.
19898 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19899 (alpha_emit_xfloating_libcall): Likewise.
19900 (emit_insxl): Likewise.
19901 (alpha_arg_type): Likewise.
19902 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19903 (arc_preferred_simd_mode): Likewise.
19904 (arc_secondary_reload): Likewise.
19905 (get_arc_condition_code): Likewise.
19906 (arc_print_operand): Likewise.
19907 (arc_legitimate_constant_p): Likewise.
19908 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19909 * config/arc/arc.md (casesi_load): Likewise.
19910 (casesi_compact_jump): Likewise.
19911 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19912 (cc_use_register): Likewise.
19913 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19914 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19915 (arm_init_iwmmxt_builtins): Likewise.
19916 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19917 (neon_expand_vector_init): Likewise.
19918 (arm_attr_length_move_neon): Likewise.
19919 (maybe_get_arm_condition_code): Likewise.
19920 (arm_emit_vector_const): Likewise.
19921 (arm_preferred_simd_mode): Likewise.
19922 (arm_output_iwmmxt_tinsr): Likewise.
19923 (thumb1_output_casesi): Likewise.
19924 (thumb2_output_casesi): Likewise.
19925 (arm_emit_load_exclusive): Likewise.
19926 (arm_emit_store_exclusive): Likewise.
19927 (arm_expand_compare_and_swap): Likewise.
19928 (arm_evpc_neon_vuzp): Likewise.
19929 (arm_evpc_neon_vzip): Likewise.
19930 (arm_evpc_neon_vrev): Likewise.
19931 (arm_evpc_neon_vtrn): Likewise.
19932 (arm_evpc_neon_vext): Likewise.
19933 (arm_validize_comparison): Likewise.
19934 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19935 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19936 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19937 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19938 (c6x_preferred_simd_mode): Likewise.
19939 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19940 (epiphany_rtx_costs): Likewise.
19941 * config/epiphany/predicates.md (proper_comparison_operator):
19943 * config/frv/frv.c (condexec_memory_operand): Likewise.
19944 (frv_emit_move): Likewise.
19945 (output_move_single): Likewise.
19946 (output_condmove_single): Likewise.
19947 (frv_hard_regno_mode_ok): Likewise.
19948 (frv_matching_accg_mode): Likewise.
19949 * config/h8300/h8300.c (split_adds_subs): Likewise.
19950 (h8300_rtx_costs): Likewise.
19951 (h8300_print_operand): Likewise.
19952 (compute_mov_length): Likewise.
19953 (output_logical_op): Likewise.
19954 (compute_logical_op_length): Likewise.
19955 (compute_logical_op_cc): Likewise.
19956 (h8300_shift_needs_scratch_p): Likewise.
19957 (output_a_shift): Likewise.
19958 (compute_a_shift_length): Likewise.
19959 (compute_a_shift_cc): Likewise.
19960 (expand_a_rotate): Likewise.
19961 (output_a_rotate): Likewise.
19962 * config/i386/i386.c (classify_argument): Likewise.
19963 (function_arg_advance_32): Likewise.
19964 (function_arg_32): Likewise.
19965 (function_arg_64): Likewise.
19966 (function_value_64): Likewise.
19967 (ix86_gimplify_va_arg): Likewise.
19968 (ix86_legitimate_constant_p): Likewise.
19969 (put_condition_code): Likewise.
19970 (split_double_mode): Likewise.
19971 (ix86_avx256_split_vector_move_misalign): Likewise.
19972 (ix86_expand_vector_logical_operator): Likewise.
19973 (ix86_split_idivmod): Likewise.
19974 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19975 (ix86_build_const_vector): Likewise.
19976 (ix86_build_signbit_mask): Likewise.
19977 (ix86_match_ccmode): Likewise.
19978 (ix86_cc_modes_compatible): Likewise.
19979 (ix86_expand_branch): Likewise.
19980 (ix86_expand_sse_cmp): Likewise.
19981 (ix86_expand_sse_movcc): Likewise.
19982 (ix86_expand_int_sse_cmp): Likewise.
19983 (ix86_expand_vec_perm_vpermi2): Likewise.
19984 (ix86_expand_vec_perm): Likewise.
19985 (ix86_expand_sse_unpack): Likewise.
19986 (ix86_expand_int_addcc): Likewise.
19987 (ix86_split_to_parts): Likewise.
19988 (ix86_vectorize_builtin_gather): Likewise.
19989 (ix86_vectorize_builtin_scatter): Likewise.
19990 (avx_vpermilp_parallel): Likewise.
19991 (inline_memory_move_cost): Likewise.
19992 (ix86_tieable_integer_mode_p): Likewise.
19993 (x86_maybe_negate_const_int): Likewise.
19994 (ix86_expand_vector_init_duplicate): Likewise.
19995 (ix86_expand_vector_init_one_nonzero): Likewise.
19996 (ix86_expand_vector_init_one_var): Likewise.
19997 (ix86_expand_vector_init_concat): Likewise.
19998 (ix86_expand_vector_init_interleave): Likewise.
19999 (ix86_expand_vector_init_general): Likewise.
20000 (ix86_expand_vector_set): Likewise.
20001 (ix86_expand_vector_extract): Likewise.
20002 (emit_reduc_half): Likewise.
20003 (ix86_emit_i387_round): Likewise.
20004 (ix86_mangle_type): Likewise.
20005 (ix86_expand_round_sse4): Likewise.
20006 (expand_vec_perm_blend): Likewise.
20007 (canonicalize_vector_int_perm): Likewise.
20008 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
20009 (expand_vec_perm_1): Likewise.
20010 (expand_vec_perm_interleave3): Likewise.
20011 (expand_vec_perm_even_odd_pack): Likewise.
20012 (expand_vec_perm_even_odd_1): Likewise.
20013 (expand_vec_perm_broadcast_1): Likewise.
20014 (ix86_vectorize_vec_perm_const_ok): Likewise.
20015 (ix86_expand_vecop_qihi): Likewise.
20016 (ix86_expand_mul_widen_hilo): Likewise.
20017 (ix86_expand_sse2_abs): Likewise.
20018 (ix86_expand_pextr): Likewise.
20019 (ix86_expand_pinsr): Likewise.
20020 (ix86_preferred_simd_mode): Likewise.
20021 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
20022 * config/i386/sse.md (*andnot<mode>3): Likewise.
20023 (<mask_codefor><code><mode>3<mask_name>): Likewise.
20024 (*<code><mode>3): Likewise.
20025 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
20026 (ia64_expand_atomic_op): Likewise.
20027 (ia64_arg_type): Likewise.
20028 (ia64_mode_to_int): Likewise.
20029 (ia64_scalar_mode_supported_p): Likewise.
20030 (ia64_vector_mode_supported_p): Likewise.
20031 (expand_vec_perm_broadcast): Likewise.
20032 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
20033 (iq2000_function_arg_advance): Likewise.
20034 (iq2000_function_arg): Likewise.
20035 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
20036 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
20037 (m68k_libcall_value): Likewise.
20038 (m68k_function_value): Likewise.
20039 (sched_attr_op_type): Likewise.
20040 * config/mcore/mcore.c (mcore_output_move): Likewise.
20041 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
20043 (microblaze_function_arg): Likewise.
20044 * config/mips/mips.c (mips16_build_call_stub): Likewise.
20045 (mips_print_operand): Likewise.
20046 (mips_mode_ok_for_mov_fmt_p): Likewise.
20047 (mips_vector_mode_supported_p): Likewise.
20048 (mips_preferred_simd_mode): Likewise.
20049 (mips_expand_vpc_loongson_even_odd): Likewise.
20050 (mips_expand_vec_unpack): Likewise.
20051 (mips_expand_vi_broadcast): Likewise.
20052 (mips_expand_vector_init): Likewise.
20053 (mips_expand_vec_reduc): Likewise.
20054 (mips_expand_msa_cmp): Likewise.
20055 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
20056 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
20057 (cc_flags_for_mode): Likewise.
20058 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20059 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20060 (nds32_output_casesi_pc_relative): Likewise.
20061 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20062 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20063 (nvptx_gen_unpack): Likewise.
20064 (nvptx_gen_pack): Likewise.
20065 (nvptx_gen_shuffle): Likewise.
20066 (nvptx_gen_wcast): Likewise.
20067 (nvptx_preferred_simd_mode): Likewise.
20068 * config/pa/pa.c (pa_secondary_reload): Likewise.
20069 * config/pa/predicates.md (base14_operand): Likewise.
20070 * config/powerpcspe/powerpcspe-c.c
20071 (altivec_resolve_overloaded_builtin): Likewise.
20072 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20074 (rs6000_preferred_simd_mode): Likewise.
20075 (output_vec_const_move): Likewise.
20076 (rs6000_expand_vector_extract): Likewise.
20077 (rs6000_split_vec_extract_var): Likewise.
20078 (reg_offset_addressing_ok_p): Likewise.
20079 (rs6000_legitimate_offset_address_p): Likewise.
20080 (rs6000_legitimize_address): Likewise.
20081 (rs6000_emit_set_const): Likewise.
20082 (rs6000_const_vec): Likewise.
20083 (rs6000_emit_move): Likewise.
20084 (spe_build_register_parallel): Likewise.
20085 (rs6000_darwin64_record_arg_recurse): Likewise.
20086 (swap_selector_for_mode): Likewise.
20087 (spe_init_builtins): Likewise.
20088 (paired_init_builtins): Likewise.
20089 (altivec_init_builtins): Likewise.
20090 (do_load_for_compare): Likewise.
20091 (rs6000_generate_compare): Likewise.
20092 (rs6000_expand_float128_convert): Likewise.
20093 (emit_load_locked): Likewise.
20094 (emit_store_conditional): Likewise.
20095 (rs6000_output_function_epilogue): Likewise.
20096 (rs6000_handle_altivec_attribute): Likewise.
20097 (rs6000_function_value): Likewise.
20098 (emit_fusion_gpr_load): Likewise.
20099 (emit_fusion_p9_load): Likewise.
20100 (emit_fusion_p9_store): Likewise.
20101 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20102 (fusion_gpr_mem_load): Likewise.
20103 (fusion_addis_mem_combo_load): Likewise.
20104 (fusion_addis_mem_combo_store): Likewise.
20105 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20106 (fusion_gpr_mem_load): Likewise.
20107 (fusion_addis_mem_combo_load): Likewise.
20108 (fusion_addis_mem_combo_store): Likewise.
20109 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20111 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20112 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20113 (rs6000_preferred_simd_mode): Likewise.
20114 (output_vec_const_move): Likewise.
20115 (rs6000_expand_vector_extract): Likewise.
20116 (rs6000_split_vec_extract_var): Likewise.
20117 (reg_offset_addressing_ok_p): Likewise.
20118 (rs6000_legitimate_offset_address_p): Likewise.
20119 (rs6000_legitimize_address): Likewise.
20120 (rs6000_emit_set_const): Likewise.
20121 (rs6000_const_vec): Likewise.
20122 (rs6000_emit_move): Likewise.
20123 (rs6000_darwin64_record_arg_recurse): Likewise.
20124 (swap_selector_for_mode): Likewise.
20125 (paired_init_builtins): Likewise.
20126 (altivec_init_builtins): Likewise.
20127 (rs6000_expand_float128_convert): Likewise.
20128 (emit_load_locked): Likewise.
20129 (emit_store_conditional): Likewise.
20130 (rs6000_output_function_epilogue): Likewise.
20131 (rs6000_handle_altivec_attribute): Likewise.
20132 (rs6000_function_value): Likewise.
20133 (emit_fusion_gpr_load): Likewise.
20134 (emit_fusion_p9_load): Likewise.
20135 (emit_fusion_p9_store): Likewise.
20136 * config/rx/rx.c (rx_gen_move_template): Likewise.
20137 (flags_from_mode): Likewise.
20138 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20139 (s390_slb_comparison): Likewise.
20140 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20141 (s390_vector_mode_supported_p): Likewise.
20142 (s390_cc_modes_compatible): Likewise.
20143 (s390_match_ccmode_set): Likewise.
20144 (s390_canonicalize_comparison): Likewise.
20145 (s390_emit_compare_and_swap): Likewise.
20146 (s390_branch_condition_mask): Likewise.
20147 (s390_rtx_costs): Likewise.
20148 (s390_secondary_reload): Likewise.
20149 (__SECONDARY_RELOAD_CASE): Likewise.
20150 (s390_expand_cs): Likewise.
20151 (s390_preferred_simd_mode): Likewise.
20152 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20153 * config/sh/sh.c (sh_print_operand): Likewise.
20154 (dump_table): Likewise.
20155 (sh_secondary_reload): Likewise.
20156 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20157 * config/sh/sh.md (casesi_worker_1): Likewise.
20158 (casesi_worker_2): Likewise.
20159 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20160 (fcc_comparison_operator): Likewise.
20161 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20162 (emit_soft_tfmode_cvt): Likewise.
20163 (sparc_preferred_simd_mode): Likewise.
20164 (output_cbranch): Likewise.
20165 (sparc_print_operand): Likewise.
20166 (sparc_expand_vec_perm_bmask): Likewise.
20167 (vector_init_bshuffle): Likewise.
20168 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20169 (spu_vector_mode_supported_p): Likewise.
20170 (spu_expand_insv): Likewise.
20171 (spu_emit_branch_or_set): Likewise.
20172 (spu_handle_vector_attribute): Likewise.
20173 (spu_builtin_splats): Likewise.
20174 (spu_builtin_extract): Likewise.
20175 (spu_builtin_promote): Likewise.
20176 (spu_expand_sign_extend): Likewise.
20177 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20178 (tilegx_simd_int): Likewise.
20179 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20180 (tilepro_simd_int): Likewise.
20181 * config/v850/v850.c (const_double_split): Likewise.
20182 (v850_print_operand): Likewise.
20183 (ep_memory_offset): Likewise.
20184 * config/vax/vax.c (vax_rtx_costs): Likewise.
20185 (vax_output_int_move): Likewise.
20186 (vax_output_int_add): Likewise.
20187 (vax_output_int_subtract): Likewise.
20188 * config/visium/predicates.md (visium_branch_operator): Likewise.
20189 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20190 (visium_print_operand_address): Likewise.
20191 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20192 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20193 (xtensa_expand_conditional_branch): Likewise.
20194 (xtensa_copy_incoming_a7): Likewise.
20195 (xtensa_output_literal): Likewise.
20196 * dfp.c (decimal_real_maxval): Likewise.
20197 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20200 Alan Hayward <alan.hayward@arm.com>
20201 David Sherwood <david.sherwood@arm.com>
20203 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20204 (mode_nunits_inline): Likewise.
20205 (mode_inner_inline): Likewise.
20206 (mode_unit_size_inline): Likewise.
20207 (mode_unit_precision_inline): Likewise.
20208 (emit_insn_modes_h): Likewise. Also emit a #define of the
20210 (emit_mode_wider): Add an E_ prefix to mode names.
20211 (emit_mode_complex): Likewise.
20212 (emit_mode_inner): Likewise.
20213 (emit_mode_adjustments): Likewise.
20214 (emit_mode_int_n): Likewise.
20215 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20216 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20217 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20218 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20219 (CRC32_BUILTIN, ENTRY): Likewise.
20220 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20221 (aarch64_pop_regs): Likewise.
20222 (aarch64_process_components): Likewise.
20223 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20224 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20225 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20226 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20227 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20228 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20230 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20231 (pcmpistr): Likewise.
20232 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20233 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20234 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20236 * config/rl78/rl78.c (mduc_regs): Likewise.
20237 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20238 (htm_expand_builtin): Likewise.
20239 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20240 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20241 * config/xtensa/xtensa.c (print_operand): Likewise.
20242 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20243 (NUM_MODE_VECTOR_INT): Likewise.
20244 * genoutput.c (null_operand): Likewise.
20245 (output_operand_data): Likewise.
20246 * genrecog.c (print_parameter_value): Likewise.
20247 * lra.c (debug_operand_data): Likewise.
20249 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20251 * dwarf2out.c (scompare_loc_descriptor_wide)
20252 (scompare_loc_descriptor_narrow): New functions, split out from...
20253 (scompare_loc_descriptor): ...here.
20254 * expmed.c (emit_store_flag_int): New function, split out from...
20255 (emit_store_flag): ...here.
20257 2017-08-30 Richard Biener <rguenther@suse.de>
20259 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20260 (dwarf2out_early_finish): Move setting of AT_pubnames from
20261 early debug output to early finish.
20263 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20265 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20266 and -mdata-region to the assembler.
20268 2017-08-30 Richard Biener <rguenther@suse.de>
20270 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20272 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20273 (dwarf2out_early_global_decl): Only generate a DIE for the
20274 abstract origin if it doesn't already exist or is a declaration DIE.
20275 (resolve_addr): Do not add the linkage name twice when
20276 generating a stub DIE for the DW_TAG_GNU_call_site target.
20278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20280 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20281 Use machine_mode rather than int for arg1_mode.
20283 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20286 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20287 that the second argument of the built-in functions to unpack
20288 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20289 switch statement instead a lot of if statements.
20290 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20291 Allow 64-bit values to be in Altivec registers as well as
20292 traditional floating point registers.
20293 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20295 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20297 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20298 MAX_REGS_PER_ADDRESS == 1.
20300 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20302 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20303 * config/i386/i386.c (ix86_option_override_internal): Simplify
20304 setting of opts->x_flag_entry.
20306 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20307 Jakub Jelinek <jakub@redhat.com>
20308 Richard Biener <rguenther@suse.de>
20310 PR tree-optimization/81503
20311 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20312 folded constant fits in the target type; reorder tests for clarity.
20314 2017-08-29 Martin Liska <mliska@suse.cz>
20316 * passes.def: Include pass_lower_switch.
20317 * stmt.c (dump_case_nodes): Remove and move to
20318 tree-switch-conversion.
20319 (case_values_threshold): Likewise.
20320 (expand_switch_as_decision_tree_p): Likewise.
20321 (emit_case_decision_tree): Likewise.
20322 (expand_case): Likewise.
20323 (balance_case_nodes): Likewise.
20324 (node_has_low_bound): Likewise.
20325 (node_has_high_bound): Likewise.
20326 (node_is_bounded): Likewise.
20327 (emit_case_nodes): Likewise.
20328 (struct simple_case_node): New struct.
20329 (add_case_node): Remove.
20330 (emit_case_dispatch_table): Use vector instead of case_list.
20331 (reset_out_edges_aux): Remove.
20332 (compute_cases_per_edge): Likewise.
20333 (expand_case): Build list of simple_case_node.
20334 (expand_sjlj_dispatch_table): Use it.
20335 * tree-switch-conversion.c (struct case_node): Moved from
20336 stmt.c and adjusted.
20337 (emit_case_nodes): Likewise.
20338 (node_has_low_bound): Likewise.
20339 (node_has_high_bound): Likewise.
20340 (node_is_bounded): Likewise.
20341 (case_values_threshold): Likewise.
20342 (reset_out_edges_aux): Likewise.
20343 (compute_cases_per_edge): Likewise.
20344 (add_case_node): Likewise.
20345 (dump_case_nodes): Likewise.
20346 (balance_case_nodes): Likewise.
20347 (expand_switch_as_decision_tree_p): Likewise.
20348 (emit_jump): Likewise.
20349 (emit_case_decision_tree): Likewise.
20350 (try_switch_expansion): Likewise.
20351 (do_jump_if_equal): Likewise.
20352 (emit_cmp_and_jump_insns): Likewise.
20353 (fix_phi_operands_for_edge): New function.
20354 (record_phi_operand_mapping): Likewise.
20355 (class pass_lower_switch): New pass.
20356 (pass_lower_switch::execute): New function.
20357 (make_pass_lower_switch): Likewise.
20358 (conditional_probability):
20359 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20360 * tree-pass.h: Add make_pass_lower_switch.
20362 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20365 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20368 2017-08-29 Richard Biener <rguenther@suse.de>
20370 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20371 we do not add a DW_AT_inline attribute twice.
20372 (gen_subprogram_die): Remove code setting DW_AT_inline on
20373 DECL_ABSTRACT_P nodes.
20375 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20377 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20378 calls to internal functions.
20379 (gimplify_modify_expr): Likewise.
20380 * tree-call-cdce.c (use_internal_fn): Likewise.
20381 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20382 (convert_to_divmod): Set the nothrow flag.
20383 * tree-if-conv.c (predicate_mem_writes): Likewise.
20384 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20385 (vectorizable_call): Likewise.
20386 (vectorizable_store): Likewise.
20387 (vectorizable_load): Likewise.
20388 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20389 (vect_recog_mask_conversion_pattern): Likewise.
20391 2017-08-29 Martin Liska <mliska@suse.cz>
20394 * gcc.c (driver_handle_option): Add new argument.
20395 * opts-common.c (handle_option): Pass
20396 target_option_override_hook.
20397 * opts-global.c (lang_handle_option): Add new option.
20398 (set_default_handlers): Add new argument.
20399 (decode_options): Likewise.
20400 * opts.c (target_handle_option): Likewise.
20401 (common_handle_option): Call target_option_override_hook.
20402 * opts.h (struct cl_option_handler_func): Add hook for
20403 target option override.
20404 (struct cl_option_handlers): Likewise.
20405 (set_default_handlers): Add new argument.
20406 (decode_options): Likewise.
20407 (common_handle_option): Likewise.
20408 (target_handle_option): Likewise.
20409 * toplev.c (toplev::main): Pass targetm.target_option.override
20412 2017-08-29 Richard Biener <rguenther@suse.de>
20413 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20415 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20416 life to the active subtree.
20418 2017-08-28 Jeff Law <law@redhat.com>
20420 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20421 derive_equivalences.
20422 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20423 Code moved into....
20424 (edge_info::derive_equivalences): New private member function
20426 * tree-ssa-dom.c (class edge_info): Changed from a struct
20427 to a class. Add ctor/dtor, methods and data members.
20428 (edge_info::edge_info): Renamed from allocate_edge_info.
20429 Initialize additional members.
20430 (edge_info::~edge_info): New.
20431 (free_dom_edge_info): Delete the edge info.
20432 (record_edge_info): Use new class & associated member functions.
20433 Tighten forms for testing for edge equivalences.
20434 (record_temporary_equivalences): Iterate over the simple
20435 equivalences rather than assuming there's only one per edge.
20436 (cprop_into_successor_phis): Iterate over the simple
20437 equivalences rather than assuming there's only one per edge.
20438 (optimize_stmt): Use operand_equal_p rather than pointer
20439 equality for mini-DSE code.
20441 2017-08-28 Nathan Sidwell <nathan@acm.org>
20443 * gcc.c (execute): Fold SIGPIPE handling into switch
20444 statement. Adjust internal error message.
20446 2017-08-28 Richard Biener <rguenther@suse.de>
20449 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20450 Do nothing for removed DIEs.
20452 2017-08-28 Richard Biener <rguenther@suse.de>
20454 PR tree-optimization/81977
20455 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20458 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20461 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20462 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20463 using targetm.gen_mem_thread_fence.
20465 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20468 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20469 predicate to register_operand. Reorder operands.
20470 (*btr<mode>): Ditto.
20471 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20472 (*btr<mode>_mask): Ditto.
20474 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20476 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20477 * config/rs6000/xmmintrin.h: New file.
20478 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20480 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20483 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20484 parameter and_insn and return it.
20485 (recombine_lvx_pattern): Insert a copy to ensure availability of
20486 the base register of the copied masking operation at the point of
20487 the instruction replacement.
20488 (recombine_stvx_pattern): Likewise.
20490 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20492 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20493 undocumented switches.
20494 (-mpower9-dform-vector): Likewise.
20495 (-mpower9-dform): Likewise.
20496 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20497 comments to delete references to -mpower9-dform* switches.
20498 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20499 Delete reference to -mpower9-dform* switches, test for
20500 -mpower9-vector instead.
20501 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20502 (OTHER_P9_VECTOR_MASKS): Likewise.
20503 (POWERPC_MASKS): Likewise.
20504 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20505 tests against -mpower9-dform* to -mpower9-vector. Delete code
20506 that checked for -mpower9-dform* consistancy with other options.
20507 Add test for -mpower9-misc to enable other power9 switches.
20508 (rs6000_init_hard_regno_mode_ok): Likewise.
20509 (rs6000_option_override_internal): Likewise.
20510 (rs6000_emit_prologue): Likewise.
20511 (rs6000_emit_epilogue): Likewise.
20512 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20513 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20514 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20516 (emit_fusion_p9_store): Likewise.
20517 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20518 resetting these macros if the assembler does not support ISA 3.0
20520 (TARGET_P9_DFORM_VECTOR): Likewise.
20521 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20522 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20524 2017-08-25 Alan Modra <amodra@gmail.com>
20526 PR rtl-optimization/81747
20527 * cse.c (cse_extended_basic_block): Don't attempt to record
20528 equivalences for degenerate conditional branches that branch
20529 to their fall-through.
20531 2017-08-24 Martin Sebor <msebor@redhat.com>
20533 PR middle-end/81908
20534 * gimple-fold.c (size_must_be_zero_p): New function.
20535 (gimple_fold_builtin_memory_op): Call it.
20537 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20539 * config/rs6000/mm_malloc.h: New file.
20541 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20543 PR tree-optimization/81913
20544 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20545 analysis when either IVs in condition can wrap.
20547 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20549 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20550 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20552 2017-08-24 Richard Biener <rguenther@suse.de>
20555 * targhooks.c (default_target_can_inline_p): Properly
20556 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20557 is present and always compare.
20558 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20559 infer -mfpmath=sse from TARGET_SSE_P.
20560 (ix86_can_inline_p): Properly use target_option_default_node when
20561 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20563 2017-08-24 Richard Biener <rguenther@suse.de>
20566 * dwarf2out.c (output_die): Handle flag_generate_offload like
20568 (output_comp_unit): Likewise.
20569 (gen_array_type_die): Likewise.
20570 (dwarf2out_early_finish): Likewise.
20571 (note_variable_value_in_expr): Likewise.
20572 (dwarf2out_finish): Likewise. Adjust assert.
20573 * cgraphunit.c (symbol_table::compile): Move setting of
20574 flag_generate_offload earlier ...
20575 (symbol_table::finalize_compilation_unit): ... here, before
20576 early debug finalization.
20578 2017-08-24 Richard Biener <rguenther@suse.de>
20580 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20581 and ipa-fnsummary.h.
20582 (ix86_can_inline_p): When ix86_fpmath flags do not match
20583 check whether the callee uses FP math at all.
20585 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20587 PR middle-end/81931
20588 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20589 instead of TYPE_PRECISION.
20591 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20592 Alan Hayward <alan.hayward@arm.com>
20593 David Sherwood <david.sherwood@arm.com>
20595 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20597 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20599 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20601 (ix86_frame::stack_realign_allocate): New field.
20602 (struct machine_frame_state): Modify comments.
20603 (machine_frame_state::sp_realigned_fp_end): New field.
20604 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20605 layout calculation.
20606 (sp_valid_at): Add assertion to assure no attempt to access invalid
20607 offset of a realigned stack.
20608 (fp_valid_at): Likewise.
20609 (choose_baseaddr): Modify comments.
20610 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20611 ix86_expand_prologue.
20612 (ix86_expand_prologue): Modify stack realignment and allocation.
20613 (ix86_expand_epilogue): Modify comments.
20614 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20615 avx2_runtime, avx512f, and avx512f_runtime.
20617 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20619 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20620 (mstackrealign): Do not init to -1.
20621 * config/i386/i386.c (ix86_option_override_internal):
20622 Check opts_set, not opts when setting default value of
20623 opts->x_ix86_force_align_arg_pointer.
20625 2017-08-23 Richard Biener <rguenther@suse.de>
20627 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20628 lang_hooks.decl_printable_name.
20629 * print-rtl-function.c (print_rtx_function): Likewise.
20630 * tree-pretty-print.c (dump_function_header): Likewise.
20632 2017-08-23 Richard Biener <rguenther@suse.de>
20635 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20636 -g0 at compile-time.
20638 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20640 PR middle-end/19706
20641 * doc/sourcebuild.texi (Other hardware attributes):
20644 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20646 PR middle-end/19706
20647 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20648 Fix single-use check.
20650 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20652 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20654 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20656 * doc/install.texi: Modify to add more details on running selected
20659 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20661 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20662 is combined with -mabi=ms.
20663 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20666 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20668 PR tree-optimization/81488
20669 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20670 and cached_basis fields.
20671 (MAX_SPREAD): New constant.
20672 (alloc_cand_and_find_basis): Initialize new fields.
20673 (clear_visited): New function.
20674 (create_phi_basis_1): Rename from create_phi_basis, set visited
20675 and cached_basis fields.
20676 (create_phi_basis): New wrapper function.
20677 (phi_add_costs_1): Rename from phi_add_costs, add spread
20678 parameter, set visited field, short-circuit when limits reached.
20679 (phi_add_costs): New wrapper function.
20680 (record_phi_increments_1): Rename from record_phi_increments, set
20682 (record_phi_increments): New wrapper function.
20683 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20684 (phi_incr_cost): New wrapper function.
20685 (all_phi_incrs_profitable_1): Rename from
20686 all_phi_incrs_profitable, set visited field.
20687 (all_phi_incrs_profitable): New wrapper function.
20689 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20690 Alan Hayward <alan.hayward@arm.com>
20691 David Sherwood <david.sherwood@arm.com>
20693 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20694 that takes the outer and inner modes.
20695 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20696 comparison as the canonical test for a paradoxical subreg.
20697 * combine.c (simplify_set): Use paradoxical_subreg_p.
20698 (make_extraction): Likewise.
20699 (force_to_mode): Likewise.
20700 (rtx_equal_for_field_assignment_p): Likewise.
20701 (gen_lowpart_for_combine): Likewise.
20702 (simplify_comparison): Likewise.
20703 * cse.c (equiv_constant): Likewise.
20704 * expmed.c (store_bit_field_1): Likewise.
20705 * final.c (alter_subreg): Likewise.
20706 * fwprop.c (propagate_rtx): Likewise.
20707 (forward_propagate_subreg): Likewise.
20708 * ira-conflicts.c (ira_build_conflicts): Likewise.
20709 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20710 * lra-constraints.c (curr_insn_transform): Likewise.
20711 (split_reg): Likewise.
20712 * lra-eliminations.c (move_plus_up): Likewise.
20713 (lra_eliminate_regs_1): Likewise.
20714 * recog.c (general_operand): Likewise.
20715 * ree.c (combine_reaching_defs): Likewise.
20716 * reload.c (push_reload): Likewise.
20717 (find_reloads): Likewise.
20718 * reload1.c (elimination_effects): Likewise.
20719 (compute_reload_subreg_offset): Likewise.
20720 (choose_reload_regs): Likewise.
20721 * rtlanal.c (subreg_lsb_1): Likewise.
20722 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20723 (simplify_subreg): Likewise.
20724 * var-tracking.c (track_loc_p): Likewise.
20725 * emit-rtl.c (byte_lowpart_offset): Likewise.
20726 (paradoxical_subreg_p): Delete out-of-line definition.
20728 2017-08-22 Jeff Law <law@redhat.com>
20730 PR tree-optimization/81741
20731 PR tree-optimization/71947
20732 * tree-ssa-dom.c: Include tree-inline.h.
20733 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20734 equivalences if one is more expensive to compute than the other.
20735 * tree-ssa-scopedtables.h (class const_or_copies): Make
20736 record_const_or_copy_raw method private.
20737 (class avail_exprs_stack): New method simplify_binary_operation.
20738 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20739 avail_exprs_stack::simplify_binary_operation as needed.
20740 (avail_exprs_stack::simplify_binary_operation): New function.
20742 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20744 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20745 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20746 (DOT_SYMBOLS): Likewise.
20747 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20748 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20749 (RS6000_ABI_NAME): Likewise.
20750 (TARGET_CMODEL): Likewise.
20751 (TOC_SECTION_ASM_OP): Likewise.
20752 (SET_CMODEL): New macro.
20753 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20755 2017-08-22 Richard Biener <rguenther@suse.de>
20757 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20758 to allow for free-lang-data replacements similar to verify_type_variant.
20760 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20761 Alan Hayward <alan.hayward@arm.com>
20762 David Sherwood <david.sherwood@arm.com>
20764 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20765 VOIDmode for the LABEL_REF.
20767 2017-08-22 Richard Biener <rguenther@suse.de>
20769 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20771 2017-08-22 Richard Biener <rguenther@suse.de>
20773 * common.opt (feliminate-dwarf2-dups): Ignore.
20774 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20775 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20776 same_die_p_wrap, compute_section_prefix,
20777 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20778 (comdat_symbol_id, comdat_symbol_number): Likewise.
20779 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20781 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20782 (output_die): Mark unreachable path unreachable.
20783 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20784 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20785 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20786 (dwarf2out_early_finish): Likewise.
20788 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20790 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20792 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20795 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20796 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20797 (avr_attribute_table) <io, io_low, address>: Initialize
20798 .decl_required with true.
20800 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20802 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20803 undocumented debugging options.
20804 (-mvsx-scalar-double): Likewise.
20805 (-mallow-df-permute): Likewise.
20806 (-mvectorize-builtins): Likewise.
20807 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20808 (rs6000_builtin_vectorized_function): Likewise.
20809 (rs6000_builtin_md_vectorized_function): Likewise.
20810 (rs6000_opt_vars): Likewise.
20812 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20815 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20816 (*btrq_imm): Rename from *btrq.
20817 (*btcq_imm): Rename from *btcq.
20818 (btsc): New code attribute.
20819 (*<btsc><mode>): New insn pattern.
20820 (*btr<mode>): Ditto.
20821 (*<btsc><mode>_mask): New insn_and_split pattern.
20822 (*btr<mode>_mask): Ditto.
20824 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20825 Alan Hayward <alan.hayward@arm.com>
20826 David Sherwood <david.sherwood@arm.com>
20828 * function.c (pad_below): Simplify padding calculation.
20830 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20831 Alan Hayward <alan.hayward@arm.com>
20832 David Sherwood <david.sherwood@arm.com>
20834 * target.def (function_prologue): Remove frame size argument.
20835 (function_epilogue): Likewise.
20836 * doc/tm.texi: Regenerate.
20837 * final.c (final_start_function): Update call to function_prologue.
20838 (final_end_function): Update call to function_epilogue.
20839 (default_function_pro_epilogue): Remove frame size argument.
20840 * output.h (default_function_pro_epilogue): Likewise.
20841 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20842 (arm_output_function_prologue): Likewise.
20843 * config/frv/frv.c (frv_function_prologue): Likewise.
20844 (frv_function_epilogue): Likewise.
20845 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20846 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20847 (ia64_output_function_epilogue): Likewise.
20848 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20849 (m32r_output_function_epilogue): Likewise.
20850 * config/microblaze/microblaze.c (microblaze_function_prologue)
20851 (microblaze_function_epilogue): Likewise.
20852 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20853 (mips_output_function_epilogue): Likewise.
20854 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20855 (mmix_target_asm_function_epilogue): Likewise.
20856 * config/msp430/msp430.c (msp430_start_function): Likewise.
20857 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20858 (nds32_asm_function_epilogue): Likewise.
20859 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20860 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20861 (pa_output_function_epilogue): Likewise.
20862 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20863 (rs6000_output_function_epilogue): Likewise.
20864 * config/rl78/rl78.c (rl78_start_function): Likewise.
20865 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20866 (rs6000_output_function_epilogue): Likewise.
20867 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20868 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20869 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20870 (sparc_asm_function_epilogue): Likewise.
20872 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20874 * tree.h (type_has_mode_precision_p): New function.
20875 * convert.c (convert_to_integer_1): Use it.
20876 * expr.c (expand_expr_real_2): Likewise.
20877 (expand_expr_real_1): Likewise.
20878 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20879 * match.pd: Likewise.
20880 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20881 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20882 * tree-tailcall.c (process_assignment): Likewise.
20883 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20884 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20885 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20886 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20887 (vectorizable_assignment): Likewise.
20888 (vectorizable_shift): Likewise.
20889 (vectorizable_operation): Likewise.
20890 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20892 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20894 * match.pd: Add pow (C, x) simplification.
20896 2017-08-21 Richard Biener <rguenther@suse.de>
20898 PR tree-optimization/81900
20899 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20900 for blocks with abnormal predecessors.
20901 (compute_antic): Do not set visited flag prematurely.
20903 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20906 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20908 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20910 * stor-layout.h (vector_type_mode): Move to...
20911 * tree.h (vector_type_mode): ...here.
20912 * stor-layout.c (vector_type_mode): Move to...
20913 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20915 2017-08-21 Richard Biener <rguenther@suse.de>
20917 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20918 register_external_die hooks.
20919 (debug_false_tree_charstarstar_uhwistar): Declare.
20920 (debug_nothing_tree_charstar_uhwi): Likewise.
20921 * debug.c (do_nothing_debug_hooks): Adjust.
20922 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20923 (debug_nothing_tree_charstar_uhwi): Likewise.
20924 * dbxout.c (dbx_debug_hooks): Adjust.
20925 (xcoff_debug_hooks): Likewise.
20926 * sdbout.c (sdb_debug_hooks): Likewise.
20927 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20928 * dwarf2out.c (macinfo_label_base): New global.
20929 (dwarf2out_register_external_die): New function for the
20930 register_external_die hook.
20931 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20932 (dwarf2_debug_hooks): Use them.
20933 (dwarf2_lineno_debug_hooks): Adjust.
20934 (struct die_struct): Add with_offset flag.
20935 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20936 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20937 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20938 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20939 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20940 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20941 defining section names for the early LTO debug variants.
20942 (reset_indirect_string): New helper.
20943 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20944 (print_dw_val): Add support for offsetted symbol references.
20945 (get_ultimate_context): Split out from is_cxx.
20946 (is_cxx): Use get_ultimate_context.
20947 (is_fortran): Add decl overload.
20948 (compute_comp_unit_symbol): Split out worker from
20949 compute_section_prefix.
20950 (compute_section_prefix): Call compute_comp_unit_symbol and
20951 set comdat_type_p here.
20952 (output_die): Skip DIE symbol output for the LTO added one.
20953 Handle DIE symbol references with offset.
20954 (output_comp_unit): Guard section name mangling properly.
20955 For LTO debug sections emit a symbol at the section beginning
20956 which we use to refer to its DIEs.
20957 (add_abstract_origin_attribute): For DIEs registered via
20958 dwarf2out_register_external_die directly refer to the early
20959 DIE rather than indirectly through the shadow one we created.
20960 Remove obsolete call to dwarf2out_abstract_function for
20961 non-function/block origins.
20962 (gen_array_type_die): When generating early LTO debug do
20963 not emit DW_AT_string_length.
20964 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20965 late when in LTO. As suggested place a gcc_unreachable for
20966 the DECL_ABSTRACT_P case.
20967 (gen_subprogram_die): Avoid another specification DIE
20968 for early built declarations/definitions for the late LTO case.
20969 (gen_variable_die): Add type references for late duplicated VLA dies
20971 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20972 we have the abstract instance already.
20973 (process_scope_var): Adjust decl DIE contexts in LTO which
20974 first puts them in limbo.
20975 (gen_decl_die): Do not generate type DIEs late apart from
20976 types for VLAs or for decls we do not yet have a DIE. Do not
20977 call dwarf2out_abstract_function late.
20978 (dwarf2out_early_global_decl): Make sure to create DIEs
20979 for abstract instances of a decl first.
20980 (dwarf2out_late_global_decl): Adjust comment.
20981 (output_macinfo_op): With multiple macro sections use
20982 macinfo_label_base to distinguish labels.
20983 (output_macinfo): Likewise. Update macinfo_label_base.
20984 Pass in the line info label.
20985 (note_variable_value_in_expr): When generating LTO resolve
20986 all variable values here by generating DIEs as needed.
20987 (init_sections_and_labels): Add early LTO debug flag parameter
20988 and generate different sections and names if set. Add generation
20989 counter for the labels so we can have multiple of them.
20990 (reset_dies): Helper to allow DIEs to be output multiple times.
20991 (dwarf2out_finish): When outputting DIEs to the fat part of an
20992 LTO object first reset DIEs.
20993 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20994 (modified_type_die): Check for decl_ultimate_origin being self
20996 (gen_type_die_with_usage): Likewise.
20997 (gen_typedef_die): Allow decl_ultimate_origin being self.
20998 (set_decl_abstract_flags): Remove.
20999 (set_block_abstract_flags): Likewise.
21000 (dwarf2out_abstract_function): Treat the early generated DIEs
21001 as the abstract copy and only add DW_AT_inline and
21002 DW_AT_artificial here and call set_decl_origin_self.
21003 If the DIE has an abstract origin don't do anything.
21004 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
21005 if we have none yet (Go fails to build one, PR78628).
21006 (variably_modified_type_p): Prevent endless recursion for Ada
21007 cyclic pointer types.
21008 * lto-streamer-in.c: Include debug.h.
21009 (dref_queue): New global.
21010 (lto_read_tree_1): Stream in DIE references.
21011 (lto_input_tree): Register DIE references.
21012 (input_function): Stream DECL_DEBUG_ARGS.
21013 * lto-streamer-out.c: Include debug.h.
21014 (lto_write_tree_1): Output DIE references.
21015 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
21016 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
21017 (output_function): Stream DECL_DEBUG_ARGS.
21018 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
21019 Stream DECL_ABSTRACT_ORIGIN.
21020 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
21021 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
21022 DECL_CONTEXT for file-scope decls.
21023 * lto-streamer.h (struct dref_entry): Declare.
21024 (dref_queue): Likewise.
21025 * cfgexpand.c (pass_expand::execute): Do not call the
21026 outlining_inline_function hook here.
21027 * lto-wrapper.c (debug_obj): New global.
21028 (tool_cleanup): Unlink it if required.
21029 (debug_objcopy): New function.
21030 (run_gcc): Handle early debug sections in the IL files by
21031 extracting them to separate files, partially linkin them and
21032 feeding the result back as result to the linker.
21033 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
21034 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
21035 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
21036 sections into a separate segment.
21037 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
21039 (darwin_asm_dwarf_section): Likewise.
21040 (darwin_asm_output_dwarf_offset): Likewise.
21041 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
21043 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21044 Alan Hayward <alan.hayward@arm.com>
21045 David Sherwood <david.sherwood@arm.com>
21047 * read-md.h (md_reader::record_potential_iterator_use): Replace
21048 pointer argument with an rtx and an index.
21049 * read-rtl.c (iterator_group::apply_iterator): Likewise.
21050 (apply_mode_iterator): Likewise.
21051 (apply_code_iterator): Likewise.
21052 (apply_int_iterator): Likewise.
21053 (apply_subst_iterator): Likewise.
21054 (record_iterator_use): Likewise.
21055 (record_attribute_use): Likewise.
21056 (md_reader::record_potential_iterator_use): Likewise. Update calls
21057 to record_iterator_use and apply_iterator.
21058 (iterator_use): Replace ptr with x and index.
21059 (attribute_use): Likewise.
21060 (apply_attribute_uses): Update calls to apply_iterator.
21061 (apply_iterators): Likewise. Update initialization of iterator_use.
21062 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21063 and record_potential_iterator_use.
21064 (rtx_reader::read_rtx_operand): Likewise.
21066 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21067 Alan Hayward <alan.hayward@arm.com>
21068 David Sherwood <david.sherwood@arm.com>
21070 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21073 2017-08-21 Richard Biener <rguenther@suse.de>
21075 PR middle-end/81884
21076 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21077 at struct end conservatively when comparing common bases.
21079 2017-08-21 Richard Biener <rguenther@suse.de>
21081 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21082 (mem_ref_in_stmt): Remove.
21083 (determine_max_movement): Use ref index to get at the reference.
21084 (invariantness_dom_walker::before_dom_children): Deal with
21085 lim data already initialized.
21086 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21088 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21090 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21091 (TARGET_ISA_ROUND): Ditto.
21092 (TARGET_ROUND): Ditto.
21093 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21094 * config/i386/i386.md: Ditto.
21095 * config/i386/sse.md: Ditto.
21096 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21097 with OPTION_MASK_ISA_SSE4_1.
21099 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21102 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21103 __builtin_ia32_lzcnt_u16.
21105 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21108 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21109 (rs6000_set_current_function): Rewrite function to use it.
21111 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21114 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21115 and TYPE_WARN_IF_NOT_ALIGN.
21116 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21117 (handle_warn_if_not_align): New.
21118 (place_union_field): Call handle_warn_if_not_align.
21119 (place_field): Call handle_warn_if_not_align.
21120 Copy TYPE_WARN_IF_NOT_ALIGN.
21121 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21122 (layout_type): Likewise.
21123 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21125 (tree_decl_common): Add warn_if_not_align.
21126 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21127 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21128 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21129 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21130 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21131 * doc/extend.texi: Document warn_if_not_aligned attribute.
21132 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21134 2017-08-17 Martin Liska <mliska@suse.cz>
21137 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21138 (get_data_dependence): Use it as pointer type.
21139 (distribute_loop): Likewise.
21141 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21143 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21144 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21145 (p8_vmrgow_<mode>_direct): New define_insn.
21146 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21147 handle endianness for vmrgew and vmrgow permute patterns.
21149 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21151 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21152 * config/rs6000/rs6000-cpus.def: Remove comment.
21153 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21154 (POWERPC_MASKS): Likewise.
21155 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21156 use of TARGET_VSX_TIMODE.
21157 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21158 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21159 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21160 (rs6000_option_override_internal): Remove dead code.
21161 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21162 (rs6000_legitimize_reload_address): Likewise.
21163 (rs6000_legitimate_address_p): Likewise.
21164 (rs6000_opt_masks): Delete "vsx-timode".
21165 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21166 from function comment.
21167 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21168 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21169 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21171 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21172 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21173 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21174 (VSX_TI): Likewise.
21176 (define_peephole2): Likewise.
21178 2017-08-17 Martin Sebor <msebor@redhat.com>
21181 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21182 past the end of an array.
21183 (test_pp_format): Add test cases.
21185 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21187 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21188 missing ECF_NOTHROW flags.
21190 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21193 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21194 operands residing in integer registers.
21195 (*vsx_le_perm_load_<mode>): Likewise.
21196 (*vsx_le_perm_store_<mode>): Likewise.
21197 (define_peephole2): Add peepholes to optimize the above.
21199 2017-08-17 Marek Polacek <polacek@redhat.com>
21201 PR middle-end/81814
21202 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21203 to mimic what shorten_compare did. Change the return type to bool.
21204 (fold_cond_expr_with_comparison): Update call to
21205 operand_equal_for_comparison_p.
21206 (fold_ternary_loc): Likewise.
21208 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21210 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21212 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21214 2017-08-17 Richard Biener <rguenther@suse.de>
21216 * tree-ssa-structalias.c (solve_graph): When propagating
21217 to successors update the graphs succ edges and avoid duplicate work.
21219 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21222 * config/i386/i386.c (ix86_option_override_internal): Save target
21223 specific options after ix86_stack_protector_guard_reg was changed.
21225 2017-08-17 Richard Biener <rguenther@suse.de>
21227 PR tree-optimization/81827
21228 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21229 (new_var_info): Initialize it conservatively.
21230 (get_call_vi): Mark register vars.
21231 (new_scalar_tmp_constraint_exp): Likewise.
21232 (handle_rhs_call): Likewise.
21233 (handle_const_call): Likewise.
21234 (create_function_info_for): Likewise.
21235 (solve_constraints): Sort varinfos to separate register from
21236 non-register vars to pack points-to solution bitmaps during
21239 2017-08-17 Marek Polacek <polacek@redhat.com>
21241 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21243 2017-08-17 Richard Biener <rguenther@suse.de>
21245 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21246 to true when overflow is undefined and we saturated the result.
21248 2017-08-17 Alan Modra <amodra@gmail.com>
21251 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21252 Don't use store multiple if only one reg needs saving.
21253 (interesting_frame_related_regno): New function.
21254 (rs6000_frame_related): Don't emit frame info for regs that
21256 (rs6000_emit_epilogue): Likewise.
21258 2017-08-16 Nathan Sidwell <nathan@acm.org>
21260 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21261 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21262 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21263 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21264 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21265 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21266 (verify_type): Adjust for TYPE_BINFO move.
21267 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21268 process TYPE_BINFO directly.
21269 (hash_tree): Likewise.
21270 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21272 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21275 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21277 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21279 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21282 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21283 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21284 Add ix86_binary_operator_ok to insn constraint.
21285 (*iordi_1_bts): Ditto.
21286 (*xordi_1_btc): Ditto.
21287 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21288 Update corresponding peephole2 pattern.
21292 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21294 PR tree-optimization/81832
21295 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21296 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21298 2017-08-16 Marek Polacek <polacek@redhat.com>
21301 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21302 regarding size_zero_node.
21304 2017-08-16 Martin Liska <mliska@suse.cz>
21307 * config.gcc: Respect previously set extra_objs in case
21310 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21312 PR tree-optimization/81835
21313 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21314 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21315 not depend on the phi.
21317 2017-08-16 Alan Modra <amodra@gmail.com>
21319 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21322 2017-08-16 Alan Modra <amodra@gmail.com>
21324 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21325 (save_reg_p): ..into this. Update all callers.
21326 (first_reg_to_save): Simplify.
21328 2017-08-16 Alan Modra <amodra@gmail.com>
21330 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21333 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21337 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21338 constant count if that count is less than 32.
21340 2017-08-15 Nathan Sidwell <nathan@acm.org>
21342 * gcc.c (execute): Emit friendlier message if inferior is killed
21343 by an external cause.
21345 2017-08-15 Richard Biener <rguenther@suse.de>
21347 PR tree-optimization/81790
21348 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21349 CONSTRUCTORs from simplifying and VN.
21351 2017-08-14 Martin Sebor <msebor@redhat.com>
21353 * builtin-attrs.def: Add comments.
21355 2017-08-14 Martin Sebor <msebor@redhat.com>
21358 * doc/extend.texi (attribute nonstring): Document new attribute.
21360 2017-08-14 Martin Sebor <msebor@redhat.com>
21363 * tree-diagnostic.c (default_tree_printer): Handle %G.
21364 * gimple-pretty-print.h (percent_G_format): Declare new function.
21365 * gimple-pretty-print.c (percent_G_format): Define.
21366 * tree-pretty-print.c (percent_K_format): Add argument.
21368 2017-08-14 Martin Sebor <msebor@redhat.com>
21370 PR translation/79998
21371 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21372 Remove a stray space.
21374 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21377 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21378 (*iordi_1_bts): Ditto.
21379 (*xordi_1_btc): Ditto.
21381 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21384 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21385 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21387 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21388 quoted strings, and make more translator-friendly.
21389 (darwin_rs6000_override_options): Likewise.
21390 (rs6000_option_override_internal): Likewise.
21391 (rs6000_return_in_memory): Fix overlong line.
21392 (init_cmulative_args): Use quoted strings, and make more
21393 translator-friendly.
21394 (rs6000_pass_by_reference): Fix overlong line.
21395 (def_builtin): Use quoted strings.
21396 (altivec_expand_predicate_builtin): Use quoted strings, and make
21397 more translator-friendly.
21398 (htm_expand_builtin): Use quoted strings.
21399 (cpu_expand_builtin): Use quoted strings, and make more
21400 translator-friendly.
21401 (altivec_expand_builtin): Likewise.
21402 (paired_expand_predicate_builtin): Likewise.
21403 (rs6000_invalid_builtin): Likewise.
21404 (builtin_function_type): Use quoted strings.
21405 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21406 more translator-friendly.
21407 (rs6000_trampoline_init): Likewise.
21408 (rs6000_handle_altivec_attribute): Likewise.
21409 (rs6000_inner_target_options): Use quoted strings.
21410 (rs6000_disable_incompatible_switches): Likewise.
21411 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21412 strings, and make more translator-friendly.
21413 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21415 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21417 PR tree-optimization/81799
21418 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21419 cond_expr to simple gimple operand.
21421 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21423 PR middle-end/46932
21424 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21426 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21430 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21431 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21432 TARGET_GASISR_PROLOGUES.
21433 * config/avr/avr.c (avr_option_override): Same.
21434 (avr_pass_pre_proep::execute): Same.
21436 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21439 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21440 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21442 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21444 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21445 define_insn_and_split. Split to a memory load from 0 in
21446 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21447 using PTR mode iterator.
21448 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21449 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21450 (*add_tp_<mode>): Redefine as define_insn_and_split.
21451 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21452 address space. Merge with *add_tp_x32 using PTR mode iterator.
21453 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21454 Split to an add with a memory load from 0 in
21455 DEFAULT_TLS_SEG_REG address space.
21457 2017-08-12 Andrew Pinski <apinski@cavium.com>
21459 * config/aarch64/aarch64-option-extensions.def (rdma):
21460 Fix feature string to what Linux prints out in /proc/cpuinfo.
21462 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21465 * dwarf2out.c (modified_type_die): For C typedef types that have
21466 an ultimate origin, process the ultimate origin instead of the
21468 (gen_typedef_die): Assert that input DECLs have no ultimate
21470 (gen_type_die_with_usage): For typedef variants that have an
21471 ultimate origin, just call gen_decl_die on the original DECL.
21472 (process_scope_var): Avoid creating DIEs for local typedefs and
21473 concrete static variables.
21475 2017-08-12 Alan Modra <amodra@gmail.com>
21479 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21480 match gnu-user.h startfile.
21481 (ENDFILE_LINUX_SPEC): Similarly.
21483 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21486 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21488 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21490 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21491 * config/aarch64/aarch64.md (mov<mode>): Change.
21492 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21493 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21494 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21496 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21498 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21499 for storage order barriers.
21501 2017-08-11 Martin Liska <mliska@suse.cz>
21504 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21505 variables of void type.
21507 2017-08-11 Martin Liska <mliska@suse.cz>
21509 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21510 TARGET_SUPPORTS_ALIASES.
21511 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21512 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21513 (optimize_weakref): Likewise.
21514 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21515 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21516 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21518 2017-08-11 Martin Liska <mliska@suse.cz>
21521 * config/i386/i386.c (make_resolver_func): Do complete
21522 refactoring of the function.
21524 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21527 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21528 * config/i386/i386.c (ix86_stack_protect_guard): Use
21529 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21530 * doc/invoke.texi (x86 Options): Document
21531 -mstack-protector-guard-symbol= option.
21533 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21535 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21536 * config/i386/i386.c (ix86_split_stack_guard): New function.
21537 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21538 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21539 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21540 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21541 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21542 (split_stack_space_check): Call ix86_split_stack_guard.
21544 2017-08-10 Martin Sebor <msebor@redhat.com>
21546 * print-tree.c (print_node): Print location using the established
21548 Replace spaces with colons.
21549 (debug_raw, debug): Ditto.
21551 2017-08-10 Martin Sebor <msebor@redhat.com>
21554 * pretty-print.c (pp_format): Correct the handling of %s precision.
21556 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21559 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21561 (ix86_finalize_stack_frame_flags): This. Also clear
21562 frame_pointer_needed if -fno-omit-frame-pointer is used without
21564 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21565 with ix86_finalize_stack_frame_flags.
21566 (ix86_expand_epilogue): Likewise.
21567 (ix86_expand_split_stack_prologue): Likewise.
21568 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21570 2017-08-10 Martin Liska <mliska@suse.cz>
21573 * c-attribs.c (handle_target_attribute):
21574 Report warning for an empty string argument of target attribute.
21576 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21579 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21581 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21582 or DECL_NONLOCAL labels.
21583 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21584 or DECL_NONLOCAL labels here.
21586 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21588 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21589 to indicate when early gimple folding has been disabled.
21590 (rs6000_gimple_fold_builtin): Add debug content.
21591 (rs6000_invalid_builtin): Fix whitespace.
21592 (rs6000_expand_builtin): Fix whitespace.
21593 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21595 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21598 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21599 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21601 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21603 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21604 (qdf24xx): Likewise.
21605 * config/aarch64/aarch64.md: Include falkor.md.
21606 * config/aarch64/falkor.md: New.
21608 2017-08-09 Marek Polacek <polacek@redhat.com>
21611 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21612 * diagnostic.c (emit_diagnostic): Add a comment.
21613 (emit_diagnostic_valist): New function.
21615 2017-08-09 Marek Polacek <polacek@redhat.com>
21618 * input.c (make_location): New overload.
21619 * input.h (make_location): Declare.
21621 2017-08-08 Alan Modra <amodra@gmail.com>
21622 H.J. Lu <hongjiu.lu@intel.com>
21625 * gcc.c (NO_PIE_SPEC): Delete.
21626 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21628 (LINK_PIE_SPEC): ..to here.
21629 (LINK_COMMAND_SPEC): Support -no-pie.
21630 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21631 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21633 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21634 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21635 (ENDFILE_CRTEND_SPEC): Similarly.
21637 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21640 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21641 (mstack-protector-guard-offset=): Ditto.
21642 * config/i386/i386.c (ix86_option_override): Handle
21643 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21645 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21646 ix86_stack_protect_guard_offset variables.
21647 (TARGET_STACK_PROTECT_GUARD): Always define.
21648 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21649 and -mstack-protector-guard-offset= options.
21651 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21653 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21654 boundary case for the last candidate.
21656 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21658 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21659 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21661 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21663 PR middle-end/19706
21664 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21665 * config/aarch64/aarch64-builtins.c
21666 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21667 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21668 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21670 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21671 Andrew Pinski <pinskia@gmail.com>
21673 PR middle-end/19706
21674 * internal-fn.def (XORSIGN): New.
21675 * optabs.def (xorsign_optab): New.
21676 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21677 (convert_expand_mult_copysign): New.
21678 (pass_optimize_widening_mul::execute): Call
21679 convert_expand_mult_copysign.
21681 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21683 PR tree-optimization/81354
21684 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21685 Insert on edges rather than explicitly creating landing pads.
21686 (analyze_candidates_and_replace): Commit edge inserts.
21688 2017-08-08 Richard Biener <rguenther@suse.de>
21690 PR middle-end/81719
21691 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21692 (expand_simple_operations): Also look through ADDR_EXPRs with
21693 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21695 2017-08-08 Richard Biener <rguenther@suse.de>
21697 PR tree-optimization/81723
21698 * tree-vect-slp.c (struct bst_traits): New hash traits.
21699 (bst_fail): New global.
21700 (vect_build_slp_tree_2): New worker, split out from ...
21701 (vect_build_slp_tree): ... this now wrapping it with using
21702 bst_fail set to cache SLP tree build fails. Properly handle
21704 (vect_analyze_slp_instance): Allocate and free bst_fail.
21706 2017-08-08 Martin Liska <mliska@suse.cz>
21709 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21710 LABEL_DECLs that can be from a different function.
21712 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21714 PR tree-optimization/81744
21715 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21716 loop's number of iterations.
21718 2017-08-08 Martin Liska <mliska@suse.cz>
21720 * asan.c: Include header files.
21721 * attribs.c (build_decl_attribute_variant): New function moved
21723 (build_type_attribute_qual_variant): Likewise.
21724 (cmp_attrib_identifiers): Likewise.
21725 (simple_cst_list_equal): Likewise.
21726 (omp_declare_simd_clauses_equal): Likewise.
21727 (attribute_value_equal): Likewise.
21728 (comp_type_attributes): Likewise.
21729 (build_type_attribute_variant): Likewise.
21730 (lookup_ident_attribute): Likewise.
21731 (remove_attribute): Likewise.
21732 (merge_attributes): Likewise.
21733 (merge_type_attributes): Likewise.
21734 (merge_decl_attributes): Likewise.
21735 (merge_dllimport_decl_attributes): Likewise.
21736 (handle_dll_attribute): Likewise.
21737 (attribute_list_equal): Likewise.
21738 (attribute_list_contained): Likewise.
21739 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21740 (lookup_attribute_by_prefix): Likewise.
21741 * bb-reorder.c: Include header files.
21742 * builtins.c: Likewise.
21743 * calls.c: Likewise.
21744 * cfgexpand.c: Likewise.
21745 * cgraph.c: Likewise.
21746 * cgraphunit.c: Likewise.
21747 * convert.c: Likewise.
21748 * dwarf2out.c: Likewise.
21749 * final.c: Likewise.
21750 * fold-const.c: Likewise.
21751 * function.c: Likewise.
21752 * gimple-expr.c: Likewise.
21753 * gimple-fold.c: Likewise.
21754 * gimple-pretty-print.c: Likewise.
21755 * gimple.c: Likewise.
21756 * gimplify.c: Likewise.
21757 * hsa-common.c: Likewise.
21758 * hsa-gen.c: Likewise.
21759 * internal-fn.c: Likewise.
21760 * ipa-chkp.c: Likewise.
21761 * ipa-cp.c: Likewise.
21762 * ipa-devirt.c: Likewise.
21763 * ipa-fnsummary.c: Likewise.
21764 * ipa-inline.c: Likewise.
21765 * ipa-visibility.c: Likewise.
21767 * lto-cgraph.c: Likewise.
21768 * omp-expand.c: Likewise.
21769 * omp-general.c: Likewise.
21770 * omp-low.c: Likewise.
21771 * omp-offload.c: Likewise.
21772 * omp-simd-clone.c: Likewise.
21773 * opts-global.c: Likewise.
21774 * passes.c: Likewise.
21775 * predict.c: Likewise.
21776 * sancov.c: Likewise.
21777 * sanopt.c: Likewise.
21778 * symtab.c: Likewise.
21779 * toplev.c: Likewise.
21780 * trans-mem.c: Likewise.
21781 * tree-chkp.c: Likewise.
21782 * tree-eh.c: Likewise.
21783 * tree-into-ssa.c: Likewise.
21784 * tree-object-size.c: Likewise.
21785 * tree-parloops.c: Likewise.
21786 * tree-profile.c: Likewise.
21787 * tree-ssa-ccp.c: Likewise.
21788 * tree-ssa-live.c: Likewise.
21789 * tree-ssa-loop.c: Likewise.
21790 * tree-ssa-sccvn.c: Likewise.
21791 * tree-ssa-structalias.c: Likewise.
21792 * tree-ssa.c: Likewise.
21793 * tree-streamer-in.c: Likewise.
21794 * tree-vectorizer.c: Likewise.
21795 * tree-vrp.c: Likewise.
21796 * tsan.c: Likewise.
21797 * ubsan.c: Likewise.
21798 * varasm.c: Likewise.
21799 * varpool.c: Likewise.
21800 * tree.c: Remove functions moved to attribs.[ch].
21801 * tree.h: Likewise.
21802 * config/aarch64/aarch64.c: Add attrs.h header file.
21803 * config/alpha/alpha.c: Likewise.
21804 * config/arc/arc.c: Likewise.
21805 * config/arm/arm.c: Likewise.
21806 * config/avr/avr.c: Likewise.
21807 * config/bfin/bfin.c: Likewise.
21808 * config/c6x/c6x.c: Likewise.
21809 * config/cr16/cr16.c: Likewise.
21810 * config/cris/cris.c: Likewise.
21811 * config/darwin.c: Likewise.
21812 * config/epiphany/epiphany.c: Likewise.
21813 * config/fr30/fr30.c: Likewise.
21814 * config/frv/frv.c: Likewise.
21815 * config/ft32/ft32.c: Likewise.
21816 * config/h8300/h8300.c: Likewise.
21817 * config/i386/winnt.c: Likewise.
21818 * config/ia64/ia64.c: Likewise.
21819 * config/iq2000/iq2000.c: Likewise.
21820 * config/lm32/lm32.c: Likewise.
21821 * config/m32c/m32c.c: Likewise.
21822 * config/m32r/m32r.c: Likewise.
21823 * config/m68k/m68k.c: Likewise.
21824 * config/mcore/mcore.c: Likewise.
21825 * config/microblaze/microblaze.c: Likewise.
21826 * config/mips/mips.c: Likewise.
21827 * config/mmix/mmix.c: Likewise.
21828 * config/mn10300/mn10300.c: Likewise.
21829 * config/moxie/moxie.c: Likewise.
21830 * config/msp430/msp430.c: Likewise.
21831 * config/nds32/nds32-isr.c: Likewise.
21832 * config/nds32/nds32.c: Likewise.
21833 * config/nios2/nios2.c: Likewise.
21834 * config/nvptx/nvptx.c: Likewise.
21835 * config/pa/pa.c: Likewise.
21836 * config/pdp11/pdp11.c: Likewise.
21837 * config/powerpcspe/powerpcspe.c: Likewise.
21838 * config/riscv/riscv.c: Likewise.
21839 * config/rl78/rl78.c: Likewise.
21840 * config/rx/rx.c: Likewise.
21841 * config/s390/s390.c: Likewise.
21842 * config/sh/sh.c: Likewise.
21843 * config/sol2.c: Likewise.
21844 * config/sparc/sparc.c: Likewise.
21845 * config/spu/spu.c: Likewise.
21846 * config/stormy16/stormy16.c: Likewise.
21847 * config/tilegx/tilegx.c: Likewise.
21848 * config/tilepro/tilepro.c: Likewise.
21849 * config/v850/v850.c: Likewise.
21850 * config/vax/vax.c: Likewise.
21851 * config/visium/visium.c: Likewise.
21852 * config/xtensa/xtensa.c: Likewise.
21854 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21857 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21858 constraints since the -mupper-regs-* switches have been
21860 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21861 into a vector from a double word element that was extracted from
21862 another vector, and eliminate extra XXPERMDI instructions.
21863 (vsx_concat_<mode>_2): Likewise.
21864 (vsx_concat_<mode>_3): Likewise.
21865 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21866 concat to allow optimizing inserts from previous extracts.
21868 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21870 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21871 memory reference to a SSP offset in TLS address space.
21872 (ix86_print_operand) <case '@'>: Remove.
21873 (ix86_print_operand_punct_valid_p): Remove '@' code.
21874 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21875 UNSPEC_SP_TLS_TEST.
21876 (stack_tls_protect_set_<mode>): Remove.
21877 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21878 (stack_tls_protect_test_<mode>): Remove.
21879 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21881 2017-08-07 Olivier Hainque <hainque@adacore.com>
21884 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21886 2017-08-07 Douglas Rupp <rupp@adacore.com>
21888 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21889 variable was referenced as multidir in command.
21891 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21894 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21897 2017-08-07 Martin Liska <mliska@suse.cz>
21899 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21900 * config/rl78/rl78.c: Add include of attribs.h.
21901 * config/sh/sh.c: Likewise.
21902 * config/v850/v850.c: Likewise.
21904 2017-08-07 Tom de Vries <tom@codesourcery.com>
21906 PR middle-end/78266
21907 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21909 2017-08-07 Martin Liska <mliska@suse.cz>
21911 * config/mips/mips.c: Include attribs.h.
21913 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21916 * doc/invoke.texi: Document change in behvaior for -Ofast for
21919 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21921 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21923 (aarch64_pop_regs): Likewise.
21924 (aarch64_gen_load_pair): Likewise.
21925 (aarch64_save_callee_saves): Likewise.
21926 (aarch64_restore_callee_saves): Likewise.
21928 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21930 * config/i386/i386.c: Revert the last change.
21932 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21935 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21937 (ix86_finalize_stack_frame_flags): This. Also clear
21938 frame_pointer_needed if -fno-omit-frame-pointer is used without
21940 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21941 with ix86_finalize_stack_frame_flags.
21942 (ix86_expand_epilogue): Likewise.
21943 (ix86_expand_split_stack_prologue): Likewise.
21945 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21948 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21949 to P_AES for Westmere.
21951 2017-08-07 Jonathan Yong <10walls@gmail.com>
21953 * config/i386/mingw.opt (fset-stack-executable): Removed.
21954 * config/i386/cygming.opt (fset-stack-executable): Moved
21956 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21958 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21960 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21962 2017-08-07 Marek Polacek <polacek@redhat.com>
21964 PR middle-end/81737
21965 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21967 2017-08-07 Martin Liska <mliska@suse.cz>
21969 * attribs.h (canonicalize_attr_name): New function.
21970 (cmp_attribs): Move from c-format.c and adjusted.
21971 (is_attribute_p): Moved from tree.h.
21972 * tree-inline.c: Add new includes.
21973 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21974 (private_is_attribute_p): Remove.
21975 (private_lookup_attribute): Likewise.
21976 (private_lookup_attribute_by_prefix): Simplify.
21977 (remove_attribute): Use is_attribute_p.
21978 * tree.h: Remove removed declarations.
21980 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21982 PR middle-end/81698
21983 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21984 instead of computing it in the function. Formatting fix.
21985 (expand_case): Don't rely on default_edge being the first edge,
21986 clear it if removing it, pass default_edge to
21987 emit_case_dispatch_table.
21988 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21991 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21993 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21994 insn in the function, emit NOP after the insn.
21996 2017-08-06 Tom de Vries <tom@codesourcery.com>
21998 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
22001 2017-08-06 Tom de Vries <tom@codesourcery.com>
22003 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
22006 2017-08-04 Yury Gribov <tetra2005@gmail.com>
22008 PR tree-optimization/57371
22009 * match.pd: New pattern.
22011 2017-08-04 Marek Polacek <polacek@redhat.com>
22013 PR middle-end/81695
22014 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
22015 perform the computation in offset_int.
22017 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22019 PR tree-optimization/81136
22020 * tree-vectorizer.h: Include tree-hash-traits.h.
22021 (vec_base_alignments): New typedef.
22022 (vec_info): Add a base_alignments field.
22023 (vect_record_base_alignments): Declare.
22024 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
22026 (DR_IS_CONDITIONAL_IN_STMT): New macro.
22027 (create_data_ref): Add an is_conditional_in_stmt argument.
22028 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
22029 the is_conditional_in_stmt field.
22030 (data_ref_loc): Add an is_conditional_in_stmt field.
22031 (get_references_in_stmt): Set the is_conditional_in_stmt field.
22032 (find_data_references_in_stmt): Update call to create_data_ref.
22033 (graphite_find_data_references_in_stmt): Likewise.
22034 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
22035 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
22036 (vect_record_base_alignment): New function.
22037 (vect_record_base_alignments): Likewise.
22038 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
22039 for nested statements even if we fail to compute a misalignment.
22040 Use pooled base alignments for unconditional references.
22041 (vect_find_same_alignment_drs): Compare base addresses instead
22043 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
22044 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
22046 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22048 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
22049 Add an explicit name for the enum. Use auto_vec for slp_instances
22050 and grouped_stores.
22051 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
22053 (_bb_vec_info): Add a constructor and destructor.
22054 (vinfo_for_stmt): Return NULL for uids of -1 as well.
22055 (destroy_loop_vec_info): Delete.
22056 (vect_destroy_datarefs): Likewise.
22057 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
22058 (vec_info::vec_info): New function.
22059 (vec_info::~vec_info): Likewise.
22060 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22061 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22062 destroy_loop_vec_info.
22063 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22064 (_loop_vec_info::_loop_vec_info): ...this.
22065 (destroy_loop_vec_info): Replace with...
22066 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22067 the stmt_vec_infos. Leave handling of vec_info information to its
22068 destructor. Remove explicit vector releases.
22069 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22070 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22071 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22072 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22073 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22074 (destroy_bb_vec_info): Replace with...
22075 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22076 information to its destructor.
22077 (vect_slp_analyze_bb_1): Use new and delete instead of
22078 new_bb_vec_info and destroy_bb_vec_info.
22079 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22082 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22084 * tree-data-ref.h (subscript): Add access_fn field.
22085 (data_dependence_relation): Add could_be_independent_p.
22086 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22087 (same_access_functions): Move to tree-data-ref.c.
22088 * tree-data-ref.c (ref_contains_union_access_p): New function.
22089 (access_fn_component_p): Likewise.
22090 (access_fn_components_comparable_p): Likewise.
22091 (dr_analyze_indices): Add a reference to access_fn_component_p.
22092 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22094 (constant_access_functions): Likewise.
22095 (add_other_self_distances): Likewise.
22096 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22097 (initialize_data_dependence_relation): Use XCNEW and remove
22098 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22099 of access functions that have the same type. Allow the
22100 subsequence to end with different bases in some circumstances.
22101 Record the chosen access functions in SUB_ACCESS_FN.
22102 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22103 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22104 (subscript_dependence_tester_1): Likewise dra and drb.
22105 (build_classic_dist_vector): Update calls accordingly.
22106 (subscript_dependence_tester): Likewise.
22107 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22108 DDR_COULD_BE_INDEPENDENT_P.
22109 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22110 comp_alias_ddrs instead of may_alias_ddrs.
22111 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22113 (vect_analyze_data_ref_dependence): Use it if
22114 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22115 distance vectors if that fails.
22116 (dependence_distance_ge_vf): New function.
22117 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22118 LOOP_VINFO_MAY_ALIAS_DDRS.
22120 2017-08-04 Richard Biener <rguenther@suse.de>
22122 PR middle-end/81705
22123 * fold-const.c (fold_binary_loc): Properly restrict
22124 minus_var0 && minus_var1 case when associating undefined overflow
22127 2017-08-04 Tom de Vries <tom@codesourcery.com>
22129 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22131 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22133 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22134 Don't start diagnostic messages with a capital letter.
22135 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22137 (rs6000_invalid_builtin): Likewise.
22138 (rs6000_trampoline_init): Likewise.
22140 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22143 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22144 after setting changeable df flags.
22146 2017-08-03 Richard Biener <rguenther@suse.de>
22148 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22149 up if the use is in USE - X.
22151 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22153 * toplev.c (dumpfile.h): New include.
22154 (internal_error_reentered): New static function. Use it...
22155 (internal_error_function): ...here to handle reentered internal_error.
22157 2017-08-03 Richard Biener <rguenther@suse.de>
22159 PR middle-end/81148
22160 * fold-const.c (split_tree): Add minus_var and minus_con
22161 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22162 here but always use minus_*.
22163 (associate_trees): Assert we never associate with MINUS_EXPR
22164 and NULL first operand. Do not recurse for PLUS_EXPR operands
22165 when associating as MINUS_EXPR either.
22166 (fold_binary_loc): Track minus_var and minus_con.
22168 2017-08-03 Tom de Vries <tom@codesourcery.com>
22171 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22172 ACCEL_COMPILER, apply finish_options on
22173 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22175 2017-08-03 Tom de Vries <tom@codesourcery.com>
22178 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22179 function_entry_patch_area_size > 0.
22181 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22184 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22185 instead of 10??LU, perform unit multiplication in wide_int,
22186 don't change alloc_object_size_limit if the limit is larger
22189 PR tree-optimization/81655
22190 PR tree-optimization/81588
22191 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22192 the case when ranges[i].low and high are 1 for unsigned type with
22195 PR middle-end/81052
22196 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22197 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22199 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22201 * tree-vrp.h: Add include guard.
22203 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22206 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22207 (ud2): New insn pattern.
22208 * config/i386/i386.c (ix86_expand_epilogue):
22209 For naked functions, generate ud2 instead of trap insn.
22211 2017-08-02 Marek Polacek <polacek@redhat.com>
22214 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22216 2017-08-02 Tom de Vries <tom@codesourcery.com>
22217 Cesar Philippidis <cesar@codesourcery.com>
22219 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22220 Add missing edge probabilities.
22222 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22224 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22225 Correct endianness.
22227 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22229 PR middle-end/79499
22230 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22231 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22232 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22233 prologue_seq sequences - if any.
22235 2017-08-02 Richard Biener <rguenther@suse.de>
22237 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22238 via vectors if supported, integer extracts via punning if supported
22239 or otherwise vector extracts.
22241 2017-08-02 Richard Biener <rguenther@suse.de>
22243 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22245 (bitmap_insert_into_set): ... this.
22247 2017-08-02 Richard Biener <rguenther@suse.de>
22249 PR tree-optimization/81633
22251 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22253 PR tree-optimization/71752
22254 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22256 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22258 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22259 (machine_function::call_ms2sysv_pad_out): Remove field.
22260 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22261 (ix86_compute_frame_layout): Likewise.
22263 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22266 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22267 attribute with interrupt attribute.
22269 2017-08-01 Andrew Pinski <apinski@cavium.com>
22271 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22272 BIT_INSERT_EXPR's operand 1
22273 to see if the types precision matches.
22275 2017-08-01 Martin Liska <mliska@suse.cz>
22277 PR middle-end/70140
22278 * builtins.c (expand_builtin_memcpy_args): Remove.
22279 (expand_builtin_memcpy): Call newly added function
22280 expand_builtin_memory_copy_args.
22281 (expand_builtin_memcpy_with_bounds): Likewise.
22282 (expand_builtin_mempcpy): Remove last argument.
22283 (expand_builtin_mempcpy_with_bounds): Likewise.
22284 (expand_builtin_memory_copy_args): New function created from
22285 expand_builtin_mempcpy_args with small modifications.
22286 (expand_builtin_mempcpy_args): Remove.
22287 (expand_builtin_stpcpy): Remove unused argument.
22288 (expand_builtin): Likewise.
22289 (expand_builtin_with_bounds): Likewise.
22291 2017-08-01 Martin Liska <mliska@suse.cz>
22294 Make mempcpy more optimal (PR middle-end/70140).
22296 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22299 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22300 __builtin_vec_cmpne verify both arguments are compatible vectors
22301 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22302 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22303 move computation of aligned to after checking the argument types.
22307 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22308 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22311 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22312 Doug Rupp <rupp@adacore.com>
22313 Olivier Hainque <hainque@adacore.com>
22315 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22316 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22318 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22319 for TARGET_OS_CPP_BUILTIN.
22320 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22321 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22323 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22324 passing required abi options to the assembler for EABI configurations.
22325 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22326 of .text.hot and .text.unlikely sections for kernel modules when
22327 using ARM style exceptions.
22328 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22329 options. Add EXTRA_CC1_SPEC.
22330 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22332 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22334 (ARM_TARGET2_DWARF_FORMAT): Define.
22335 * config/arm/t-vxworks: Adjust multilib control to removal of the
22336 Diab command line options.
22338 2017-08-01 Martin Liska <mliska@suse.cz>
22340 PR gcov-profile/81561
22341 * gcov.c (unblock): Make unblocking safe as we need to preserve
22342 index correspondence of blocks and block_lists.
22344 2017-08-01 Richard Biener <rguenther@suse.de>
22346 PR tree-optimization/81181
22347 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22348 (compute_antic): ... end of iteration here.
22350 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22352 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22353 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22354 (ftree-slp-vectorize): Likewise.
22355 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22356 can no longer be set independent of flag_tree_loop_vectorize.
22357 * omp-general.c (emp_max_vf): Likewise.
22358 * opts.c (enable_fdo_optimizations): Remove references to
22359 flag_tree_vectorize, these are now implicit.
22360 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22361 and leave it for the options machinery.
22363 2017-08-01 Martin Liska <mliska@suse.cz>
22365 PR middle-end/70140
22366 * builtins.c (expand_builtin_memcpy_args): Remove.
22367 (expand_builtin_memcpy): Call newly added function
22368 expand_builtin_memory_copy_args.
22369 (expand_builtin_memcpy_with_bounds): Likewise.
22370 (expand_builtin_mempcpy): Remove last argument.
22371 (expand_builtin_mempcpy_with_bounds): Likewise.
22372 (expand_builtin_memory_copy_args): New function created from
22373 expand_builtin_mempcpy_args with small modifications.
22374 (expand_builtin_mempcpy_args): Remove.
22375 (expand_builtin_stpcpy): Remove unused argument.
22376 (expand_builtin): Likewise.
22377 (expand_builtin_with_bounds): Likewise.
22379 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22382 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22383 print "ds:" only for immediates in generic address space.
22385 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22388 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22389 (ix86_function_ok_for_sibcall): Return false for naked functions.
22391 2017-08-01 Richard Biener <rguenther@suse.de>
22393 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22394 (compute_antic): Seed worklist with exit block predecessors.
22395 * cfganal.c (dfs_find_deadend): For a cycle return the source
22396 of the edge closing it.
22398 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22400 * config/aarch64/aarch64.c
22401 (aarch64_can_const_movi_rtx_p): Move 0 check.
22403 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22405 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22406 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22408 * match.pd: Ditto in address comparison pattern.
22410 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22412 PR tree-optimization/81627
22413 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22414 closed ssa form for store-store chain.
22416 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22418 PR tree-optimization/81620
22419 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22420 for store-store chain.
22422 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22424 PR tree-optimization/81588
22425 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22426 ranges[i].in_p, invert comparison code ccode. For >/>=,
22427 swap rhs1 and rhs2 and comparison code unconditionally,
22428 for </<= don't do that. Don't swap rhs1/rhs2 again if
22429 ranges[i].in_p, instead invert comparison code ccode if
22430 opcode or oe->rank is BIT_IOR_EXPR.
22433 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22434 a direct optab to conversion optab.
22435 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22436 with GET_MODE_INNER as last argument instead of optab_handler.
22437 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22438 vector extraction if possible and optab is available.
22439 * expr.c (store_constructor): Use convert_optab_handler instead
22440 of optab_handler. Use vector initialization from smaller
22441 vectors if possible and optab is available.
22442 * tree-vect-stmts.c (vectorizable_load): Likewise.
22443 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22444 now have two modes.
22445 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22446 of vec_init from half-sized vectors with the same element mode.
22447 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22448 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22449 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22450 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22451 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22452 after mode in gen_vec_extract* calls.
22453 (vec_extract<mode>): Renamed to ...
22454 (vec_extract<mode><ssescalarmodelower>): ... this.
22455 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22456 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22457 element mode after mode in gen_vec_init* calls.
22458 (VEC_INIT_HALF_MODE): New mode iterator.
22459 (vec_init<mode>): Renamed to ...
22460 (vec_init<mode><ssescalarmodelower>): ... this.
22461 (vec_init<mode><ssehalfvecmodelower>): New expander.
22462 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22463 (vec_extractv2sfsf): ... this.
22464 (vec_initv2sf): Renamed to ...
22465 (vec_initv2sfsf): ... this.
22466 (vec_extractv2si): Renamed to ...
22467 (vec_extractv2sisi): ... this.
22468 (vec_initv2si): Renamed to ...
22469 (vec_initv2sisi): ... this.
22470 (vec_extractv4hi): Renamed to ...
22471 (vec_extractv4hihi): ... this.
22472 (vec_initv4hi): Renamed to ...
22473 (vec_initv4hihi): ... this.
22474 (vec_extractv8qi): Renamed to ...
22475 (vec_extractv8qiqi): ... this.
22476 (vec_initv8qi): Renamed to ...
22477 (vec_initv8qiqi): ... this.
22478 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22479 (vec_init<mode>): Renamed to ...
22480 (vec_init<mode><VEC_base_l>): ... this.
22481 (vec_extract<mode>): Renamed to ...
22482 (vec_extract<mode><VEC_base_l>): ... this.
22483 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22484 (vec_initv2sfsf): ... this.
22485 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22486 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22487 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22488 element mode after mode in gen_vec_init* calls.
22489 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22490 (vec_init<mode><Vel>): ... this.
22491 (vec_extract<mode>): Renamed to ...
22492 (vec_extract<mode><Vel>): ... this.
22493 * config/aarch64/iterators.md (Vel): New mode attribute.
22494 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22495 Add element mode after mode in gen_vec_extract* calls.
22496 * config/s390/vector.md (non_vec_l): New mode attribute.
22497 (vec_extract<mode>): Renamed to ...
22498 (vec_extract<mode><non_vec_l>): ... this.
22499 (vec_init<mode>): Renamed to ...
22500 (vec_init<mode><non_vec_l>): ... this.
22501 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22502 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22504 * config/arm/iterators.md (V_elem_l): New mode attribute.
22505 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22506 (vec_extract<mode><V_elem_l>): ... this.
22507 (vec_extractv2di): Renamed to ...
22508 (vec_extractv2didi): ... this.
22509 (vec_init<mode>): Renamed to ...
22510 (vec_init<mode><V_elem_l>): ... this.
22511 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22512 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22513 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22514 Add element mode after gen_vec_extract* calls.
22515 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22516 (vec_init<mode><unitmode>): ... this.
22517 (vec_extract<mode>): Renamed to ...
22518 (vec_extract<mode><unitmode>): ... this.
22519 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22520 (vec_init<mode><unitmode>): ... this.
22521 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22522 (vec_initv2sfsf): ... this.
22523 (vec_extractv2sf): Renamed to ...
22524 (vec_extractv2sfsf): ... this.
22525 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22526 Add element mode after gen_vec_extract* calls.
22527 * config/mips/mips.md (unitmode): New mode iterator.
22528 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22529 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22530 * config/spu/spu.md (inner_l): New mode attribute.
22531 (vec_init<mode>): Renamed to ...
22532 (vec_init<mode><inner_l>): ... this.
22533 (vec_extract<mode>): Renamed to ...
22534 (vec_extract<mode><inner_l>): ... this.
22535 * config/sparc/sparc.md (veltmode): New mode iterator.
22536 (vec_init<VMALL:mode>): Renamed to ...
22537 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22538 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22539 (vec_initv2sisi): ... this.
22540 (vec_initv2sf): Renamed to ...
22541 (vec_initv2sfsf): ... this.
22542 (vec_extractv2sf): Renamed to ...
22543 (vec_extractv2sfsf): ... this.
22544 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22545 (vec_init<mode>): Renamed to ...
22546 (vec_init<mode><VEC_base_l>): ... this.
22547 (vec_extract<mode>): Renamed to ...
22548 (vec_extract<mode><VEC_base_l>): ... this.
22549 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22550 (vec_initv2sfsf): ... this.
22551 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22552 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22553 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22554 gen_vec_init* calls.
22556 2017-08-01 Richard Biener <rguenther@suse.de>
22558 PR tree-optimization/81297
22559 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22560 TREE_OVERFLOW from INTEGER_CSTs.
22562 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22564 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22566 2017-07-31 Carl Love <cel@us.ibm.com>
22568 * config/rs6000/rs6000-c: Add support for built-in functions
22569 vector signed char vec_xl_be (signed long long, signed char *);
22570 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22571 vector signed int vec_xl_be (signed long long, signed int *);
22572 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22573 vector signed long long vec_xl_be (signed long long, signed long long *);
22574 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22575 vector signed short vec_xl_be (signed long long, signed short *);
22576 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22577 vector double vec_xl_be (signed long long, double *);
22578 vector float vec_xl_be (signed long long, float *);
22579 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22580 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22581 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22583 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22584 (altivec_expand_builtin): Add switch statement to call
22585 altivec_expand_xl_be for each builtin.
22586 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22587 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22588 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22589 __builtin_vsx_le_be_v16qi.
22590 * doc/extend.texi: Update the built-in documentation file for the
22591 new built-in functions.
22593 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22596 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22598 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22600 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22602 * config.gcc: Add z14.
22603 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22604 CPU model numbers for z13s and z14.
22605 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22607 * config/s390/s390-opts.h (enum processor_type): Rename
22608 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22609 * config/s390/s390.c (processor_table): Add field for CPU name to
22610 be passed to Binutils.
22611 (s390_asm_output_machine_for_arch): Use the new field in
22612 processor_table for Binutils.
22613 (s390_expand_builtin): Replace arch12 with z14.
22614 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22615 (s390_get_sched_attrmask): Likewise.
22616 (s390_get_unit_mask): Likewise.
22617 * config/s390/s390.opt: Add z14 to processor_type enum.
22619 2017-07-31 Martin Jambor <mjambor@suse.cz>
22622 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22623 regardless of optimization level.
22625 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22626 Martin Liska <mliska@suse.cz>
22628 * predict.def: Remove old comment and adjust probability.
22629 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22630 PREDICT statements.
22632 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22635 * config/i386/i386.c (ix86_function_naked): New function.
22636 (ix86_can_use_return_insn_p): Return false for naked functions.
22637 (ix86_expand_prologue): Skip prologue for naked functions.
22638 (ix86_expand_epilogue): Skip epilogue for naked functions
22639 and emit trap instruction.
22640 (ix86_warn_func_return): New function.
22641 (ix86_attribute_table): Add "naked" attribute specification.
22642 (TARGET_WARN_FUNC_RETURN): Define.
22643 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22645 2017-07-31 Martin Liska <mliska@suse.cz>
22647 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22648 (dump_gimple_bb_header): Always dump BB info.
22649 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22651 2017-07-31 Martin Liska <mliska@suse.cz>
22654 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22655 also with current_function_decl non-null equality.
22657 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22660 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22661 change type to the element type, instead add eltype variable and
22662 use it where we are interested in the element type.
22664 PR tree-optimization/81603
22665 * ipa-polymorphic-call.c
22666 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22667 offset arithmetic in offset_int, bail out if the resulting bit offset
22668 doesn't fit into shwi.
22670 2017-07-31 Martin Liska <mliska@suse.cz>
22672 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22673 (gimplify_save_expr): Fix comment.
22675 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22678 * config/i386/i386.c (ix86_function_arg): Update arguments for
22680 (ix86_compute_frame_layout): Set the initial stack offset to
22681 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22682 INCOMING_FRAME_SP_OFFSET.
22683 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22684 stack before exception handler returns.
22685 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22686 the 'ERROR_CODE' for exception handler.
22688 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22690 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22691 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22692 (ASM_OUTPUT_REG_POP): Ditto.
22693 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22694 instead of asm_fprintf to output pure string.
22696 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22698 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22699 to imported_module_or_decl hook.
22700 (debug_nothing_tree_tree_tree_bool): Remove.
22701 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22702 * debug.c (do_nothing_debug_hooks): Use
22703 debug_nothing_tree_tree_tree_bool_bool instead of
22704 debug_nothing_tree_tree_tree_bool.
22705 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22706 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22707 * sdbout.c (sdb_debug_hooks): Likewise.
22708 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22709 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22711 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22712 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22713 attribute, don't add anything.
22715 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22717 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22718 (fold_build2_stat_loc): Likewise.
22719 (fold_build3_stat_loc): Likewise.
22720 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22721 (fold_build1_loc): Remove macro.
22722 (fold_build2_loc): Likewise.
22723 (fold_build3_loc): Likewise.
22725 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22727 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22728 (gimple_build_debug_bind_source_stat): Likewise.
22729 * gimple.h (gimple_build_debug_bind): Remove macro.
22730 (gimple_build_debug_bind_source): Likewise.
22732 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22734 * bitmap.c (bitmap_alloc): Adjust.
22735 (bitmap_gc_alloc): Likewise.
22736 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22738 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22740 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22741 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22742 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22743 (bitmap_gc_alloc_stat): Likewise.
22744 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22746 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22748 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22749 * rtl.h (shallow_copy_rtx): Remove macro.
22751 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22753 * emit-rtl.c (gen_raw_REG): Adjust.
22754 * gengenrtl.c (gendef): Likewise.
22755 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22756 * rtl.h (rtx_alloc): Remove macro.
22758 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22760 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22761 (build_tree_list_stat): Likewise.
22762 * tree.h (build_tree_list): Remove macro.
22763 (build_tree_list_vec): Likewise.
22765 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22767 * tree.c (make_vector_stat): Remove _stat from name.
22768 (build_vector_stat): Likewise.
22769 * tree.h (make_vector_stat): Remove macro.
22770 (build_vector_stat): Likewise.
22772 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22774 * tree.h (build_var_debug_value): Remove prototype.
22776 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22778 * tree.c (tree_cons_stat): Remove _stat from name.
22779 * tree.h (tree_cons): Remove macro.
22781 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22783 * tree.c (build_vl_exp_stat): Remove _stat from name.
22784 * tree.h (build_vl_exp): Remove macro.
22786 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22788 * tree.c (build_decl_stat): Remove _stat from name.
22789 * tree.h (build_decl): Remove macro.
22791 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22793 * gimple.c (gimple_build_with_ops_stat): Adjust.
22794 (gimple_alloc_stat): Remove _stat from name.
22795 * gimple.h (gimple_alloc): Remove macro.
22797 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22799 * tree.c (make_tree_vec_stat): Remove _stat from name.
22800 (grow_tree_vec_stat): Likewise.
22801 * tree.h (make_tree_vec_stat): Adjust prototype.
22802 (grow_tree_vec_stat): Likewise.
22803 (make_tree_vec): Remove macro.
22804 (grow_tree_vec): Likewise.
22806 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22808 * fold-const.c (fold_build1_stat_loc): Adjust.
22809 (fold_build2_stat_loc): Likewise.
22810 (fold_build3_stat_loc): Likewise.
22811 * tree.c (build0_stat): Remove _stat from name.
22812 (build1_stat): Likewise.
22813 (build2_stat): Likewise.
22814 (build3_stat): Likewise.
22815 (build4_stat): Likewise.
22816 (build5_stat): Likewise.
22817 * tree.h (build1_loc): Remove macro, and rename _stat function
22819 (build2_loc): Likewise.
22820 (build3_loc): Likewise.
22821 (build4_loc): Likewise.
22822 (build5_loc): Likewise.
22824 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22826 * tree.c (make_int_cst_stat): Remove _stat from name.
22827 * tree.h (make_int_cst_stat): Adjust prototype.
22828 (make_int_cst): Remove macro.
22830 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22832 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22833 * tree.h (make_tree_binfo_stat): Adjust prototype.
22834 (make_tree_binfo): Remove.
22836 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22838 * tree.c (copy_node_stat): Rename to copy_node.
22839 (build_distinct_type_copy): Adjust.
22840 * tree.h (copy_node_stat): Adjust prototype.
22841 (copy_node): Remove macro.
22843 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22845 * tree.c (make_node_stat): rename to make_node.
22846 (build_tree_list_stat): Adjust.
22847 (build0_stat): Likewise.
22848 (build2_stat): Likewise.
22849 (build3_stat): Likewise.
22850 (build4_stat): Likewise.
22851 (build5_stat): Likewise.
22852 (build_decl_stat): Likewise.
22853 * tree.h (make_node_stat): Adjust prototype.
22854 (make_node): remove macro.
22856 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22858 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22859 (PPC_FEATURE2_SCV): Likewise.
22860 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22862 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22864 * config/aarch64/aarch64.c
22865 (aarch64_internal_mov_immediate): Add new special pattern.
22866 * config/aarch64/aarch64.md (*movdi_aarch64):
22867 Add reg/32bit const mov case.
22869 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22870 Richard Sandiford <richard.sandiford@linaro.org>
22872 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22873 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22874 Add integer and movi cases.
22875 (movi-split-hf-df-sf split, fp16): New.
22876 (enabled): Added TARGET_FP_F16INST.
22877 * config/aarch64/iterators.md (GPF_HF): New.
22878 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22880 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22882 * config/aarch64/aarch64.c
22883 (aarch64_simd_container_mode): Add prototype.
22884 (aarch64_expand_mov_immediate): Add HI support.
22885 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22886 (aarch64_can_const_movi_rtx_p): New.
22887 (aarch64_preferred_reload_class):
22888 Remove restrictions of using FP registers for certain SIMD operations.
22889 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22890 (aarch64_valid_floating_const): Add integer move validation.
22891 (aarch64_simd_imm_scalar_p): Remove.
22892 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22893 (aarch64_legitimate_constant_p): Expand list of supported cases.
22894 * config/aarch64/aarch64-protos.h
22895 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22896 (aarch64_reinterpret_float_as_int): New.
22897 (aarch64_simd_imm_scalar_p): Remove.
22898 * config/aarch64/constraints.md (Uvi): New.
22899 (Dd): Split into Ds and new Dd.
22900 * config/aarch64/aarch64.md (*movsi_aarch64):
22902 (*movdi_aarch64): Add SIMD mov case.
22904 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22906 * tree-predcom.c: (struct chain): Handle store-store chain in which
22907 stores for elimination only store loop invariant values.
22908 (execute_pred_commoning_chain): Ditto.
22909 (prepare_initializers_chain_store_elim): Ditto.
22910 (prepare_finalizers): Ditto.
22911 (is_inv_store_elimination_chain): New function.
22912 (initialize_root_vars_store_elim_1): New function.
22914 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22916 * tree-predcom.c: Revise general description of the pass.
22917 (enum chain_type): New enum type for store elimination.
22918 (struct chain): New field supporting store elimination.
22919 (struct component): Ditto.
22920 (dump_chain): Dump store-stores chain.
22921 (release_chain): Release resources.
22922 (split_data_refs_to_components): Compute and create component
22923 contains only stores for elimination.
22924 (get_chain_last_ref_at): New function.
22925 (make_invariant_chain): Initialization.
22926 (make_rooted_chain): Specify chain type in parameter and record it.
22927 (add_looparound_copies): Skip for store-stores chain.
22928 (determine_roots_comp): Compute type of chain and pass it to
22930 (initialize_root_vars_store_elim_2): New function.
22931 (finalize_eliminated_stores): New function.
22932 (remove_stmt): Handle store for elimination.
22933 (execute_pred_commoning_chain): Execute predictive commoning on
22934 store-store chains.
22935 (determine_unroll_factor): Skip unroll for store-stores chain.
22936 (prepare_initializers_chain_store_elim): New function.
22937 (prepare_initializers_chain): Hanlde store-store chain.
22938 (prepare_finalizers_chain, prepare_finalizers): New function.
22939 (tree_predictive_commoning_loop): Return integer value indicating
22940 if loop is unrolled or lcssa form is corrupted.
22941 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22943 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22945 * tree-predcom.c (initialize_root): Delete.
22946 (execute_pred_commoning_chain): Initialize root vars and replace
22947 reference of non-combined chain directly, rather than call above
22950 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22952 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22953 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22955 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22957 * tree-predcom.c (struct chain): New field init_seq.
22958 (release_chain): Release init_seq.
22959 (prepare_initializers_chain): Record intialization stmts in above
22961 (insert_init_seqs): New function.
22962 (tree_predictive_commoning_loop): Call insert_init_seqs.
22964 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22966 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22968 2017-07-28 Richard Biener <rguenther@suse.de>
22970 * match.pd: Remove superfluous :c.
22971 * genmatch.c (simplify::id): Add member.
22972 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22974 (current_id): New global.
22975 (dt_node::parent): Move from ...
22976 (dt_operand::parent): ... here. Add for_id member.
22977 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22978 (decision_tree::find_node): Relax order requirement when
22979 merging DT_TRUE nodes to ones inbetween the current simplify
22980 and the one we try to merge with. Add diagnostic whenever
22981 we need to enforce pattern order by not merging.
22982 (decision_tree::insert): Set current_id.
22983 (decision_tree::print_node): Dump parent node and for_id.
22984 (parser::last_id): Add member.
22985 (parser::push_simplify): Assign unique id.
22986 (parser::parser): Initialize last_id.
22988 2017-07-28 Martin Liska <mliska@suse.cz>
22991 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22992 gimple_build_debug_bind.
22994 2017-07-28 Richard Biener <rguenther@suse.de>
22996 PR tree-optimization/81502
22997 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22999 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
23001 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
23002 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
23003 for BIT_FIELD_REF args.
23004 * fold-const.c (make_bit_field_ref): Likewise.
23005 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
23007 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23010 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
23011 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
23012 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
23013 Or it into SANITIZER_UNDEFINED.
23014 * ubsan.c: Include gimple-fold.h and varasm.h.
23015 (ubsan_expand_ptr_ifn): New function.
23016 (instrument_pointer_overflow): New function.
23017 (maybe_instrument_pointer_overflow): New function.
23018 (instrument_object_size): Formatting fix.
23019 (pass_ubsan::execute): Call instrument_pointer_overflow
23020 and maybe_instrument_pointer_overflow.
23021 * internal-fn.c (expand_UBSAN_PTR): New function.
23022 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
23023 * sanitizer.def (__ubsan_handle_pointer_overflow,
23024 __ubsan_handle_pointer_overflow_abort): New builtins.
23025 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
23026 * internal-fn.def (UBSAN_PTR): New internal function.
23027 * opts.c (sanitizer_opts): Add pointer-overflow.
23028 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
23029 * fold-const.c (build_range_check): Compute pointer range check in
23030 integral type if pointer arithmetics would be needed. Formatting
23033 2017-07-28 Martin Liska <mliska@suse.cz>
23036 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
23037 parameters that are of a variable-length.
23039 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
23041 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
23043 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
23044 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
23045 (CRT_CALL_STATIC_FUNCTION): Likewise.
23046 (ASM_DEFAULT_SPEC): New define.
23047 (ASM_SPEC32): Likewise.
23048 (ASM_SPEC64): Likewise.
23049 (ASM_SPEC_COMMON): Likewise.
23050 (ASM_SPEC): Likewise.
23051 (INVALID_64BIT): Likewise.
23052 (LINK_OS_DEFAULT_SPEC): Likewise.
23053 (LINK_OS_SPEC32): Likewise.
23054 (LINK_OS_SPEC64): Likewise.
23055 (POWERPC_LINUX): Likewise.
23056 (PTRDIFF_TYPE): Likewise.
23057 (RESTORE_FP_PREFIX): Likewise.
23058 (RESTORE_FP_SUFFIX): Likewise.
23059 (SAVE_FP_PREFIX): Likewise.
23060 (SAVE_FP_SUFFIX): Likewise.
23061 (SIZE_TYPE): Likewise.
23062 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23063 (TARGET_64BIT): Likewise.
23064 (TARGET_64BIT): Likewise.
23065 (TARGET_AIX): Likewise.
23066 (WCHAR_TYPE_SIZE): Likewise.
23067 (WCHAR_TYPE): Undefine.
23068 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23069 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23070 (CPP_OS_RTEMS_SPEC): Delete.
23071 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23072 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23074 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23076 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23078 PR tree-optimization/81578
23079 * tree-parloops.c (build_new_reduction): Bail out if
23080 reduction_code isn't one of the standard OpenMP reductions.
23081 Move the details printing after that decision.
23083 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23085 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23086 related to reload_in_progress.
23087 (splat_input_operand): Likewise.
23088 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23090 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23092 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23093 (TARGET_INSTANTIATE_DECLS): Likewise.
23094 (legitimate_indexed_address_p): Delete reload_in_progress code.
23095 (rs6000_debug_legitimate_address_p): Likewise.
23096 (rs6000_eliminate_indexed_memrefs): Likewise.
23097 (rs6000_emit_le_vsx_store): Likewise.
23098 (rs6000_emit_move_si_sf_subreg): Likewise.
23099 (rs6000_emit_move): Likewise.
23100 (register_to_reg_type): Likewise.
23101 (rs6000_pre_atomic_barrier): Likewise.
23102 (rs6000_machopic_legitimize_pic_address): Likewise.
23103 (rs6000_allocate_stack_temp): Likewise.
23104 (rs6000_address_for_fpconvert): Likewise.
23105 (rs6000_address_for_altivec): Likewise.
23106 (rs6000_secondary_memory_needed_rtx): Delete function.
23107 (rs6000_check_sdmode): Likewise.
23108 (rs6000_alloc_sdmode_stack_slot): Likewise.
23109 (rs6000_instantiate_decls): Likewise.
23110 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23111 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23112 Delete reload_in_progress.
23113 (*vec_reload_and_plus_<mptrsize>): Likewise.
23114 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23115 (vsx_div_v2di): Likewise.
23116 (vsx_udiv_v2di): Likewise.
23118 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23120 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23121 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23122 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23123 (rs6000_debug_reg_global): Delete print of LRA status.
23124 (rs6000_option_override_internal): Delete dead LRA related code.
23125 (rs6000_lra_p): Delete function.
23126 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23128 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23130 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23131 * config/riscv/rtems.h: New file.
23133 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23134 Sudakshina Das <sudi.das@arm.com>
23136 * config/aarch64/aarch64.md
23137 (define_split for and<mode>3nr_compare): Move
23138 non aarch64_logical_operand to a register.
23139 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23140 register immediate operand to a register.
23141 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23143 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23145 PR middle-end/81564
23146 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23148 2017-07-27 Richard Biener <rguenther@suse.de>
23150 PR tree-optimization/81573
23151 PR tree-optimization/81494
23152 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23153 multi defuse cycle case.
23155 2017-07-27 Richard Biener <rguenther@suse.de>
23157 PR tree-optimization/81571
23158 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23161 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23163 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23164 earlier and only if MASK_FPU is set. Adjust formatting.
23166 2017-07-27 Martin Liska <mliska@suse.cz>
23168 * opt-functions.awk: Add validation of value of Init.
23169 * optc-gen.awk: Pass new argument.
23171 2017-07-27 Martin Liska <mliska@suse.cz>
23173 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23174 Fix wrong condition.
23176 2017-07-27 Martin Liska <mliska@suse.cz>
23178 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23179 BBs and edges seen by autoFDO.
23181 2017-07-27 Richard Biener <rguenther@suse.de>
23183 PR tree-optimization/81502
23184 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23185 with incompatible but same sized type.
23186 (execute_update_addresses_taken): Likewise.
23188 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23190 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23191 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23193 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23196 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23197 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23198 Change s_operand to memory_operand.
23200 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23202 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23203 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23204 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23205 Emit instructions rather than returning an expression. Handle TFmode
23206 and KFmode by casting to TImode.
23207 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23208 (rs6000_emit_le_vsx_store): Likewise.
23209 * config/rs6000/vsx.md (VSX_TI): New iterator.
23210 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23211 (*vsx_le_undo_permute_<mode>): Likewise.
23212 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23213 emit the split sequence.
23214 (*vsx_le_perm_store_<mode>): Likewise.
23216 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23218 PR tree-optimization/81555
23219 PR tree-optimization/81556
23220 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23221 if true, force CHANGED for the recursive invocation.
23222 (reassociate_bb): Remember original length of ops array, pass
23223 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23225 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23226 attributes for noipa attribute. For naked attribute use
23227 lookup_attribute first before lookup_attribute_spec.
23228 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23230 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23231 for functions with noipa attribute.
23232 (cgraph_externally_visible_p): Return true for functions with noipa
23234 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23235 for functions with noipa attribute.
23236 * doc/extend.texi: Document noipa function attribute.
23237 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23238 also for functions with noipa attribute.
23239 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23241 2017-07-26 Andrew Pinski <apinski@cavium.com>
23243 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23244 vec_unalign_load_cost and vec_unalign_store_cost.
23246 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23248 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23249 -mvsx-small-integer option.
23250 (ISA_3_0_MASKS_IEEE): Likewise.
23251 (OTHER_VSX_VECTOR_MASKS): Likewise.
23252 (POWERPC_MASKS): Likewise.
23253 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23254 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23255 code, only testing for DImode being allowed in non-VSX floating
23257 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23258 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23260 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23261 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23262 TARGET_P8_VECTOR test.
23263 (rs6000_secondary_reload_simple_move): Likewise.
23264 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23265 since TARGET_P9_VECTOR was already tested.
23266 (rs6000_opt_masks): Remove -mvsx-small-integer.
23267 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23268 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23270 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23271 test for TARGET_VEXTRACTUB was used, and that uses
23273 (p9 extract splitter): Likewise.
23274 (vsx_extract_<mode>_di_p9): Likewise.
23275 (vsx_extract_<mode>_store_p9): Likewise.
23276 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23277 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23278 the elimination of TARGET_VSX_SMALL_INTEGER.
23279 (vsx_extract_<mode>_p8): Likewise.
23280 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23281 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23282 (vsx_set_<mode>_p9): Likewise.
23283 (vsx_set_v4sf_p9): Likewise.
23284 (vsx_set_v4sf_p9_zero): Likewise.
23285 (vsx_insert_extract_v4sf_p9): Likewise.
23286 (vsx_insert_extract_v4sf_p9_2): Likewise.
23287 * config/rs6000/rs6000.md (sign extend splitter): Change
23288 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23289 (floatsi<mode>2_lfiwax_mem): Likewise.
23290 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23291 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23292 since a test for TARGET_P9_VECTOR was used.
23293 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23294 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23295 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23296 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23297 TARGET_P8_VECTOR test.
23298 (fix_trunc<mode>si2_stfiwx): Likewise.
23299 (fix_trunc<mode>si2_internal): Likewise.
23300 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23301 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23303 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23304 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23305 TARGET_P8_VECTOR test.
23306 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23307 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23308 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23310 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23311 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23312 since a test for TARGET_P9_VECTOR was used.
23313 (splitter for loading small constants): Likewise.
23315 2017-07-26 Andrew Pinski <apinski@cavium.com>
23317 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23320 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23323 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23324 (fp_valid_at): Likewise.
23326 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23328 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23329 (qdf24xx_addrcost_table): Likewise.
23330 (cortexa57_tunings): Update to use generic_branch_cost.
23331 (cortexa72_tunings): Likewise.
23332 (cortexa73_tunings): Likewise.
23333 (qdf24xx_tunings): Likewise.
23335 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23337 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23338 (thunderx2t99_branch_cost): Likewise.
23339 (cortexa35_tunings): Update to use generic_branch_cost.
23340 (cortexa53_tunings): Likewise.
23341 (cortexa57_tunings): Likewise.
23342 (cortexa72_tunings): Likewise.
23343 (cortexa73_tunings): Likewise.
23344 (thunderx2t99_tunings): Likewise.
23346 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23348 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23349 (sparc_option_override): Honour MASK_FSMULD.
23350 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23351 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23352 * config/sparc/sparc.opt (mfsmuld): New option.
23353 * doc/invoke.texi (mfsmuld): Document option.
23355 2017-07-26 Marek Polacek <polacek@redhat.com>
23357 PR middle-end/70992
23358 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23360 2017-07-26 Richard Biener <rguenther@suse.de>
23362 * gimple-match-head.c (do_valueize): Return OP if valueize
23364 (get_def): New helper to get at the def stmt of a SSA name
23365 if valueize allows.
23366 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23367 do_valueize to get at the def stmt.
23368 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23370 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23372 PR middle-end/46932
23373 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23375 2017-07-26 Martin Liska <mliska@suse.cz>
23378 * function.c (expand_function_start): Make expansion of
23379 nonlocal_goto_save_area after parm_birth_insn.
23381 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23383 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23384 from all CPU target flags enable members.
23386 2017-07-26 Richard Biener <rguenther@suse.de>
23388 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23389 (decision_tree::gen): Make 'type' const.
23390 (write_predicate): Likewise.
23392 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23394 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23395 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23396 (rs6000_option_override_internal): Likewise.
23397 (rs6000_expand_vector_set): Likewise.
23398 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23399 (TARGET_UPPER_REGS_SF): Likewise.
23400 (TARGET_UPPER_REGS_DI): Likewise.
23401 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23402 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23403 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23404 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23405 (Splitters for DI constants in Altivec registers): Likewise.
23406 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23407 (vsx_set_v4sf_p9): Likewise.
23408 (vsx_set_v4sf_p9_zero): Likewise.
23409 (vsx_insert_extract_v4sf_p9): Likewise.
23410 (vsx_insert_extract_v4sf_p9_2): Likewise.
23412 2017-07-25 Carl Love <cel@us.ibm.com>
23414 * doc/extend.texi: Update the built-in documentation file for the
23415 existing built-in functions
23416 vector signed char vec_cnttz (vector signed char);
23417 vector unsigned char vec_cnttz (vector unsigned char);
23418 vector signed short vec_cnttz (vector signed short);
23419 vector unsigned short vec_cnttz (vector unsigned short);
23420 vector signed int vec_cnttz (vector signed int);
23421 vector unsigned int vec_cnttz (vector unsigned int);
23422 vector signed long long vec_cnttz (vector signed long long);
23423 vector unsigned long long vec_cnttz (vector unsigned long long);
23425 2017-07-25 Andrew Pinski <apinski@cavium.com>
23427 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23428 accesses where the use is for the first operand of a BIT_INSERT.
23430 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23433 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23434 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23436 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23438 * config/i386/gstabs.h: Delete.
23439 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23441 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23443 * config/i386/i386.c (ix86_decompose_address): Do not check for
23444 register RTX when looking at index_reg or base_reg.
23445 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23447 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23449 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23450 to update EH info here.
23452 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23454 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23456 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23458 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23460 2017-07-25 Torsten Duwe <duwe@suse.de>
23462 * common.opt: Introduce -fpatchable-function-entry
23463 command line option, and its variables function_entry_patch_area_size
23464 and function_entry_patch_area_start.
23465 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23466 including a two-value parser.
23467 * target.def (print_patchable_function_entry): New target hook.
23468 * targhooks.h (default_print_patchable_function_entry): New function.
23469 * targhooks.c (default_print_patchable_function_entry): Likewise.
23470 * toplev.c (process_options): Switch off IPA-RA if
23471 patchable function entries are being generated.
23472 * varasm.c (assemble_start_function): Look at the
23473 patchable-function-entry command line switch and current
23474 function attributes and maybe generate NOP instructions by
23475 calling the print_patchable_function_entry hook.
23476 * doc/extend.texi: Document patchable_function_entry attribute.
23477 * doc/invoke.texi: Document -fpatchable_function_entry
23478 command line option.
23479 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23481 * doc/tm.texi: Re-generate.
23483 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23486 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23487 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23489 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23491 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23493 2017-07-25 Richard Biener <rguenther@suse.de>
23495 PR tree-optimization/81455
23496 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23497 not walk in cycles when looking for guards.
23499 2017-07-25 Richard Biener <rguenther@suse.de>
23501 PR tree-optimization/81529
23502 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23503 when optimizing backedge uses.
23505 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23507 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23509 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23510 to dl_section_ref. On AIX, append an expression to subtract
23511 the size of the section length to dl_section_ref.
23513 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23515 * configure.ac: If any of the config.* scripts fail, exit 1.
23516 * configure: Regenerate.
23518 2017-07-25 Richard Biener <rguenther@suse.de>
23520 PR middle-end/81546
23521 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23522 of immediate uses, be more verbose on errors.
23524 2017-07-25 Richard Biener <rguenther@suse.de>
23526 PR tree-optimization/81510
23527 * tree-vect-loop.c (vect_is_simple_reduction): When the
23528 reduction stmt is not inside the loop bail out.
23530 2017-07-25 Richard Biener <rguenther@suse.de>
23532 PR tree-optimization/81303
23533 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23534 profitability check against LOOP_VINFO_NITERSM1.
23536 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23538 * domwalk.c (cmp_bb_postorder): Simplify.
23539 (sort_bbs_postorder): New function. Use it...
23540 (dom_walker::walk): ...here to optimize common cases.
23542 2017-07-25 Martin Liska <mliska@suse.cz>
23545 * ipa-visibility.c (function_and_variable_visibility): Make the
23546 redirection just on target that supports aliasing.
23547 Fix GNU coding style.
23549 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23552 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23554 * config/arm/rtems.h: Likewise.
23555 * config/bfin/rtems.h: Likewise.
23556 * config/i386/rtemself.h: Likewise.
23557 * config/lm32/rtems.h: Likewise.
23558 * config/m32c/rtems.h: Likewise.
23559 * config/m68k/rtemself.h: Likewise.
23560 * config/microblaze/rtems.h: Likewise.
23561 * config/mips/rtems.h: Likewise.
23562 * config/moxie/rtems.h: Likewise.
23563 * config/nios2/rtems.h: Likewise.
23564 * config/powerpcspe/rtems.h: Likewise.
23565 * config/rs6000/rtems.h: Likewise.
23566 * config/rtems.h: Likewise.
23567 * config/sh/rtems.h: Likewise.
23568 * config/sh/rtemself.h: Likewise.
23569 * config/sparc/rtemself.h: Likewise.
23571 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23574 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23575 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23576 * tree-ssa-structalias.c (alias_get_name): Same.
23578 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23581 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23582 instructions if no du chain is found.
23584 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23586 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23588 2017-07-25 Richard Biener <rguenther@suse.de>
23590 PR middle-end/81505
23591 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23594 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23596 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23597 upper-regs options.
23598 (ISA_2_7_MASKS_SERVER): Likewise.
23599 (ISA_3_0_MASKS_IEEE): Likewise.
23600 (OTHER_P8_VECTOR_MASKS): Likewise.
23601 (OTHER_VSX_VECTOR_MASKS): Likewise.
23602 (POWERPC_MASKS): Likewise.
23603 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23604 duplicate list of options.
23605 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23606 explicit -mupper-regs options.
23607 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23608 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23609 alias for -mupper-regs-df.
23610 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23611 (rs6000_init_hard_regno_mode_ok): Likewise.
23612 (rs6000_option_override_internal): Likewise.
23613 (rs6000_opt_masks): Likewise.
23614 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23615 options in terms of whether -mvsx or -mpower8-vector was used.
23616 (TARGET_UPPER_REGS_DI): Likewise.
23617 (TARGET_UPPER_REGS_SF): Likewise.
23618 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23619 -mupper-regs-* options.
23621 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23623 * passes.c (emergency_dump_function): Print some empty lines and a
23624 header before the RTL dump.
23626 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23628 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23630 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23633 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23634 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23636 2017-07-24 Carl Love <cel@us.ibm.com>
23638 * config/rs6000/rs6000-c.c: Add support for built-in functions
23639 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23640 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23641 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23642 vec_extract_fp_from_shortl): Add defines for the two builtins.
23643 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23644 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23646 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23647 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23648 * doc/extend.texi: Update the built-in documentation file for the
23649 new built-in function.
23651 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23654 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23656 * doc/generic.texi: Likewise.
23657 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23658 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23660 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23662 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23663 (aarch64_mls_elt_merge<mode>): Likewise.
23665 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23667 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23668 having __cxa_atexit.
23670 2017-07-23 Michael Collison <michael.collison@arm.com>
23672 * config/arm/arm.c (arm_option_override): Deprecate
23673 use of -mstructure-size-boundary.
23674 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23675 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23677 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23680 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23681 Reduce cost estimate for direct moves.
23683 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23686 * config/i386/i386.c (ix86_option_override_internal): Disable
23687 BMI, BMI2 and TBM instructions for -m16.
23689 2017-07-21 Carl Love <cel@us.ibm.com>
23691 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23692 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23693 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23694 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23695 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23696 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23697 VMULOSW): New enum "unspec" values.
23698 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23699 altivec_vmulosw): New patterns.
23700 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23701 VMULOSW): Add definitions.
23703 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23705 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23706 (qdf24xx): Likewise.
23707 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23708 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23709 (AARCH64_FL_V8_1): Renumber.
23710 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23711 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23712 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23713 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23714 rdma to feature modifiers list.
23716 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23718 PR middle-end/56727
23719 * ipa-visibility (function_and_variable_visibility): Convert
23720 recursive PLT call to direct call if appropriate.
23722 2017-07-21 Andrew Pinski <apinski@cavium.com>
23724 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23725 operand 1 to see if the types precision matches.
23726 * fold-const.c (operand_equal_p): Likewise.
23728 2017-07-21 Richard Biener <rguenther@suse.de>
23730 PR tree-optimization/81303
23731 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23732 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23733 (vect_peeling_hash_get_lowest_cost): Adjust.
23734 (vect_enhance_data_refs_alignment): Likewise. Use
23735 vect_get_peeling_costs_all_drs to compute the penalty for no
23736 peeling to match up costs.
23738 2017-07-21 Richard Biener <rguenther@suse.de>
23740 PR tree-optimization/81500
23741 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23742 we didn't identify a reduction path.
23744 2017-07-21 Tom de Vries <tom@codesourcery.com>
23745 Cesar Philippidis <cesar@codesourcery.com>
23747 PR gcov-profile/81442
23748 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23751 2017-07-21 Tom de Vries <tom@codesourcery.com>
23754 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23756 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23757 nvptx_override_options_after_change.
23759 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23761 * dwarf2out.c (output_file_names): Avoid double testing for
23762 dwarf_version >= 5.
23764 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23766 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23768 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23770 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23772 (try_crossjump_to_edge): Do not punt on partitioned functions.
23774 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23776 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23777 Put all BBs reachable only via paths crossing cold region to cold
23779 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23781 2016-07-21 Richard Biener <rguenther@suse.de>
23783 PR tree-optimization/81303
23784 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23785 into account prologue and epilogue iterations when raising
23786 min_profitable_iters to sth at least covering one vector iteration.
23788 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23790 * config/arm/arm.c (arm_test_cpu_arch_dat):
23793 2017-07-20 Nathan Sidwell <nathan@acm.org>
23795 Remove TYPE_METHODS.
23796 * tree.h (TYPE_METHODS): Delete.
23797 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23798 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23799 (dbxout_type_methods): Scan TYPE_FIELDS.
23800 (dbxout_type): Don't check TYPE_METHODS here.
23801 * function.c (use_register_for_decl): Always ignore register for
23802 class types when not optimizing.
23803 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23804 * tree.c (free_lang_data_in_type): Stitch out member functions and
23805 templates from TYPE_FIELDS.
23806 (build_distinct_type_copy, verify_type_variant,
23807 verify_type): Member fns are on TYPE_FIELDS.
23808 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23809 * tree-pretty-print.c (dump_generic_node): Likewise.
23811 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23814 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23815 V2TImode and V4TImode.
23816 (ix86_expand_vector_extract): Likewise.
23817 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23818 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23819 (ssescalarmode): Handle V4TImode and V2TImode.
23820 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23821 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23822 (VEXTRACTI128_MODE): New mode iterator.
23823 (splitter for *vec_extractv?ti first element): New.
23824 (VEC_INIT_MODE): New mode iterator.
23825 (vec_init<mode>): Consolidate 3 expanders into one using
23826 VEC_INIT_MODE mode iterator.
23828 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23830 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23831 non_spilled_static_chain_regno_p.
23833 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23835 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23837 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23839 * bb-reorder.c (connect_traces): Allow copying of blocks within
23842 2017-07-20 Richard Biener <rguenther@suse.de>
23844 * gimple.h (gimple_phi_result): Add gphi * overload.
23845 (gimple_phi_result_ptr): Likewise.
23846 (gimple_phi_arg): Likewise. Adjust index assert to only
23847 allow actual argument accesses rather than all slots available
23849 (gimple_phi_arg_def): Add gphi * overload.
23850 * tree-phinodes.c (make_phi_node): Initialize only actual
23852 (resize_phi_node): Clear memory not covered by old node,
23853 do not initialize excess argument slots.
23854 (reserve_phi_args_for_new_edge): Initialize new argument slot
23857 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23859 PR tree-optimization/81388
23861 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23863 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23864 by removing computation of may_be_zero.
23866 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23867 Tom de Vries <tom@codesourcery.com>
23869 PR middle-end/81030
23870 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23871 when gimple level profile disagrees with what RTL expander did.
23873 2017-07-20 Richard Biener <rguenther@suse.de>
23875 PR tree-optimization/61171
23876 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23877 (vect_analyze_stmt): Add slp instance parameter.
23878 (vectorizable_reduction): Likewise.
23879 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23880 (vect_is_simple_reduction): Deal with chains not detected
23881 as SLP reduction chain, specifically not properly associated
23882 chains containing a mix of plus/minus.
23883 (get_reduction_op): Remove.
23884 (get_initial_defs_for_reduction): Simplify, pass in whether
23885 this is a reduction chain, pass in the SLP node for the PHIs.
23886 (vect_create_epilog_for_reduction): Get the SLP instance as
23888 (vectorizable_reduction): Get the SLP instance as arg.
23889 During analysis remember the SLP node with the PHIs in the
23890 instance. Simplify getting at the vectorized reduction PHIs.
23891 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23892 through SLP instance.
23893 (vect_slp_analyze_operations): Likewise.
23894 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23895 (vect_transform_stmt): Likewise.
23897 2017-07-20 Tom de Vries <tom@codesourcery.com>
23899 PR tree-optimization/81489
23900 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23901 read of phi arg location to before loop that modifies phi.
23903 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23905 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23908 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23910 PR middle-end/81331
23911 * except.c (execute): Fix ordering issue.
23913 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23915 PR rtl-optimization/81423
23916 * combine.c (make_compound_operation_int): Don't try to optimize
23917 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23919 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23921 PR rtl-optimization/81423
23922 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23923 with a constant that is -1 in the truncated to mode.
23925 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23927 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23928 (determine_unlikely_bbs): ... here.
23929 * predict.h (propagate_unlikely_bbs_forward): Declare.
23930 * cfgexpand.c (pass_expand::execute): Use it.
23931 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23933 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23934 propagate_unlikely_bbs_forward.
23936 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23938 PR middle-end/81331
23939 * except.c (maybe_add_nop_after_section_switch): New function.
23942 2017-07-19 Tom de Vries <tom@codesourcery.com>
23944 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23946 2017-07-19 Tom de Vries <tom@codesourcery.com>
23948 * gimple.h (gimple_phi_arg): Make assert more strict.
23950 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23952 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23953 * config/rs6000/mmintrin.h: New file.
23954 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23956 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23958 PR tree-optimization/81346
23959 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23961 2017-07-19 Tom de Vries <tom@codesourcery.com>
23963 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23965 2017-07-19 Tom de Vries <tom@codesourcery.com>
23967 * config/nvptx/nvptx-modes.def: Add V2DImode.
23968 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23969 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23970 (nvptx_output_mov_insn): Handle lack of mov.b128.
23971 (nvptx_print_operand): Handle 'H' and 'L' codes.
23972 (nvptx_vector_mode_supported): Allow V2DImode.
23973 (nvptx_preferred_simd_mode): New function.
23974 (nvptx_data_alignment): New function.
23975 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23976 nvptx_preferred_simd_mode.
23977 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23979 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23981 2017-07-19 Tom de Vries <tom@codesourcery.com>
23983 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23984 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23985 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23986 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23987 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23988 (mov<VECIM>_insn): New define_insn.
23989 (define_expand "mov<VECIM>): New define_expand.
23991 2017-07-19 Tom de Vries <tom@codesourcery.com>
23993 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23995 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23997 PR tree-optimization/81346
23998 * fold-const.h (fold_div_compare, range_check_type): Declare.
23999 * fold-const.c (range_check_type): New function.
24000 (build_range_check): Use range_check_type.
24001 (fold_div_compare): No longer static, rewritten into
24002 a match.pd helper function.
24003 (fold_comparison): Don't call fold_div_compare here.
24004 * match.pd (X / C1 op C2): New optimization using fold_div_compare
24005 as helper function.
24007 2017-07-19 Nathan Sidwell <nathan@acm.org>
24009 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
24010 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
24011 * tree.c (find_decls_types_r, verify_type): Use
24012 TYPE_{MIN,MAX}_VALUE_RAW.
24013 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
24014 (hash_tree): Likewise.
24015 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
24017 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
24020 2017-07-18 Tom de Vries <tom@codesourcery.com>
24022 PR middle-end/81464
24023 * omp-expand.c (expand_omp_for_static_chunk): Handle
24024 equal-argument loop exit phi.
24026 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
24029 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
24030 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
24031 operand 2 predicate.
24032 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
24033 operand 2 predicate.
24034 (ror,rol -> rorx splitters): Use const_int_operand as
24035 operand 2 predicate.
24037 2017-06-18 Richard Biener <rguenther@suse.de>
24039 PR tree-optimization/81410
24040 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
24041 the gap in the ! slp_perm SLP case after each group.
24043 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24045 PR middle-end/81463
24046 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
24049 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24051 PR middle-end/81462
24052 * predict.c (set_even_probabilities): Cleanup; do not affect
24053 probabilities that are already known.
24054 (combine_predictions_for_bb): Call even when count is set.
24056 2017-07-18 Nathan Sidwell <nathan@acm.org>
24058 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24061 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24064 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24065 optimization for loop niter analysis.
24067 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24070 * config/avr/avr.c (avr_optimize_casesi): Don't use
24071 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24073 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24075 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24076 body_cost_vec from _vect_peel_extended_info.
24077 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24078 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24081 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24083 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24085 2017-07-18 Richard Biener <rguenther@suse.de>
24087 PR tree-optimization/80620
24088 PR tree-optimization/81403
24089 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24090 info when re-using a VN table entry.
24092 2017-07-18 Richard Biener <rguenther@suse.de>
24094 PR tree-optimization/81418
24095 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24096 vectype_in. Verify that with lane-reducing reduction operations
24097 we have a single def-use cycle.
24099 2017-07-17 Carl Love <cel@us.ibm.com>
24101 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24103 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24104 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24105 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24106 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24107 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24108 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24109 VMULOSW): New enum "unspec" values.
24110 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24111 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24112 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24113 altivec_vmulosw): New patterns.
24114 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24115 VMULOSW): Add definitions.
24117 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24119 * config/alpha/alpha.c: Include predict.h.
24121 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24123 * tree-vrp.c (compare_assert_loc): Fix comparison function
24124 to return predictable results.
24126 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24128 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24130 (subdi3): Likewise.
24131 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24132 * doc/invoke.texi (mexpand-adddi): Update text.
24134 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24136 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24137 that also clobbers the CC register. The old expand code is moved
24139 (*arc_clzsi2): ... here.
24140 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24141 the CC register. The old expand code is moved to ...
24142 (arc_ctzsi2): ... here.
24144 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24146 * config/arc/arc.opt (mindexed-loads): Use initial value
24147 TARGET_INDEXED_LOADS_DEFAULT.
24148 (mauto-modify-reg): Use initial value
24149 TARGET_AUTO_MODIFY_REG_DEFAULT.
24150 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24151 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24152 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24153 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24155 2017-07-17 Martin Liska <mliska@suse.cz>
24158 * opts.c (finish_options): Do not allow -fgnu-tm
24159 w/ -fsanitize={kernel-,}address. Say sorry.
24161 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24164 * tree-loop-distribution.c (classify_partition): Only assert on
24165 numer of iterations.
24166 (merge_dep_scc_partitions): Delete prameter. Update function call.
24167 (distribute_loop): Remove code handling loop with unknown niters.
24168 (pass_loop_distribution::execute): Skip loop with unknown niters.
24170 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24173 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24174 function sort_partitions_by_post_order.
24176 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24178 PR tree-optimization/81374
24179 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24180 the max index of basic blocks, rather than number of basic blocks.
24182 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24184 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24186 (arc_legitimate_pic_operand_p): Likewise.
24187 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24189 (arc_needs_pcl_p): Likewise.
24190 (arc_legitimate_pc_offset_p): Likewise.
24191 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24192 function is also used in constrains.md.
24193 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24194 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24195 PLUS. Only return true/false in known cases, otherwise assert.
24196 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24197 is already called in arc_legitimate_constant_p.
24198 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24200 (LEGITIMATE_PIC_OPERAND_P): Use
24201 arc_raw_symbolic_reference_mentioned_p function.
24202 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24207 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24208 Andrew Burgess <andrew.burgess@embecosm.com>
24210 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24211 (arc_return_address_register): New function.
24212 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24213 (arc_handle_fndecl_attribute): Add naked attribute.
24214 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24215 (TARGET_WARN_FUNC_RETURN): Likewise.
24216 (arc_allocate_stack_slots_for_args): New function.
24217 (arc_warn_func_return): Likewise.
24218 (machine_function): Change type fn_type.
24219 (arc_compute_function_type): Consider new naked function type,
24220 change function return type.
24221 (arc_must_save_register): Adapt to handle new
24222 arc_compute_function_type's return type.
24223 (arc_expand_prologue): Likewise.
24224 (arc_expand_epilogue): Likewise.
24225 (arc_return_address_regs): Delete.
24226 (arc_return_address_register): New function.
24227 (arc_epilogue_uses): Use above function.
24228 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24229 (arc_function_type): Change encoding, add naked type.
24230 (ARC_INTERRUPT_P): Change to handle the new encoding.
24231 (ARC_FAST_INTERRUPT_P): Likewise.
24232 (ARC_NORMAL_P): Define.
24233 (ARC_NAKED_P): Likewise.
24234 (arc_compute_function_type): Delete prototype.
24235 * config/arc/arc.md (in_ret_delay_slot): Use
24236 arc_return_address_register function.
24237 (simple_return): Likewise.
24238 (p_return_i): Likewise.
24240 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24242 PR tree-optimization/81428
24243 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24244 can't be built for those types.
24246 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24248 Remove stuff dead since r239246.
24250 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24251 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24252 (avr_inform_devices): Remove dead stuff.
24254 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24256 * config/arm/arm_neon.h: Fix softp typo.
24258 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24260 PR tree-optimization/81365
24261 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24262 aggregate moves onto bb predecessor edges, make sure there are no
24263 loads that could alias the lhs in between the start of bb and the
24266 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24269 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24270 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24271 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24273 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24275 PR tree-optimization/81396
24276 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24277 (init_symbolic_number): Initialize it to 1.
24278 (perform_symbolic_merge): Add n_ops from both operands into the new
24280 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24281 without base_addr as useless if they need more than one operation.
24282 (bswap_replace): Handle !bswap case for NULL base_addr.
24284 2017-07-17 Tom de Vries <tom@codesourcery.com>
24287 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24290 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24292 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24293 conditional builtin define __FIX_LEON3FT_B2BST.
24295 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24297 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24298 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24301 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24303 PR rtl-optimization/81424
24304 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24305 to remove potential trapping from operands if -fnon-call-exceptions.
24307 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24309 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24310 profile_proability for scalling.
24311 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24313 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24315 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24317 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24319 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24320 fixpoint arithmetics.
24322 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24324 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24325 fixpoint arithmetics.
24327 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24329 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24330 fixpoint arithmetics.
24332 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24334 * profile-count.h (profile_probability::from_reg_br_prob_note,
24335 profile_probability::to_reg_br_prob_note): New functions.
24336 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24337 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24338 * predict.c (probability_reliable_p): Update.
24339 (edge_probability_reliable_p): Update.
24340 (br_prob_note_reliable_p): Update.
24341 (invert_br_probabilities): Update.
24342 (add_reg_br_prob_note): New function.
24343 (combine_predictions_for_insn): Update.
24344 * asan.c (asan_clear_shadow): Update.
24345 * cfgbuild.c (compute_outgoing_frequencies): Update.
24346 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24347 (update_br_prob_note): Update.
24348 (rtl_verify_edges): Update.
24349 (purge_dead_edges): Update.
24350 (fixup_reorder_chain): Update.
24351 * emit-rtl.c (try_split): Update.
24352 * ifcvt.c (cond_exec_process_insns): Update.
24353 (cond_exec_process_if_block): Update.
24354 (dead_or_predicable): Update.
24355 * internal-fn.c (expand_addsub_overflow): Update.
24356 (expand_neg_overflow): Update.
24357 (expand_mul_overflow): Update.
24358 * loop-doloop.c (doloop_modify): Update.
24359 * loop-unroll.c (compare_and_jump_seq): Update.
24360 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24361 * predict.h: Update.
24362 * reorg.c (mostly_true_jump): Update.
24364 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24365 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24366 * config/arc/arc.c: (emit_unlikely_jump): Update.
24367 * config/arm/arm.c: (emit_unlikely_jump): Update.
24368 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24369 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24370 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24371 (ix86_print_operand): Update.
24372 (ix86_split_fp_branch): Update.
24373 (predict_jump): Update.
24374 * config/ia64/ia64.c (ia64_print_operand): Update.
24375 * config/mmix/mmix.c (mmix_print_operand): Update.
24376 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24377 (rs6000_expand_split_stack_prologue): Update.
24378 * config/rs6000/rs6000.c: Update.
24379 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24380 (s390_expand_vec_movstr): Update.
24381 (s390_expand_cs_tdsi): Update.
24382 (s390_expand_split_stack_prologue): Update.
24383 * config/sh/sh.c (sh_print_operand): Update.
24384 (expand_cbranchsi4): Update.
24385 (expand_cbranchdi4): Update.
24386 * config/sparc/sparc.c (output_v9branch): Update.
24387 * config/spu/spu.c (get_branch_target): Update.
24388 (ea_load_store_inline): Update.
24389 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24390 * config/tilepro/tilepro.c: Update.
24392 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24394 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24395 (gimplify_save_expr): Likewise.
24397 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24399 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24401 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24403 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24405 * timevar.def (TV_IPA_FNSUMMARY): Define.
24407 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24409 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24410 to back store errata sensitive sequence from being generated.
24411 (sqrtdf2_fix): Likewise.
24413 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24415 * tree-ssa-threadupdate.c (compute_path_counts,
24416 update_joiner_offpath_counts): Use profile_probability.
24418 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24421 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24423 * config/arm/arm-c.c (arm_cpu_builtins): Define
24424 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24426 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24428 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24429 array entries to represent __ieee128 versions of the
24430 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24431 scalar_extract_sig, and scalar_insert_exp built-in functions.
24432 (altivec_resolve_overloaded_builtin): Add special case handling
24433 for the __builtin_scalar_insert_exp function, as represented by
24434 the P9V_BUILTIN_VEC_VSIEDP constant.
24435 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24436 exponent support for __ieee128 argument.
24437 (VSESQP): Add scalar extract signature support for __ieee128
24439 (VSTDCNQP): Add scalar test negative support for __ieee128
24441 (VSIEQP): Add scalar insert exponent support for __int128 argument
24442 with __ieee128 result.
24443 (VSIEQPF): Add scalar insert exponent support for __ieee128
24444 argument with __ieee128 result.
24445 (VSTDCQP): Add scalar test data class support for __ieee128
24447 (VSTDCNQP): Add overload support for scalar test negative with
24448 __ieee128 argument.
24449 (VSTDCQP): Add overload support for scalar test data class
24450 __ieee128 argument.
24451 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24452 UNSPEC_VSX_SXSIGDP.
24453 (UNSPEC_VSX_SIEXPQP): New constant.
24454 (xsxexpqp): New insn for VSX scalar extract exponent quad
24456 (xsxsigqp): New insn for VSX scalar extract significand quad
24458 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24459 precision with floating point argument.
24460 (xststdcqp): New expand for VSX scalar test data class quad
24462 (xststdcnegqp): New expand for VSX scalar test negative quad
24464 (xststdcqp): New insn to match expansions for VSX scalar test data
24465 class quad precision and VSX scalar test negative quad precision.
24466 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24467 special case operand checking to enforce that second operand of
24468 VSX scalar test data class with quad precision argument is a 7-bit
24470 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24471 prototypes and descriptions of __ieee128 versions of
24472 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24473 scalar_test_data_class, and scalar_test_neg built-in functions.
24475 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24477 PR tree-optimization/81162
24478 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24479 replace a negate with an add.
24481 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24483 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24485 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24487 * config/arm/arm-c.c (arm_cpu_builtins): Define
24488 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24490 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24492 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24493 (armv8-r): Set ARM Cortex-R52 as default CPU.
24494 * config/arm/arm-tables.opt: Regenerate.
24495 * config/arm/arm-tune.md: Regenerate.
24496 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24498 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24499 extension for -mcpu=cortex-r52.
24501 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24503 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24504 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24505 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24506 (fp-armv8): Define it as FP_ARMv8 only.
24507 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24508 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24510 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24511 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24512 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24513 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24514 than TARGET_FPU_ARMV8.
24515 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24516 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24517 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24519 * config/arm/neon.md (neon_vrint): Likewise.
24520 (neon_vcvt): Likewise.
24521 (neon_<fmaxmin_op><mode>): Likewise.
24522 (<fmaxmin><mode>3): Likewise.
24523 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24524 * config/arm/predicates.md (arm_cond_move_operator): Check against
24525 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24527 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24529 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24530 to top of function.
24532 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24534 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24535 loop in comment with memset.
24537 2017-07-14 Martin Liska <mliska@suse.cz>
24539 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24540 * dwarf2out.c (is_java): Remove the function.
24541 (output_pubname): Remove usage of the function.
24542 (lower_bound_default): Remove usage of DW_LANG_Java.
24543 (gen_compile_unit_die): Likewise.
24544 * gcc.c: Remove compiler defaults for .java and .zip files.
24545 * gimple-expr.c (remove_suffix): Change as there's no longer
24546 extension than 4-letter one.
24547 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24548 (gimplify_save_expr): Likewise.
24549 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24550 as it's possible even for other languages than Java.
24551 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24552 * lto-opts.c (lto_write_options): Remove reference to Java.
24553 * opts.c (strip_off_ending): Update file extension handling.
24554 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24555 * tree-eh.c (lower_resx): Likewise.
24556 * tree.c (free_lang_data_in_type): Remove dead code.
24557 (find_decls_types_r): Likewise.
24558 (build_common_builtin_nodes): Remove Java from a comment.
24559 (verify_type): Remove dead code.
24560 * varasm.c (assemble_external): Remove Java from a comment.
24562 2017-07-14 Martin Liska <mliska@suse.cz>
24564 * opts.c (finish_options): Add quotes.
24565 (common_handle_option): Likewise.
24567 2017-07-14 Martin Liska <mliska@suse.cz>
24569 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24570 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24571 Remove N_SO_PASCAL.
24572 * dwarf2out.c (lower_bound_default): Do not handle
24574 (gen_compile_unit_die): Likewise.
24575 * gcc.c: Remove default extension binding for GNU Pascal.
24576 * stmt.c: Remove Pascal language from a comment.
24577 * xcoffout.c: Likewise.
24579 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24582 * diagnostic-show-locus.c (fixit_cmp): New function.
24583 (layout::layout): Sort m_fixit_hints.
24584 (column_range::column_range): Assert that the values are valid.
24585 (struct char_span): New struct.
24586 (correction::overwrite): New method.
24587 (struct source_line): New struct.
24588 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24589 calls in terms of classes source_line and char_span, and
24590 correction::overwrite.
24591 (selftest::test_overlapped_fixit_printing_2): New function.
24592 (selftest::diagnostic_show_locus_c_tests): Call it.
24594 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24596 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24597 early if there is no lhs.
24599 2017-07-13 Martin Liska <mliska@suse.cz>
24601 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24602 (gen_reference_type_die): Likewise.
24603 * stor-layout.c: Remove Pascal-related comment.
24605 2017-07-13 Martin Liska <mliska@suse.cz>
24607 * opts.c (finish_options): Add quotes to error messages.
24608 (parse_sanitizer_options): Likewise.
24610 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24612 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24614 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24616 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24618 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24620 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24622 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24624 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24627 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24628 provides the hardware capability bits, define the macro
24629 __BUILTIN_CPU_SUPPORTS__.
24630 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24631 if GLIBC does not provide the hardware capability bits. Add a
24632 gcc_unreachable call if the built-in cpu function is neither
24633 __builtin_cpu_is nor __builtin_cpu_supports.
24634 (rs6000_get_function_versions_dispatcher): Change the warning
24635 that an old GLIBC is used which does not export the capability
24636 bits to be an error.
24637 * doc/extend.texi (target_clones attribute): Document the
24638 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24639 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24640 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24641 the macros defined by GCC if the newer GLIBC is available.
24643 2017-07-12 Jeff Law <law@redhat.com>
24645 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24646 remaining includes slightly.
24647 * config/riscv/riscv-builtins.c: Include profile-count.h.
24649 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24652 * config/avr/avr.c (avr_set_current_function): In diagnostic
24653 messages: Quote keywords and (parts of) identifiers.
24654 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24657 2017-07-12 Carl Love <cel@us.ibm.com>
24659 * config/rs6000/rs6000-c.c: Add support for built-in functions
24660 vector bool char vec_revb (vector bool char);
24661 vector bool short vec_revb (vector short char);
24662 vector bool int vec_revb (vector bool int);
24663 vector bool long long vec_revb (vector bool long long);
24664 * doc/extend.texi: Update the built-in documentation file for the
24665 new built-in functions.
24667 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24669 * config/s390/s390.md: Remove movcc splitter.
24671 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24673 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24674 load/store on condition.
24676 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24679 * config/avr/avr.c (avr_encode_section_info)
24680 [progmem && !TREE_READONLY]: Error if progmem object needs
24683 2017-07-11 Michael Collison <michael.collison@arm.com>
24685 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24688 2017-07-11 Carl Love <cel@us.ibm.com>
24690 * config/rs6000/rs6000-c.c: Add support for builtins
24691 vector unsigned int vec_parity_lsbb (vector signed int);
24692 vector unsigned int vec_parity_lsbb (vector unsigned int);
24693 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24694 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24695 vector unsigned long long vec_parity_lsbb (vector signed long long);
24696 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24697 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24698 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24699 * doc/extend.texi: Update the built-in documentation file for the
24700 new built-in functions.
24702 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24704 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24705 (layout::m_primary_loc): New field.
24706 (layout::layout): Initialize new field. Move location filtering
24707 logic from here to...
24708 (layout::maybe_add_location_range): ...this new method. Add
24709 support for filtering to just the lines already specified by other
24711 (layout::will_show_line_p): New method.
24712 (gcc_rich_location::add_location_if_nearby): New method.
24713 (selftest::test_add_location_if_nearby): New test function.
24714 (selftest::diagnostic_show_locus_c_tests): Call it.
24715 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24718 2017-07-11 Tom de Vries <tom@codesourcery.com>
24720 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24721 (bb_first_real_insn): New function.
24722 (nvptx_single): Add extra initialization of broadcasted condition
24725 2017-07-11 Nathan Sidwell <nathan@acm.org>
24727 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24729 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24731 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24732 Binutils doc as TEXI will mess them up.
24733 * doc/invoke.texi (AVR Options): Same here.
24735 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24737 * config/sparc/sparc.opt (mfix-ut700): New option.
24738 (mfix-gr712rc): Likewise.
24739 (sparc_fix_b2bst): New variable.
24740 * doc/invoke.texi (SPARC options): Document them.
24741 (ARM options): Fix warnings.
24742 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24743 instructions to prevent sequences that can trigger the store-store
24744 errata for certain LEON3FT processors.
24745 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24746 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24747 * config/sparc/sparc.md (fix_b2bst): New attribute.
24748 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24750 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24753 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24755 (*rsqrtsf2_sse): Ditto.
24757 (div<mode>3): Macroize insn from divdf3 and divsf3
24758 using MODEF mode iterator.
24760 2017-07-10 Martin Sebor <msebor@redhat.com>
24762 PR tree-optimization/80397
24763 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24764 instead of testing for equality to INTEGER_TYPE.
24766 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24768 * config.gcc: Remove uclibc from arc target spec.
24770 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24772 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24774 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24777 * lto-wrapper.c (remove_option): New function.
24778 (merge_and_complain): Merge PIC/PIE options more realistically.
24780 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24782 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24786 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24787 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24788 * config.in: Regenerate.
24789 * configure: Regenerate.
24790 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24791 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24792 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24793 (TARGET_GASISR_PROLOGUES): ...target mask.
24794 * common/config/avr/avr-common.c
24795 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24796 Set -mgas-isr-prologues.
24797 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24798 INSERT_PASS_BEFORE for it.
24799 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24800 * config/avr/avr.c (avr_option_override)
24801 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24802 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24803 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24804 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24805 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24807 (make_avr_pass_pre_proep): New function.
24808 (emit_push_sfr) <treg>: Add argument to function and use it
24809 instead of TMP_REG.
24810 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24811 and set machine->gasisr.yes.
24812 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24813 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24814 __gcc_isr.n_pushed to .L__stack_usage.
24815 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24816 (avr_asm_final_postscan_insn): ...this new static function.
24817 * config/avr/avr.h (machine_function)
24818 <is_no_gccisr, use_L__stack_usage>: New fields.
24819 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24820 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24821 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24822 (gasisr, *gasisr): New expander and insn.
24823 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24824 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24825 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24827 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24829 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24832 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24834 Move jump-tables out of .text again.
24837 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24838 (ASM_OUTPUT_ADDR_VEC): New function.
24839 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24840 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24841 INSN_ADDRESSes as asm comment.
24842 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24843 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24844 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24845 * config/avr/avr.md (*tablejump): Adjust comment.
24846 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24847 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24849 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24850 (avr_output_addr_vec): New proto.
24851 (avr_log_t) <insn_addresses>: New field.
24853 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24856 * config/i386/i386.c (ix86_function_arg_advance): Set
24857 outgoing_args_on_stack to true if there are outgoing arguments
24859 (ix86_function_arg): Likewise.
24860 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24861 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24862 * config/i386/i386.h (machine_function): Add
24863 outgoing_args_on_stack.
24865 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24867 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24868 supporting pthreds.
24869 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24871 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24873 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24874 (REAL_H): Remove $(MACHMODE_H).
24875 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24877 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24878 $(MACHMODE_H) and double-int.h.
24879 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24881 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24884 2017-07-07 Andrew Pinski <apinski@cavium.com>
24886 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24887 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24889 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24891 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24892 Add warning if GCC was not configured to link against a GLIBC that
24893 exports the hardware capability bits.
24894 (make_resolver_func): Make resolver function private and not a
24895 COMDAT function. Create the name with clone_function_name instead
24896 of make_unique_name.
24899 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24900 correct operand in doing the split.
24902 2017-07-07 Carl Love <cel@us.ibm.com>
24904 * config/rs6000/rs6000-c: Add support for built-in function
24905 vector unsigned short vec_pack_to_short_fp32 (vector float,
24907 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24908 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24909 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24910 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24911 (convert_4f32_8i16): Add define_expand.
24912 * doc/extend.texi: Update the built-in documentation file for the
24913 new built-in function.
24915 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24917 * config/sparc/m8.md: New file.
24918 * config/sparc/sparc.md: Include m8.md.
24920 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24922 * config/sparc/sparc.opt: New option -mvis4b.
24923 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24924 (sparc_option_override): Handle VIS4B.
24925 (enum sparc_builtins): Define
24926 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24927 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24928 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24929 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24930 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24931 (check_constant_argument): New function.
24932 (sparc_vis_init_builtins): Define builtins
24933 __builtin_vis_dictunpack{8,16,32},
24934 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24935 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24936 __builtin_vis_fpcmpde{8,16,32}shl and
24937 __builtin_vis_fpcmpur{8,16,32}shl.
24938 (sparc_expand_builtin): Check that the constant operands to
24939 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24940 constant and in range.
24941 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24943 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24944 (SPARC_IMM5_P): Likewise.
24945 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24946 (enabled): Handle vis4b.
24947 (UNSPEC_DICTUNPACK): New unspec.
24948 (UNSPEC_FPCMPSHL): Likewise.
24949 (UNSPEC_FPUCMPSHL): Likewise.
24950 (UNSPEC_FPCMPDESHL): Likewise.
24951 (UNSPEC_FPCMPURSHL): Likewise.
24952 (cpu_feature): New CPU feature `vis4b'.
24953 (dictunpack{8,16,32}): New insns.
24954 (FPCSMODE): New mode iterator.
24955 (fpcscond): New code iterator.
24956 (fpcsucond): Likewise.
24957 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24958 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24959 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24960 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24961 * config/sparc/constraints.md: Define constraints `q' for unsigned
24962 2-bit integer constants and `t' for unsigned 5-bit integer
24964 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24966 (imm5_operand_dictunpack16): Likewise.
24967 (imm5_operand_dictunpack32): Likewise.
24968 (imm2_operand): Likewise.
24969 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24970 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24971 ditunpack* and fpcmp*shl builtins.
24973 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24975 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24976 * config.in: Add HAVE_AS_SPARC6 define.
24977 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24979 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24981 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24982 (CPP_CPU_SPEC): Handle m8.
24983 (ASM_CPU_SPEC): Likewise.
24984 * config/sparc/sparc-opts.h (enum processor_type): Add
24986 * config/sparc/sparc.c (m8_costs): New struct.
24987 (sparc_option_override): Handle TARGET_CPU_m8.
24988 (sparc32_initialize_trampoline): Likewise.
24989 (sparc64_initialize_trampoline): Likewise.
24990 (sparc_issue_rate): Likewise.
24991 (sparc_register_move_cost): Likewise.
24992 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24993 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24994 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24995 (CPP_CPU_SPEC): Handle M8.
24996 (ASM_CPU_SPEC): Likewise.
24997 (AS_M8_FLAG): Define.
24998 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24999 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
25000 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
25002 * configure: Regenerate.
25003 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
25006 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25008 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
25010 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
25011 ("*movdi_insn_sp32"): Do not set v3pipe.
25012 ("*movsi_insn"): Likewise.
25013 ("*movdi_insn_sp64"): Likewise.
25014 ("*movsf_insn"): Likewise.
25015 ("*movdf_insn_sp32"): Likewise.
25016 ("*movdf_insn_sp64"): Likewise.
25017 ("*zero_extendsidi2_insn_sp64"): Likewise.
25018 ("*sign_extendsidi2_insn"): Likewise.
25019 ("*mov<VM32:mode>_insn"): Likewise.
25020 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25021 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25022 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25023 ("<vlop:code><VL:mode>3"): Likewise.
25024 ("*not_<vlop:code><VL:mode>3"): Likewise.
25025 ("*nand<VL:mode>_vis"): Likewise.
25026 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
25027 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
25028 ("one_cmpl<VL:mode>2"): Likewise.
25029 ("faligndata<VM64:mode>_vis"): Likewise.
25030 ("alignaddrsi_vis"): Likewise.
25031 ("alignaddrdi_vis"): Likweise.
25032 ("alignaddrlsi_vis"): Likewise.
25033 ("alignaddrldi_vis"): Likewise.
25034 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25035 ("bmaskdi_vis"): Likewise.
25036 ("bmasksi_vis"): Likewise.
25037 ("bshuffle<VM64:mode>_vis"): Likewise.
25038 ("cmask8<P:mode>_vis"): Likewise.
25039 ("cmask16<P:mode>_vis"): Likewise.
25040 ("cmask32<P:mode>_vis"): Likewise.
25041 ("pdistn<P:mode>_vis"): Likewise.
25042 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25044 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25046 * config/sparc/sparc.md ("subtype"): New insn attribute.
25047 ("*wrgsr_sp64"): Set insn subtype.
25048 ("*rdgsr_sp64"): Likewise.
25049 ("alignaddrsi_vis"): Likewise.
25050 ("alignaddrdi_vis"): Likewise.
25051 ("alignaddrlsi_vis"): Likewise.
25052 ("alignaddrldi_vis"): Likewise.
25053 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25054 ("fexpand_vis"): Likewise.
25055 ("fpmerge_vis"): Likewise.
25056 ("faligndata<VM64:mode>_vis"): Likewise.
25057 ("bshuffle<VM64:mode>_vis"): Likewise.
25058 ("cmask8<P:mode>_vis"): Likewise.
25059 ("cmask16<P:mode>_vis"): Likewise.
25060 ("cmask32<P:mode>_vis"): Likewise.
25061 ("fchksm16_vis"): Likewise.
25062 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25063 ("fmean16_vis"): Likewise.
25064 ("fp<plusminus_insn>64_vis"): Likewise.
25065 ("<plusminus_insn>v8qi3"): Likewise.
25066 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25067 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25068 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25069 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25070 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25071 ("*movqi_insn"): Likewise.
25072 ("*movhi_insn"): Likewise.
25073 ("*movsi_insn"): Likewise.
25074 ("movsi_pic_gotdata_op"): Likewise.
25075 ("*movdi_insn_sp32"): Likewise.
25076 ("*movdi_insn_sp64"): Likewise.
25077 ("movdi_pic_gotdata_op"): Likewise.
25078 ("*movsf_insn"): Likewise.
25079 ("*movdf_insn_sp32"): Likewise.
25080 ("*movdf_insn_sp64"): Likewise.
25081 ("*zero_extendhisi2_insn"): Likewise.
25082 ("*zero_extendqihi2_insn"): Likewise.
25083 ("*zero_extendqisi2_insn"): Likewise.
25084 ("*zero_extendqidi2_insn"): Likewise.
25085 ("*zero_extendhidi2_insn"): Likewise.
25086 ("*zero_extendsidi2_insn_sp64"): Likewise.
25087 ("ldfsr"): Likewise.
25088 ("prefetch_64"): Likewise.
25089 ("prefetch_32"): Likewise.
25090 ("tie_ld32"): Likewise.
25091 ("tie_ld64"): Likewise.
25092 ("*tldo_ldub_sp32"): Likewise.
25093 ("*tldo_ldub1_sp32"): Likewise.
25094 ("*tldo_ldub2_sp32"): Likewise.
25095 ("*tldo_ldub_sp64"): Likewise.
25096 ("*tldo_ldub1_sp64"): Likewise.
25097 ("*tldo_ldub2_sp64"): Likewise.
25098 ("*tldo_ldub3_sp64"): Likewise.
25099 ("*tldo_lduh_sp32"): Likewise.
25100 ("*tldo_lduh1_sp32"): Likewise.
25101 ("*tldo_lduh_sp64"): Likewise.
25102 ("*tldo_lduh1_sp64"): Likewise.
25103 ("*tldo_lduh2_sp64"): Likewise.
25104 ("*tldo_lduw_sp32"): Likewise.
25105 ("*tldo_lduw_sp64"): Likewise.
25106 ("*tldo_lduw1_sp64"): Likewise.
25107 ("*tldo_ldx_sp64"): Likewise.
25108 ("*mov<VM32:mode>_insn"): Likewise.
25109 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25110 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25112 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25114 * config/sparc/sparc.md ("type"): New insn type viscmp.
25115 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25117 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25118 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25119 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25120 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25122 ("n7_vis_logical_11cycle"): Likewise.
25123 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25124 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25125 * config/sparc/niagara.md ("niag_vis"): Likewise.
25126 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25127 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25129 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25131 * config/sparc/sparc.md: New instruction type `bmask'.
25132 (bmaskdi_vis): Use the `bmask' type.
25133 (bmasksi_vis): Likewise.
25134 * config/sparc/ultra3.md (us3_array): Likewise.
25135 * config/sparc/niagara7.md (n7_array): Likewise.
25136 * config/sparc/niagara4.md (n4_array): Likewise.
25137 * config/sparc/niagara2.md (niag2_vis): Likewise.
25138 (niag3_vis): Likewise.
25139 * config/sparc/niagara.md (niag_vis): Likewise.
25141 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25143 * ipa-comdats.c: Remove optimize check from gate.
25144 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25145 for functions not optimized.
25146 (ipa_fn_summary_read): Skip optimize check.
25147 (ipa_fn_summary_write): Likewise.
25148 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25150 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25152 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25153 (check_callers): Give up on caller which is not optimized.
25154 (inline_small_functions): Likewise.
25155 (ipa_inline): Do not give up when not optimizing.
25156 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25157 away unoptimizes cdtors.
25158 (whole_program_function_and_variable_visibility): Do
25159 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25160 * ipa.c (process_references): Do not check optimize.
25161 (symbol_table::remove_unreachable_nodes): Update optimize check.
25162 (set_writeonly_bit): Update optimize check.
25163 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25164 (pass_ipa_single_use::gate): Remove.
25166 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25168 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25169 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25170 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25171 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25172 permute_load, permute_store, adjust_extract, adjust_splat,
25173 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25174 replace_swap_with_copy, dump_swap_insn_table,
25175 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25176 recombine_lvx_pattern, recombine_stvx_pattern,
25177 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25178 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25179 to file rs6000-p8swap.c.
25180 * config/rs6000/rs6000-p8swap.c: New file.
25181 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25182 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25183 and rs6000*-*-* targets.
25185 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25187 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25189 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25191 * lto-wrapper.c (merge_and_complain): Do not merge
25192 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25193 fsigned_zeros, ftrapping_math, fwrapv.
25194 (append_compiler_options): Do not track these options.
25195 (append_linker_options): Likewie
25197 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25199 * cgraphunit.c (cgraph_node::finalize_function): When
25200 !flag_toplevel_reorde set no_reorder flag.
25201 (varpool_node::finalize_decl): Likewise.
25202 (symbol_table::compile): Drop no toplevel reorder path.
25204 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25206 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25207 edges; zero probability is not better than uninitialized.
25209 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25211 * asan.h (asan_sanitize_allocas_p): Declare.
25212 * asan.c (asan_sanitize_allocas_p): New function.
25213 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25214 (handle_builtin_alloca): Likewise.
25215 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25216 if !asan_sanitize_allocas_p.
25217 * params.def (asan-instrument-allocas): Add new option.
25218 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25219 * opts.c (common_handle_option): Disable allocas sanitization for
25222 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25224 * asan.c: Include gimple-fold.h.
25225 (get_last_alloca_addr): New function.
25226 (handle_builtin_stackrestore): Likewise.
25227 (handle_builtin_alloca): Likewise.
25228 (asan_emit_allocas_unpoison): Likewise.
25229 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25230 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25231 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25232 (instrument_builtin_call): Pass gimple iterator to
25233 get_mem_refs_of_builtin_call.
25234 (last_alloca_addr): New global.
25235 * asan.h (asan_emit_allocas_unpoison): Declare.
25236 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25237 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25238 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25239 if function calls alloca.
25240 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25241 * gimple-fold.h (replace_call_with_value): Declare.
25242 * internal-fn.c: Include asan.h.
25243 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25244 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25246 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25248 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25249 (C_SELFTEST_FLAGS): New.
25250 (CPP_SELFTEST_FLAGS): New.
25251 (SELFTEST_DEPS): New, from deps of s-selftest.
25252 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25253 (CPP_SELFTEST_DEPS): New.
25254 (selftest): Add dependency on s-selftest-c++.
25255 (s-selftest): Rename to...
25256 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25257 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25258 than SELFTEST_FLAGS.
25259 (selftest-gdb): Rename to...
25260 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25262 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25263 (selftest-valgrind): Rename to...
25264 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25266 (selftest-valgrind): Reintroduce as an alias for
25267 selftest-c-valgrind.
25268 (s-selftest-c++): New.
25269 (selftest-c++-gdb): New.
25270 (selftest-c++-valgrind): New.
25272 2017-07-06 Olivier Hainque <hainque@adacore.com>
25274 * gcc.c (process_command): When deciding if undefined variables
25275 should be ignored when processing specs, accept "gcc -v" as well.
25277 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25279 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25280 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25282 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25284 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25285 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25286 * config/arm/arm-tables.opt: Regenerate.
25287 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25289 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25291 2017-07-06 Carl Love <cel@us.ibm.com>
25293 * ChangeLog: Clean up from mid air collision
25295 2017-07-06 Carl Love <cel@us.ibm.com>
25297 * config/rs6000/rs6000-c.c: Add support for built-in functions
25298 vector signed int vec_subc (vector signed int, vector signed int);
25299 vector signed __int128 vec_subc (vector signed __int128,
25300 vector signed __int128);
25301 vector unsigned __int128 vec_subc (vector unsigned __int128,
25302 vector unsigned __int128);
25303 vector signed int vec_sube (vector signed int, vector signed int,
25304 vector signed int);
25305 vector unsigned int vec_sube (vector unsigned int,
25306 vector unsigned int,
25307 vector unsigned int);
25308 vector signed __int128 vec_sube (vector signed __int128,
25309 vector signed __int128,
25310 vector signed__int128);
25311 vector unsigned __int128 vec_sube (vector unsigned __int128,
25312 vector unsigned __int128,
25313 vector unsigned __int128);
25314 vector signed int vec_subec (vector signed int, vector signed int,
25315 vector signed int);
25316 vector unsigned int vec_subec (vector unsigned int,
25317 vector unsigned int,
25318 vector unsigned int);
25319 vector signed __int128 vec_subec (vector signed __int128,
25320 vector signed __int128,
25321 vector signed__int128);
25322 vector unsigned __int128 vec_subec (vector unsigned __int128,
25323 vector unsigned __int128,
25324 vector unsigned __int128);
25325 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25326 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25327 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25328 BU_ALTIVEC_OVERLOAD_X definitions.
25329 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25330 * doc/extend.texi: Update the built-in documentation file for the new
25331 built-in functions.
25333 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25336 * diagnostic-show-locus.c (layout::layout): Use start and finish
25337 spelling location for the start and finish of each range.
25338 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25339 Add unused aspect param.
25340 * input.c (expand_location_1): Add "aspect" param, and use it
25341 to access the correct part of the location.
25342 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25344 (expand_location_to_spelling_point): Likewise.
25345 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25346 param, and pass it to expand_location_1.
25348 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25350 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25351 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25352 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25353 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25354 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25355 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25356 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25357 _mm_maskz_getmant_ss): New intrinsics.
25358 (__builtin_ia32_getexpss128_mask): Changed to ...
25359 __builtin_ia32_getexpss128_round ... this.
25360 (__builtin_ia32_getexpsd128_mask): Changed to ...
25361 __builtin_ia32_getexpsd128_round ... this.
25362 * config/i386/i386-builtin-types.def
25363 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25364 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25365 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25366 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25367 __builtin_ia32_getmantss_mask_round): New builtins.
25368 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25369 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25370 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25371 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25372 * config/i386/sse.md
25373 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25374 avx512f_sgetexp<mode><mask_scalar_name>
25375 <round_saeonly_scalar_name> ... this.
25376 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25377 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25378 vgetexp<ssescalarmodesuffix>
25379 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25380 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25381 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25382 avx512f_vgetmant<mode><mask_scalar_name>
25383 <round_saeonly_scalar_name> ... this.
25384 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25385 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25386 vgetmant<ssescalarmodesuffix>
25387 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25388 %0<mask_scalar_operand4>, %1, %2
25389 <round_saeonly_scalar_mask_op4>, %3} ... this.
25390 * config/i386/subst.md (mask_scalar_operand4,
25391 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25392 round_saeonly_scalar_nimm_predicate): New subst attributes.
25394 2017-07-06 Julia Koval <julia.koval@intel.com>
25396 * config/i386/i386.c (ix86_erase_embedded_rounding):
25397 Remove code for old rounding pattern.
25399 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25401 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25403 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25405 * doc/sourcebuild.texi (Test Directives, Variants of
25406 dg-require-support): Add documentation for dg-require-stack-check.
25408 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25410 * config/i386/subst.md (mask_scalar, round_scalar,
25411 round_saeonly_scalar): New meta-templates.
25412 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25413 round_scalar_mask_operand3, round_scalar_mask_op3,
25414 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25415 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25416 round_saeonly_scalar_constraint,
25417 round_saeonly_scalar_prefix): New subst attribute.
25418 * config/i386/sse.md
25419 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25420 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25421 <round_scalar_name> ... this.
25422 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25423 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25424 <round_scalar_name> ... this.
25425 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25426 <sse>_vm<code><mode>3<mask_scalar_name>
25427 <round_saeonly_scalar_name> ... this.
25428 (v<plusminus_mnemonic><ssescalarmodesuffix>
25429 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25430 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25431 v<plusminus_mnemonic><ssescalarmodesuffix>
25432 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25433 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25434 (v<multdiv_mnemonic><ssescalarmodesuffix>
25435 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25436 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25437 v<multdiv_mnemonic><ssescalarmodesuffix>
25438 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25439 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25440 (v<maxmin_float><ssescalarmodesuffix>
25441 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25442 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25443 v<maxmin_float><ssescalarmodesuffix>
25444 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25445 %0<mask_scalar_operand3>, %1, %<iptr>2
25446 <round_saeonly_scalar_mask_op3>} ... this.
25448 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25450 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25451 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25453 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25454 Alan Hayward <alan.hayward@arm.com>
25455 David Sherwood <david.sherwood@arm.com>
25457 * combine.c (simplify_if_then_else): Remove "enum" before
25459 * compare-elim.c (can_eliminate_compare): Likewise.
25460 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25462 (aarch64_lookup_simd_builtin_type): Likewise.
25463 (aarch64_simd_builtin_type): Likewise.
25464 (aarch64_init_simd_builtin_types): Likewise.
25465 (aarch64_simd_expand_args): Likewise.
25466 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25468 (aarch64_reverse_mask): Likewise.
25469 (aarch64_simd_emit_reg_reg_move): Likewise.
25470 (aarch64_gen_adjusted_ldpstp): Likewise.
25471 (aarch64_ccmp_mode_to_code): Likewise.
25472 (aarch64_operands_ok_for_ldpstp): Likewise.
25473 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25474 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25476 (aarch64_min_divisions_for_recip_mul): Likewise.
25477 (aarch64_reassociation_width): Likewise.
25478 (aarch64_get_condition_code_1): Likewise.
25479 (aarch64_simd_emit_reg_reg_move): Likewise.
25480 (aarch64_simd_attr_length_rglist): Likewise.
25481 (aarch64_reverse_mask): Likewise.
25482 (aarch64_operands_ok_for_ldpstp): Likewise.
25483 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25484 (aarch64_gen_adjusted_ldpstp): Likewise.
25485 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25487 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25488 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25489 (arm_lookup_simd_builtin_type): Likewise.
25490 (arm_simd_builtin_type): Likewise.
25491 (arm_init_simd_builtin_types): Likewise.
25492 (arm_expand_builtin_args): Likewise.
25493 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25494 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25495 (ft32_setup_incoming_varargs): Likewise.
25496 (ft32_function_arg): Likewise.
25497 (ft32_function_arg_advance): Likewise.
25498 (ft32_pass_by_reference): Likewise.
25499 (ft32_arg_partial_bytes): Likewise.
25500 (ft32_valid_pointer_mode): Likewise.
25501 (ft32_addr_space_pointer_mode): Likewise.
25502 (ft32_addr_space_legitimate_address_p): Likewise.
25503 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25505 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25506 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25507 (iamcu_alignment): Likewise.
25508 (canonicalize_vector_int_perm): Likewise.
25509 (ix86_noce_conversion_profitable_p): Likewise.
25510 (ix86_mpx_bound_mode): Likewise.
25511 (ix86_operands_ok_for_move_multiple): Likewise.
25512 * config/microblaze/microblaze-protos.h
25513 (microblaze_expand_conditional_branch_reg): Likewise.
25514 * config/microblaze/microblaze.c
25515 (microblaze_expand_conditional_branch_reg): Likewise.
25516 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25518 (rs6000_reassociation_width): Likewise.
25519 (rs6000_invalid_binary_op): Likewise.
25520 (fusion_p9_p): Likewise.
25521 (emit_fusion_p9_load): Likewise.
25522 (emit_fusion_p9_store): Likewise.
25523 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25525 (riscv_hard_regno_mode_ok_p): Likewise.
25526 (riscv_address_insns): Likewise.
25527 (riscv_split_symbol): Likewise.
25528 (riscv_legitimize_move): Likewise.
25529 (riscv_function_value): Likewise.
25530 (riscv_hard_regno_nregs): Likewise.
25531 (riscv_expand_builtin): Likewise.
25532 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25533 (riscv_build_integer): Likewise.
25534 (riscv_split_integer): Likewise.
25535 (riscv_legitimate_constant_p): Likewise.
25536 (riscv_cannot_force_const_mem): Likewise.
25537 (riscv_regno_mode_ok_for_base_p): Likewise.
25538 (riscv_valid_base_register_p): Likewise.
25539 (riscv_valid_offset_p): Likewise.
25540 (riscv_valid_lo_sum_p): Likewise.
25541 (riscv_classify_address): Likewise.
25542 (riscv_legitimate_address_p): Likewise.
25543 (riscv_address_insns): Likewise.
25544 (riscv_load_store_insns): Likewise.
25545 (riscv_force_binary): Likewise.
25546 (riscv_split_symbol): Likewise.
25547 (riscv_force_address): Likewise.
25548 (riscv_legitimize_address): Likewise.
25549 (riscv_move_integer): Likewise.
25550 (riscv_legitimize_const_move): Likewise.
25551 (riscv_legitimize_move): Likewise.
25552 (riscv_address_cost): Likewise.
25553 (riscv_subword): Likewise.
25554 (riscv_output_move): Likewise.
25555 (riscv_canonicalize_int_order_test): Likewise.
25556 (riscv_emit_int_order_test): Likewise.
25557 (riscv_function_arg_boundary): Likewise.
25558 (riscv_pass_mode_in_fpr_p): Likewise.
25559 (riscv_pass_fpr_single): Likewise.
25560 (riscv_pass_fpr_pair): Likewise.
25561 (riscv_get_arg_info): Likewise.
25562 (riscv_function_arg): Likewise.
25563 (riscv_function_arg_advance): Likewise.
25564 (riscv_arg_partial_bytes): Likewise.
25565 (riscv_function_value): Likewise.
25566 (riscv_pass_by_reference): Likewise.
25567 (riscv_setup_incoming_varargs): Likewise.
25568 (riscv_print_operand): Likewise.
25569 (riscv_elf_select_rtx_section): Likewise.
25570 (riscv_save_restore_reg): Likewise.
25571 (riscv_for_each_saved_reg): Likewise.
25572 (riscv_register_move_cost): Likewise.
25573 (riscv_hard_regno_mode_ok_p): Likewise.
25574 (riscv_hard_regno_nregs): Likewise.
25575 (riscv_class_max_nregs): Likewise.
25576 (riscv_memory_move_cost): Likewise.
25577 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25578 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25579 (rl78_addr_space_address_mode): Likewise.
25580 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25582 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25583 (rs6000_reassociation_width): Likewise.
25584 (rs6000_invalid_binary_op): Likewise.
25585 (fusion_p9_p): Likewise.
25586 (emit_fusion_p9_load): Likewise.
25587 (emit_fusion_p9_store): Likewise.
25588 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25589 (ok_for_simple_move_operands): Likewise.
25590 (ok_for_simple_move_strict_operands): Likewise.
25591 (ok_for_simple_arith_logic_operands): Likewise.
25592 (visium_legitimize_reload_address): Likewise.
25593 (visium_select_cc_mode): Likewise.
25594 (output_cbranch): Likewise.
25595 (visium_split_double_move): Likewise.
25596 (visium_expand_copysign): Likewise.
25597 (visium_expand_int_cstore): Likewise.
25598 (visium_expand_fp_cstore): Likewise.
25599 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25600 (visium_function_arg): Likewise.
25601 (visium_function_arg_advance): Likewise.
25602 (visium_libcall_value): Likewise.
25603 (visium_setup_incoming_varargs): Likewise.
25604 (visium_legitimate_constant_p): Likewise.
25605 (visium_legitimate_address_p): Likewise.
25606 (visium_legitimize_address): Likewise.
25607 (visium_secondary_reload): Likewise.
25608 (visium_register_move_cost): Likewise.
25609 (visium_memory_move_cost): Likewise.
25610 (prepare_move_operands): Likewise.
25611 (ok_for_simple_move_operands): Likewise.
25612 (ok_for_simple_move_strict_operands): Likewise.
25613 (ok_for_simple_arith_logic_operands): Likewise.
25614 (visium_function_value_1): Likewise.
25615 (rtx_ok_for_offset_p): Likewise.
25616 (visium_legitimize_reload_address): Likewise.
25617 (visium_split_double_move): Likewise.
25618 (visium_expand_copysign): Likewise.
25619 (visium_expand_int_cstore): Likewise.
25620 (visium_expand_fp_cstore): Likewise.
25621 (visium_split_cstore): Likewise.
25622 (visium_select_cc_mode): Likewise.
25623 (visium_split_cbranch): Likewise.
25624 (output_cbranch): Likewise.
25625 (visium_print_operand_address): Likewise.
25626 * expmed.c (flip_storage_order): Likewise.
25627 * expmed.h (emit_cstore): Likewise.
25628 (flip_storage_order): Likewise.
25629 * genrecog.c (validate_pattern): Likewise.
25630 * hsa-gen.c (gen_hsa_addr): Likewise.
25631 * internal-fn.c (expand_arith_overflow): Likewise.
25632 * ira-color.c (allocno_copy_cost_saving): Likewise.
25633 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25634 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25635 (process_invariant_for_inheritance): Likewise.
25636 * lra-eliminations.c (move_plus_up): Likewise.
25637 * omp-low.c (lower_oacc_reductions): Likewise.
25638 * simplify-rtx.c (simplify_subreg): Likewise.
25639 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25640 (TARGET_CHKP_BOUND_MODE): Likewise..
25641 * targhooks.c (default_chkp_bound_mode): Likewise.
25642 (default_setup_incoming_vararg_bounds): Likewise.
25643 * targhooks.h (default_chkp_bound_mode): Likewise.
25644 (default_setup_incoming_vararg_bounds): Likewise.
25645 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25646 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25647 (have_whole_vector_shift): Likewise.
25648 * tree-vect-stmts.c (vectorizable_load): Likewise.
25649 * doc/tm.texi: Regenerate.
25651 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25653 Graceful degrade if Binutils PR21472 is not available.
25656 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25657 .rodata in flash test fails.
25658 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25659 * confgure: Regenerate.
25660 * config.in: Regenerate.
25661 * config/avr/avr.c (avr_asm_named_section)
25662 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25663 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25664 (avr_asm_init_sections): Same.
25666 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25668 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25669 (fma<VH:mode>4_intrinsic): Likewise.
25670 (*fmsub<VCVTF:mode>4): Likewise.
25671 (*fmsub<VH:mode>4_intrinsic): Likewise.
25673 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25676 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25677 Don't depend on "optimize > 0".
25678 (out_movhi_r_mr, out_movqi_mr_r): Same.
25679 (out_movhi_mr_r, out_movqi_r_mr): Same.
25680 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25681 io_address_operand on "optimize > 0".
25683 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25685 * tree-loop-distribution.c: Add general explanantion on the pass.
25686 (generate_loops_for_partition): Mark distributed loop.
25687 (pg_add_dependence_edges): New parameter. Handle alias data
25688 dependence specially and record it in the parameter if asked.
25689 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25690 (init_partition_graph_vertices, add_partition_graph_edge): New.
25691 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25692 (free_partition_graph_vdata, build_partition_graph): New.
25693 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25694 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25695 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25696 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25697 (version_for_distribution_p, finalize_partitions): New.
25698 (distribute_loop): Handle alias data dependence specially. Factor
25699 out loop fusion code as functions and call these functions.
25701 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25703 * tree-loop-distribution.c (classify_partition): New parameter and
25704 better handle reduction statement.
25705 (rdg_build_partitions): Revise comment.
25706 (distribute_loop): Compute statements in all partitions and pass it
25707 to classify_partition.
25709 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25711 * tree-loop-distribution.c (enum partition_type): New.
25712 (struct partition): New field type.
25713 (partition_merge_into): Add parameter. Update partition type.
25714 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25715 (build_rdg_partition_for_vertex): Compute partition type.
25716 (rdg_build_partitions): Dump partition type.
25717 (distribute_loop): Update calls to partition_merge_into.
25719 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25721 * tree-loop-distribution.c (struct ddr_hasher): New.
25722 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25724 (classify_partition): Call get_data_dependence.
25725 (pg_add_dependence_edges): Ditto.
25726 (distribute_loop): Release data dependence hash table.
25728 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25730 * tree-loop-distribution.c (ref_base_address): Delete.
25731 (similar_memory_accesses): Rename ...
25732 (share_memory_accesses): ... to this. Check if partitions access
25733 the same memory reference.
25734 (distribute_loop): Call share_memory_accesses.
25736 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25738 * tree-loop-distribution.c (struct partition): New field recording
25739 its data reference.
25740 (partition_alloc, partition_free): Init and release data refs.
25741 (partition_merge_into): Merge data refs.
25742 (build_rdg_partition_for_vertex): Collect data refs for partition.
25743 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25745 (distribute_loop): Remve data refs from vertice data of partition
25748 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25750 * tree-loop-distribution.c (params.h): Include header file.
25751 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25752 (datarefs_vec): New global var.
25753 (create_rdg_vertices): Use datarefs_vec directly.
25754 (free_rdg): Don't free data references.
25755 (build_rdg): Update use. Don't free data references.
25756 (distribute_loop): Compute global variable for data references.
25757 Bail out if there are too many data references.
25759 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25761 * tree-loop-distribution.c (loop_nest): New global var.
25762 (build_rdg): Use loop directly, rather than loop nest.
25763 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25765 (distribute_loop): Compute global variable loop nest. Update use.
25767 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25769 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25770 (partition_merge_into): New parameter. Dump reason for fusion.
25771 (distribute_loop): Update use of partition_merge_into.
25773 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25775 * tree-loop-distribution.c (bb_top_order_index): New.
25776 (bb_top_order_index_size, bb_top_order_cmp): New.
25777 (stmts_from_loop): Use topological order.
25778 (pass_loop_distribution::execute): Compute and release topological
25779 order for basic blocks.
25781 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25783 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25786 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25788 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25789 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25790 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25791 * internal-fn.def (LOOP_DIST_ALIAS): New.
25792 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25793 (fold_loop_internal_call): ... this.
25794 (vect_loop_dist_alias_call): New function.
25795 (set_uid_loop_bbs): Call fold_loop_internal_call.
25796 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25799 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25802 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25803 Require dead FLAGS_REG at the beginning of a peephole.
25805 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25808 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25809 arguments in the call to __builtin_ia32_sbb_u32.
25810 (_subborrow_u64): Swap _X and _Y arguments in the call to
25811 __builtin_ia32_sbb_u64.
25813 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25816 * tree-vrp.c (compare_assert_loc): Turn into a function template
25817 with stable template parameter. Only test if a->e is NULL,
25818 !a->e == !b->e has been verified already. Use e == NULL or
25819 e != NULL instead of e or ! e tests. If stable is true, don't use
25820 iterative_hash_expr, on the other side allow a or b or both NULL
25821 and sort the NULLs last.
25822 (process_assert_insertions): Sort using compare_assert_loc<false>
25823 instead of compare_assert_loc, later sort using
25824 compare_assert_loc<true> before calling process_assert_insertions_for
25825 in a loop. Use break instead of continue once seen NULL pointer.
25827 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25829 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25830 Cortex-R7 and Cortex-R8 processors.
25832 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25834 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25835 uninitialized while src is not.
25837 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25839 * common/config/arm/arm-common.c: Adjust include path for
25841 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25842 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25843 (arm-cpu-data.h): Likewise.
25844 (arm-cpu-cdata.h): Likewise.
25845 * config/arm/arm-cpu.h: Delete.
25846 * config/arm/arm-cpu-cdata.h: Delete.
25847 * config/arm/arm-cpu-data.h: Delete.
25849 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25851 * config/arm/arm-cpus.in (cortex-a55): New.
25852 (cortex-a75): Likewise.
25853 (cortex-a75.cortex-a55): Likewise.
25854 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25856 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25857 * config/arm/arm-cpu-cdata.h: Regenerate.
25858 * config/arm/arm-cpu-data.h: Regenerate.
25859 * config/arm/arm-cpu.h: Regenerate.
25860 * config/arm/arm-tables.opt: Regenerate.
25861 * config/arm/arm-tune.md: Regenerate.
25863 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25865 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25867 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25869 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25872 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25874 PR tree-optimization/81292
25875 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25876 full_string_p, also call adjust_related_strinfos if the adjustment
25877 is simple, otherwise invalidate related strinfos.
25879 2017-07-04 Martin Liska <mliska@suse.cz>
25882 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25883 newly created variable as DECL_IGNORED_P.
25885 2017-07-04 Martin Liska <mliska@suse.cz>
25888 * ipa-inline.c (inline_small_functions):
25889 Use xstrdup_for_dump.
25891 2017-07-04 Tom de Vries <tom@codesourcery.com>
25893 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25895 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25898 * config/darwin.c (darwin_function_switched_text_sections):
25901 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25903 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25905 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25907 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25909 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25911 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25912 min_profitable_iters, and th as inclusive lower bounds.
25913 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25914 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25915 for min_profitable_iters and min_profitable_estimate.
25916 (vect_transform_loop): Treat th as an inclusive lower bound.
25917 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25919 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25922 * config/darwin.c (darwin_function_switched_text_sections):
25923 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25924 in two pieces, and suppress the use of buf.
25926 2017-07-03 Nathan Sidwell <nathan@acm.org>
25928 * hash-table.h (hash_table_mod1): Fix indentation.
25930 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25932 PR middle-end/81290
25933 * predict.c (force_edge_cold): Be more careful about propagation
25935 * profile-count.h (profile_probability::guessed,
25936 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25938 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25940 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25942 * doc/invoke.texi (rcpc architecture extension): Document it.
25944 2017-07-03 Richard Biener <rguenther@suse.de>
25946 PR tree-optimization/60510
25947 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25948 the scalar reduction PHI and use it.
25949 (vectorizable_reduction): Properly guard the single_defuse_cycle
25950 path for non-SLP reduction chains where we cannot use it.
25951 Rework reduc_def/index and vector type deduction. Rework
25952 vector operand gathering during reduction op code-gen.
25953 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25954 chains dissolve the chain and leave it to non-SLP reduction
25957 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25959 * tree-data-ref.h (dr_alignment): Declare.
25960 * tree-data-ref.c (dr_alignment): New function.
25961 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25962 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25964 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25966 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25968 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25969 and base_misalignment fields.
25970 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25971 * tree-data-ref.c: Include builtins.h.
25972 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25973 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25974 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25975 * tree-vect-data-refs.c: Include tree-cfg.h.
25976 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25977 fields instead of calculating an alignment here.
25978 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25979 innermost_loop_behavior fields.
25981 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25983 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25985 (DR_STEP_ALIGNMENT): New macro.
25986 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25987 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25988 (create_data_ref): Print it.
25989 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25990 to tell whether the step preserves vector (mis)alignment.
25991 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25992 Move the check for an integer step and generalise to all INTEGER_CST.
25993 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25994 Print the outer step alignment.
25996 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25998 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25999 with offset_alignment.
26000 (DR_ALIGNED_TO): Delete.
26001 (DR_OFFSET_ALIGNMENT): New macro.
26002 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
26003 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
26004 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
26005 (create_data_ref): Likewise.
26006 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26007 (vect_analyze_data_refs): Likewise.
26008 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
26009 creating dummy innermost behavior.
26011 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26013 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
26014 with a "innermost_loop_behavior *" and refeence tree.
26015 * tree-data-ref.c (dr_analyze_innermost): Likewise.
26016 (create_data_ref): Update call accordingly.
26017 * tree-predcom.c (find_looparound_phi): Likewise.
26019 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26021 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
26022 fields with dr_wrt_vec_loop.
26023 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
26024 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
26025 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
26026 (vect_dr_behavior): New function.
26027 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26028 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
26029 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
26030 track whether the step preserves the misalignment.
26031 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26032 Use vect_dr_behavior.
26033 (vect_setup_realignment): Update call accordingly.
26034 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
26035 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
26036 call to vect_create_addr_base_for_vector_ref.
26037 (vect_create_cond_for_align_checks): Likewise.
26038 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
26039 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
26040 (vect_recog_mask_conversion_pattern): Likewise.
26041 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
26042 (new_stmt_vec_info): Remove redundant zeroing.
26044 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26046 * common/config/arm/arm-common.c (arm_be8_option): New function.
26047 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
26048 (ISA_ARMv6): Add isa_bit_be8.
26049 * config/arm/arm.h (arm_be8_option): Add prototype.
26050 (BE8_SPEC_FUNCTION): New define.
26051 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
26052 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
26053 (mlittle-endian): Similarly.
26054 (mbe8, mbe32): New options.
26055 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
26056 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
26058 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26060 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26062 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26064 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26065 (cleanup_tree_cfg_bb): Use it.
26066 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26068 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26070 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26073 * loop-doloop.c (add_test): Update profile.
26075 2017-07-03 Martin Liska <mliska@suse.cz>
26078 * sanopt.c (rewrite_usage_of_param): New function.
26079 (sanitize_rewrite_addressable_params): Likewise.
26080 (pass_sanopt::execute): Call rewrite_usage_of_param.
26082 2017-07-03 Richard Biener <rguenther@suse.de>
26084 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26085 back to using VIEW_CONVERT_EXPR.
26087 2017-07-03 Martin Liska <mliska@suse.cz>
26090 * doc/extend.texi: Document when a resolver function is
26091 generated for target_clones.
26093 2017-07-03 Martin Liska <mliska@suse.cz>
26095 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26096 and shadow memory of auto variables which are subject of
26097 use-after-scope sanitization.
26098 (asan_expand_mark_ifn): Add do set only when is_poison.
26100 2016-07-03 Richard Biener <rguenther@suse.de>
26102 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26104 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26105 (vectorizable_reduction): Perform reduction PHI creation when
26106 visiting a reduction PHI and adjust and simplify code generation
26107 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26108 (vect_transform_loop): Visit reduction PHIs.
26109 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26110 defs into the SLP tree.
26111 (vect_build_slp_tree): Reduction defs terminate the recursion.
26112 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26114 (vect_get_vec_defs_for_stmt_copy): Export.
26115 (vect_get_vec_defs): Likewise.
26116 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26118 (vect_get_vec_defs_for_stmt_copy): Declare.
26119 (vect_get_vec_defs): Likewise.
26121 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26123 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26124 parameter with a "loop" parameter and use it instead of the
26125 loop containing DR_STMT. Don't check simple_iv when doing
26126 BB analysis. Describe the two analysis modes in the comment.
26128 2017-07-03 Tom de Vries <tom@codesourcery.com>
26130 PR tree-optimization/69468
26131 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26132 (find_same_succ_bb): Handle ignore_edge_flags.
26134 2017-07-03 Tom de Vries <tom@codesourcery.com>
26136 PR tree-optimization/81192
26137 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26139 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26141 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26143 2017-07-03 Tom de Vries <tom@codesourcery.com>
26145 PR tree-optimization/81192
26146 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26147 BB_SAME_SUCC (bb) == NULL.
26149 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26151 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26154 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26156 * dumpfile.c: Include profile-count.h
26157 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26159 (insert_cond_bb): Update profile.
26160 * tree-cfg.h (insert_cond_bb): Update prototype.
26161 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26162 * tree-dump.c: Do not include tree-cfg.
26164 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26166 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26168 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26170 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26173 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26175 * tree-complex.c (expand_complex_div_wide): update profile.
26177 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26178 Alan Hayward <alan.hayward@arm.com>
26179 David Sherwood <david.sherwood@arm.com>
26181 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26182 (CORETYPES_H): New define.
26183 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26184 (insn-modes-inline.h, s-modes-inline-h): New rules.
26185 (generated_files): Add insn-modes-inline.h.
26186 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26187 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26188 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26189 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26190 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26191 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26192 (build/gencodes.o, build/genconditions.o): Likewise.
26193 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26194 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26195 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26196 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26197 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26198 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26199 * coretypes.h: Include everything up to real.h for generators.
26200 Include insn-modes.h first. Include wide-int-print.h after
26201 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26202 * machmode.h: Don't include insn-modes.h here.
26203 * function-tests.c: Remove includes of signop.h, machmode.h,
26204 double-int.h and wide-int.h.
26206 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26208 * optc-save-gen.awk: Likewise.
26209 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26210 * godump.c: Remove include of wide-int-print.h.
26211 * pretty-print.h: Likewise.
26212 * wide-int-print.cc: Likewise.
26213 * wide-int.cc: Likewise.
26214 * hash-map-tests.c: Remove include of signop.h.
26215 * hash-set-tests.c: Likewise.
26216 * rtl-tests.c: Likewise.
26217 * mkconfig.sh: Remove include of machmode.h.
26218 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26220 (emit_insn_modes_inline_h): ...this new function. Emit the code
26221 into an insn-modes-inline.h header file, adding appropriate
26222 include guards and end comments.
26223 (emit_insn_modes_c_header): Remove include of machmode.h.
26224 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26226 (main): Handle -i flag and call emit_insn_modes_inline_h when
26229 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26231 * tree-ssa-strlen.c (strinfo): Rename the length field to
26232 nonzero_chars. Add a full_string_p field.
26233 (compare_nonzero_chars, zero_length_string_p): New functions.
26234 (get_addr_stridx): Add an offset_out parameter.
26235 Use compare_nonzero_chars.
26236 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26237 (new_strinfo): Update after above changes to strinfo.
26238 (set_endptr_and_length): Set full_string_p.
26239 (get_string_length): Update after above changes to strinfo.
26240 (unshare_strinfo): Update call to new_strinfo.
26241 (maybe_invalidate): Likewise.
26242 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26243 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26244 as a uhwi instead of an shwi. Update after above changes to
26245 strinfo and new_strinfo.
26246 (zero_length_string): Assert that chainsi contains full strings.
26247 Use zero_length_string_p. Update call to new_strinfo.
26248 (adjust_related_strinfos): Update after above changes to strinfo.
26249 Copy full_string_p from origsi.
26250 (adjust_last_stmt): Use zero_length_string_p.
26251 (handle_builtin_strlen): Update after above changes to strinfo and
26252 new_strinfo. Install the lhs as the string length if the previous
26253 entry didn't describe a full string.
26254 (handle_builtin_strchr): Update after above changes to strinfo
26256 (handle_builtin_strcpy): Likewise.
26257 (handle_builtin_strcat): Likewise.
26258 (handle_builtin_malloc): Likewise.
26259 (handle_pointer_plus): Likewise.
26260 (handle_builtin_memcpy): Likewise. Track nonzero characters
26261 that aren't necessarily followed by a nul terminator.
26262 (handle_char_store): Likewise.
26264 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26266 PR tree-optimization/80769
26267 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26268 for malloc and calloc. Document the new invariant that all related
26269 strinfos have delayed lengths or none do.
26270 (verify_related_strinfos): Move earlier in file.
26271 (set_endptr_and_length): New function, split out from...
26272 (get_string_length): ...here. Also set the lengths of related
26274 (zero_length_string): Assert that chainsi has known (rather than
26276 (adjust_related_strinfos): Likewise.
26278 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26280 PR tree-optimization/81136
26281 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26282 assert that two references with the same misalignment have the same
26283 compile-time misalignment if those compile-time misalignments
26286 2017-07-01 Andi Kleen <ak@linux.intel.com>
26288 * print-tree.c (print_node): Print all attributes.
26290 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26292 * cfg.c (scale_bbs_frequencies): New function.
26293 * cfg.h (scale_bbs_frequencies): Declare it.
26294 * cfgloopanal.c (single_likely_exit): Cleanup.
26295 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26297 (scale_loop_profile): Likewise.
26298 (loop_version): Likewise.
26299 (create_empty_loop_on_edge): Update.
26300 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26301 scale_loop_frequencies, scale_loop_profile, loopify,
26302 loop_version): Update prototypes.
26303 * modulo-sched.c (sms_schedule): Update.
26304 * predict.c (unlikely_executed_edge_p): Also check probability.
26305 (probably_never_executed_edge_p): Fix typo.
26306 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26307 * tree-parloops.c (gen_parallel_loop): Update.
26308 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26309 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26310 * tree-ssa-loop-split.c (split_loop): Update.
26311 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26312 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26313 (vect_loop_versioning): Update.
26314 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26316 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26318 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26320 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26322 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26323 to keep profile consistent.
26325 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26327 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26328 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26329 * profile-count.h (max_safe_multiplier): Make unsigned.
26330 (profile_count::guessed_zero): New.
26332 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26334 * bb-reorder.c (fix_up_crossing_landing_pad,
26335 fix_crossing_conditional_branches): Use make_single_succ_edge
26336 to keep profile consistent.
26338 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26340 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26343 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26346 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26347 the right scopes, make sure cond_jump isn't preserved between multiple
26348 iterations. Search for fallthru edge whenever there are 3+ edges and
26349 use find_fallthru_edge for it.
26351 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26353 Patch by Alexander Monakov <amonakov@ispras.ru>
26354 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26355 probabilities consistently.
26357 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26359 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26360 emit_cmp_and_jump_insns.
26362 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26365 * tree-inline.c (expand_call_inline): Combine profile statuses.
26367 2017-06-30 Andrew Pinski <apinski@cavium.com>
26369 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26370 fold_stmt returned true.
26372 2017-06-30 Nathan Sidwell <nathan@acm.org>
26374 * ggc.h (empty_string): Delete.
26375 * cfgexpand.c (expand_asm_stmt): Use plain "".
26376 * optabs.c (expand_asm_memory_barrier): Likewise.
26377 * stringpool.c (empty_string): Delete.
26378 (digit_vector, digit_string): Delete.
26379 (ggc_alloc_string): Use plain "", don't optimize single digit
26380 strings. Use ggc_alloc_atomic.
26382 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26384 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26385 comparison set and one other set, use the cost of the non-comparison
26388 2017-06-30 Nathan Sidwell <nathan@acm.org>
26390 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26393 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26395 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26396 loops. Remove now unneeded calls to gimple_switch_set_label() that
26397 just set removed labels to NULL_TREE.
26399 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26401 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26402 (set_range_info): ...here. Only call set_range_info_raw if domain
26404 (set_nonzero_bits): Call set_range_info_raw.
26405 * tree-ssanames.h (set_range_info_raw): New.
26407 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26410 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26411 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26412 of nonimmediate_operand and <store_mask_constraint> instead of m
26413 for the input operand. For V8FI iterator, always split if input
26414 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26415 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26416 <store_mask_predicate> instead of register_operand and
26417 <store_mask_constraint> instead of v for the input operand. Make
26418 sure both operands aren't MEMs for if not <mask_applied>.
26420 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26422 * lto-wrapper.c (copy_file) Close both file descriptors before
26425 2017-06-30 Martin Liska <mliska@suse.cz>
26428 * multiple_target.c (create_dispatcher_calls): Make ifunc
26429 also for function that don't have calls or are not referenced.
26431 2017-06-30 Richard Biener <rguenther@suse.de>
26433 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26434 analyze the first scalar stmt. Move vector type computation
26435 for the BB case here from ...
26436 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26437 live operation processing in the SLP case properly.
26439 2017-06-30 Richard Biener <rguenther@suse.de>
26441 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26443 2017-06-30 Martin Liska <mliska@suse.cz>
26446 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26447 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26449 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26451 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26454 2017-06-30 Richard Biener <rguenther@suse.de>
26456 PR tree-optimization/81249
26457 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26458 condition reduction result to original scalar type.
26460 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26462 * profile-count.h (enum profile_quality): Fix typos and whitespace
26465 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26467 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26468 type for branch probabilities.
26470 2017-06-29 Julian Brown <julian@codesourcery.com>
26471 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26473 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26474 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26475 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26476 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26478 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26480 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26481 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26482 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26483 CC usage from generic code to here.
26484 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26485 CC usage into the target macros.
26487 2017-06-29 Maya Rashish <coypu@sdf.org>
26489 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26492 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26494 * arm/arm-builtins.c: Include profile-count.h
26495 * except.c (sjlj_emit_function_enter): Use
26496 profile_probability::unlikely.
26498 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26500 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26501 and tocrel_offset be pointer args rather than implicitly using
26503 (legitimate_constant_pool_address_p, rs6000_emit_move,
26504 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26505 tocrel_offset and use in toc_relative_expr_p call.
26506 (print_operand, print_operand_address): Use static tocrel_base_oac
26507 and tocrel_offset_oac.
26508 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26511 2017-06-29 Maya Rashish <coypu@sdf.org>
26513 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26515 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26517 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26518 objects, take into account only the alignment of 'op0' and 'mode1' if
26521 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26523 * ccmp.c (ccmp_tree_comparison_p): New function.
26524 (ccmp_candidate_p): Update to use above function.
26525 (get_compare_parts): New function.
26526 (expand_ccmp_next): Update to use new functions.
26527 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26529 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26530 take mode as argument.
26531 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26532 * expr.c (expand_expr_real_1): Pass mode as argument.
26534 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26536 * combine.c (combine_instructions): Print insns to dump_file, together
26539 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26541 * asan.c (asan_emit_stack_protection): Update.
26542 (create_cond_insert_point): Update.
26543 * auto-profile.c (afdo_propagate_circuit): Update.
26544 * basic-block.h (struct edge_def): Turn probability to
26545 profile_probability.
26546 (EDGE_FREQUENCY): Update.
26547 * bb-reorder.c (find_traces_1_round): Update.
26548 (better_edge_p): Update.
26549 (sanitize_hot_paths): Update.
26550 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26551 (make_single_succ_edge): Update.
26552 (check_bb_profile): Update.
26553 (dump_edge_info): Update.
26554 (update_bb_profile_for_threading): Update.
26555 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26556 probabilitycount to 0.
26557 * cfgbuild.c (compute_outgoing_frequencies): Update.
26558 * cfgcleanup.c (try_forward_edges): Update.
26559 (outgoing_edges_match): Update.
26560 (try_crossjump_to_edge): Update.
26561 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26562 (expand_gimple_tailcall): Update.
26563 (construct_init_block): Use make_single_succ_edge.
26564 (construct_exit_block): Use make_single_succ_edge.
26565 * cfghooks.c (verify_flow_info): Update.
26566 (redirect_edge_succ_nodup): Update.
26567 (split_edge): Update.
26568 (account_profile_record): Update.
26569 * cfgloopanal.c (single_likely_exit): Update.
26570 * cfgloopmanip.c (scale_loop_profile): Update.
26571 (set_zero_probability): Remove.
26572 (duplicate_loop_to_header_edge): Update.
26573 * cfgloopmanip.h (loop_version): Update prototype.
26574 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26575 (force_nonfallthru_and_redirect): Update.
26576 (update_br_prob_note): Update.
26577 (rtl_verify_edges): Update.
26578 (purge_dead_edges): Update.
26579 (rtl_lv_add_condition_to_bb): Update.
26580 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26581 * cgraphunit.c (init_lowered_empty_function): Update.
26582 (cgraph_node::expand_thunk): Update.
26583 * cilk-common.c: Include profile-count.h
26584 * dojump.c (inv): Remove.
26585 (jumpifnot): Update.
26586 (jumpifnot_1): Update.
26587 (do_jump_1): Update.
26589 (do_jump_by_parts_greater_rtx): Update.
26590 (do_compare_rtx_and_jump): Update.
26591 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26592 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26593 * dwarf2cfi.c: Include profile-count.h
26594 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26595 (sjlj_emit_dispatch_table): Likewise.
26596 * explow.c: Include profile-count.h
26597 * expmed.c (emit_store_flag_force): Update.
26598 (do_cmp_and_jump): Update.
26599 * expr.c (compare_by_pieces_d::generate): Update.
26600 (compare_by_pieces_d::finish_mode): Update.
26601 (emit_block_move_via_loop): Update.
26602 (store_expr_with_bounds): Update.
26603 (store_constructor): Update.
26604 (expand_expr_real_2): Update.
26605 (expand_expr_real_1): Update.
26606 * expr.h (try_casesi, try_tablejump): Update prototypes.
26607 * gimple-pretty-print.c (dump_probability): Update.
26608 (dump_profile): New.
26609 (dump_gimple_label): Update.
26610 (dump_gimple_bb_header): Update.
26611 * graph.c (draw_cfg_node_succ_edges): Update.
26612 * hsa-gen.c (convert_switch_statements): Update.
26613 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26614 (find_if_case_1): Update.
26615 (find_if_case_2): Update.
26616 * internal-fn.c (expand_arith_overflow_result_store): Update.
26617 (expand_addsub_overflow): Update.
26618 (expand_neg_overflow): Update.
26619 (expand_mul_overflow): Update.
26620 (expand_vector_ubsan_overflow): Update.
26621 * ipa-cp.c (good_cloning_opportunity_p): Update.
26622 * ipa-split.c (split_function): Use make_single_succ_edge.
26623 * ipa-utils.c (ipa_merge_profiles): Update.
26624 * loop-doloop.c (add_test): Update.
26625 (doloop_modify): Update.
26626 * loop-unroll.c (compare_and_jump_seq): Update.
26627 (unroll_loop_runtime_iterations): Update.
26628 * lra-constraints.c (lra_inheritance): Update.
26629 * lto-streamer-in.c (input_cfg): Update.
26630 * lto-streamer-out.c (output_cfg): Update.
26631 * mcf.c (adjust_cfg_counts): Update.
26632 * modulo-sched.c (sms_schedule): Update.
26633 * omp-expand.c (expand_omp_for_init_counts): Update.
26634 (extract_omp_for_update_vars): Update.
26635 (expand_omp_ordered_sink): Update.
26636 (expand_omp_for_ordered_loops): Update.
26637 (expand_omp_for_generic): Update.
26638 (expand_omp_for_static_nochunk): Update.
26639 (expand_omp_for_static_chunk): Update.
26640 (expand_cilk_for): Update.
26641 (expand_omp_simd): Update.
26642 (expand_omp_taskloop_for_outer): Update.
26643 (expand_omp_taskloop_for_inner): Update.
26644 * omp-simd-clone.c (simd_clone_adjust): Update.
26645 * optabs.c (expand_doubleword_shift): Update.
26646 (expand_abs): Update.
26647 (emit_cmp_and_jump_insn_1): Update.
26648 (expand_compare_and_swap_loop): Update.
26649 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26650 * predict.c (predictable_edge_p): Update.
26651 (edge_probability_reliable_p): Update.
26652 (set_even_probabilities): Update.
26653 (combine_predictions_for_insn): Update.
26654 (combine_predictions_for_bb): Update.
26655 (propagate_freq): Update.
26656 (estimate_bb_frequencies): Update.
26657 (force_edge_cold): Update.
26658 * profile-count.c (profile_count::dump): Add missing space into dump.
26659 (profile_count::debug): Add newline.
26660 (profile_count::differs_from_p): Explicitly convert to unsigned.
26661 (profile_count::stream_in): Update.
26662 (profile_probability::dump): New member function.
26663 (profile_probability::debug): New member function.
26664 (profile_probability::differs_from_p): New member function.
26665 (profile_probability::differs_lot_from_p): New member function.
26666 (profile_probability::stream_in): New member function.
26667 (profile_probability::stream_out): New member function.
26668 * profile-count.h (profile_count_quality): Rename to ...
26669 (profile_quality): ... this one.
26670 (profile_probability): New.
26671 (profile_count): Update.
26672 * profile.c (compute_branch_probabilities): Update.
26673 * recog.c (peep2_attempt): Update.
26674 * sched-ebb.c (schedule_ebbs): Update.
26675 * sched-rgn.c (find_single_block_region): Update.
26676 (compute_dom_prob_ps): Update.
26677 (schedule_region): Update.
26678 * sel-sched-ir.c (compute_succs_info): Update.
26679 * stmt.c (struct case_node): Update.
26680 (do_jump_if_equal): Update.
26681 (get_outgoing_edge_probs): Update.
26682 (conditional_probability): Update.
26683 (emit_case_dispatch_table): Update.
26684 (expand_case): Update.
26685 (expand_sjlj_dispatch_table): Update.
26686 (emit_case_nodes): Update.
26687 * targhooks.c: Update.
26688 * tracer.c (better_p): Update.
26689 (find_best_successor): Update.
26690 * trans-mem.c (expand_transaction): Update.
26691 * tree-call-cdce.c: Update.
26692 * tree-cfg.c (gimple_split_edge): Upate.
26693 (move_sese_region_to_fn): Upate.
26694 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26695 * tree-eh.c (lower_resx): Upate.
26696 (cleanup_empty_eh_move_lp): Upate.
26697 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26698 * tree-inline.c (copy_edges_for_bb): Update.
26699 (copy_cfg_body): Update.
26700 * tree-parloops.c (gen_parallel_loop): Update.
26701 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26702 (gimple_gen_time_profiler): Update.
26703 * tree-ssa-dce.c (remove_dead_stmt): Update.
26704 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26705 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26706 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26707 (unloop_loops): Update.
26708 (try_peel_loop): Update.
26709 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26710 * tree-ssa-loop-split.c (connect_loops): Update.
26711 (split_loop): Update.
26712 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26713 (hoist_guard): Update.
26714 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26715 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26716 (value_replacement): Update.
26717 * tree-ssa-reassoc.c (branch_fixup): Update.
26718 * tree-ssa-tail-merge.c (replace_block_by): Update.
26719 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26720 (create_edge_and_update_destination_phis): Update.
26721 (compute_path_counts): Update.
26722 (recompute_probabilities): Update.
26723 (update_joiner_offpath_counts): Update.
26724 (freqs_to_counts_path): Update.
26725 (duplicate_thread_path): Update.
26726 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26727 (struct switch_conv_info): Update.
26728 (gen_inbound_check): Update.
26729 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26730 (vect_do_peeling): Update.
26731 (vect_loop_versioning): Update.
26732 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26733 (optimize_mask_stores): Update.
26734 * ubsan.c (ubsan_expand_null_ifn): Update.
26735 * value-prof.c (gimple_divmod_fixed_value): Update.
26736 (gimple_divmod_fixed_value_transform): Update.
26737 (gimple_mod_pow2): Update.
26738 (gimple_mod_pow2_value_transform): Update.
26739 (gimple_mod_subtract): Update.
26740 (gimple_mod_subtract_transform): Update.
26741 (gimple_ic): Update.
26742 (gimple_stringop_fixed_value): Update.
26743 (gimple_stringops_transform): Update.
26744 * value-prof.h: Update.
26746 2017-06-29 Carl Love <cel@us.ibm.com>
26748 * config/rs6000/rs6000-c.c: Add support for built-in functions
26749 vector signed int vec_signed (vector float);
26750 vector signed long long vec_signed (vector double);
26751 vector signed int vec_signed2 (vector double, vector double);
26752 vector signed int vec_signede (vector double);
26753 vector signed int vec_signedo (vector double);
26754 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26755 instruction generator.
26756 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26757 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26758 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26760 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26761 vunsignede_v2df): Add define_expands.
26762 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26763 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26764 VEC_UNSIGNEDO): Add definitions.
26765 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26766 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26767 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26768 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26769 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26770 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26771 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26772 vec_unsignede, vec_unsignedo): Add builtin defines.
26773 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26775 * doc/extend.texi: Update the built-in documentation file for the
26776 new built-in functions.
26778 2017-06-29 Richard Biener <rguenther@suse.de>
26780 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26781 reduction chains to LOOP_VINFO_REDUCTIONS.
26782 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26783 SLP reductions after processing reduction chains.
26785 2017-06-29 Nathan Sidwell <nathan@acm.org>
26787 * builtins.c (fold_builtin_FUNCTION): Use
26788 lang_hooks.decl_printable_name.
26790 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26792 PR middle-end/81194
26793 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26794 with only one label.
26795 * stmt.c (expand_case): Assert NCASES is greater than one.
26797 2017-06-29 Richard Biener <rguenther@suse.de>
26799 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26801 (group_case_labels): Likewise.
26802 (find_taken_edge): Push sanity checking on val to workers...
26803 (find_taken_edge_cond_expr): ... here
26804 (find_taken_edge_switch_expr): ... and here, handle cases
26805 with just a default label.
26806 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26807 (group_case_labels): Likewise.
26808 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26809 group_case_labels does anything cleanup the CFG again.
26811 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26813 PR tree-optimization/81196
26814 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26815 exit condition comparing two IVs.
26817 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26819 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26820 profile to the dummy entry at the end of the list of architectures.
26821 * config/arm/arm-cpu-cdata.h: Regenerated.
26823 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26824 Michael Collison <michael.collison@arm.com>
26827 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26829 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26830 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26831 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26832 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26833 of shift when the shift amount is masked with constant equal to
26834 the size of the mode.
26835 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26838 2017-06-29 Martin Liska <mliska@suse.cz>
26840 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26842 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26845 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26848 (edge_growth_cache_entry::edge_growth_cache_entry): New
26850 (reset_edge_growth_cache): Update to use constructor.
26852 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26854 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26855 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26856 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26858 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26860 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26861 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26863 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26865 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26866 (*-linux-uclibc*): Add t-uclibc tmake_file.
26867 * config/t-musl: New.
26868 * config/t-uclibc: New.
26870 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26872 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26874 (gen_comm_data): Emit architectural setting of arch_prof.
26875 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26877 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26878 (armv8-m.base, armv8-m.main): Likewise.
26879 * arm-protos.h (arm_build_target): Add profile field.
26880 (arch_option): Likewise.
26881 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26883 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26884 arm_active_target.profile.
26886 2017-06-28 Richard Biener <rguenther@suse.de>
26888 PR middle-end/81227
26889 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26890 TYPE_OVERFLOW_WRAPS.
26891 * match.pd (negate_expr_p): Likewise.
26892 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26893 fold_build2, not fold_binary.
26895 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26897 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26898 Convert memory address to Pmode.
26899 (aarch64_print_operand): Assert MEM operands are always Pmode.
26901 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26904 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26905 Remove redundant if.
26906 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26907 * config/arm/aarch-common-protos.h
26908 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26909 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26911 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26914 * multiple_target.c (create_dispatcher_calls): Set the default
26915 clone to be static, not public.
26917 2017-06-28 Richard Biener <rguenther@suse.de>
26919 * tree-vect-loop.c (vectorizable_reduction): Move special
26920 cond reduction IV var creation ...
26921 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26922 parameter. Use STMT_VINFO_VECTYPE.
26923 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26926 2017-06-28 Martin Liska <mliska@suse.cz>
26929 * ipa-visibility.c (non_local_p): Handle visibility.
26931 2017-06-28 Martin Liska <mliska@suse.cz>
26934 * common.opt: Add IntegerRange to various options.
26935 * opt-functions.awk (integer_range_info): New function.
26936 * optc-gen.awk: Add integer_range_info to cl_options struct.
26937 * opts-common.c (decode_cmdline_option): Handle
26938 CL_ERR_INT_RANGE_ARG.
26939 (cmdline_handle_error): Likewise.
26940 * opts.c (print_filtered_help): Show valid interval in
26941 when --help is provided.
26942 * opts.h (struct cl_option): Add range_min and range_max fields.
26943 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26945 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26947 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26948 (x * C EQ/NE y * C): New transformation.
26950 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26952 * genmultilib (combination_space): Accept '+' in option names.
26954 2017-06-28 Martin Liska <mliska@suse.cz>
26957 * asan.c (instrument_derefs): Bail out inner references
26958 that are hard register variables.
26960 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26963 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26964 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26966 2017-06-28 Richard Biener <rguenther@suse.de>
26968 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26969 (vect_get_slp_defs): Adjust.
26970 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26972 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26974 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26975 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26976 (vectorizable_reduction): Adjust.
26977 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26979 (vect_get_slp_defs): Likewise.
26980 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26981 (vectorizable_bswap): Adjust.
26982 (vectorizable_call): Likewise.
26983 (vectorizable_conversion): Likewise.
26984 (vectorizable_assignment): Likewise.
26985 (vectorizable_shift): Likewise.
26986 (vectorizable_operation): Likewise.
26987 (vectorizable_store): Likewise.
26988 (vectorizable_condition): Likewise.
26989 (vectorizable_comparison): Likewise.
26991 2017-06-28 Michael Collison <michael.collison@arm.com>
26994 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26996 (arm_gen_movmemqi): Removed unused variable 'i'.
26997 Convert 'for' loop into 'while' loop.
26998 (arm_expand_prologue): Remove last unnecessary set of insn.
26999 (thumb_pop): Remove unused variable 'pushed_words'.
27000 (thumb_exit): Remove last unnecessary set of regs_to_pop.
27002 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27004 * config/s390/predicates.md: Use s390_rel_address_ok_p.
27005 * config/s390/s390-protos.h: Add prototype of
27006 s390_rel_address_ok_p.
27007 * config/s390/s390.c (s390_got_symbol): New function.
27008 (s390_rel_address_ok_p): New function.
27009 (legitimize_pic_address): Use s390_rel_address_ok_p.
27010 (s390_load_got): Use s390_got_symbol.
27011 (s390_option_override): Issue error if
27012 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
27013 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
27015 * config/s390/s390.opt: New option mpic-data-is-text-relative.
27017 2017-06-27 Andrew Pinski <apinski@cavium.com>
27019 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
27020 (X * copysign (1.0, X)): New pattern.
27021 (X * copysign (1.0, -X)): New pattern.
27022 (copysign (-1.0, CST)): New pattern.
27024 2017-06-27 Joseph Myers <joseph@codesourcery.com>
27026 * genmultilib (combination_space): Remove variable.
27027 Validate reuse rules against regular expression for any sequence
27028 of multilib options in any order.
27030 2017-06-27 Michael Collison <michael.collison@arm.com>
27032 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
27033 call aarch64_split_simd_combine.
27034 * (aarch64_combine_internal<mode>): Delete pattern.
27035 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
27036 Allow register and subreg operands.
27038 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27040 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
27041 specific need, just fallback on defaults.
27042 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
27044 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27045 Olivier Hainque <hainque@adacore.com>
27047 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
27049 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
27050 targets. Pick a default if no particular attempt applied.
27051 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
27054 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27056 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
27057 (x86_64-wrs-vxworks7): Likewise.
27059 2017-06-27 Marek Polacek <polacek@redhat.com>
27062 * ubsan.c (instrument_null): Check get_base_address's result for null.
27064 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27066 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27068 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27070 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27071 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27072 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27073 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27074 New function types.
27075 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27076 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27077 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27078 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27079 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27080 BUILT_IN_FEUPDATEENV): New builtins.
27081 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27082 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27083 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27084 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27086 (builtin_structptr_types): Adjust size.
27087 * tree.c (builtin_structptr_types): Add four entries.
27089 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27090 Olivier Hainque <hainque@adacore.com>
27092 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27093 (TLS_SYM): New local macro, forcing reference to __tls__ on
27094 link command lines for VxWorks 7 RTPs, triggering initialization
27096 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27097 OS features TLS support, true for RTPs on VxWorks 7.
27098 * config/vxworks.c (vxworks_override_options): Setup emutls
27101 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27103 * predict.c (test_prediction_value_range): Use -1U instead of -1
27104 to avoid narrowing conversion warning.
27105 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27106 to avoid narrowing conversion warning.
27107 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27109 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27111 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27113 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27114 64bit configurations.
27115 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27116 (SIZE_TYPE): Likewise.
27117 * config/vxworks.c (vxworks_emutls_var_fields): Use
27118 long_unsigned_type_node instead of unsigned_type_node as the offset
27119 field type, which is "pointer" mode in emutls.c.
27121 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27124 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27126 PR middle-end/81207
27127 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27128 gimple_vuse copying separately from gimple_vdef copying.
27130 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27132 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27135 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27136 Olivier Hainque <hainque@adacore.com>
27138 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27139 for all vxworks7 targets.
27140 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27141 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27142 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27143 variations for VX6/VX7 and 32/64bits later on in ...
27144 (VXWORKS_LIB_SPEC): Leverage new macros.
27145 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27146 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27148 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27150 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27151 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27153 2017-06-26 Carl Love <cel@us.ibm.com>
27155 * config/rs6000/rs6000-c.c: Add support for built-in functions
27156 vector bool char vec_reve (vector bool char);
27157 vector signed char vec_reve (vector signed char);
27158 vector unsigned char vec_reve (vector unsigned char);
27159 vector bool int vec_reve (vector bool int);
27160 vector signed int vec_reve (vector signed int);
27161 vector unsigned int vec_reve (vector unsigned int);
27162 vector bool long long vec_reve (vector bool long long);
27163 vector signed long long vec_reve (vector signed long long);
27164 vector unsigned long long vec_reve (vector unsigned long long);
27165 vector bool short vec_reve (vector bool short);
27166 vector signed short vec_reve (vector signed short);
27167 vector double vec_reve (vector double);
27168 vector float vec_reve (vector float);
27169 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27170 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27171 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27172 (altivec_vreve): New pattern.
27173 * config/rs6000/altivec.h (vec_reve): New define.
27174 * doc/extend.texi (vec_rev): Update the built-in documentation file
27175 for the new built-in functions.
27177 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27179 PR tree-optimization/71815
27180 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27182 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27184 (slsr_process_phi): Likewise.
27185 (replace_uncond_cands_and_profitable_phis): Don't replace a
27186 multiply candidate with a stride of 1 (copy or cast).
27187 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27189 (lowest_cost_path): Likewise.
27190 (total_savings): Likewise.
27192 2017-06-26 Richard Biener <rguenther@suse.de>
27195 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27196 Use def_builtin_pure for all gather builtins.
27198 2017-06-26 Richard Biener <rguenther@suse.de>
27200 PR tree-optimization/81203
27201 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27202 non-dominating BBs.
27204 2017-06-26 Marek Polacek <polacek@redhat.com>
27207 * doc/invoke.texi: Document -Wmultistatement-macros.
27209 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27211 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27212 arm_neon_ok_no_float_abi effective target.
27214 2017-06-26 Richard Biener <rguenther@suse.de>
27216 PR tree-optimization/80928
27217 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27218 (copy_bbs): Set BB_DUPLICATED flag early.
27219 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27221 (execute_on_shrinking_pred): Likewise.
27222 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27223 BB_DUPLICATED blocks.
27224 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27225 iterate over all PHIs considering removal of *gsi.
27227 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27229 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27232 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27234 * config/rs6000/rs6000-string.c: (expand_block_clear,
27235 do_load_for_compare, select_block_compare_mode,
27236 compute_current_alignment, expand_block_compare,
27237 expand_strncmp_align_check, expand_strn_compare,
27238 expand_block_move, rs6000_output_load_multiple)
27239 Move functions related to string/block move/compare
27240 to a separate file.
27241 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27242 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27243 for this function which is now used in two files.
27244 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27245 * config.gcc: Add rs6000-string.o to extra_objs for
27246 targets powerpc*-*-* and rs6000*-*-*.
27248 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27251 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27252 32-bit, since indexed is not valid for DImode.
27253 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27254 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27255 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27256 (define_peephole2 for Altivec d-form store): Likewise.
27259 * multiple_target.c (create_dispatcher_calls): Only create the
27260 dispatcher call if the function is the default clone of a
27261 versioned function.
27263 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27265 PR middle-end/80902
27266 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27267 a call, force the call to not be a tail call.
27269 2017-06-23 Jeff Law <law@redhat.com>
27271 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27274 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27276 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27277 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27278 handling for early expansion of vector shifts (sl,sr,sra,rl).
27279 (builtin_function_type): Add vector shift right instructions
27280 to the unsigned argument list.
27282 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27284 rtl-optimizatoin/79286
27285 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27286 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27287 trap. PIC register plus a const unspec without offset can never trap.
27289 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27291 * tree.h (builtin_structptr_type): New type.
27292 (builtin_structptr_types): Declare new array.
27293 * tree.c (builtin_structptr_types): New array.
27294 (free_lang_data, build_common_tree_nodes): Use it.
27296 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27299 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27302 2017-06-22 Matt Turner <mattst88@gmail.com>
27304 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27305 Lake models to skylake case. Assume skylake for unknown
27306 models with clflushopt.
27308 2017-06-22 Jeff Law <law@redhat.com>
27310 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27311 frame sizes that do not satisfy aarch64_uimm12_shift.
27313 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27315 * profile-count.h (apply_probability,
27316 apply_scale, probability_in): Fix checks for zero.
27318 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27320 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27321 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27323 2016-06-22 Richard Biener <rguenther@suse.de>
27325 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27326 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27327 REDUC_MAX_EXPR support.
27328 (vectorizable_reduction): Likewise.
27329 (vect_create_epilog_for_reduction): Likewise.
27331 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27333 * match.pd (A / (1 << B) -> A >> B): New.
27334 * generic-match-head.c: Include optabs-tree.h.
27335 * gimple-match-head.c: Likewise.
27336 * optabs-tree.h (target_supports_op_p): New.
27337 * optabs-tree.c (target_supports_op_p): New.
27339 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27341 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27342 $gcc_cv_ld --help output.
27343 (gcc_cv_ld_demangle): Likewise.
27344 (gcc_cv_ld_eh_frame_hdr): Likewise.
27345 (gcc_cv_ld_pie): Likewise.
27346 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27347 (gcc_cv_ld_buildid): Likewise.
27348 (gcc_cv_ld_sysroot): Likewise.
27349 (ld_bndplt_support): Likewise.
27350 (ld_pushpopstate_support): Likewise.
27351 * configure: Regenerate.
27352 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27354 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27357 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27358 operands indexes to avoid gap between operands and match_dups.
27360 2017-06-21 Andrew Pinski <apinski@cavium.com>
27362 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27363 Increment Arith_shift and Arith_shift_reg by 1.
27364 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27366 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27367 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27368 (aarch64_strip_extend): Add new argument and test for it.
27369 (aarch64_cheap_mult_shift_p): New function.
27370 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27371 add a cost if it is true.
27372 Update calls to aarch64_strip_extend.
27373 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27375 2017-06-21 Andrew Pinski <apinski@cavium.com>
27377 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27379 (thunderxt88): Likewise.
27380 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27381 (thunderx_prefetch_tune): New variable.
27382 (thunderx2t99_prefetch_tune): Update for the correct values.
27383 (thunderxt88_tunings): New variable.
27384 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27385 generic_prefetch_tune.
27386 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27388 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27390 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27391 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27392 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27393 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27394 (aarch64_atomic_cas<mode>, GPI): Likewise.
27396 2017-06-21 Martin Liska <mliska@suse.cz>
27398 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27399 statements on cold and hot labels.
27400 * predict.c (tree_estimate_probability_bb): Remove the
27401 prediction from this place.
27403 2017-06-21 Martin Liska <mliska@suse.cz>
27405 PR tree-optimization/79489
27406 * gimplify.c (maybe_add_early_return_predict_stmt): New
27408 (gimplify_return_expr): Call the function.
27409 * predict.c (tree_estimate_probability_bb): Remove handling
27411 * predict.def: Update comment about early return predictor.
27412 * gimple-predict.h (is_gimple_predict): New function.
27413 * predict.def: Change default value of early return to 66.
27414 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27416 * passes.def: Put pass_strip_predict_hints to the beginning of
27419 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27421 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27422 FUNCTION_DECL declarations.
27423 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27425 (dwaf2out_decl): Likewise.
27426 * godump.c (go_early_global_decl): Skip call to the real debug hook
27427 for FUNCTION_DECL declarations.
27428 * passes.c (rest_of_decl_compilation): Skip call to the
27429 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27430 -fdump-go-spec is passed.
27432 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27434 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27435 Reorder for compactness.
27436 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27437 (def_builtin_pure, def_builtin_pure2): New functions.
27438 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27440 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27442 * match.pd (nop_convert): New predicate.
27443 ((A +- CST1) +- CST2): Allow some NOP conversions.
27445 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27448 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27449 with ctors/dtors if GOVD_SHARED is set.
27451 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27453 * config/aarch64/aarch64.md (movti_aarch64):
27454 Emit mov rather than orr.
27455 (movtf_aarch64): Likewise.
27456 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27457 Emit mov rather than orr.
27459 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27461 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27462 Swap alternatives, make integer dup more expensive.
27464 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27466 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27467 Return true for non-tls symbols.
27469 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27471 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27472 (cortex-a75): Likewise.
27473 (cortex-a75.cortex-a55): Likewise.
27474 * config/aarch64/aarch64-tune.md: Regenerate.
27475 * doc/invoke.texi (-mtune): Document new values for -mtune.
27477 2017-06-21 Tom de Vries <tom@codesourcery.com>
27479 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27480 stack_size feature.
27481 (Effective-Target Keywords, Other attributes): Suggest using
27482 dg-add-options stack_size feature to get stack limit in stack_size
27483 effective target documentation.
27485 2017-06-21 Julian Brown <julian@codesourcery.com>
27486 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27488 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27489 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27490 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27492 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27493 attribute type list for neon_multiply.
27494 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27495 attribute type list for neon_multiply.
27496 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27497 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27498 attribute type list for neon_multiply.
27499 * config/arm/types.md (crypto_pmull): Add.
27500 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27501 attribute type list.
27503 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27505 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27508 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27510 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27511 to make sure not to dereference a NULL cost_classes_ptr pointer.
27513 2017-06-20 Carl Love <cel@us.ibm.com>
27515 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27516 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27517 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27518 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27519 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27520 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27521 VMULOSW): New enum "unspec" values.
27522 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27523 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27524 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27525 altivec_vmulosw): New patterns.
27526 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27527 VMULOSW): Add definitions.
27529 2017-06-20 Julia Koval <julia.koval@intel.com>
27531 * config/i386/i386.c: Fix rounding expand for new pattern.
27532 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27534 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27536 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27537 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27539 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27541 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27544 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27546 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27547 architecture, then by implementer ID.
27548 * config/aarch64/aarch64-tune.md: Regenerate.
27550 2017-06-20 Richard Biener <rguenther@suse.de>
27552 PR middle-end/81097
27553 * fold-const.c (split_tree): Fold to type before negating.
27555 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27557 * diagnostic-show-locus.c
27558 (selftest::test_fixit_deletion_affecting_newline): New function.
27559 (selftest::diagnostic_show_locus_c_tests): Call it.
27561 2017-06-20 Andreas Schwab <schwab@suse.de>
27564 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27567 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27569 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27570 __ARM_FEATURE_COPROC according to support.
27572 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27574 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27575 Rewritten to avoid overflow for > 32-bit pointers.
27578 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27579 by removing enum keyword.
27580 (ubsan_type_descriptor): Likewise. Formatting fix.
27583 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27584 splitter): Require TARGET_SSE2 in the condition.
27586 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27589 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27590 for doing vector set of SFmode on ISA 3.0.
27591 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27592 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27594 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27595 SFmode value into a V4SF variable that was extracted from another
27596 V4SF variable without converting the element to double precision
27597 and back to single precision vector format.
27598 (vsx_insert_extract_v4sf_p9_2): Likewise.
27600 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27602 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27603 in UWHI to avoid undefined overflow.
27606 * ubsan.h (enum ubsan_encode_value_phase): New.
27607 (ubsan_encode_value): Change second argument to
27608 enum ubsan_encode_value_phase with default value of
27609 UBSAN_ENCODE_VALUE_GENERIC.
27610 * ubsan.c (ubsan_encode_value): Change second argument to
27611 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27612 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27613 create_tmp_var_raw instead of create_tmp_var and use a
27615 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27616 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27617 ubsan_encode_value callers.
27620 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27621 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27622 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27624 2017-06-19 Richard Biener <rguenther@suse.de>
27626 PR middle-end/81118
27627 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27628 estimates if we changed anything.
27630 2017-06-19 Richard Biener <rguenther@suse.de>
27632 PR tree-optimization/80887
27633 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27634 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27635 simplified lookups, then reset mprts_hook.
27636 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27638 (try_to_simplify): Likewise.
27640 2017-06-19 Martin Liska <mliska@suse.cz>
27643 * gimplify.c (gimplify_switch_expr):
27644 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27646 2017-06-19 Martin Liska <mliska@suse.cz>
27648 * doc/install.texi: Document that PGO runs in 4 stages.
27650 2017-06-19 Martin Liska <mliska@suse.cz>
27653 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27654 to dispatcher function name.
27655 * multiple_target.c (replace_function_decl): New function.
27656 (create_dispatcher_calls): Redirect both edges and references.
27658 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27660 * profile-count.c (profile_count::dump): Dump quality.
27661 (profile_count::differs_from_p): Update for unsigned val.
27662 * profile-count.h (profile_count_quality): New enum.
27663 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27665 2017-06-19 Richard Biener <rguenther@suse.de>
27667 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27668 struct function as arg.
27669 (estimate_numbers_of_iterations): Export overload with loop arg.
27670 (free_numbers_of_iterations_estimates_loop): Use an overload of
27671 free_numbers_of_iterations_estimates instead.
27672 * tree-cfg.c (remove_bb): Adjust.
27673 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27674 * tree-parloops.c (gen_parallel_loop): Likewise.
27675 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27677 (tree_unroll_loops_completely): Likewise.
27678 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27679 Use an overload instead and export.
27680 (estimated_loop_iterations): Adjust.
27681 (max_loop_iterations): Likewise.
27682 (likely_max_loop_iterations): Likewise.
27683 (estimate_numbers_of_iterations): Take struct function as arg
27685 (loop_exits_before_overflow): Adjust.
27686 (free_numbers_of_iterations_estimates_loop): Use an overload.
27687 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27688 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27690 2017-06-19 Richard Biener <rguenther@suse.de>
27693 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27694 RANGE_EXPR conservatively.
27696 2017-06-16 Carl Love <cel@us.ibm.com>
27698 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27699 definitions for vec_float, vec_float2, vec_floato,
27700 vec_floate built-ins.
27701 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27702 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27704 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27705 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27706 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27707 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27708 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27709 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27710 vec_floato): Add builtin defines.
27711 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27712 Update the built-in documentation file for the new built-in
27715 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27717 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27718 (mthumb): Mark as the negative of -marm.
27720 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27722 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27724 (ARM Options, -mtune): Document that this accepts the same
27725 extension options as -mcpu.
27726 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27728 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27730 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27731 permitted extensions.
27733 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27735 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27736 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27737 (armv8-m.main): Add option +nodsp.
27738 * config/arm/arm-cpu-cdata.h: Regenerated.
27740 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27742 * config/arm/t-fuchsia: New file.
27743 * config.gcc (arm*-*-fuchsia*): Use it.
27745 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27747 * config/arm/t-symbian: Rewrite for new option infrastructure.
27749 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27751 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27752 (MULTILIB_REQUIRED): Likewise.
27754 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27756 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27757 (MULTILIB_RESUE): Likewise.
27758 (MULTILIB_MATCHES): Likewise.
27759 (MULTLIB_REQUIRED): Likewise.
27761 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27763 * config/arm/t-rtems: Rewrite for new option framework.
27765 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27767 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27768 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27769 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27770 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27771 * config/arm/t-multilib: ... here.
27772 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27773 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27774 armv7-a and armv8*-a when A-profile libraries have not been built.
27775 * config/arm/t-rmprofile: Rewrite.
27777 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27779 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27780 with a backslash. Remove the backslash after substituting unescaped
27782 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27784 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27786 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27787 the driver to pass through the default setting of -mfloat-abi.
27788 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27790 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27791 (all_feat_combs): New rule.
27792 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27794 * config/arm/t-aprofile: Rewrite.
27796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27798 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27799 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27800 fpu is not specified by the user/command-line.
27801 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27802 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27803 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27804 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27805 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27806 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27808 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27810 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27811 * config/arm/t-arm-elf: Rewritten.
27813 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27815 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27816 have some floating-point instructions.
27817 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27818 (TARGET_MAYBE_HARD_FLOAT): New macro.
27819 * config/arm/arm-builtins.c (arm_init_builtins): Use
27820 TARGET_MAYBE_HARD_FLOAT.
27821 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27823 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27825 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27826 (configargs.h): Include it.
27827 (arm_print_hint_for_fpu_option): New function.
27828 (arm_parse_fpu_option): New function.
27829 (candidate_extension): New class.
27830 (arm_canon_for_multilib): New function.
27831 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27832 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27833 (ARCH_CANONICAL_SPECS): New macro.
27834 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27836 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27838 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27839 are set after handling multilib fragments. Set target_cpu_default2
27842 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27844 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27846 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27847 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27849 (check_arch): Likewise.
27850 * config/arm/arm.c (arm_configure_build_target): Handle
27851 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27852 options in the default.
27854 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27856 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27857 when an option is an alias of another.
27858 * config/arm/parsecpu.awk (optalias): New parser token.
27859 (gen_comm_data): Mark non-alias options as such. Emit entries
27860 for extension aliases.
27861 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27862 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27863 (armv6kz, armv6zk, armv6t2): Likewise.
27864 (armv7): Make vfpv3-d16 an alias.
27865 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27867 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27868 Sort in canonical order.
27869 (armv8-a): Sort in canonical order.
27870 (armv8.1-a, armv8.2-a): Likewise.
27871 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27873 (cortex-a9): Sort in canonical order.
27874 * config/arm/arm.c (selftests.h): Include it.
27875 (arm_test_cpu_arch_data): New function.
27876 (arm_run_self_tests): New function.
27877 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27878 (targetm): Move declaration to the end of the file.
27879 * arm-cpu-cdata.h: Regenerated.
27881 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27883 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27884 call to target_mode_check describing the type of option passed.
27885 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27886 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27887 arm_parse_cpu_option_name to match parameters against list of
27889 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27890 arm_arch_core_flags data structure.
27891 * config/arm/arm-cpu_cdata.h: Regenerated.
27893 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27895 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27897 (arm_print_hint_for_cpu_option): Likewise.
27898 (arm_print_hint_for_arch_option): Likewise.
27899 (arm_parse_cpu_option_name): Likewise.
27900 (arm_parse_arch_option_name): Likewise.
27901 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27902 of entries in the all_fpus list.
27903 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27904 (arm_parse_cpu_option_name): Declare.
27905 (arm_parse_arch_option_name): Declare.
27906 (arm_parse_option_features): Declare.
27907 (arm_intialize_isa): Declare.
27908 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27910 (gen_comm_data): ... here. Make definitions non-static.
27911 * config/arm/arm-cpu-data.h: Regenerated.
27912 * config/arm/arm-cpu-cdata.h: Regenerated.
27914 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27916 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27917 (cpu_arch_extension): New structure.
27918 (cpu_arch_option, arch_option, cpu_option): New structures.
27919 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27920 architecture types.
27921 (gen_data): Generate new format data tables.
27922 * config/arm/arm.c (cpu_tune): New structure.
27923 (cpu_option, processors): Delete.
27924 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27925 (arm_print_hint_for_cpu_option): ... this and ...
27926 (arm_print_hint_for_arch_option): ... this.
27927 (arm_parse_arch_cpu_name): Delete. Replace with ...
27928 (arm_parse_cpu_option_name): ... this and ...
27929 (arm_parse_arch_option_name): ... this.
27930 (arm_unrecognized_feature): Change type of target parameter to
27932 (arm_parse_arch_cpu_features): Delete. Replace with ...
27933 (arm_parse_option_features): ... this.
27934 (arm_configure_build_target): Rework to use new configuration data
27936 (arm_print_tune_info): Rework for new configuration data tables.
27937 * config/arm/arm-cpu-data.h: Regenerated.
27938 * config/arm/arm-cpu.h: Regenerated.
27940 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27942 * Makefile.in (OBJS): Move sbitmap.o from here ...
27943 (OBJS-libcommon): ... to here.
27945 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27947 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27948 (ISA_ALL_CRYPTO): New macro.
27949 (ISA_ALL_SIMD): New macro
27950 (ISA_ALL_FP): New macro.
27951 * config/arm/arm.c (fpu_bitlist): Update initializer.
27952 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27954 (arm9e): Add fpu. Add option for nofp
27955 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27956 (arm926ej-s, arm1026ej-s): Likewise.
27957 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27958 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27959 neon-fp16, neon-vfpv4, nofp and nosimd.
27960 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27961 (cortex-a8): Add fpu. Add option for nofp.
27962 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27963 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27964 (cortex-r4f): Add fpu.
27965 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27966 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27968 (cortex-r8): Likewise.
27969 (cortex-m4): Add fpu. Add option for nofp.
27970 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27971 (cortex-a17.cortex-a7): Likewise.
27972 (cortex-a32): Add fpu. Add options for crypto and nofp.
27973 (cortex-a35, cortex-a53): Likewise.
27974 (cortex-a57): Add fpu. Add option for crypto.
27975 (cortex-a72, cortex-a73): Likewise.
27976 (exynos-m1): Likewise.
27977 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27978 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27979 (cortex-m33): Add fpu. Add option for nofp.
27980 * config/arm/arm-cpu-cdata.h: Regenerated
27981 * config/arm/arm-cpu-data.h: Regenerated.
27983 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27985 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27986 (armv5te, armv5tej): Likewise.
27987 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27988 (armv7): Add options fp and vfpv3-d16.
27989 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27990 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27992 (armv7ve): Likewise.
27993 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27994 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27995 (armv8-a): Add nocrypto option.
27996 (armv8.1-a, armv8.2-a): Likewise.
27997 (armv8-m.main): add options fp, fp.dp and nofp.
27999 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28001 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
28003 (armv8-a+crc): Delete.
28004 (armv8.1-a): Add options simd, crypto and nofp.
28005 (armv8.2-a): Add options fp16, simd, crypto and nofp.
28006 (armv8.2-a+fp16): Delete.
28007 (armv8-m.main): Add option dsp.
28008 (armv8-m.main+dsp): Delete.
28009 (cortex-a8): Add fpu. Add nofp option.
28010 (cortex-a9): Add fpu. Add nofp and nosimd options.
28011 * config/arm/parsecpu.awk (gen_data): Generate option tables and
28012 link to main cpu and architecture data structures.
28013 (gen_comm_data): Only put isa attributes from the main architecture
28015 (option): New statement for architecture and CPU entries.
28016 * arm.c (struct cpu_option): New structure.
28017 (struct processors): Add entry for options.
28018 (arm_unrecognized_feature): New function.
28019 (arm_parse_arch_cpu_name): Ignore any characters after the first
28021 (arm_parse_arch_cpu_feature): New function.
28022 (arm_configure_build_target): Separate out any CPU and architecture
28023 features and parse separately. Don't error out if -mfpu=auto is
28024 used with only an architecture string.
28025 (arm_print_asm_arch_directives): New function.
28026 (arm_file_start): Call it.
28027 * config/arm/arm-cpu-cdata.h: Regenerated.
28028 * config/arm/arm-cpu-data.h: Likewise.
28029 * config/arm/arm-tables.opt: Likewise.
28031 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28033 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
28034 assembler when it is not -mfpu=auto.
28036 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28038 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
28039 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
28040 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
28041 (ASM_CPU_SPEC): Rewrite.
28042 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
28043 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
28044 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
28045 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
28046 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
28047 copied string is NUL-terminated. Also strip any characters prefixed
28049 (arm_rewrite_selected_arch): New function.
28050 (arm_rewrite_march): New function.
28052 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28054 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
28055 (x_arm_cpu_string, x_arm_tune_string): Likewise.
28056 (march, mcpu, mtune): Convert to string-based options.
28057 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
28058 (arm_parse_arch_cpu_name): New function.
28059 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28060 identify selected architecture or CPU.
28061 (arm_option_save): New function.
28062 (TARGET_OPTION_SAVE): Redefine.
28063 (arm_option_restore): Restore string options.
28064 (arm_option_print): Print string options.
28066 2017-06-16 Martin Sebor <msebor@redhat.com>
28068 PR tree-optimization/80933
28069 PR tree-optimization/80934
28070 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28071 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28072 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28073 (gimple_fold_builtin): Call them.
28075 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28077 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28078 as unlikely; update profile.
28080 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28082 * predict.c (force_edge_cold): Handle declaring edges impossible
28085 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28087 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28089 (try_unroll_loop_completely): Fix reporting.
28091 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28093 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28095 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28098 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28099 if given a non-constant argument for an intrinsic which requires a
28102 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28104 * profile.c (compare_freqs): New function.
28105 (branch_prob): Sort edge list.
28106 (find_spanning_tree): Assume that the list is priority sorted.
28108 2017-06-16 Richard Biener <rguenther@suse.de>
28110 PR tree-optimization/81090
28111 * passes.def (pass_record_bounds): Remove.
28112 * tree-pass.h (make_pass_record_bounds): Likewise.
28113 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28114 make_pass_record_bounds): Likewise.
28115 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28116 not free niter estimates at the beginning but at the end.
28117 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28119 2017-06-16 Richard Biener <rguenther@suse.de>
28121 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28122 initializer to workaround ICE in host GCC 4.8.
28124 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28126 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28128 (clone_inlined_nodes): Update.
28130 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28132 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28133 prefetch settings, and enable prefetching by default at -O3.
28135 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28137 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28138 Set flag_prefetch_loop_arrays according to tuning data.
28140 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28142 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28143 New tune structure.
28144 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28145 [Unrelated to main purpose of the patch] Place the pointer field last
28146 to enable type checking errors when tune structure are wrongly merged.
28147 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28148 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28149 (thunderx2t99_prefetch_tune): New tune constants.
28150 (tune_params *_tunings): Update all tunings (no functional change).
28151 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28152 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28153 from tunings structures.
28155 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28158 * ubsan.c (instrument_null): Add T argument, use it instead
28159 of computing it based on IS_LHS.
28160 (instrument_object_size): Likewise.
28161 (pass_ubsan::execute): Adjust instrument_null and
28162 instrument_object_size callers to pass gimple_get_lhs or
28163 gimple_assign_rhs1 result to it. Use instrument_null instead of
28164 calling get_base_address and instrument_mem_ref. Handle
28165 aggregate call arguments for object-size sanitization.
28167 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28169 PR tree-optimization/81089
28170 * tree-vrp.c (is_masked_range_test): Validate operands of
28173 2017-06-15 Martin Sebor <msebor@redhat.com>
28176 * dumpfile.c (dump_register): Avoid calling memset to initialize
28177 a class with a default ctor.
28178 * gcc.c (struct compiler): Remove const qualification.
28179 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28180 * hash-table.h: Ditto.
28181 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28183 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28184 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28186 * params.h (struct param_info): Make struct members non-const.
28187 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28188 with default initialization.
28189 * vec.h (vec_copy_construct, vec_default_construct): New helper
28191 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28192 with vec_copy_construct.
28193 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28194 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28195 * doc/invoke.texi (-Wclass-memaccess): Document.
28197 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28199 * emit-rtl.h (is_leaf): Update comment about local
28200 register allocator.
28202 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28205 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28206 for a variable to have a section before checking if the section has a
28208 Set section to.persistent if persistent attribute is set.
28209 Warn if .persistent attribute is used on an automatic variable.
28211 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28213 PR rtl-optimization/80474
28214 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28216 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28218 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28221 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28223 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28224 (casesi): Emit bounds checking as RTL.
28225 (casesi_internal_mips16_<mode>): Remove bounds checking.
28227 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28229 * config/xtensa/xtensa.c (xtensa_option_override): Append
28230 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28231 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28232 xtensa_doloop_hooks): Define unconditionally.
28233 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28235 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28236 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28237 for it in xtensa_option_override.
28238 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28239 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28241 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28243 * doc/cppopts.texi: Document '-' special value to -MF.
28245 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28247 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28248 (cortex_a53_fconst): Likewise.
28249 (cortex_a53_fpmul): Likewise.
28250 (cortex_a53_f_load_64): Likewise.
28251 (cortex_a53_f_load_many): Likewise.
28252 (cortex_a53_advsimd_alu): Likewise.
28253 (cortex_a53_advsimd_alu_q): Likewise.
28254 (cortex_a53_advsimd_mul): Likewise.
28255 (cortex_a53_advsimd_mul_q): Likewise.
28256 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28257 Add missing fmul, r2f_cvt and fconst cases.
28259 2017-06-14 Richard Biener <rguenther@suse.de>
28261 PR middle-end/81088
28262 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28264 (fold_binary_loc): When associating do not treat pre-existing
28265 TREE_OVERFLOW on literal constants as a reason to allow
28266 TREE_OVERFLOW on associated literal constants.
28268 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28270 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28271 (MASK_FEATURES): New macro.
28272 * config/sparc/sparc.c (sparc_option_override): Remove the special
28273 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28275 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28277 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28278 a division of 0 if non-call exceptions are enabled.
28280 2017-06-14 Andrew Pinski <apinski@cavium.com>
28281 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28284 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28285 Improve vector initialization code gen for only variable case.
28287 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28289 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28291 2017-06-14 Richard Biener <rguenther@suse.de>
28293 PR tree-optimization/81083
28294 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28297 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28299 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28300 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28301 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28302 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28303 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28304 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28306 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28308 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28309 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28311 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28313 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28315 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28317 * config/rs6000/t-rtems: Don't handle SPE.
28319 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28321 * config/rs6000/t-linux: Don't handle SPE.
28323 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28325 * config/rs6000/eabispe.h: Delete file.
28327 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28329 * config/rs6000/t-spe: Delete file.
28331 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28333 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28334 (rs6000_legitimate_offset_address_p): Return false for anything in
28335 V2SImode or V2SFmode.
28337 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28339 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28340 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28341 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28343 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28344 (rs6000_legitimate_offset_address_p): Ditto.
28345 (rs6000_emit_move): Ditto.
28346 (rs6000_init_builtins): Remove V4HI_type_node.
28348 2017-06-13 Martin Liska <mliska@suse.cz>
28351 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28352 (gate_asan): Likewise.
28353 * asan.h (asan_no_sanitize_address_p): Remove the function.
28354 (sanitize_flags_p): New function.
28355 * builtins.def: Fix coding style.
28356 * common.opt: Use renamed enum value.
28357 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28358 * doc/extend.texi: Document no_sanitize attribute.
28359 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28360 to SANITIZE_UNDEFINED_NONDEFAULT.
28361 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28362 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28363 Use sanitize_flags_p.
28364 * gimplify.c (gimplify_function_tree): Likewise.
28365 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28366 * opts.c (parse_no_sanitize_attribute): New function.
28367 (common_handle_option): Use renamed enum value.
28368 * opts.h (parse_no_sanitize_attribute): Declare.
28369 * tree.c (sanitize_flags_p): New function.
28370 * tree.h: Declared here.
28371 * tsan.c: Use sanitize_flags_p.
28372 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28373 (instrument_mem_ref): Likewise.
28374 (instrument_bool_enum_load): Likewise.
28375 (do_ubsan_in_current_function): Remove the function.
28376 (pass_ubsan::execute): Use sanitize_flags_p.
28377 * ubsan.h: Remove do_ubsan_in_current_function
28378 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28379 (dump_function_to_file): Use it here.
28381 2017-06-13 Martin Jambor <mjambor@suse.cz>
28383 PR tree-optimization/80803
28384 PR tree-optimization/81063
28385 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28386 (propagate_subaccesses_across_link): Enqueue subtree whenever
28387 necessary instead of relying on the caller.
28389 2017-06-13 Martin Jambor <mjambor@suse.cz>
28391 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28392 that have a first_link.
28393 (sort_and_splice_var_accesses): Do not check first_link before
28395 (subtree_mark_written_and_enqueue): Likewise.
28396 (propagate_all_subaccesses): Likewise and do not stop at first
28397 parent with a first_link.
28399 2017-06-13 Martin Jambor <mjambor@suse.cz>
28401 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28404 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28406 * match.pd: New pattern.
28408 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28410 * tree-vrp.c (is_masked_range_test): New function.
28411 (register_edge_assert_for): Determine ranges for
28414 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28416 PR tree-optimization/67328
28417 * fold-const.c (maskable_range_p): New function.
28418 (build_range_check): Generate bittests if possible.
28420 2017-06-13 Martin Liska <mliska@suse.cz>
28422 * gimple-pretty-print.c (dump_probability): Add new argument.
28423 (dump_edge_probability): Dump both probability and count.
28424 (dump_gimple_label): Likewise.
28425 (dump_gimple_bb_header): Likewise.
28427 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28430 * config/avr/avr-devices.c: Fix indentation.
28431 * config/avr/gen-avr-mmcu-specs.c: Dito.
28433 2017-06-13 Richard Biener <rguenther@suse.de>
28435 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28436 instead get vector type from stmt_info.
28437 (vectorizable_reduction): Adjust. Remove dead code.
28439 2017-06-13 Richard Biener <rguenther@suse.de>
28441 PR middle-end/81065
28442 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28443 case of C * (x * C2 + C3).
28444 (fold_addr_of_array_ref_difference): Properly fold index difference.
28446 2017-06-12 David S. Miller <davem@davemloft.net>
28449 * config/sparc/sparc.md (return expander): Emit frame blockage if
28450 function uses alloca.
28452 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28454 * combine.c (make_field_assignment): Check len rather than the mode
28455 precision when calling force_to_mode.
28457 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28459 Support multilibs and devices that see flash in RAM address range.
28462 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28463 (avr_mcu_t) <flash_pm_offset>: New field.
28464 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28465 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28466 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28467 (AVR_TINY_PM_OFFSET): Remove macro.
28468 * config/avr/avr.opt (-mshort-calls): New option.
28469 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28470 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28471 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28472 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28473 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28474 instead of avr_arch->have_jmp_call.
28475 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28476 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28477 avr_arch->flash_pm_offset to define.
28478 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28479 new field flash_pm_offset. Add entry for avrxmega3.
28480 (avr_texinfo): Add entry for avrxmega3.
28481 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28482 attiny212, attiny214,
28483 attiny412, attiny414, attiny416, attiny417,
28484 attiny814, attiny816, attiny817,
28485 attiny1614, attiny1616, attiny1617,
28486 attiny3214, attiny3216, attiny3217.
28487 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28488 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28489 (avr_print_operand_address) [AVR_TINY]: Same.
28490 (avr_asm_init_sections) <readonly_data_section>: Only patch
28491 callback if avr_arch->flash_pm_offset = 0.
28492 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28493 for rodata if avr_arch->flash_pm_offset != 0.
28494 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28495 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28496 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28497 (m_options): Append opt_rcall.
28498 (m_dirnames): Append dir_rcall.
28499 * config/avr/t-multilib: Regenerate.
28501 * configure.ac [target=avr]: Check whether avrxmega3 default
28502 linker description file works as needed.
28503 * configure: Regenerate.
28504 * doc/avr-mmcu.texi: Regenerate.
28505 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28506 <__AVR_ARCH__>: Document avrxmega3 and 103.
28507 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28508 <__AVR_SHORT_CALLS__>: Document it.
28509 <__AVR_PM_BASE_ADDRESS__>: Document it.
28510 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28511 (AVR Variable Attributes) <progmem>: Document this is
28512 not needed for avrxmega3.
28513 (AVR Named Address Spaces) <__flash>: Dito.
28515 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28517 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28519 2017-06-12 Doug Rupp <rupp@adacore.com>
28521 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28522 Append vxworks-stdint.h to the tm_file list.
28523 * config/vxworks-stdint.h: New file.
28525 2017-06-12 Martin Liska <mliska@suse.cz>
28527 PR tree-optimization/81041
28528 * tree-profile.c (gimple_gen_ic_func_profiler):
28529 Create an extra BB in profile-generate
28530 (gimple_gen_time_profiler): Likewise.
28532 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28534 PR tree-optimization/81003
28535 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28536 (update_range_test): Use it instead of force_gimple_operand_gsi.
28538 2017-06-12 Richard Biener <rguenther@suse.de>
28540 PR tree-optimization/81053
28541 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28542 with backedge value not defined in loop. Simplify def stmt
28545 2017-06-11 Tom de Vries <tom@codesourcery.com>
28548 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28550 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28551 nvptx_cannot_force_const_mem.
28553 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28555 * opts.c (finish_options): Move test for flag_split_stack after
28556 it has been initialized.
28558 2017-06-11 Jason Merrill <jason@redhat.com>
28560 * tree.h (id_equal): New.
28561 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28562 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28563 instead of strcmp of IDENTIFIER_POINTER.
28565 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28567 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28568 (mark_all_inlined_calls_cdtor): Fix formating.
28569 (inline_transform): Rescale profile before inlining.
28571 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28573 * cgraph.h (cgraph_edge::clone): Update prototype.
28574 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28575 (cgraph_node::create_clone): Update.
28576 (cgraph_node::create_version_clone): Update.
28577 * tree-inline.c (copy_bb): Update.
28578 (expand_call_inline): Update.
28580 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28582 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28583 factored out from ...
28584 (rs6000_emit_prologue): ... here.
28586 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28588 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28589 factored out from ...
28590 (rs6000_emit_prologue): ... here.
28592 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28594 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28596 (handle_missing_profiles): Fix computation of tp_first_run.
28597 (counts_to_freqs): Do not touch freqs when count is 0.
28599 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28601 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28604 2017-06-10 Tom de Vries <tom@codesourcery.com>
28606 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28607 attributes): Document signal effective target.
28609 2017-06-10 Tom de Vries <tom@codesourcery.com>
28611 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28612 Document effective target stack_size.
28614 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28616 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28617 to the edit_context if they can be auto-applied.
28619 2017-06-9 Ian Lance Taylor <iant@golang.org>
28621 * opts.c (finish_options): If -fsplit-stack, disable implicit
28622 -forder-blocks-and-partition.
28623 * doc/invoke.texi (Optimize Options): Document that when using
28624 -fsplit-stack -forder-blocks-and-partition is not implicitly
28627 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28629 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28630 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28631 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28632 * builtins.def (abort, trap, unreachable): Declare cold.
28633 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28634 * tree-core.h (ECF_COLD): New.
28635 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28636 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28638 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28640 * predict.c (unlikely_executed_stmt_p): Cleanup.
28642 2017-06-09 Richard Biener <rguenther@suse.de>
28644 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28645 model if the ref is always written to.
28647 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28649 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28651 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28653 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28656 2017-06-09 Tom de Vries <tom@codesourcery.com>
28659 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28660 "target cannot support label values" when encountering LABEL_REF.
28662 2017-06-09 Martin Liska <mliska@suse.cz>
28664 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28665 (gimple_gen_ic_func_profiler): Emit direct comparison
28666 of __gcov_indirect_call_callee with NULL.
28667 (gimple_gen_time_profiler): Change probability from
28668 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28670 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28672 * profile.c (edge_gcov_counts): Turn to pointer.
28673 (compute_branch_probabilities, compute_branch_probabilities): Update.
28674 (branch_prob): Do not clear edge_gcov_count.
28675 * profile.h (edge_gcov_counts): Turn to pointer.
28676 (edge_gcov_count): Update.
28678 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28680 * gimple.h (gimple_check_failed): Mark cold.
28682 2017-06-09 Richard Biener <rguenther@suse.de>
28684 PR tree-optimization/66623
28685 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28686 refactor check_reduction into two parts, properly computing
28687 whether we have to check reduction validity for outer loop
28690 2017-06-09 Richard Biener <rguenther@suse.de>
28692 PR tree-optimization/79483
28693 * graphite-scop-detection.c (order): New global.
28694 (get_order): Compute bb to order mapping that satisfies code
28695 generation constraints.
28696 (cmp_pbbs): New helper.
28697 (build_scops): Start domwalk at entry block, sort generated
28700 2017-06-09 Richard Biener <rguenther@suse.de>
28702 PR middle-end/81007
28703 * ipa-polymorphic-call.c
28704 (ipa_polymorphic_call_context::restrict_to_inner_class):
28705 Skip FIELD_DECLs with error_mark_node type.
28706 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28709 2017-06-09 Martin Liska <mliska@suse.cz>
28711 * predict.c (struct branch_predictor): New struct.
28712 (test_prediction_value_range): New test.
28713 (predict_c_tests): New function.
28714 * selftest-run-tests.c (selftest::run_tests): Run the function.
28715 * selftest.h: Declare new tests.
28717 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28720 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28721 gen_add3_insn did not fail.
28722 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28723 r0, construct that number in a temporary reg and add that reg to r0.
28724 If asked to put the result in r0 as well, fail.
28726 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28728 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28729 for early expansion of vec_eqv.
28731 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28733 PR middle-end/81005
28734 * ubsan.c (instrument_null): Avoid pointless code temporary.
28735 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28737 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28741 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28744 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28747 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28749 * predict.c (unlikely_executed_edge_p): Move ahead.
28750 (probably_never_executed_edge_p): Use it.
28752 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28754 PR middle-end/79988
28755 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28756 gimple_call_builtin_p call.
28758 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28760 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28761 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28762 rtl_check_failed_type2, rtl_check_failed_code1,
28763 rtl_check_failed_code2, rtl_check_failed_code_mode,
28764 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28765 rtvec_check_failed_bounds, rtl_check_failed_flag,
28766 _fatal_insn_not_found, _fatal_insn): Likewise.
28767 * tree.h (tree_contains_struct_check_failed,
28768 tree_check_failed, tree_not_check_failed,
28769 tree_class_check_failed, tree_range_check_failed,
28770 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28771 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28772 tree_operand_check_failed, omp_clause_check_failed,
28773 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28776 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28778 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28779 flag_branch_probabilities.
28780 * ipa-inline.c (edge_badness): Likewise.
28781 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28782 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28783 * predict.c (maybe_hot_frequency_p): Likewise.
28784 (probably_never_executed): Likewise.
28785 * sched-ebb.c (schedule_ebbs): Likewise.
28786 * sched-rgn.c (find_single_block_region): Likewise.
28787 * tracer.c (tail_duplicate): Likewise.
28789 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28791 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28792 longer requires x_flag_profile_use.
28794 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28796 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28797 instead of flag_reorder_blocks_and_partition.
28798 * dbxout.c (dbxout_function_end): Likewise.
28799 * dwarf2out.c (gen_subprogram_die): Likewise.
28800 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28801 * hw-doloop.c (reorg_loops): Likewise.
28802 * varasm.c (assemble_start_function,
28803 assemble_end_function): Likewise.
28804 (decide_function_section): Do not check for
28805 flag_reorder_blocks_and_partition.
28807 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28809 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28811 (chkp_get_hard_register_fake_addr_expr): Ditto.
28812 (chkp_build_addr_expr): Add check for hard reg case.
28813 (chkp_parse_array_and_component_ref): Ditto.
28814 (chkp_find_bounds_1): Ditto.
28815 (chkp_process_stmt): Don't generate bounds store for
28818 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28820 * predict.c (maybe_hot_bb_p): Do not check profile status.
28821 (maybe_hot_edge_p): Likewise.
28822 (probably_never_executed): Check for zero counts even if profile
28824 (unlikely_executed_edge_p): New function.
28825 (unlikely_executed_stmt_p): New function.
28826 (unlikely_executed_bb_p): New function.
28827 (set_even_probabilities): Use unlikely predicates.
28828 (combine_predictions_for_bb): Likewise.
28829 (predict_paths_for_bb): Likewise.
28830 (predict_paths_leading_to_edge): Likewise.
28831 (determine_unlikely_bbs): New function.
28832 (estimate_bb_frequencies): Use it.
28833 (compute_function_frequency): Use zero counts even if profile is
28835 * profile-count.h: Fix typo.
28837 2017-08-08 Julia Koval <julia.koval@intel.com>
28839 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28840 _mm512_mask_cvtsepi16_storeu_epi8,
28841 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28842 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28843 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28844 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28845 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28846 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28847 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28848 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28849 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28850 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28851 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28852 __builtin_ia32_pmovuswb256mem_mask,
28853 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28854 __builtin_ia32_pmovwb512mem_mask): New builtins.
28856 2017-08-08 Julia Koval <julia.koval@intel.com>
28858 PR target/73350,80862
28859 * config/i386/subst.md (round): Fix round pattern.
28860 * config/i386/i386.c (ix86_erase_embedded_rounding):
28861 Fix erasing rounding for the fixed pattern.
28863 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28865 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28867 2017-06-08 Martin Liska <mliska@suse.cz>
28869 PR gcov-profile/80911
28870 * gcov.c (block_info::block_info): New constructor.
28872 2017-06-07 Carl Love <cel@us.ibm.com>
28874 * config/rs6000/rs6000-c: The return type of the following
28875 built-in functions was implemented as int not long long. Fix sign
28876 of return value for the unsigned version of vec_mulo and vec_mule.
28877 vector unsigned long long vec_bperm (vector unsigned long long,
28878 vector unsigned char)
28879 vector signed long long vec_mule (vector signed int,
28881 vector unsigned long long vec_mule (vector unsigned int,
28882 vector unsigned int)
28883 vector signed long long vec_mulo (vector signed int,
28885 vector unsigned long long vec_mulo (vector unsigned int,
28886 vector unsigned int)
28887 * doc/extend.texi: Fix the documentation for the built-in
28890 2017-06-07 Carl Love <cel@us.ibm.com>
28893 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28896 2017-06-07 Carl Love <cel@us.ibm.com>
28898 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28899 support, Generate doublehv for signed int/float for BE case only.
28901 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28903 * doc/invoke.texi (mcx16): Rewrite.
28905 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28907 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28908 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28909 *mov<mode>_softfloat, and an anonymous splitter): Use
28910 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28912 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28914 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28916 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28917 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28918 (rs6000_debug_reg_global): Adjust.
28919 (rs6000_init_hard_regno_mode_ok): Adjust.
28920 (rs6000_dbx_register_number): Adjust.
28921 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28922 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28923 Remove SPE_ACC and SPEFSCR.
28924 (REG_ALLOC_ORDER): Ditto.
28925 (FRAME_POINTER_REGNUM): Change to 111.
28926 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28927 (REG_CLASS_NAMES): Ditto.
28928 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28929 (REGISTER_NAMES): Ditto.
28930 (ADDITIONAL_REG_NAMES): Ditto.
28931 (rs6000_reg_names): Ditto.
28932 * config/rs6000/rs6000.md: Renumber some register number
28935 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28937 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28939 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28940 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28942 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28943 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28944 Delete the SPE high registers.
28945 (REG_ALLOC_ORDER): Ditto.
28946 (enum reg_class): Remove SPE_HIGH_REGS.
28947 (REG_CLASS_NAMES): Ditto.
28948 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28949 (REGISTER_NAMES): Ditto.
28950 (rs6000_reg_names): Ditto.
28951 * doc/tm.texi.in: Remove SPE as example.
28952 * doc/tm.texi: Regenerate.
28954 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28956 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28957 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28958 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28959 * config/rs6000/rs6000.md (type): Remove "brinc".
28961 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28963 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28964 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28965 * config/rs6000/linuxspe.h: Delete file.
28966 * config/rs6000/rs6000.md: Don't include spe.md.
28967 * config/rs6000/spe.h: Delete file.
28968 * config/rs6000/spe.md: Delete file.
28969 * config/rs6000/t-rs6000: Remove spe.md.
28971 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28973 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28974 (reg_or_none500mem_operand): Delete.
28975 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28976 instead of reg_or_none500mem_operand.
28978 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28980 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28981 handling of SPE flags.
28982 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28984 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28986 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28988 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28989 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28990 paired_divv2sf3): Similar.
28991 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28992 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28993 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28995 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28996 Rename the paired_* instruction patterns.
28997 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28999 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
29000 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
29001 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
29002 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
29003 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
29004 PAIRED_VECTOR_MODE.
29005 (struct machine_function): Delete field spe_insn_chain_scanned_p.
29006 (spe_func_has_64bit_regs_p): Delete.
29007 (spe_expand_predicate_builtin): Delete.
29008 (spe_expand_evsel_builtin): Delete.
29009 (TARGET_DWARF_REGISTER_SPAN): Do not define.
29010 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
29011 (invalid_e500_subreg): Delete.
29012 (rs6000_legitimize_address): Always force_reg op2 as well, for
29013 paired single memory accesses.
29014 (rs6000_member_type_forces_blk): Delete.
29015 (rs6000_spe_function_arg): Delete.
29016 (rs6000_expand_unop_builtin): Delete SPE handling.
29017 (rs6000_expand_binop_builtin): Ditto.
29018 (spe_expand_stv_builtin): Delete.
29019 (bdesc_2arg_spe): Delete.
29020 (spe_expand_builtin): Delete.
29021 (spe_expand_predicate_builtin): Delete.
29022 (spe_expand_evsel_builtin): Delete.
29023 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
29024 (spe_init_builtins): Delete.
29025 (spe_func_has_64bit_regs_p): Delete.
29026 (savres_routine_name): Delete "info" parameter. Adjust callers.
29027 (rs6000_emit_stack_reset): Ditto.
29028 (rs6000_dwarf_register_span): Delete.
29029 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
29030 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
29031 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
29033 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
29035 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
29036 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
29037 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
29038 mulv2sf3, divv2sf3): Delete expanders.
29040 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29042 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
29044 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29046 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
29047 * config/rs6000/rs6000.c: Ditto.
29049 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29051 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
29052 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
29053 comparison_operator.
29055 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29057 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
29058 * config/rs6000/rs6000.opt: Ditto.
29059 * config/rs6000/t-rtems: Ditto.
29061 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29063 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29064 TARGET_E500_SINGLE by 0, simplify.
29065 * config/rs6000/rs6000.c: Ditto.
29066 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29067 (spe_build_register_parallel): Delete.
29068 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29069 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29070 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29071 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29072 (E500_CONVERT): Delete.
29073 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29075 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29077 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29078 * config/rs6000/dfp.md: Ditto.
29079 (negdd2, *negdd2_fpr): Merge.
29080 (absdd2, *absdd2_fpr): Merge.
29081 (negtd2, *negtd2_fpr): Merge.
29082 (abstd2, *abstd2_fpr): Merge.
29083 * config/rs6000/e500.h: Delete file.
29084 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29085 TARGET_FPRS by 1 and simplify.
29086 * config/rs6000/rs6000-c.c: Ditto.
29087 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29088 TARGET_DF_SPE by 0.
29089 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29091 * config/rs6000/rs6000.md: Ditto.
29092 (floatdidf2, *floatdidf2_fpr): Merge.
29093 (move_from_CR_gt_bit): Delete.
29094 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29095 (E500_CR_IOR_COMPARE): Delete.
29096 (All patterns that require !TARGET_FPRS): Delete.
29097 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29099 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29101 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29103 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29105 * graphds.c (add_edge): Intitialize edge's attached data.
29106 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29107 pointer parameter. Call pointed function on each edge during
29108 graph traversing. Skip traversing the edge when the function
29110 (graphds_dfs, graphds_scc): Ditto.
29111 (for_each_edge): New parameter. Pass the new parameter to callback
29113 * graphds.h (skip_edge_callback): New function pointer type.
29114 (graphds_dfs, graphds_scc): New function pointer parameter.
29115 (graphds_edge_callback, for_each_edge): New parameter.
29117 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29119 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29120 out code checking if runtime alias check is possible to below ...
29121 Call the new function.
29122 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29123 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29125 2017-06-07 Marek Polacek <polacek@redhat.com>
29128 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29129 TYPE_OVERFLOW_WRAPS check.
29131 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29133 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29134 if versioning is required.
29135 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29136 peeling with the check for versioning.
29138 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29140 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29141 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29142 Set true to new parameter if new ssa variable is defined.
29143 (vect_gen_vector_loop_niters): Refactor. Set range information
29144 for the new vector loop bound variable.
29145 (vect_do_peeling): Ditto.
29147 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29149 * tree-affine.c (ssa.h): Include header file.
29150 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29151 has wrapping overflow behavior.
29153 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29155 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29157 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29159 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29160 (tree_to_aff_combination): ... here.
29162 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29164 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29165 reg_pressure model function.
29166 (ivopts_global_cost_for_size): Delete.
29167 (determine_set_costs, iv_ca_recount_cost): Call new model function
29168 ivopts_estimate_reg_pressure.
29170 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29172 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29173 expensive than udiv. Remove floating point cases from mod.
29175 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29177 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29178 Increase idiv cost.
29180 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29182 * config/aarch64/aarch64.md
29183 (copysignsf3): Fix mask generation.
29185 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29187 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29189 (class dump_manager): Add register_dumps method.
29190 * dumpfile.c: Include langhooks.h.
29191 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29192 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29193 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29194 (dump_manager::dump_register): Start with 512 entries instead of 32.
29195 (dump_manager::register_dumps): New method.
29196 * toplev.c (general_init): Instead of invoking register_dumps
29197 langhook, invoke register_dumps method on the dump manager.
29198 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29201 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29203 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29204 Say that # requires an associated define_split to exist, and that
29205 the define_split must be suitable for use after register allocation.
29207 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29209 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29210 (compute_outgoing_frequencies): Also initialize zero counts.
29211 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29212 around loops; preserve more of profile when nothing changes.
29214 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29216 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29218 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29219 * config/arm/arm-cpu-cdata.h: Regenerate.
29220 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29221 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29222 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29223 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29224 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29226 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29228 * config/arm/t-rmprofile: Likewise.
29229 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29231 2017-06-06 David S. Miller <davem@davemloft.net>
29234 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29235 blockage if function uses alloca.
29237 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29239 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29240 New "uid" fields to hold pretty-print IDs of group and ref.
29241 Memory references are now identified as <group_id>:<ref_id>
29242 instead of using [random] addresses.
29243 (dump_mem_details): Simplify, no functional change.
29244 (dump_mem_ref): Simplify and make output more concise.
29245 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29246 (find_or_create_group): Initialize group uid.
29247 (record_ref): Initialize ref uid. Improve debug output.
29248 (prune_group_by_reuse, should_issue_prefetch_p,)
29249 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29250 (mark_nontemporal_store, determine_loop_nest_reuse):
29251 Improve debug output.
29253 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29255 * dbgcnt.def (prefetch): New debug counter.
29256 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29257 (schedule_prefetches): Stop issueing prefetches if debug counter
29260 2017-06-06 Tom de Vries <tom@codesourcery.com>
29262 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29263 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29265 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29267 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29268 Use aarch64_reg_or_zero predicate for operand 4.
29269 (aarch64_compare_and_swap<mode> define_insn_and_split):
29270 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29271 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29273 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29275 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29276 (arm_compute_save_core_reg_mask): This.
29277 (thumb1_compute_save_reg_mask): Rename into ...
29278 (thumb1_compute_save_core_reg_mask): This.
29279 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29280 (arm_compute_frame_layout): Likewise.
29282 2017-06-06 Richard Biener <rguenther@suse.de>
29284 PR tree-optimization/80974
29285 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29286 keep or clear leaders SSA info.
29288 2017-06-06 Tom de Vries <tom@codesourcery.com>
29290 * config/nvptx/nvptx.c (split_mode_p): New function.
29291 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29293 2017-06-06 Tom de Vries <tom@codesourcery.com>
29295 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29297 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29300 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29303 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29305 * shrink-wrap.c (handle_simple_exit): Update profile.
29306 (try_shrink_wrapping): Upate profile.
29308 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29310 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29311 (tree_guess_outgoing_edge_probabilities): New.
29312 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29313 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29315 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29317 * ipa-split.c (split_function): Initialize return bb profile.
29319 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29321 * profile.c (compute_branch_probabilities): Also initialize
29322 EXIT_BLOCK profile.
29324 2017-06-06 Richard Biener <rguenther@suse.de>
29326 PR tree-optimization/80928
29327 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29328 (vect_analyze_loop_operations): Properly guard analysis for
29330 (vect_transform_loop): Likewise.
29331 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29332 (vect_model_induction_cost): Do not cost for pure SLP.
29333 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29334 of induction in inner loop vectorization.
29335 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29336 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29337 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29339 (vect_analyze_slp_cost_1): Cost induction.
29340 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29341 (vect_get_slp_vect_defs): Likewise.
29342 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29343 (vect_transform_stmt): Handle SLP reductions.
29344 * tree-vectorizer.h (vectorizable_induction): Adjust.
29346 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29348 * config/rs6000/rs6000.c (make_resolver_func): Update
29349 init_lowered_empty_function call.
29351 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29353 * doc/invoke.texi: Document the -fprofile-abs-path option.
29354 * common.opt (fprofile-abs-path): New option.
29355 * gcov-io.h (gcov_write_filename): Declare.
29356 * gcov-io.c (gcov_write_filename): New function.
29357 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29358 * profile.c (output_location): Likewise.
29360 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29362 * shring-wrap.c: Revert accidental commit.
29364 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29366 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29368 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29370 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29372 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29373 profile in callgraph edge.
29374 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29375 (apply_scale): Likewise.
29376 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29377 Also scale profile when inlining function with zero profile.
29378 (initialize_cfun): Update exit block profile even when it is zero.
29379 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29380 when profile is read.
29382 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29384 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29385 (CLONE_*): New constants to define the processors we can generate
29386 code for with the target_clone attribute.
29387 (rs6000_clone_map): New array to identify which clone processors
29388 the current program is running on.
29389 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29390 target_clone attribute.
29391 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29392 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29393 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29394 (cpu_expand_builtin): Add support for target_clone attribute.
29395 (rs6000_valid_attribute_p): Allow "default" attribute.
29396 (get_decl_name): New debug function to simplify printing the
29397 current function name in debugging statements.
29398 (rs6000_clone_priority): New functions to support the target_clone
29399 attribute, and be able to generate code to switch between ISA 2.05
29400 through ISA 3.0 (power6 through power9).
29401 (rs6000_compare_version_priority): Likewise.
29402 (rs6000_get_function_versions_dispatcher): Likewise.
29403 (make_resolver_func): Likewise.
29404 (add_condition_to_bb): Likewise.
29405 (dispatch_function_versions): Likewise.
29406 (rs6000_generate_version_dispatcher_body): Likewise.
29407 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29408 (fusion_gpr_load_p): Fix a spacing issue.
29409 * doc/extend.texi (Common Function Attributes): Document that the
29410 PowerPC supports the target_clone attribute.
29412 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29414 * config/arm/arm.h: explain F symbol found in description of ARM
29415 register allocation in its legend.
29417 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29419 * config/mips/frame-header-opt.c: Include profile-count.h.
29420 * config/riscv/riscv.c: Include profile-count.h
29422 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29424 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29426 (sm_set_flag_if_changed): Add bbs field.
29427 (execute_sm_if_changed_flag_set): Pass BBS.
29428 (execute_sm): Update.
29430 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29432 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29435 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29437 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29438 (peephole2): New peephole2 to emit the above.
29439 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29441 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29443 * config/aarch64/aarch64.c (define_peephole2 above
29444 *sub_<shift>_<mode>): New peephole.
29446 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29448 * config/i386/i386.c (make_resolver_func): Update.
29449 * Makefile.in: Add profile-count.h and profile-count.o
29450 * auto-profile.c (afdo_indirect_call): Update to new API.
29451 (afdo_set_bb_count): Update.
29452 (afdo_propagate_edge): Update.
29453 (afdo_propagate_circuit): Update.
29454 (afdo_calculate_branch_prob): Update.
29455 (afdo_annotate_cfg): Update.
29456 * basic-block.h: Include profile-count.h
29457 (struct edge_def): Turn count to profile_count.
29458 (struct basic_block_def): Likewie.
29459 (REG_BR_PROB_BASE): Move to profile-count.h
29460 (RDIV): Move to profile-count.h
29461 * bb-reorder.c (max_entry_count): Turn to profile_count.
29462 (find_traces): Update.
29463 (rotate_loop):Update.
29464 (connect_traces):Update.
29465 (sanitize_hot_paths):Update.
29466 * bt-load.c (migrate_btr_defs): Update.
29467 * cfg.c (RDIV): Remove.
29468 (init_flow): Use alloc_block.
29469 (alloc_block): Uninitialize count.
29470 (unchecked_make_edge): Uninitialize count.
29471 (check_bb_profile): Update.
29472 (dump_edge_info): Update.
29473 (dump_bb_info): Update.
29474 (update_bb_profile_for_threading): Update.
29475 (scale_bbs_frequencies_int): Update.
29476 (scale_bbs_frequencies_gcov_type): Update.
29477 (scale_bbs_frequencies_profile_count): New.
29478 * cfg.h (update_bb_profile_for_threading): Update.
29479 (scale_bbs_frequencies_profile_count): Declare.
29480 * cfgbuild.c (compute_outgoing_frequencies): Update.
29481 (find_many_sub_basic_blocks): Update.
29482 * cfgcleanup.c (try_forward_edges): Update.
29483 (try_crossjump_to_edge): Update.
29484 * cfgexpand.c (expand_gimple_tailcall): Update.
29485 (construct_exit_block): Update.
29486 * cfghooks.c (verify_flow_info): Update.
29487 (dump_bb_for_graph): Update.
29488 (split_edge): Update.
29489 (make_forwarder_block): Update.
29490 (duplicate_block): Update.
29491 (account_profile_record): Update.
29492 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29493 (get_estimated_loop_iterations): Update.
29494 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29495 (single_likely_exit): Update.
29496 * cfgloopmanip.c (scale_loop_profile): Update.
29498 (set_zero_probability): Update.
29499 (lv_adjust_loop_entry_edge): Update.
29500 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29501 (purge_dead_edges): Update.
29502 (rtl_account_profile_record): Update.
29503 * cgraph.c (cgraph_node::create): Uninitialize count.
29504 (symbol_table::create_edge): Uninitialize count.
29505 (cgraph_update_edges_for_call_stmt_node): Update.
29506 (cgraph_edge::dump_edge_flags): Update.
29507 (cgraph_node::dump): Update.
29508 (cgraph_edge::maybe_hot_p): Update.
29509 * cgraph.h: Include profile-count.h
29510 (create_clone), create_edge, create_indirect_edge): Update.
29511 (cgraph_node): Turn count to profile_count.
29512 (cgraph_edge0: Likewise.
29513 (make_speculative, clone): Update.
29514 (create_edge): Update.
29515 (init_lowered_empty_function): Update.
29516 * cgraphclones.c (cgraph_edge::clone): Update.
29517 (duplicate_thunk_for_node): Update.
29518 (cgraph_node::create_clone): Update.
29519 * cgraphunit.c (cgraph_node::analyze): Update.
29520 (cgraph_node::expand_thunk): Update.
29521 * final.c (dump_basic_block_info): Update.
29522 * gimple-streamer-in.c (input_bb): Update.
29523 * gimple-streamer-out.c (output_bb): Update.
29524 * graphite.c (print_global_statistics): Update.
29525 (print_graphite_scop_statistics): Update.
29526 * hsa-brig.c: Include basic-block.h.
29527 * hsa-dump.c: Include basic-block.h.
29528 * hsa-gen.c (T sum_slice): Update.
29529 (convert_switch_statements):Update.
29530 * hsa-regalloc.c: Include basic-block.h.
29531 * ipa-chkp.c (chkp_produce_thunks): Update.
29532 * ipa-cp.c (struct caller_statistics): Update.
29533 (init_caller_stats): Update.
29534 (gather_caller_stats): Update.
29535 (ipcp_cloning_candidate_p): Update.
29536 (good_cloning_opportunity_p): Update.
29537 (get_info_about_necessary_edges): Update.
29538 (dump_profile_updates): Update.
29539 (update_profiling_info): Update.
29540 (update_specialized_profile): Update.
29541 (perhaps_add_new_callers): Update.
29542 (decide_about_value): Update.
29543 (ipa_cp_c_finalize): Update.
29544 * ipa-devirt.c (struct odr_type_warn_count): Update.
29545 (struct decl_warn_count): Update.
29546 (struct final_warning_record): Update.
29547 (possible_polymorphic_call_targets): Update.
29548 (ipa_devirt): Update.
29549 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29550 * ipa-icf.c (sem_function::merge): Update.
29551 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29552 * ipa-inline.c (compute_uninlined_call_time): Update.
29553 (compute_inlined_call_time): Update.
29554 (want_inline_small_function_p): Update.
29555 (want_inline_self_recursive_call_p): Update.
29556 (edge_badness): Update.
29557 (lookup_recursive_calls): Update.
29558 (recursive_inlining): Update.
29559 (inline_small_functions): Update.
29560 (dump_overall_stats): Update.
29561 (dump_inline_stats): Update.
29562 * ipa-profile.c (ipa_profile_generate_summary): Update.
29563 (ipa_propagate_frequency): Update.
29564 (ipa_profile): Update.
29565 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29566 * ipa-utils.c (ipa_merge_profiles): Update.
29567 * loop-doloop.c (doloop_modify): Update.
29568 * loop-unroll.c (report_unroll): Update.
29569 (unroll_loop_runtime_iterations): Update.
29570 * lto-cgraph.c (lto_output_edge): Update.
29571 (lto_output_node): Update.
29572 (input_node): Update.
29573 (input_edge): Update.
29574 (merge_profile_summaries): Update.
29575 * lto-streamer-in.c (input_cfg): Update.
29576 * lto-streamer-out.c (output_cfg): Update.
29577 * mcf.c (create_fixup_graph): Update.
29578 (adjust_cfg_counts): Update.
29579 (sum_edge_counts): Update.
29580 * modulo-sched.c (sms_schedule): Update.
29581 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29582 * predict.c (maybe_hot_count_p): Update.
29583 (probably_never_executed): Update.
29584 (dump_prediction): Update.
29585 (combine_predictions_for_bb): Update.
29586 (propagate_freq): Update.
29587 (handle_missing_profiles): Update.
29588 (counts_to_freqs): Update.
29589 (rebuild_frequencies): Update.
29590 (force_edge_cold): Update.
29591 * predict.h: Include profile-count.h
29592 (maybe_hot_count_p, counts_to_freqs): UPdate.
29593 * print-rtl-function.c: Do not include cfg.h
29594 * print-rtl.c: Include basic-block.h
29595 * profile-count.c: New file.
29596 * profile-count.h: New file.
29597 * profile.c (is_edge_inconsistent): Update.
29598 (correct_negative_edge_counts): Update.
29599 (is_inconsistent): Update.
29600 (set_bb_counts): Update.
29601 (read_profile_edge_counts): Update.
29602 (compute_frequency_overlap): Update.
29603 (compute_branch_probabilities): Update; Initialize and deinitialize
29605 (branch_prob): Update.
29606 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29607 (edge_gcov_count): New.
29608 (bb_gcov_count): New.
29609 * shrink-wrap.c (try_shrink_wrapping): Update.
29610 * tracer.c (better_p): Update.
29611 * trans-mem.c (expand_transaction): Update.
29612 (ipa_tm_insert_irr_call): Update.
29613 (ipa_tm_insert_gettmclone_call): Update.
29614 * tree-call-cdce.c: Update.
29615 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29616 (gimple_duplicate_sese_tail): Update.
29617 (gimple_account_profile_record): Update.
29618 (execute_fixup_cfg): Update.
29619 * tree-inline.c (copy_bb): Update.
29620 (copy_edges_for_bb): Update.
29621 (initialize_cfun): Update.
29622 (freqs_to_counts): Update.
29623 (copy_cfg_body): Update.
29624 (expand_call_inline): Update.
29625 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29626 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29627 (try_unroll_loop_completely): Update.
29628 (try_peel_loop): Update.
29629 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29630 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29631 * tree-ssa-loop-split.c (connect_loops): Update.
29632 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29633 * tree-ssa-reassoc.c (branch_fixup): Update.
29634 * tree-ssa-tail-merge.c (replace_block_by): Update.
29635 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29636 (compute_path_counts): Update.
29637 (update_profile): Update.
29638 (recompute_probabilities): Update.
29639 (update_joiner_offpath_counts): Update.
29640 (estimated_freqs_path): Update.
29641 (freqs_to_counts_path): Update.
29642 (clear_counts_path): Update.
29643 (ssa_fix_duplicate_block_edges): Update.
29644 (duplicate_thread_path): Update.
29645 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29646 (struct switch_conv_info): Update.
29647 * tree-tailcall.c (decrease_profile): Update.
29648 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29649 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29650 * value-prof.c (check_counter): Update.
29651 (gimple_divmod_fixed_value): Update.
29652 (gimple_mod_pow2): Update.
29653 (gimple_mod_subtract): Update.
29654 (gimple_ic_transform): Update.
29655 (gimple_stringop_fixed_value): Update.
29656 * value-prof.h (gimple_ic): Update.
29658 2017-06-02 Carl Love <cel@us.ibm.com>
29660 * config/rs6000/rs6000-c: Add support for built-in functions
29661 vector double vec_doublee (vector signed int);
29662 vector double vec_doublee (vector unsigned int);
29663 vector double vec_doublee (vector float);
29664 vector double vec_doubleh (vector signed int);
29665 vector double vec_doubleh (vector unsigned int);
29666 vector double vec_doubleh (vector float);
29667 vector double vec_doublel (vector signed int);
29668 vector double vec_doublel (vector unsigned int);
29669 vector double vec_doublel (vector float);
29670 vector double vec_doubleo (vector signed int);
29671 vector double vec_doubleo (vector unsigned int);
29672 vector double vec_doubleo (vector float);.
29673 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29674 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29676 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29677 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29678 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29680 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29681 vec_doublel, vec_doubleh.
29682 * doc/extend.texi: Update the built-in documentation file for the
29683 new built-in functions.
29685 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29688 * ipa-inline-analysis.c (free_growth_caches): Set
29689 edge_removal_hook_holder to NULL after removing it.
29691 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29693 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29694 comparision with zero.
29696 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29697 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29698 for early expansion of vec_min and vec_max builtins.
29699 (builtin_function_type): Add min/max unsigned variants to those
29700 identified as having unsigned arguments.
29702 2017-06-02 Olivier Hainque <hainque@adacore.com>
29704 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29706 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29708 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29709 Use VALL_F16 iterator rather than VALL.
29711 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29713 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29714 Emit CBNZ inside loop when doing a strong exchange and comparing
29715 against zero. Generate the CC flags after the loop.
29717 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29719 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29720 (dl_section_ref): New.
29721 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29722 On AIX, append an expression to subtract the size of the
29723 section length to dl_section_ref.
29725 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29727 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29728 for early expansion of vector absolute builtins.
29730 2017-06-02 Richard Biener <rguenther@suse.de>
29732 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29733 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29735 2017-06-02 Richard Biener <rguenther@suse.de>
29737 PR tree-optimization/80948
29738 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29739 stmt order as well.
29741 2017-06-02 Richard Biener <rguenther@suse.de>
29743 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29745 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29746 for inductions as relevant.
29748 2017-06-02 Richard Biener <rguenther@suse.de>
29750 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29751 (vectorizable_induction): ... this. Remove dead code.
29753 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29755 * builtins. (expand_builtin_alloca): Remove second parameter and
29756 infer its value from the first parameter instead.
29757 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29759 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29761 PR rtl-optimization/80903
29762 * loop-doloop.c (add_test): Unshare sequence.
29764 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29766 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29768 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29770 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29772 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29773 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29774 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29775 (xlogue_layout::get_stub_rtx): Make static.
29776 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29777 (xlogue_layout::compute_stub_managed_regs): Rename to...
29778 (xlogue_layout::count_stub_managed_regs): ...this.
29779 (xlogue_layout::is_stub_managed_reg): New function.
29780 (xlogue_layout::m_stub_names): Rename to...
29781 (xlogue_layout::s_stub_names): ...this, make static.
29782 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29783 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29784 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29785 xlogue_layout::s_stub_names): Instantiate statics.
29786 (stub_managed_regs): Remove.
29787 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29788 (disable_call_ms2sysv_xlogues): Rename to...
29789 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29790 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29792 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29793 change after reload_completed.
29794 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29796 (ix86_expand_prologue): Likewise.
29797 (ix86_expand_epilogue): Likewise.
29798 (ix86_expand_split_stack_prologue): Likewise.
29799 (ix86_compute_frame_layout): Remove frame parameter ...
29800 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29801 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29803 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29804 (ix86_frame): Move from here ...
29805 * config/i386/i386.h (ix86_frame): ... to here.
29806 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29807 complete ix86_frame data structure instead. Remove some_ld_name.
29809 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29811 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29812 symbols that hold a DECL_VALUE_EXPR.
29814 2017-06-01 Martin Jambor <mjambor@suse.cz>
29816 PR tree-optimization/80898
29817 * tree-sra.c (process_subtree_disqualification): Removed.
29818 (disqualify_candidate): Do not acll
29819 process_subtree_disqualification.
29820 (subtree_mark_written_and_enqueue): New function.
29821 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29822 RHS has been disqualified and re-queue LHS if necessary. Apart
29823 from that, ignore disqualified RHS.
29825 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29827 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29828 address fetch for z10 or later.
29830 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29832 * config/arc/arc.md (tst_movb): Add guard when splitting.
29834 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29836 * config/arc/arc.c (arc_can_eliminate): Test against
29837 arc_frame_pointer_needed.
29839 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29841 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29842 to prevent store reordering.
29843 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29844 (type): Add block type.
29845 (stack_tie): Define special instruction to be used in
29848 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29850 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29851 constraint. It is not valid for the pattern.
29852 (noncommutative_binary_comparison): Likewise.
29854 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29856 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29859 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29861 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29862 be used by the reg-alloc.
29864 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29866 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29867 reg-alloc when having mul64 or mul32x16 instructions.
29868 (mulsidi3): Likewise.
29869 (umulsidi3): Likewise.
29870 (mulsi32x16): New pattern.
29871 (mulsi64): Likewise.
29872 (mulsidi64): Likewise.
29873 (umulsidi64): Likewise.
29874 (MUL32x16_REG): Define.
29875 (mul64_600): Use MUL32x16_REG.
29876 (mac64_600): Likewise.
29877 (umul64_600): Likewise.
29878 (umac64_600): Likewise.
29880 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29882 * config/arc/arc.md (mulsi3_700): Make it commutative.
29884 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29886 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29888 (*sign_extendsidi2_insn): Likewise for movstosw.
29890 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29892 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29893 the type of the input discriminant value. Convert the
29894 discriminant value of signedness vary.
29896 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29898 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29899 Add to -Wall section.
29901 2017-06-01 Richard Biener <rguenther@suse.de>
29903 PR middle-end/66313
29904 * fold-const.c (fold_plusminus_mult_expr): If the factored
29905 factor may be zero use a wrapping type for the inner operation.
29906 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29907 and handle moved defs.
29908 (process_assignment): Properly guard the unary op case. Return a
29909 tri-state indicating that moving the stmt before the call may allow
29910 to continue. Pass through to_move.
29911 (find_tail_calls): Handle moving unrelated defs before
29914 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29917 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29918 splitter result in the canonical way.
29920 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29922 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29923 also for 32bit target. Update insn attributes.
29924 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29926 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29928 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29929 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29930 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29931 (_mm_maskz_min_ss): New intrinsics.
29933 2017-05-31 Martin Liska <mliska@suse.cz>
29935 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29936 Change comment style to one we normally use.
29937 (vectorizable_reduction): Likewise.
29938 (vectorizable_induction): Likewise.
29939 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29940 (vectorizable_call): Likewise.
29941 (vectorizable_simd_clone_call): Likewise.
29942 (vectorizable_conversion): Likewise.
29943 (vectorizable_assignment): Likewise.
29944 (vectorizable_shift): Likewise.
29945 (vectorizable_operation): Likewise.
29946 (vectorizable_store): Likewise.
29947 (vectorizable_load): Likewise.
29948 * tree-vectorizer.h: Likewise.
29950 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29952 * passes.c (emergency_dump_function): New.
29953 * tree-pass.h (emergency_dump_function): Declare.
29954 * plugin.c (plugins_internal_error_function): Remove.
29955 * plugin.h (plugins_internal_error_function): Remove declaration.
29956 * toplev.c (internal_error_function): New static function. Use it...
29957 (general_init): ...here.
29959 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29961 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29962 (arc_rtx_costs): Add costs for new patterns.
29963 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29964 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29966 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29968 * tree-ssa-strlen.c (get_next_strinfo): New function.
29969 (get_stridx_plus_constant): Use it.
29970 (zero_length_string): Likewise.
29971 (adjust_related_strinfos): Likewise.
29972 (adjust_last_stmt): Likewise.
29974 2017-05-31 Richard Biener <rguenther@suse.de>
29977 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29978 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29980 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29982 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29983 loop_vinfo argument and use of dependence distance vectors.
29984 Check instead whether the two references differ only in their
29985 initial value and assume that they have the same alignment if the
29986 difference is a multiple of the vector alignment.
29987 (vect_analyze_data_refs_alignment): Update call accordingly.
29989 2017-05-31 Martin Liska <mliska@suse.cz>
29992 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29994 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29996 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29997 (create_intersect_range_checks): Move from ...
29998 * tree-data-ref.c (create_intersect_range_checks_index)
29999 (create_intersect_range_checks): ... to here.
30000 (create_runtime_alias_checks): New function factored from ...
30001 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
30002 here. Call above function.
30003 * tree-data-ref.h (create_runtime_alias_checks): New function.
30005 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30007 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
30008 segment length for dr_b and compute it in wide_int.
30010 2017-05-31 Richard Biener <rguenther@suse.de>
30012 PR tree-optimization/80906
30013 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
30014 and pass through iv_map.
30015 (copy_bb_and_scalar_dependences): Adjust.
30016 (translate_pending_phi_nodes): Likewise.
30017 (copy_loop_close_phi_args): Handle code-generating IVs instead
30020 2017-05-30 David Malcolm <dmalcolm@redhat.com>
30022 * diagnostic-color.c (color_dict): Add "type-diff".
30023 (parse_gcc_colors): Update comment.
30024 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
30025 -fdiagnostics-show-template-tree and -fno-elide-type.
30026 (GCC_COLORS): Add type-diff to example.
30028 (-fdiagnostics-show-template-tree): New.
30029 (-fno-elide-type): New.
30030 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
30031 the pp_format_decoder callback. Call any m_format_postprocessor's
30033 (pretty_printer::pretty_printer): Initialize
30034 m_format_postprocessor.
30035 (pretty_printer::~pretty_printer): Delete any
30036 m_format_postprocessor.
30037 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
30038 (class format_postprocessor): New class.
30039 (struct pretty_printer::format_decoder): Document the new parameters.
30040 (struct pretty_printer::m_format_postprocessor): New field.
30041 * tree-diagnostic.c (default_tree_printer): Update for new
30042 bool and const char ** params.
30043 * tree-diagnostic.h (default_tree_printer): Likewise.
30045 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
30047 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
30048 (lwa_operand): Delete rs6000_gen_cell_microcode test.
30049 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30050 rs6000_gen_cell_microcode code.
30051 (rs6000_final_prescan_insn): Delete.
30052 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
30053 "warn-cell-microcode" entries.
30054 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
30055 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
30056 throughout. Change cc_reg_not_micro_cr0_operand to
30057 cc_reg_not_cr0_operand throughout.
30058 (*extendhi<mode>2_noload): Delete.
30059 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30060 (mwarn-cell-microcode): Delete.
30061 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30062 -mgen-cell-microcode and -mwarn-cell-microcode.
30064 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30067 * config/i386/constraints.md (Yd): New constraint.
30069 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30070 and (?Yd, r) alternatives. Update insn attributes.
30071 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30072 and (?*Yd, r) alternatives. Update insn attributes.
30073 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30075 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30077 * gimplify.c (gimplify_modify_expr): Don't create a
30078 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30081 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30083 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30085 2017-05-30 Richard Biener <rguenther@suse.de>
30087 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30088 and reduc_def fields.
30089 (STMT_VINFO_REDUC_TYPE): New define.
30090 (STMT_VINFO_REDUC_DEF): Likewise.
30091 (vect_force_simple_reduction): Adjust prototype.
30092 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30093 (vect_is_simple_reduction): Remove check_reduction argument.
30094 (vect_force_simple_reduction): Adjust and set
30095 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30096 (vectorizable_reduction): Do not re-do reduction analysis
30097 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30098 * tree-parloops.c (gather_scalar_reductions): Adjust.
30100 2017-05-30 Richard Biener <rguenther@suse.de>
30102 PR middle-end/80901
30103 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30106 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30108 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30109 Introduce unknown_misalignment parameter and remove vf.
30110 (vect_peeling_hash_get_lowest_cost):
30111 Pass unknown_misalignment parameter.
30112 (vect_enhance_data_refs_alignment):
30113 Fix unsupportable data ref treatment.
30115 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30117 * tree-vect-data-refs.c (vect_get_data_access_cost):
30118 Workaround for SLP handling.
30119 (vect_enhance_data_refs_alignment):
30120 Compute costs for doing no peeling at all, compare to the best
30121 peeling costs so far and avoid peeling if cheaper.
30123 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30125 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30126 Return peeling info and set costs to zero for unlimited cost
30128 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30129 with unknown misalignment. Compute and costs for unknown
30130 misalignment, compare them to the costs for known misalignment
30131 and choose the cheapest for peeling.
30133 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30135 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30136 (vect_get_peeling_costs_all_drs): Create function.
30137 (vect_peeling_hash_get_lowest_cost):
30138 Use vect_get_peeling_costs_all_drs.
30139 (vect_peeling_supportable): Create function.
30140 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30142 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30144 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30145 DR_HAS_NEGATIVE_STEP.
30146 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30147 (vect_enhance_data_refs_alignment): Use.
30148 (vect_duplicate_ssa_name_ptr_info): Use.
30149 * tree-vectorizer.h (dr_misalignment): Use.
30150 (known_alignment_for_access_p): Use.
30152 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30155 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30157 (has_section_name): New function.
30159 2017-05-30 Martin Liska <mliska@suse.cz>
30162 * auto-profile.c (get_function_decl_from_block): Fix
30165 2017-05-30 Richard Biener <rguenther@suse.de>
30167 PR middle-end/80876
30168 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30170 2017-05-30 Martin Liska <mliska@suse.cz>
30172 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30173 * dumpfile.h (struct dump_file_info): Remove ctors.
30175 2017-05-30 Martin Liska <mliska@suse.cz>
30177 * predict.def: Fix GNU coding style.
30179 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30181 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30182 Mark 'to' argument with ATTRIBUTE_UNUSED.
30184 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30186 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30187 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30188 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30191 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30193 * doc/install.texi (Options specification): Restore entry of
30194 --enable-sjlj-exceptions.
30196 2017-05-27 Michael Eager <eager@eagercon.com>
30199 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30201 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30203 * config/microblaze/microblaze.h
30204 (FIXED_REGISTERS): Update in macro.
30205 (CALL_USED_REGISTERS): Update in macro.
30207 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30209 * doc/install.texi: Add links to macOS binary distributions.
30211 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30215 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30217 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30219 2017-05-26 Martin Liska <mliska@suse.cz>
30221 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30223 2017-05-26 Martin Liska <mliska@suse.cz>
30225 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30226 always leading ';; '.
30227 (dump_bb_info): Likewise.
30228 (brief_dump_cfg): Likewise.
30229 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30230 * dumpfile.c: Remove usage of TDF_VERBOSE.
30231 * dumpfile.h (enum dump_kind): Likewise.
30232 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30233 * print-tree.c (debug_verbose): Remove.
30234 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30235 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30236 * tree-diagnostic.c (default_tree_printer): Replace
30237 TDF_DIAGNOSTIC with TDF_SLIM.
30239 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30241 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30242 in parameter loop, rather than loop_vinfo.
30243 (create_intersect_range_checks): Ditto.
30244 (vect_create_cond_for_alias_checks): Update call to above functions.
30246 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30248 PR tree-optimization/80815
30249 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30250 for merging runtime alias checks. Handle negative DR_STEPs.
30252 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30254 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30256 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30257 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30258 out code pruning runtime alias checks.
30259 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30260 factored out from above.
30261 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30263 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30265 (prune_runtime_alias_test_list): New decalaration.
30267 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30269 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30270 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30271 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30272 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30273 (operator==, comp_dr_with_seg_len_pair): Ditto.
30274 (vect_prune_runtime_alias_test_list): Ditto.
30276 2017-05-26 Martin Liska <mliska@suse.cz>
30279 * params.def: Bound partial-inlining-entry-probability param.
30281 2017-05-26 Marek Polacek <polacek@redhat.com>
30284 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30287 2017-05-26 Richard Biener <rguenther@suse.de>
30289 PR tree-optimization/80842
30290 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30293 2017-05-26 Richard Biener <rguenther@suse.de>
30295 PR tree-optimization/80844
30296 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30298 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30300 * doc/md.texi (Machine Constraints): Update x86 family
30301 machine constraints section to match 'config/i386/constraints.md'.
30303 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30305 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30307 2017-05-25 Nathan Sidwell <nathan@acm.org>
30309 * doc/invoke.texi (--enable-languages): Update documentation.
30311 2017-05-25 Martin Liska <mliska@suse.cz>
30313 * dumpfile.c: Add TDF_FOLDING.
30314 * dumpfile.h (enum dump_kind): Likewise.
30315 * genmatch.c (dt_simplify::gen_1): Use it.
30317 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30319 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30321 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30323 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30324 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30326 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30328 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30329 * match.pd (X == C): Rewrite it here.
30330 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30331 with_certain_nonzero_bits2): New predicates.
30332 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30334 2017-05-24 Nathan Sidwell <nathan@acm.org>
30336 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30339 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30342 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30344 * config/powerpcspe: New port. Files are copied from the rs6000
30345 port, with "rs6000" in filenames replaced by "powerpcspe".
30347 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30349 PR rtl-optimization/80754
30350 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30352 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30354 * config/sparc/sparc.md (length): Return the correct value for -mflat
30355 sibcalls to match output_sibcall.
30357 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30361 * config/rs6000/rs6000.c (struct machine_function): Add new field
30363 (rs6000_get_separate_components): Init that field, use it.
30364 (rs6000_components_for_bb): Use the field.
30366 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30368 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30370 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30372 PR middle-end/80823
30373 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30375 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30378 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30379 address against address_operand predicate.
30380 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30382 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30384 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30385 parameters passed indirectly.
30387 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30389 * config/i386/i386.md (*movdi_internal): Remove SSE4
30390 alternative 18 (?r, *v). Update insn attributes.
30391 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30392 Update insn attributes.
30393 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30394 Update insn attributes.
30395 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30396 alternative 1 (r, v). Remove isa attribute.
30397 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30398 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30399 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30401 2017-05-23 Tom de Vries <tom@codesourcery.com>
30403 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30406 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30408 * cgraphunit.c (symbol_table::process_new_functions): Update.
30409 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30410 (inline_generate_summary): Rename to ...
30411 (ipa_fn_summary_generate): ... this one.
30412 (inline_read_summary): Rename to ...
30413 (ipa_fn_summary_read): ... this one.
30414 (inline_write_summary): Rename to ...
30415 (ipa_fn_summary_write): ... this one.
30416 (inline_free_summary): Rename to ...
30417 (ipa_free_fn_summary): ... this one.
30418 (pass_data_local_fn_summary, pass_local_fn_summary,
30419 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30420 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30421 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30422 make_pass_ipa_fn_summary): New.
30423 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30424 inline_write_summary, inline_free_summary): Remove.
30425 (ipa_free_fn_summary) : New.
30426 * ipa-inline.c (ipa_inline): Update.
30427 (pass_ipa_inline): Do not generate summaries.
30428 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30430 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30431 and add pass_ipa_fn_summary.
30432 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30434 (make_pass_inline_parameters): Remove.
30436 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30438 * omp-low.c (struct omp_context): Remove "default_kind" member.
30441 * omp-offload.c (execute_oacc_device_lower): Remove the
30442 parallelism dimensions function attributes for unparallelized
30443 OpenACC kernels constructs.
30445 2017-05-23 Martin Liska <mliska@suse.cz>
30447 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30449 (cgraph_edge::make_speculative): Likewise.
30450 (cgraph_edge::resolve_speculation): Likewise.
30451 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30452 (cgraph_node::dump): Likewise.
30453 * cgraph.h: Likewise.
30454 * cgraphunit.c (analyze_functions): Likewise.
30455 (symbol_table::compile): Likewise.
30456 * ipa-cp.c (print_all_lattices): Likewise.
30457 (determine_versionability): Likewise.
30458 (initialize_node_lattices): Likewise.
30459 (ipcp_verify_propagated_values): Likewise.
30460 (estimate_local_effects): Likewise.
30461 (update_profiling_info): Likewise.
30462 (create_specialized_node): Likewise.
30463 (perhaps_add_new_callers): Likewise.
30464 (decide_about_value): Likewise.
30465 (decide_whether_version_node): Likewise.
30466 (identify_dead_nodes): Likewise.
30467 (ipcp_store_bits_results): Likewise.
30468 * ipa-devirt.c (dump_targets): Likewise.
30469 (ipa_devirt): Likewise.
30470 * ipa-icf.c (sem_item::dump): Likewise.
30471 (sem_function::equals): Likewise.
30472 (sem_variable::equals): Likewise.
30473 (sem_item_optimizer::read_section): Likewise.
30474 (sem_item_optimizer::execute): Likewise.
30475 (congruence_class::dump): Likewise.
30476 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30477 (dump_inline_summary): Likewise.
30478 (estimate_node_size_and_time): Likewise.
30479 (inline_analyze_function): Likewise.
30480 * ipa-inline-transform.c (inline_call): Likewise.
30481 * ipa-inline.c (report_inline_failed_reason): Likewise.
30482 (want_early_inline_function_p): Likewise.
30483 (edge_badness): Likewise.
30484 (update_edge_key): Likewise.
30485 (inline_small_functions): Likewise.
30486 * ipa-profile.c (ipa_profile): Likewise.
30487 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30488 (ipa_make_edge_direct_to_target): Likewise.
30489 (remove_described_reference): Likewise.
30490 (ipa_impossible_devirt_target): Likewise.
30491 (propagate_controlled_uses): Likewise.
30492 (ipa_print_node_params): Likewise.
30493 (ipcp_transform_function): Likewise.
30494 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30495 (propagate_pure_const): Likewise.
30496 * ipa-reference.c (generate_summary): Likewise.
30497 (read_write_all_from_decl): Likewise.
30498 (propagate): Likewise.
30499 (ipa_reference_read_optimization_summary): Likewise.
30500 * ipa-utils.c (ipa_merge_profiles): Likewise.
30501 * ipa.c (walk_polymorphic_call_targets): Likewise.
30502 (symbol_table::remove_unreachable_nodes): Likewise.
30503 (ipa_single_use): Likewise.
30504 * passes.c (execute_todo): Likewise.
30505 * predict.c (drop_profile): Likewise.
30506 * symtab.c (symtab_node::get_dump_name): New function.
30507 (symtab_node::dump_name): Likewise.
30508 (symtab_node::dump_asm_name): Likewise.
30509 (symtab_node::dump_references): Likewise.
30510 (symtab_node::dump_referring): Likewise.
30511 (symtab_node::dump_base): Likewise.
30512 (symtab_node::debug_symtab): Likewise.
30513 * tree-sra.c (convert_callers_for_node): Likewise.
30514 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30515 * value-prof.c (init_node_map): Likewise.
30517 2017-05-23 Martin Liska <mliska@suse.cz>
30519 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30520 and symtab_node::debug_symtab to symbol_table::debug.
30521 * cgraphunit.c (analyze_functions): Use the renamed function.
30522 (symbol_table::compile): Likewise.
30523 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30524 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30525 * passes.c (execute_todo): Likewise.
30526 * symtab.c (symbol_table::dump): New function.
30527 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30529 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30531 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30532 that nonconst implies exec.
30534 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30536 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30537 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30538 (inline_edge_summary_vec): Turn into ...
30539 (ipa_call_summaries): ... this one.
30540 (redirect_to_unreachable, edge_set_predicate,
30541 evaluate_properties_for_edge, inline_summary_alloc,
30542 reset_ipa_call_summary, reset_inline_summary,
30543 inline_summary_t::duplicate): Update.
30544 (inline_edge_duplication_hook): Turn to ...
30545 (ipa_call_summary_t::duplicate): ... this one.
30546 (inline_edge_removal_hook): Turn to ...
30547 (ipa_call_summary_t::remove): ... this one.
30548 (dump_inline_edge_summary): Turn to ...
30549 (dump_ipa_call_summary): ... this one.
30550 (estimate_function_body_sizes): Update.
30551 (inline_update_callee_summaries): Update.
30552 (remap_edge_change_prob): Update.
30553 (remap_edge_summaries): Update.
30554 (inline_merge_summary): Update.
30555 (do_estimate_edge_time): Update.
30556 (inline_generate_summary): Update.
30557 (inline_read_section): Update.
30558 (inline_read_summary): Update.
30559 (inline_free_summary): Update.
30560 * ipa-inline.c (can_inline_edge_p): Update.
30561 (compute_inlined_call_time): Update.
30562 (want_inline_small_function_p): Update.
30563 (edge_badness): Update.
30564 (early_inliner): Update.
30565 * ipa-inline.h (inline_edge_summary): Turn to ...
30566 (ipa_call_summary): ... this one.
30567 (ipa_call_summary_t): New class.
30568 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30569 (ipa_call_summaries): New.
30570 (inline_edge_summary): Remove.
30571 (estimate_edge_growth): Update.
30572 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30573 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30574 * ipa-split.c (execute_split_functions): Update.
30575 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30577 2017-05-23 Tom de Vries <tom@codesourcery.com>
30579 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30580 attributes): Document rdrand effective target.
30582 2017-05-23 Tom de Vries <tom@codesourcery.com>
30584 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30585 attributes): Sort alphabetically.
30587 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30589 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30591 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30594 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30595 V2DF/V2DI splat into two separate patterns, one that handles
30596 registers, and the other that only handles memory. Drop support
30597 for splatting from a GPR on ISA 2.07 and then splitting the
30598 splat into direct move and splat.
30599 (vsx_splat_<mode>_reg): Likewise.
30600 (vsx_splat_<mode>_mem): Likewise.
30602 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30604 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30606 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30608 PR middle-end/80809
30609 * omp-low.c (finish_taskreg_remap): New function.
30610 (finish_taskreg_scan): If unit size of ctx->record_type
30611 is non-constant, unshare the size expression and replace
30612 decls in it with possible outer var refs.
30614 PR middle-end/80809
30615 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30616 GOVD_SHARED rather than GOVD_PRIVATE with it.
30617 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30618 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30620 PR middle-end/80853
30621 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30622 as last argument to build_outer_var_ref for pointer bases of array
30623 section reductions.
30625 2017-05-19 Martin Sebor <msebor@redhat.com>
30627 * print-tree.c (print_node): Print DECL_READ_P flag.
30629 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30631 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30632 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30633 * cgraph.c: Likewise.
30634 * cgraphunit.c: Likewise.
30635 * gengtype.c: Likewise.
30636 * ipa-cp.c: Likewise.
30637 * ipa-devirt.c: Likewise.
30638 * ipa-icf.c: Likewise.
30639 * ipa-predicate.c: Likewise.
30640 * ipa-profile.c: Likewise.
30641 * ipa-prop.c: Likewise.
30642 * ipa-split.c: Likewise.
30644 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30645 edge_predicate_pool, dump_inline_hints,
30646 inline_summary::account_size_time, redirect_to_unreachable,
30647 edge_set_predicate, set_hint_predicate,
30648 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30649 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30650 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30651 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30652 ipa_call_summary_t::remove, initialize_growth_caches,
30653 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30654 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30655 mark_modified, unmodified_parm_1, unmodified_parm,
30656 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30657 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30658 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30659 will_be_nonconstant_predicate, record_modified_bb_info,
30660 get_minimal_bb, record_modified, param_change_prob,
30661 phi_result_unknown_predicate, predicate_for_phi_result,
30662 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30663 estimate_function_body_sizes, compute_inline_parameters,
30664 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30665 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30666 inline_update_callee_summaries, remap_edge_change_prob,
30667 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30668 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30669 inline_analyze_function, inline_summary_t::insert,
30670 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30671 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30672 inline_free_summary): Move to ipa-fnsummary.h
30673 (predicate_t): Remove.
30674 * ipa-fnsummary.c: New file.
30675 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30676 (enum inline_hints_vals, inline_hints, agg_position_info,
30677 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30678 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30679 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30680 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30681 inline_read_summary, inline_write_summary, inline_free_summary,
30682 inline_analyze_function, initialize_inline_failed,
30683 inline_merge_summary, inline_update_overall_summary,
30684 compute_inline_parameters): Move to ipa-fnsummary.h
30685 * ipa-fnsummary.h: New file.
30686 * ipa-inline-transform.h: Include ipa-inline.h.
30687 * ipa-inline.c: LIkewise.
30689 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30691 * ipa-inline.c (edge_badness): Use inlined_time instead of
30692 inline_summaries->get.
30694 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30696 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30698 2017-05-22 Nathan Sidwell <nathan@acm.org>
30700 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30701 (fdump-lang): Document 'raw' option.
30702 * dumpfile.h (TDI_tu): Delete.
30703 * dumpfile.c (dump_files): Remove translation-unit.
30704 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30706 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30708 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30709 command option from $(AWK) call.
30710 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30711 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30712 [FORMAT]: Remove handling of variable.
30713 * config/avr/t-multilib: Regenerate.
30715 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30717 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30719 (dump_inline_summary): Do not print self_time.
30720 (estimate_function_body_sizes): Do not set self_time.
30721 (compute_inline_parameters): Likewise.
30722 (inline_read_section, inline_write_summary): Do not stream self_time.
30723 * ipa-inline.h (inline_summary): Drop self_time.
30725 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30727 * ipa-inline-analysis.c (account_size_time): Rename to ...
30728 (inline_summary::account_size_time): ... this one.
30729 (reset_ipa_call_summary): Turn to ...
30730 (ipa_call_summary::reset): ... this one.
30731 (reset_inline_summary): Turn to ...
30732 (inline_summary::reset): ... this one.
30733 (inline_summary_t::remove): Update.
30734 (inline_summary_t::duplicate): Update.
30735 (ipa_call_summary_t::remove): Update.
30736 (dump_inline_summary): Update.
30737 (estimate_function_body_sizes): Update.
30738 (compute_inline_parameters): Update.
30739 (estimate_node_size_and_time): Update.
30740 (inline_merge_summary): Update.
30741 (inline_update_overall_summary): Update.
30742 (inline_read_section): Update.
30743 (inline_write_summary): Update.
30744 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30745 add account_size_time and reset member functions.
30746 (ipa_call_summary): Add reset function.
30747 * ipa-predicate.h (predicate::operator &): Constify.
30749 2017-05-22 Richard Biener <rguenther@suse.de>
30751 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30753 2017-05-19 Jason Merrill <jason@redhat.com>
30755 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30757 2017-05-19 Marek Polacek <polacek@redhat.com>
30760 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30761 TYPE_OVERFLOW_WRAPS checks.
30763 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30765 * tree-core.h (enum omp_clause_default_kind): Add
30766 "OMP_CLAUSE_DEFAULT_PRESENT".
30767 * tree-pretty-print.c (dump_omp_clause): Handle it.
30768 * gimplify.c (enum gimplify_omp_var_data): Add
30769 "GOVD_MAP_FORCE_PRESENT".
30770 (gimplify_adjust_omp_clauses_1): Map it to
30771 "GOMP_MAP_FORCE_PRESENT".
30772 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30774 * gimplify.c (oacc_default_clause): Clarify.
30776 2017-05-19 Nathan Sidwell <nathan@acm.org>
30778 LANG_HOOK_REGISTER_DUMPS
30779 * toplev.c (general_init): Call register dump lang hook.
30780 * doc/invoke.texi: Document -fdump-lang option family.
30781 * dumpfile.c (dump_files): Remove class dump here.
30782 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30783 * dumpfile.h (tree_dump_index): Remove TDI_class.
30784 * langhooks-def.h (lhd_register_dumps): Declare.
30785 (LANG_HOOKS_REGISTER_DUMPS): Define.
30786 (LANG_HOOKS_INITIALIZER): Add it.
30787 * langhooks.c (lhd_register_dumps): Define.
30788 * langhooks.h (struct lang_hooks): Add register_dumps.
30790 2017-05-19 Nathan Sidwell <nathan@acm.org>
30792 * context.h (context::set_passes): New.
30793 * context.c (context::context): Do not create pass manager.
30794 * toplev.c (general_init): Create pass manager here.
30796 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30798 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30799 use this splitter if two add or or instructions would also work for
30800 the constant we want to generate.
30802 2017-05-19 Richard Biener <rguenther@suse.de>
30805 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30806 predicate evaluation.
30808 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30810 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30812 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30815 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30817 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30818 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30819 (inline_edge_summary_vec): Turn into ...
30820 (ipa_call_summaries): ... this one.
30821 (redirect_to_unreachable, edge_set_predicate,
30822 evaluate_properties_for_edge, inline_summary_alloc,
30823 reset_ipa_call_summary, reset_inline_summary,
30824 inline_summary_t::duplicate): Update.
30825 (inline_edge_duplication_hook): Turn to ...
30826 (ipa_call_summary_t::duplicate): ... this one.
30827 (inline_edge_removal_hook): Turn to ...
30828 (ipa_call_summary_t::remove): ... this one.
30829 (dump_inline_edge_summary): Turn to ...
30830 (dump_ipa_call_summary): ... this one.
30831 (estimate_function_body_sizes): Update.
30832 (inline_update_callee_summaries): Update.
30833 (remap_edge_change_prob): Update.
30834 (remap_edge_summaries): Update.
30835 (inline_merge_summary): Update.
30836 (do_estimate_edge_time): Update.
30837 (inline_generate_summary): Update.
30838 (inline_read_section): Update.
30839 (inline_read_summary): Update.
30840 (inline_free_summary): Update.
30841 * ipa-inline.c (can_inline_edge_p): Update.
30842 (compute_inlined_call_time): Update.
30843 (want_inline_small_function_p): Update.
30844 (edge_badness): Update.
30845 (early_inliner): Update.
30846 * ipa-inline.h (inline_edge_summary): Turn to ...
30847 (ipa_call_summary): ... this one.
30848 (ipa_call_summary_t): New class.
30849 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30850 (ipa_call_summaries): New.
30851 (inline_edge_summary): Remove.
30852 (estimate_edge_growth): Update.
30853 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30854 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30855 * ipa-split.c (execute_split_functions): Update.
30856 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30858 2017-05-19 Richard Biener <rguenther@suse.de>
30860 PR middle-end/80764
30861 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30863 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30865 * config/rs6000/rs6000.c (struct machine_function): Add field
30866 fpr_is_wrapped_separately.
30867 (rs6000_get_separate_components): Use 64 components. Handle the
30868 new FPR components.
30869 (rs6000_components_for_bb): Handle the FPR components.
30870 (rs6000_emit_prologue_components): Handle the FPR components.
30871 (rs6000_emit_epilogue_components): Handle the FPR components.
30872 (rs6000_set_handled_components): Handle the FPR components.
30873 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30874 that are already separately shrink-wrapped.
30875 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30876 that are already separately shrink-wrapped.
30878 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30881 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30884 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30885 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30886 cases where the register allocator uses a move and an offsettable
30887 memory operation to/from a FPR register on ISA 2.06/2.07.
30888 (define_peephole2 for Altivec d-form store): Likewise.
30890 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30893 * config/i386/mmx.md (*mov<mode>_internal): Enable
30894 alternatives 11, 12, 13 and 14 also for 32bit targets.
30895 Remove alternatives 15, 16, 17 and 18.
30896 * config/i386/sse.md (vec_concatv2di): Change
30897 alternative (!x, *y) to (x, ?!*Yn).
30899 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30901 * dumpfile.h (enum dump_kind): Remove stray comma.
30903 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30905 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30906 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30907 predicate::num_conditions
30908 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30909 (CHANGED): turn into predicate::changed.
30910 (agg_position_info): Move to ipa-predicate.h
30911 (add_condition, predicate::add_clause, predicate::operator &=,
30912 predicate::or_with, predicate::evaluate, predicate::probability,
30913 dump_condition, dump_clause, predicate::dump,
30914 predicate::remap_after_duplication, predicate::remap_after_inlining,
30915 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30916 (evaluate_conditions_for_known_args): Update.
30917 (set_cond_stmt_execution_predicate): Update.
30918 * ipa-inline.h: Include ipa-predicate.h
30919 (condition, inline_param_summary, conditions, agg_position_info,
30920 predicate): Move to ipa-predicate.h
30921 * ipa-predicate.c: New file.
30922 * ipa-predicate.h: New file.
30924 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30926 * final.c (leaf_function_p): Check we are not in a sequence.
30928 2017-05-18 Martin Liska <mliska@suse.cz>
30930 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30931 * dumpfile.c (dump_register): Use new enum dump_kind.
30932 (get_dump_file_name): Likewise.
30933 (dump_enable_all): Likewise.
30934 (dump_switch_p_1): Likewise.
30935 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30936 * dumpfile.h (enum dump_kind): New enum type.
30937 (struct dump_file_info): Create constructor and
30938 format fields and comments.
30939 * passes.c (pass_manager::register_one_dump_file):
30941 * statistics.c (statistics_early_init): Likewise.
30942 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30943 TDF_TREE with TDF_SLIM.
30944 (gather_memory_references_ref): Likewise.
30946 2017-05-18 Martin Liska <mliska@suse.cz>
30948 * vec.h (struct vnull): Use it.
30950 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30952 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30953 (true_predicate, false_predicate, true_predicate_p,
30954 false_predicate_p): Remove.
30955 (single_cond_predicate, not_inlined_predicate): Turn to member function
30957 (add_condition): Update.
30958 (add_clause): Turn to...
30959 (predicate::add_clause): ... this one; update; allow passing NULL
30961 (and_predicates): Turn to ...
30962 (predicate::operator &=): ... this one.
30963 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30964 (or_predicates): Turn to ...
30965 (predicate::or_with): ... this one.
30966 (evaluate_predicate): Turn to ...
30967 (predicate::evaluate): ... this one.
30968 (predicate_probability): Turn to ...
30969 (predicate::probability): ... this one.
30970 (dump_condition): Update.
30971 (dump_predicate): Turn to ...
30972 (predicate::dump): ... this one.
30973 (account_size_time): Update.
30974 (edge_set_predicate): Update.
30975 (set_hint_predicate): UPdate.
30976 (evaluate_conditions_for_known_args): Update.
30977 (evaluate_properties_for_edge): Update.
30978 (remap_predicate_after_duplication): Turn to...
30979 (predicate::remap_after_duplication): ... this one.
30980 (remap_hint_predicate_after_duplication): Update.
30981 (inline_summary_t::duplicate): UPdate.
30982 (dump_inline_edge_summary): Update.
30983 (dump_inline_summary): Update.
30984 (set_cond_stmt_execution_predicate): Update.
30985 (set_switch_stmt_execution_predicate): Update.
30986 (compute_bb_predicates): Update.
30987 (will_be_nonconstant_expr_predicate): Update.
30988 (will_be_nonconstant_predicate): Update.
30989 (phi_result_unknown_predicate): Update.
30990 (predicate_for_phi_result): Update.
30991 (array_index_predicate): Update.
30992 (estimate_function_body_sizes): Update.
30993 (estimate_node_size_and_time): Update.
30994 (estimate_ipcp_clone_size_and_time): Update.
30995 (remap_predicate): Rename to ...
30996 (predicate::remap_after_inlining): ... this one.
30997 (remap_hint_predicate): Update.
30998 (inline_merge_summary): Update.
30999 (inline_update_overall_summary): Update.
31000 (estimate_size_after_inlining): Update.
31001 (read_predicate): Rename to ...
31002 (predicate::stream_in): ... this one.
31003 (read_inline_edge_summary): Update.
31004 (write_predicate): Rename to ...
31005 (predicate::stream_out): ... this one.
31006 (write_inline_edge_summary): Update.
31007 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
31008 (clause_t): Turn to uint32_t
31009 (predicate): Turn to class; implement constructor and operators
31011 (size_time_entry): Update.
31012 (inline_summary): Update.
31013 (inline_edge_summary): Update.
31015 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
31017 * fold-const.c (fold_binary_loc): Move transformation...
31018 * match.pd (C - X CMP X): ... here.
31020 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
31022 * config/sparc/sparc.c (sparc_option_override): Set function
31023 alignment for -mcpu=niagara7 to 64 to match the I$ line.
31024 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
31026 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
31028 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
31030 2017-05-18 Marek Polacek <polacek@redhat.com>
31033 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
31034 (pass_ubsan::execute): Call gimple_assign_single_p instead of
31035 gimple_assign_load_p.
31037 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
31039 PR middle-end/80692
31040 * real.c (do_compare): Give decimal_do_compare preference over
31041 comparing just the signs.
31043 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
31045 * doc/md.texi (Canonicalization of Instructions): Describe the
31046 canonical form of instructions that inherently set a condition
31049 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
31051 PR middle-end/80775
31052 * tree-cfg.c: Move deletion of unreachable case statements to after
31053 the merging of consecutive case labels.
31055 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
31057 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
31058 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31059 restoring of callee-saved registers.
31061 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31063 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31064 * config/visium/visium.c (single_set_and_flags): Likewise.
31065 * config/visium/visium.md (Substitutions): Likewise.
31067 2017-05-17 Martin Liska <mliska@suse.cz>
31069 * cfg.c: Introduce dump_flags_t type and
31070 use it instead of int type.
31072 * cfghooks.c: Likewise.
31073 * cfghooks.h (struct cfg_hooks): Likewise.
31074 * cfgrtl.c: Likewise.
31075 * cfgrtl.h: Likewise.
31076 * cgraph.c (cgraph_node::get_body): Likewise.
31077 * coretypes.h: Likewise.
31078 * domwalk.c: Likewise.
31079 * domwalk.h: Likewise.
31080 * dumpfile.c (struct dump_option_value_info): Likewise.
31081 (dump_enable_all): Likewise.
31082 (dump_switch_p_1): Likewise.
31083 (opt_info_switch_p): Likewise.
31084 * dumpfile.h (enum tree_dump_index): Likewise.
31085 (struct dump_file_info): Likewise.
31086 * genemit.c: Likewise.
31087 * generic-match-head.c: Likewise.
31088 * gengtype.c (open_base_files): Likewise.
31089 * gimple-pretty-print.c: Likewise.
31090 * gimple-pretty-print.h: Likewise.
31091 * graph.c (print_graph_cfg): Likewise.
31092 * graphite-scop-detection.c (dot_all_sese): Likewise.
31093 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31094 * loop-unroll.c (report_unroll): Likewise.
31095 * passes.c (pass_manager::register_one_dump_file): Likewise.
31096 * print-tree.c: Likewise.
31097 * statistics.c: Likewise.
31098 * tree-cfg.c: Likewise.
31099 * tree-cfg.h: Likewise.
31100 * tree-dfa.c: Likewise.
31101 * tree-dfa.h: Likewise.
31102 * tree-dump.c (dump_function): Likewise.
31103 * tree-dump.h (struct dump_info): Likewise.
31104 * tree-pretty-print.c: Likewise.
31105 * tree-pretty-print.h: Likewise.
31106 * tree-ssa-live.c: Likewise.
31107 * tree-ssa-live.h: Likewise.
31108 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31109 * tree-vect-loop.c: Likewise.
31110 * tree-vect-slp.c: Likewise.
31112 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31113 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31115 PR tree-optimization/80457
31116 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31117 of all arguments to a statement as scalar_to_vec operations.
31118 (vectorizable_call): Adjust call to vect_model_simple_cost for
31120 (vectorizable_conversion): Likewise.
31121 (vectorizable_assignment): Likewise.
31122 (vectorizable_shift): Likewise.
31123 (vectorizable_operation): Likewise.
31124 (vectorizable_comparison): Likewise.
31125 (vect_is_simple_cond): Record the def types for operands.
31126 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31127 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31128 for statement argument count.
31130 2017-05-16 Carl Love <cel@us.ibm.com>
31132 * config/rs6000/rs6000-c: Add support for built-in functions
31133 vector unsigned long long vec_bperm (vector unsigned long long,
31134 vector unsigned char)
31135 vector signed long long vec_mule (vector signed int,
31137 vector unsigned long long vec_mule (vector unsigned int,
31138 vector unsigned int)
31139 vector signed long long vec_mulo (vector signed int,
31141 vector unsigned long long vec_mulo (vector unsigned int,
31142 vector unsigned int)
31143 vector signed char vec_sldw (vector signed char,
31144 vector signed char,
31146 vector unsigned char vec_sldw (vector unsigned char,
31147 vector unsigned char,
31149 vector signed short vec_sldw (vector signed short,
31150 vector signed short,
31152 vector unsigned short vec_sldw (vector unsigned short,
31153 vector unsigned short,
31155 vector signed int vec_sldw (vector signed int,
31158 vector unsigned int vec_sldw (vector unsigned int,
31159 vector unsigned int,
31161 vector signed long long vec_sldw (vector signed long long,
31162 vector signed long long,
31164 vector unsigned long long vec_sldw (vector unsigned long long,
31165 vector unsigned long long,
31167 * config/rs6000/rs6000-c: Add support for built-in functions
31168 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31169 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31170 * doc/extend.texi: Update the built-in documentation for the
31171 new built-in functions.
31173 2017-05-16 Marek Polacek <polacek@redhat.com>
31177 * tree.c (save_expr): Don't fold the expression.
31179 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31181 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31182 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31183 and (?*y,m). Update insn attributes.
31185 2017-05-16 Martin Liska <mliska@suse.cz>
31187 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31188 flags argument of print_gimple_stmt, print_gimple_expr,
31189 print_generic_stmt and print_generic_expr.
31190 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31191 * coretypes.h: Likewise.
31192 * except.c (dump_eh_tree): Likewise.
31193 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31194 * gimple-pretty-print.h: Likewise.
31195 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31196 (backprop::push_to_worklist): Likewise.
31197 (backprop::pop_from_worklist): Likewise.
31198 (backprop::process_use): Likewise.
31199 (backprop::intersect_uses): Likewise.
31200 (note_replacement): Likewise.
31201 * gimple-ssa-store-merging.c
31202 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31203 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31204 (pass_store_merging::execute): Likewise.
31205 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31206 (ssa_base_cand_dump_callback): Likewise.
31207 (dump_incr_vec): Likewise.
31208 (replace_refs): Likewise.
31209 (replace_mult_candidate): Likewise.
31210 (create_add_on_incoming_edge): Likewise.
31211 (create_phi_basis): Likewise.
31212 (insert_initializers): Likewise.
31213 (all_phi_incrs_profitable): Likewise.
31214 (introduce_cast_before_cand): Likewise.
31215 (replace_one_candidate): Likewise.
31216 * gimplify.c (gimplify_expr): Likewise.
31217 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31218 (set_rename): Likewise.
31219 (rename_uses): Likewise.
31220 (copy_loop_phi_nodes): Likewise.
31221 (add_close_phis_to_merge_points): Likewise.
31222 (copy_loop_close_phi_args): Likewise.
31223 (copy_cond_phi_args): Likewise.
31224 (graphite_copy_stmts_from_block): Likewise.
31225 (translate_pending_phi_nodes): Likewise.
31226 * graphite-poly.c (print_pdr): Likewise.
31227 (dump_gbb_cases): Likewise.
31228 (dump_gbb_conditions): Likewise.
31229 (print_scop_params): Likewise.
31230 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31231 (build_cross_bb_scalars_use): Likewise.
31232 (gather_bbs::before_dom_children): Likewise.
31233 * hsa-dump.c (dump_hsa_immed): Likewise.
31234 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31235 (get_replacement_map): Likewise.
31236 * ipa-inline-analysis.c (dump_condition): Likewise.
31237 (estimate_function_body_sizes): Likewise.
31238 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31239 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31240 * ipa-prop.c (ipa_dump_param): Likewise.
31241 (ipa_print_node_jump_functions_for_edge): Likewise.
31242 (ipa_modify_call_arguments): Likewise.
31243 (ipa_modify_expr): Likewise.
31244 (ipa_dump_param_adjustments): Likewise.
31245 (ipa_dump_agg_replacement_values): Likewise.
31246 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31247 * ipa-pure-const.c (check_stmt): Likewise.
31248 (pass_nothrow::execute): Likewise.
31249 * ipa-split.c (execute_split_functions): Likewise.
31250 * omp-offload.c (dump_oacc_loop_part): Likewise.
31251 (dump_oacc_loop): Likewise.
31252 * trans-mem.c (tm_log_emit): Likewise.
31253 (tm_memopt_accumulate_memops): Likewise.
31254 (dump_tm_memopt_set): Likewise.
31255 (dump_tm_memopt_transform): Likewise.
31256 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31257 (print_loop): Likewise.
31258 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31259 (chkp_gather_checks_info): Likewise.
31260 (chkp_get_check_result): Likewise.
31261 (chkp_remove_check_if_pass): Likewise.
31262 (chkp_use_outer_bounds_if_possible): Likewise.
31263 (chkp_reduce_bounds_lifetime): Likewise.
31264 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31265 (chkp_mark_completed_bounds): Likewise.
31266 (chkp_register_incomplete_bounds): Likewise.
31267 (chkp_mark_invalid_bounds): Likewise.
31268 (chkp_maybe_copy_and_register_bounds): Likewise.
31269 (chkp_build_returned_bound): Likewise.
31270 (chkp_get_bound_for_parm): Likewise.
31271 (chkp_build_bndldx): Likewise.
31272 (chkp_get_bounds_by_definition): Likewise.
31273 (chkp_generate_extern_var_bounds): Likewise.
31274 (chkp_get_bounds_for_decl_addr): Likewise.
31275 * tree-chrec.c (chrec_apply): Likewise.
31276 * tree-data-ref.c (dump_data_reference): Likewise.
31277 (dump_subscript): Likewise.
31278 (dump_data_dependence_relation): Likewise.
31279 (analyze_overlapping_iterations): Likewise.
31280 * tree-inline.c (expand_call_inline): Likewise.
31281 (tree_function_versioning): Likewise.
31282 * tree-into-ssa.c (dump_defs_stack): Likewise.
31283 (dump_currdefs): Likewise.
31284 (dump_names_replaced_by): Likewise.
31285 (dump_update_ssa): Likewise.
31286 (update_ssa): Likewise.
31287 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31288 * tree-parloops.c (build_new_reduction): Likewise.
31289 (try_create_reduction_list): Likewise.
31290 (ref_conflicts_with_region): Likewise.
31291 (oacc_entry_exit_ok_1): Likewise.
31292 (oacc_entry_exit_single_gang): Likewise.
31293 * tree-pretty-print.h: Likewise.
31294 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31295 (get_scalar_evolution): Likewise.
31296 (add_to_evolution): Likewise.
31297 (get_loop_exit_condition): Likewise.
31298 (analyze_evolution_in_loop): Likewise.
31299 (analyze_initial_condition): Likewise.
31300 (analyze_scalar_evolution): Likewise.
31301 (instantiate_scev): Likewise.
31302 (number_of_latch_executions): Likewise.
31303 (gather_chrec_stats): Likewise.
31304 (final_value_replacement_loop): Likewise.
31305 (scev_const_prop): Likewise.
31306 * tree-sra.c (dump_access): Likewise.
31307 (disqualify_candidate): Likewise.
31308 (create_access): Likewise.
31309 (reject): Likewise.
31310 (maybe_add_sra_candidate): Likewise.
31311 (create_access_replacement): Likewise.
31312 (analyze_access_subtree): Likewise.
31313 (analyze_all_variable_accesses): Likewise.
31314 (sra_modify_assign): Likewise.
31315 (initialize_constant_pool_replacements): Likewise.
31316 (find_param_candidates): Likewise.
31317 (decide_one_param_reduction): Likewise.
31318 (replace_removed_params_ssa_names): Likewise.
31319 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31320 * tree-ssa-copy.c (dump_copy_of): Likewise.
31321 (copy_prop_visit_cond_stmt): Likewise.
31322 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31323 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31324 (record_equivalences_from_stmt): Likewise.
31325 * tree-ssa-dse.c (compute_trims): Likewise.
31326 (delete_dead_call): Likewise.
31327 (delete_dead_assignment): Likewise.
31328 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31329 (forward_propagate_into_cond): Likewise.
31330 (pass_forwprop::execute): Likewise.
31331 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31332 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31334 (move_computations_worker): Likewise.
31335 (execute_sm): Likewise.
31336 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31337 (remove_exits_and_undefined_stmts): Likewise.
31338 (remove_redundant_iv_tests): Likewise.
31339 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31340 (adjust_iv_update_pos): Likewise.
31341 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31342 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31343 (value_replacement): Likewise.
31344 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31345 * tree-ssa-pre.c (print_pre_expr): Likewise.
31346 (get_representative_for): Likewise.
31347 (create_expression_by_pieces): Likewise.
31348 (insert_into_preds_of_block): Likewise.
31349 (eliminate_insert): Likewise.
31350 (eliminate_dom_walker::before_dom_children): Likewise.
31351 (eliminate): Likewise.
31352 (remove_dead_inserted_code): Likewise.
31353 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31354 * tree-ssa-reassoc.c (get_rank): Likewise.
31355 (eliminate_duplicate_pair): Likewise.
31356 (eliminate_plus_minus_pair): Likewise.
31357 (eliminate_not_pairs): Likewise.
31358 (undistribute_ops_list): Likewise.
31359 (eliminate_redundant_comparison): Likewise.
31360 (update_range_test): Likewise.
31361 (optimize_range_tests_var_bound): Likewise.
31362 (optimize_vec_cond_expr): Likewise.
31363 (rewrite_expr_tree): Likewise.
31364 (rewrite_expr_tree_parallel): Likewise.
31365 (linearize_expr): Likewise.
31366 (break_up_subtract): Likewise.
31367 (linearize_expr_tree): Likewise.
31368 (attempt_builtin_powi): Likewise.
31369 (attempt_builtin_copysign): Likewise.
31370 (transform_stmt_to_copy): Likewise.
31371 (transform_stmt_to_multiply): Likewise.
31372 (dump_ops_vector): Likewise.
31373 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31374 (print_scc): Likewise.
31375 (set_ssa_val_to): Likewise.
31376 (visit_reference_op_store): Likewise.
31377 (visit_use): Likewise.
31378 (sccvn_dom_walker::before_dom_children): Likewise.
31379 (run_scc_vn): Likewise.
31380 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31382 (expr_hash_elt::print): Likewise.
31383 (const_and_copies::pop_to_marker): Likewise.
31384 (const_and_copies::record_const_or_copy_raw): Likewise.
31385 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31386 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31387 (dump_predicates): Likewise.
31388 (find_uninit_use): Likewise.
31389 (warn_uninitialized_phi): Likewise.
31390 (pass_late_warn_uninitialized::execute): Likewise.
31391 * tree-ssa.c (verify_vssa): Likewise.
31392 (verify_ssa): Likewise.
31393 (maybe_optimize_var): Likewise.
31394 * tree-vrp.c (dump_value_range): Likewise.
31395 (dump_all_value_ranges): Likewise.
31396 (dump_asserts_for): Likewise.
31397 (register_edge_assert_for_2): Likewise.
31398 (vrp_visit_cond_stmt): Likewise.
31399 (vrp_visit_switch_stmt): Likewise.
31400 (vrp_visit_stmt): Likewise.
31401 (vrp_visit_phi_node): Likewise.
31402 (simplify_cond_using_ranges_1): Likewise.
31403 (fold_predicate_in): Likewise.
31404 (evrp_dom_walker::before_dom_children): Likewise.
31405 (evrp_dom_walker::push_value_range): Likewise.
31406 (evrp_dom_walker::pop_value_range): Likewise.
31407 (execute_early_vrp): Likewise.
31409 2017-05-16 Richard Biener <rguenther@suse.de>
31411 * dwarf2out.c (loc_list_from_tree_1): Do not create
31412 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31414 2017-05-16 Richard Biener <rguenther@suse.de>
31416 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31418 (note_variable_value_in_expr): If we resolved the decl ref
31419 do not push to the stack.
31421 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31423 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31424 operations in fast-math mode.
31425 (vaddq_f16): Likewise.
31426 (vmul_f16): Likewise.
31427 (vmulq_f16): Likewise.
31428 (vsub_f16): Likewise.
31429 (vsubq_f16): Likewise.
31430 * config/arm/neon.md (add<mode>3): New.
31432 (fma:<VH:mode>3): New. Also remove outdated comment.
31435 2017-05-16 Martin Liska <mliska@suse.cz>
31439 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31440 (odr_types_equivalent_p): Likewise.
31442 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31444 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31446 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31449 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31450 non-interunit SSE move alternatives with '?'.
31451 (zero-extendsidi peephole2): New peephole to skip intermediate
31452 general register in SSE zero-extend sequence.
31454 2017-05-15 Jeff Law <law@redhat.com>
31456 * reorg.c (relax_delay_slots): Create a new variable to hold
31457 the temporary target rather than clobbering TARGET_LABEL.
31459 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31460 missing argument to extract_bit_field call.
31461 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31463 2017-05-15 Martin Liska <mliska@suse.cz>
31466 * gcc.c (process_command): Do not allow empty argument of -o option.
31468 2017-05-15 Renlin Li <renlin.li@arm.com>
31470 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31471 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31472 * config/aarch64/constraints.md (Usf): Add long call check.
31473 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31474 (call_value): Likewise.
31475 (sibcall): Likewise.
31476 (sibcall_value): Likewise.
31478 (call_value_insn): New.
31479 (sibcall_insn): Update rtx pattern.
31480 (sibcall_value_insn): Likewise.
31481 (call_internal): Remove.
31482 (call_value_internal): Likewise.
31483 (sibcall_internal): Likewise.
31484 (sibcall_value_internal): Likewise.
31485 (call_reg): Likewise.
31486 (call_symbol): Likewise.
31487 (call_value_reg): Likewise.
31488 (call_value_symbol): Likewise.
31490 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31493 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31495 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31497 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31498 compatible with CCGOCmode and with CCZmode.
31500 2017-05-14 Martin Sebor <msebor@redhat.com>
31502 PR middle-end/77671
31503 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31504 (gimple_fold_builtin_snprintf): Same.
31505 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31506 (gimple_fold_builtin_snprintf): Same.
31507 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31508 of character types.
31509 (is_call_safe): New function.
31510 (try_substitute_return_value): Call it.
31511 (try_simplify_call): New function.
31512 (pass_sprintf_length::handle_gimple_call): Call it.
31514 2017-05-14 Martin Sebor <msebor@redhat.com>
31516 PR middle-end/80669
31517 * builtins.c (expand_builtin_stpncpy): Simplify.
31519 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31521 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31522 * config/i386/i386.h
31523 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31524 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31525 (struct machine_function): Add new members call_ms2sysv,
31526 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31527 (struct machine_frame_state): New fields sp_realigned and
31528 sp_realigned_offset.
31529 * config/i386/i386.c
31530 (enum xlogue_stub): New enum.
31531 (enum xlogue_stub_sets): New enum.
31532 (class xlogue_layout): New class.
31533 (struct ix86_frame): New fields stack_realign_allocate_offset,
31534 stack_realign_offset and outlined_save_offset. Modify comments to
31535 detail stack layout when using out-of-line stubs.
31536 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31537 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31538 -mcall-ms2sysv-xlogues.
31539 (stub_managed_regs): New static variable.
31540 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31541 registers managed by out-of-line stub.
31542 (disable_call_ms2sysv_xlogues): New function.
31543 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31544 m->call_ms2sysv when appropriate and compute frame layout for
31546 (sp_valid_at, fp_valid_at): New inline functions.
31547 (choose_basereg): New function.
31548 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31550 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31551 Use align parameter of choose_baseaddr to generated aligned SSE movs
31553 (pro_epilogue_adjust_stack): Modify to track
31554 machine_frame_state::sp_realigned.
31555 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31556 (ix86_nsaved_sseregs): Likewise.
31557 (ix86_emit_save_regs): Likewise.
31558 (ix86_emit_save_regs_using_mov): Likewise.
31559 (ix86_emit_save_sse_regs_using_mov): Likewise.
31560 (get_scratch_register_on_entry): Likewise.
31561 (gen_frame_set): New function.
31562 (gen_frame_load): Likewise.
31563 (gen_frame_store): Likewise.
31564 (emit_outlined_ms2sysv_save): Likewise.
31565 (emit_outlined_ms2sysv_restore): Likewise.
31566 (ix86_expand_prologue): Modify stack re-alignment code and call
31567 emit_outlined_ms2sysv_save when appropriate.
31568 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31569 parameter rtx_insn *insn, which allows the function to be used to only
31570 generate the notes.
31571 (ix86_expand_epilogue): Modify validity checks of frame and stack
31572 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31573 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31574 * config/i386/predicates.md
31575 (save_multiple): New predicate.
31576 (restore_multiple): Likewise.
31577 * config/i386/sse.md
31578 (save_multiple<mode>): New pattern.
31579 (save_multiple_realign<mode>): Likewise.
31580 (restore_multiple<mode>): Likewise.
31581 (restore_multiple_and_return<mode>): Likewise.
31582 (restore_multiple_leave_return<mode>): Likewise.
31583 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31585 2017-05-14 Julia Koval <julia.koval@intel.com>
31587 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31588 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31589 (__builtin_ia32_xsetbv): New builtins.
31590 * config/i386/i386.c (ix86_expand_special_args_builtin):
31592 (ix86_expand_builtin): Special expand for new intrinsics.
31593 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31594 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31595 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31597 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31599 * cfganal.c (inverted_post_order_compute): Change argument type
31601 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31602 * df-core.c (rest_of_handle_df_initialize): Adjust.
31603 (rest_of_handle_df_finish): Likewise.
31604 (df_analyze_1): Likewise.
31605 (df_analyze): Likewise.
31606 (loop_inverted_post_order_compute): Change argument to be a vec *.
31607 (df_analyze_loop): Adjust.
31608 (df_get_n_blocks): Likewise.
31609 (df_get_postorder): Likewise.
31610 * df.h (struct df_d): Change field to be a vec.
31611 * lcm.c (compute_laterin): Adjust.
31612 (compute_available): Likewise.
31613 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31614 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31615 * tree-ssa-pre.c (compute_antic): Likewise.
31617 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31619 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31620 (depth_first_search::depth_first_search): Change structure init
31621 function to this constructor.
31622 (depth_first_search::add_bb): Rename function to this member.
31623 (depth_first_search::execute): Likewise.
31624 (flow_dfs_compute_reverse_finish): Adjust.
31626 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31628 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31629 (longest_simple_path): Likewise.
31630 * shrink-wrap.c (spread_components): Likewise.
31631 (disqualify_problematic_components): Likewise.
31632 (emit_common_heads_for_components): Likewise.
31633 (emit_common_tails_for_components): Likewise.
31634 (insert_prologue_epilogue_for_components): Likewise.
31636 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31638 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31641 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31643 * df-core.c (df_set_blocks): Start using auto_bitmap.
31644 (df_compact_blocks): Likewise.
31645 * df-problems.c (df_rd_confluence_n): Likewise.
31646 * df-scan.c (df_insn_rescan_all): Likewise.
31647 (df_process_deferred_rescans): Likewise.
31648 (df_update_entry_block_defs): Likewise.
31649 (df_update_exit_block_uses): Likewise.
31650 (df_entry_block_bitmap_verify): Likewise.
31651 (df_exit_block_bitmap_verify): Likewise.
31652 (df_scan_verify): Likewise.
31653 * lra-constraints.c (lra_constraints): Likewise.
31654 (undo_optional_reloads): Likewise.
31655 (lra_undo_inheritance): Likewise.
31656 * lra-remat.c (calculate_gen_cands): Likewise.
31657 (do_remat): Likewise.
31658 * lra-spills.c (assign_spill_hard_regs): Likewise.
31659 (spill_pseudos): Likewise.
31660 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31661 (bitmap_set_subtract_values): Likewise.
31663 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31665 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31666 management with auto_bitmap.
31667 (fix_inter_tick): Likewise.
31668 (fix_recovery_deps): Likewise.
31669 * ira.c (add_store_equivs): Likewise.
31670 (find_moveable_pseudos): Likewise.
31671 (split_live_ranges_for_shrink_wrap): Likewise.
31672 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31673 (rtx_reuse_manager::seen_def_p): Likewise.
31674 (rtx_reuse_manager::set_seen_def): Likewise.
31675 * print-rtl.h (class rtx_reuse_manager): Likewise.
31677 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31679 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31681 (migrate_btr_def): Likewise.
31682 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31683 * df-core.c (loop_post_order_compute): Likewise.
31684 (loop_inverted_post_order_compute): Likewise.
31685 * hsa-common.h: Likewise.
31686 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31687 * init-regs.c (initialize_uninitialized_regs): Likewise.
31688 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31689 (inline_small_functions): Likewise.
31690 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31691 * ira.c (combine_and_move_insns): Likewise.
31692 (build_insn_chain): Likewise.
31693 * loop-invariant.c (find_invariants): Likewise.
31694 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31695 * predict.c (tree_predict_by_opcode): Likewise.
31696 (predict_paths_leading_to): Likewise.
31697 (predict_paths_leading_to_edge): Likewise.
31698 (estimate_loops_at_level): Likewise.
31699 (estimate_loops): Likewise.
31700 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31701 (spread_components): Likewise.
31702 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31703 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31704 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31705 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31706 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31707 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31708 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31709 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31710 (mark_threaded_blocks): Likewise.
31711 (thread_through_all_blocks): Likewise.
31712 * tree-ssa.c (verify_ssa): Likewise.
31713 (execute_update_addresses_taken): Likewise.
31714 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31716 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31718 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31720 (post_order_compute): Likewise.
31721 (inverted_post_order_compute): Likewise.
31722 (pre_and_rev_post_order_compute_fn): Likewise.
31724 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31726 * genrecog.c (int_set::int_set): Explicitly construct our
31727 auto_vec base class.
31728 * vec.h (auto_vec::auto_vec): New constructor.
31730 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31732 * bitmap.h (class auto_bitmap): New constructor taking
31733 bitmap_obstack * argument.
31735 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31737 * bitmap.h (class auto_bitmap): Change type of m_bits to
31738 bitmap_head, and adjust ctor / dtor and member operators.
31740 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31742 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31743 when returned register mode doesn't match original mode.
31745 2017-05-12 Jeff Law <law@redhat.com>
31746 Jakub Jelinek <jakub@redhat.com>
31748 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31749 we look for cc setter after the compare-elim changes.
31750 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31751 within the vector to match what compare-elim now expects.
31752 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31753 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31755 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31756 after the compare-elim changes.
31757 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31758 the vector to match what compare-elim now expects.
31759 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31760 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31761 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31762 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31763 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31765 * config/visium/visium.c (single_set_and_flags): Fix where
31766 we look for cc setter after the compare-elim changes.
31767 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31768 with the vector to match what compare-elim now expects.
31769 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31770 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31771 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31772 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31773 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31774 (neg<mode>2_insn_set_overflow): Likewise.
31776 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31778 PR middle-end/79794
31779 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31780 maybe_expand_insn call, set ops[0].target. If still set after call,
31781 set alt_rtl. Add extra arg to recursive calls.
31782 (extract_bit_field): Add alt_rtl argument. Pass to
31784 * expmed.h (extract_bit_field): Fix prototype.
31785 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31786 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31787 to extract_bit_field_calls.
31788 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31789 Pass alt_rtl to extract_bit_field calls.
31790 * calls.c (store_unaligned_arguments_into_psuedos)
31791 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31792 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31794 * optabs.h (struct expand_operand): Add target bitfield.
31796 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31798 * compare-elim.c (try_eliminate_compare): Canonicalize
31799 operation with embedded compare to
31800 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31801 (set (reg) (operation)].
31803 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31805 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31808 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31809 cost of adding a carry flag for ADC instruction.
31810 [case MINUS]: Ignore the cost of subtracting a carry flag
31811 for SBB instruction.
31813 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31815 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31817 * config/rs6000/bmiintrin.h: New file.
31818 * config/rs6000/bmi2intrin.h: New file.
31819 * config/rs6000/x86intrin.h: New file.
31821 2017-05-12 Jeff Law <law@redhat.com>
31823 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31826 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31828 PR middle-end/80707
31829 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31831 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31833 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31834 early expansion of vector divide builtins.
31835 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31836 builtins identified as having unsigned arguments.
31838 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31840 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31841 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31842 expansion of vector logical operations (and, andc, or, xor,
31845 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31847 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31848 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31850 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31852 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31853 early GIMPLE expansion of vector multiplies.
31855 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31857 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31858 TARGET_HAVE_MOVT conditional.
31859 (movt splitter): Likewise.
31861 2017-05-12 Richard Biener <rguenther@suse.de>
31863 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31864 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31865 Fold all stmts not inplace.
31867 2017-05-12 Richard Biener <rguenther@suse.de>
31869 PR tree-optimization/80713
31870 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31871 inserted_exprs bit for not removed stmts.
31873 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31875 PR middle-end/69921
31876 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31877 parallelized" attribute for parallelized OpenACC kernels.
31878 * omp-offload.c (execute_oacc_device_lower): Use it.
31880 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31881 Set "oacc kernels" attribute.
31882 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31883 parameter. Adjust all users.
31884 (oacc_fn_attrib_kernels_p): Remove function.
31885 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31886 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31887 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31888 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31889 assert "oacc kernels" attribute is set.
31891 2017-05-11 Carl Love <cel@us.ibm.com>
31893 * config/rs6000/rs6000-c: Add support for built-in functions
31894 vector unsigned char vec_popcnt (vector signed char)
31895 vector unsigned char vec_popcnt (vector unsigned char)
31896 vector unsigned short vec_popcnt (vector signed short)
31897 vector unsigned short vec_popcnt (vector unsigned short)
31898 vector unsigned int vec_popcnt (vector signed int)
31899 vector unsigned int vec_popcnt (vector unsigned int)
31900 vector unsigned long long vec_popcnt (vector signed long long)
31901 vector unsigned long long vec_popcnt (vector unsigned long long)
31902 vector signed long long vec_slo (vector signed long long,
31903 vector signed char)
31904 vector signed long long vec_slo (vector signed long long,
31905 vector unsigned char)
31906 vector unsigned long long vec_slo (vector unsigned long long,
31907 vector signed char)
31908 vector unsigned long long vec_slo (vector unsigned long long,
31909 vector unsigned char)
31910 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31911 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31912 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31913 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31914 * doc/extend.texi: Update the built-in documentation file for the
31915 new built-in functions.
31917 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31919 * attribs.h (sorted_attr_string): Move machine independent
31920 functions for target clone support from the i386 port to common
31921 code. Rename ix86_function_versions to common_function_versions.
31922 Rename make_name to make_unique_name.
31923 (common_function_versions): Likewise.
31924 (make_unique_name): Likewise.
31925 (make_dispatcher_decl): Likewise.
31926 (is_function_default_version): Likewise.
31927 * attribs.c (attr_strcmp): Likewise.
31928 (sorted_attr_string): Likewise.
31929 (common_function_versions): Likewise.
31930 (make_unique_name): Likewise.
31931 (make_dispatcher_decl): Likewise.
31932 (is_function_default_version): Likewise.
31933 * config/i386/i386.c (attr_strcmp): Likewise.
31934 (sorted_attr_string): Likewise.
31935 (ix86_function_versions): Likewise.
31936 (make_name): Likewise.
31937 (make_dispatcher_decl): Likewise.
31938 (is_function_default_version): Likewise.
31939 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31941 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31944 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31945 Account for direct move costs for vec_construct of integer
31948 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31951 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31952 (UNSPEC_STX_ATOMIC): Ditto.
31953 (loaddi_via_sse): New insn.
31954 (storedi_via_sse): Ditto.
31955 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31956 Update corresponding peephole2 patterns.
31957 (atomic_storedi_fpu): Ditto.
31959 2017-05-11 Julia Koval <julia.koval@intel.com>
31961 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31962 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31964 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31965 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31966 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31968 2017-05-11 Nathan Sidwell <nathan@acm.org>
31970 * graphite-poly.c: Include dumpfile.h.
31972 * dumpfle.h (dump_function): Declare here ...
31973 * tree-dump.h (dump_function): ... not here.
31974 * dumpfile.c: #include tree-cfg.h.
31975 (dump_function): Move here from ...
31976 * tree-dump.c (dump_function): ... here.
31977 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31978 * graphite-poly.c: Don't include tree-dump.h.
31979 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31980 * print-tree.c: Likewise.
31981 * stor-layout.c: Likewise.
31982 * tree-nested.c: Likewise.
31984 * dumpfile.c (dump_start): Use TDF_FLAGS.
31985 (dump_enable_all): Fix TDF_KIND check thinko.
31987 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31989 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31990 array entries to represent two legal parameterizations of the
31991 overloaded __builtin_cmpb function, as represented by the
31992 P6_OV_BUILTIN_CMPB constant.
31993 (altivec_resolve_overloaded_builtin): Add special case handling
31994 for the __builtin_cmpb function, as represented by the
31995 P6_OV_BUILTIN_CMPB constant.
31996 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31997 (BU_P6_64BIT_2): New macro.
31998 (BU_P6_OVERLOAD_2): New macro
31999 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
32000 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
32001 (CMPB): Add overload support to represent both 32-bit and 64-bit
32002 compare-bytes function.
32003 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
32004 support for TARGET_CMPB.
32005 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
32006 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32007 documentation of the __builtin_cmpb overloaded built-in function.
32009 2017-05-11 Richard Biener <rguenther@suse.de>
32011 PR tree-optimization/80705
32012 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
32013 bases are not vectorizable.
32015 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32017 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
32018 when counting register pressure.
32020 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32022 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
32023 (iv_ca_more_deps): Renamed to ...
32024 (iv_ca_compare_deps): ... this.
32025 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
32027 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32029 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
32031 (determine_group_iv_costs): ... here.
32032 (find_inv_vars_cb): Record inv var if it's not recorded before.
32034 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32036 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
32037 (get_shiftadd_cost): Ditto.
32039 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32041 * tree-ssa-address.c: Include header file.
32042 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
32044 (add_to_parts): Refactor.
32045 (addr_to_parts): New parameter. Update use of move_hint_to_base.
32046 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
32049 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32051 PR tree-optimization/53090
32052 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
32054 (extract_cond_operands): Detect condition with IV on both sides
32055 and return COMP_IV_EXPR_2.
32056 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
32057 (rewrite_use_compare): Simplify by removing call to function
32058 extract_cond_operands.
32060 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32062 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32063 (extract_cond_operands): Detect condition comparing against non-
32064 invariant bound and return appropriate enum value.
32065 (find_interesting_uses_cond): Update use of extract_cond_operands.
32066 Handle its return value accordingly.
32067 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32069 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32071 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32072 nonlinear iv_use computation in loop invariant sensitive way.
32074 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32076 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32077 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32079 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32081 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32082 (dump_cand): Support iv_cand.inv_exprs.
32083 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32085 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32088 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32090 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32092 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32093 as local function. Include necessary header files.
32094 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32096 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32098 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32100 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32102 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32103 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32104 RSHIFT_EXPR and BIT_NOT_EXPR.
32106 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32108 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32109 (adjust_setup_cost): New parameter supporting round up adjustment.
32110 (struct address_cost_data): Delete.
32111 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32112 (split_address_cost, ptr_difference_cost): Delete.
32113 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32114 (struct ainc_cost_data): New struct.
32115 (get_address_cost_ainc): New function.
32116 (get_address_cost, get_computation_cost): Reimplement.
32117 (determine_group_iv_cost_address): Record inv_expr for all uses of
32119 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32120 (iv_ca_has_deps): Reimplemented to ...
32121 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32123 (iv_ca_extend): Call iv_ca_more_deps.
32125 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32127 * tree-ssa-address.c (struct mem_address): Move to header file.
32128 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32129 * tree-ssa-address.h (struct mem_address): Move from C file.
32130 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32132 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32134 * tree-affine.h (aff_combination_type): New interface.
32135 (aff_combination_zero_p): Remove static.
32136 (aff_combination_const_p): New interface.
32137 (aff_combination_singleton_var_p): New interfaces.
32139 2017-05-11 Richard Biener <rguenther@suse.de>
32141 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32142 Skip unreachable blocks and destinations.
32143 (eliminate): Move stmt removal and fixup ...
32144 (fini_eliminate): ... here. Skip inserted exprs.
32145 (pass_pre::execute): Move fini_pre after fini_eliminate.
32146 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32147 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32148 PRE to get rid of dead code that has invalid SSA form and
32149 split critical edges again.
32151 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32153 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32155 2017-05-11 Richard Biener <rguenther@suse.de>
32157 * passes.c (execute_function_todo): Verify loops if they are
32158 said to be up-to-date.
32159 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32160 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32162 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32165 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32166 handle calling assemble_external ourself.
32169 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32170 modes with zero size. Enhance comment.
32172 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32174 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32175 built-ins for vec_xl and vec_xst with short and char pointer
32178 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32180 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32181 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32182 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32183 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32184 (_mm_maskz_min_round_ss): New intrinsics.
32185 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32186 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32187 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32188 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32189 (__builtin_ia32_minss_mask_round): New builtins.
32190 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32191 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32192 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32194 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32195 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32197 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32200 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32202 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32203 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32204 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32205 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32206 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32207 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32208 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32209 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32210 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32211 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32212 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32213 (__builtin_ia32_mulss_mask_round): New builtins.
32214 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32215 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32216 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32218 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32219 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32221 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32224 2017-05-10 Julia Koval <julia.koval@intel.com>
32226 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32227 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32228 (_mm256_setr_m128i): New intrinsics.
32230 2017-05-10 Julia Koval <julia.koval@intel.com>
32232 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32233 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32234 (_mm_maskz_rcp14_ss): New intrinsics.
32235 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32236 (__builtin_ia32_rcp14ss_mask): New builtins.
32237 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32239 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32241 PR tree-optimization/51513
32242 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32243 (assert_unreachable_fallthru_edge_p): Use it.
32244 (group_case_labels_stmt): Likewise.
32245 * tree-cfg.h: Prototype it.
32246 * stmt.c: Include cfghooks.h and tree-cfg.h.
32247 (emit_case_dispatch_table) <gap_label>: New local variable.
32248 Use it to fill dispatch table gaps.
32249 Test for default_label before updating probabilities.
32250 (expand_case) <default_label>: Remove unneeded initialization.
32251 Test for unreachable default case statement and remove its edge.
32252 Set default_label accordingly.
32253 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32255 2017-05-10 Carl Love <cel@us.ibm.com>
32257 * config/rs6000/rs6000-c: Add support for built-in functions
32258 vector signed char vec_neg (vector signed char)
32259 vector signed short int vec_neg (vector short int)
32260 vector signed int vec_neg (vector signed int)
32261 vector signed long long vec_neg (vector signed long long)
32262 vector float vec_neg (vector float)
32263 vector double vec_neg (vector double)
32264 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32266 * config/rs6000/altivec.h: Add define for vec_neg
32267 * doc/extend.texi: Update the built-in documentation for the
32268 new built-in functions.
32270 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32272 PR tree-optimization/77644
32273 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32275 2017-05-10 Nathan Sidwell <nathan@acm.org>
32277 * dumpfile.h (TDI_lang_all): New.
32278 (TDF_KIND): New. Renumber others
32279 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32281 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32283 (get_dump_file_name): Adjust suffix generation.
32284 (dump_enable_all): Use TDF_KIND.
32285 * doc/invoke.texi (-fdump-lang-all): Document.
32287 * dumpfile.h: Tabify.
32289 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32292 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32293 Move member access before delete.
32295 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32297 * tree-inline.c (expand_call_inline): Split block at stmt
32300 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32303 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32304 are now unused after splitting mov{sf,sd}_hardfloat.
32305 (f32_lr2): Likewise.
32306 (f32_lm): Likewise.
32307 (f32_lm2): Likewise.
32308 (f32_li): Likewise.
32309 (f32_li2): Likewise.
32310 (f32_lv): Likewise.
32311 (f32_sr): Likewise.
32312 (f32_sr2): Likewise.
32313 (f32_sm): Likewise.
32314 (f32_sm2): Likewise.
32315 (f32_si): Likewise.
32316 (f32_si2): Likewise.
32317 (f32_sv): Likewise.
32318 (f32_dm): Likewise.
32319 (f32_vsx): Likewise.
32320 (f32_av): Likewise.
32321 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32322 For movsf, order stores so the VSX stores occur before the GPR
32323 store which encourages the register allocator to use a traditional
32324 FPR instead of a GPR. For movsd, order the stores so that the GPR
32325 store comes before the VSX stores to allow the power6 to work.
32326 This is due to the power6 not having a 32-bit integer store
32327 instruction from a FPR.
32328 (movsf_hardfloat): Likewise.
32329 (movsd_hardfloat): Likewise.
32331 2017-05-09 Martin Sebor <msebor@redhat.com>
32333 PR translation/80280
32334 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32337 PR translation/80280
32338 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32339 data member added in r247778.
32340 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32342 2017-05-09 Nathan Sidwell <nathan@acm.org>
32344 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32346 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32349 2017-05-09 Marek Polacek <polacek@redhat.com>
32351 * doc/invoke.texi: Fix typo.
32353 2017-05-09 Richard Biener <rguenther@suse.de>
32355 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32356 (vrp_val_is_min): Likewise.
32357 (set_value_range_to_value): Likewise.
32358 (set_value_range_to_nonnegative): Likewise.
32359 (gimple_assign_nonzero_p): Likewise.
32360 (gimple_stmt_nonzero_p): Likewise.
32361 (vrp_int_const_binop): Likewise. Remove unreachable case.
32362 (adjust_range_with_scev): Adjust comments.
32363 (compare_range_with_value): Likewise.
32364 (extract_range_from_phi_node): Likewise.
32365 (test_for_singularity): Likewise.
32367 2017-05-09 Richard Biener <rguenther@suse.de>
32369 * tree-vrp.c (get_single_symbol): Add assert that we don't
32370 get overflowed constants as invariant part.
32371 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32372 checks. Use wi::cmp instead of recursing for integer constants.
32373 (compare_values): Just ignore whether we assumed undefined
32374 overflow instead of failing the compare.
32375 (extract_range_for_var_from_comparison_expr): Add comment before the
32376 TREE_NO_WARNING sets.
32377 (test_for_singularity): Likewise.
32378 (extract_range_from_comparison): Do not disable optimization
32379 when we assumed undefined overflow.
32380 (extract_range_basic): Remove init of unused var.
32382 2017-05-09 Richard Biener <rguenther@suse.de>
32384 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32385 (extract_range_from_multiplicative_op_1): Adjust.
32386 (extract_range_from_binary_expr_1): Use int_const_binop.
32388 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32391 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32392 rs6000_store_data_bypass_p in seven define_bypass directives and
32393 in several comments.
32394 * config/rs6000/rs6000-protos.h: Add prototype for
32395 rs6000_store_data_bypass_p function.
32396 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32397 function implements slightly different (rs6000-specific) semantics
32398 than store_data_bypass_p, returning false rather than aborting
32399 with assertion error when arguments do not satisfy the
32400 requirements of store data bypass.
32401 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32402 rs6000_store_data_bypass_p.
32404 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32406 * config/xtensa/xtensa-protos.h
32407 (xtensa_initial_elimination_offset): New declaration.
32408 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32409 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32410 macro definition, add case for FRAME_POINTER_REGNUM when
32411 FRAME_GROWS_DOWNWARD.
32412 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32413 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32414 xtensa_initial_elimination_offset.
32416 2017-05-08 Nathan Sidwell <nathan@acm.org>
32418 * doc/invoke.texi: Alphabetize -fdump options.
32420 2017-05-08 Martin Sebor <msebor@redhat.com>
32422 PR translation/80280
32423 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32425 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32427 * target.def (compute_frame_layout): New optional target hook.
32428 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32429 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32430 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32432 * reload1.c (verify_initial_elim_offsets): Likewise.
32433 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32434 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32435 (arm_get_frame_offsets): Split up into this ...
32436 (arm_compute_frame_layout): ... and this function.
32438 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32440 * config/aarch64/constraints.md (Usa): New constraint.
32441 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32443 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32445 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32446 with_multilib_list after it has been checked.
32448 2017-05-08 Richard Biener <rguenther@suse.de>
32450 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32451 (bitmap_set_subtract_values): Likewise.
32453 2017-05-08 Richard Biener <rguenther@suse.de>
32455 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32456 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32458 (gimple_stmt_nonzero_warnv_p): Rename to ...
32459 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32461 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32462 (extract_range_basic): Adjust, do not disable propagation on
32463 strict overflow sensitive simplification.
32464 (vrp_visit_cond_stmt): Likewise.
32466 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32468 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32469 body size unconditionally.
32471 2017-05-07 Jeff Law <law@redhat.com>
32474 2017-05-06 Jeff Law <law@redhat.com>
32475 PR tree-optimization/78496
32476 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32479 PR tree-optimization/78496
32480 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32481 (simplify_stmt_using_ranges): Call it.
32482 (vrp_dom_walker::before_dom_children): Extract equivalences
32483 from an ASSERT_EXPR with an equality comparison against a
32486 2017-05-06 Jeff Law <law@redhat.com>
32488 PR tree-optimization/78496
32489 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32492 PR tree-optimization/78496
32493 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32494 (simplify_stmt_using_ranges): Call it.
32495 (vrp_dom_walker::before_dom_children): Extract equivalences
32496 from an ASSERT_EXPR with an equality comparison against a
32499 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32501 * lra-constraints.c (lra_copy_reg_equiv): New function.
32502 (split_reg): Use it to copy equivalence information from the
32503 original register to the spill register.
32505 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32507 PR rtl-optimization/75964
32508 * simplify-rtx.c (simplify_const_relational_operation): Remove
32509 invalid handling of comparisons of integer ABS.
32511 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32513 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32514 initialize to zero.
32515 (init_regs): Remove declaration.
32516 (function_arg_advance_32): Initialize error_p as boolean variable.
32518 2017-05-05 Nathan Sidwell <nathan@acm.org>
32520 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32521 lines. Use for (;;).
32523 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32525 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32526 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32527 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32528 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32529 VF=2 that require versioning.
32531 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32533 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32536 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32538 * diagnostic.h (diagnostic_override_option_index): Convert from
32539 macro to inline function.
32541 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32543 * diagnostic.c (last_module_changed_p): New function.
32544 (set_last_module): New function.
32545 (diagnostic_report_current_module): Convert macro usage to
32546 the above functions.
32547 * diagnostic.h (diagnostic_context::last_module): Strengthen
32548 from const line_map * to const line_map_ordinary *.
32549 (diagnostic_last_module_changed): Delete macro.
32550 (diagnostic_set_last_module): Delete macro.
32552 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32554 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32555 with diagnostic_report_diagnostic.
32556 (diagnostic_n_impl_richloc): Likewise.
32557 * diagnostic.h (report_diagnostic): Delete macro.
32558 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32559 with diagnostic_report_diagnostic.
32560 * substring-locations.c (format_warning_va): Likewise.
32562 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32564 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32565 save/restor of format_spec. Move option-printing code to...
32566 (print_option_information): ...this new function, and
32567 reimplement by simply printing to the pretty_printer,
32568 rather than appending to the format string.
32570 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32572 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32573 handling logic into...
32574 (update_effective_level_from_pragmas): ...this new function.
32576 2017-05-04 Andrew Waterman <andrew@sifive.com>
32578 * config/riscv/riscv.opt (mstrict-align): New option.
32579 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32580 (SLOW_UNALIGNED_ACCESS): Define.
32581 (riscv_slow_unaligned_access): Declare.
32582 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32584 (riscv_slow_unaligned_access): New variable.
32585 (rocket_tune_info): Set slow_unaligned_access to true.
32586 (optimize_size_tune_info): Set slow_unaligned_access to false.
32587 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32588 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32589 (riscv_option_override): Set riscv_slow_unaligned_access.
32590 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32592 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32594 * config/riscv/riscv.md: Unify indentation.
32596 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32601 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32603 (extendsi<mode>2): Add support for doing sign extension via
32604 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32605 don't have ISA 3.0 instructions.
32606 (extendsi<mode>2 splitter): Likewise.
32607 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32608 generate the normal insns since SImode can now go in vector
32609 registers. Disallow the special UNSPECs needed for previous
32610 machines to hide SImode being used. Add new insns
32611 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32612 (fix_trunc<mode>si2_stfiwx): Likewise.
32613 (fix_trunc<mode>si2_internal): Likewise.
32614 (fixuns_trunc<mode>si2): Likewise.
32615 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32616 (fctiw<u>z_<mode>_smallint): Likewise.
32617 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32618 of floating point to 32-bit integer from doing a direct move to
32619 the GPR registers to do a store.
32620 (fctiwz_<mode>): Break long line.
32622 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32624 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32625 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32626 (addr_list, addr_offset_valid_p): New.
32627 (split_address_groups): Check offset validity with above function.
32628 (gt-tree-ssa-loop-ivopts.h): Include header file.
32630 2017-05-05 Nathan Sidwell <nathan@acm.org>
32632 * config.gcc (arm*-*-*): Add missing 'fi'.
32634 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32636 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32637 included in -fopt-info does not matter.
32638 * doc/optinfo.texi (-fopt-info): Fix description of default
32639 behavour. Explicitly say order of options included in -fopt-info
32642 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32644 * config.gcc: Allow combinations of aprofile and rmprofile values for
32645 --with-multilib-list.
32646 * config/arm/t-multilib: New file.
32647 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32648 variables. Remove setting of ISA and floating-point ABI in
32649 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32650 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32651 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32652 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32654 * config/arm/t-rmprofile: Likewise except for the matches changes.
32655 * doc/install.texi (--with-multilib-list): Document the combination of
32656 aprofile and rmprofile values and warn about pitfalls in doing that.
32658 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32660 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32661 (movdi_aarch64): Likewise.
32663 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32665 PR tree-optimization/80632
32666 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32668 (build_arrays): Initialize it for virtual phis.
32669 (fix_phi_nodes): Use it for virtual phis.
32671 PR tree-optimization/80558
32672 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32673 [x, y] op z into [x op, y op z] for op & or | if conditions
32676 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32677 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32680 * config/arm/arm.md (use_literal_pool): Remove.
32681 (64-bit immediate split): No longer takes cost into consideration
32682 if arm_disable_literal_pool is enabled.
32683 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32684 used when arm_disable_literal_pool is enabled.
32685 (arm_max_const_double_inline_cost): Remove use of
32686 arm_disable_literal_pool.
32687 (push_minipool_fix): Add assert.
32688 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32689 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32690 (no_literal_pool_sf_immediate): New.
32692 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32694 PR tree-optimization/80613
32695 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32696 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32698 2017-05-05 Richard Biener <rguenther@suse.de>
32700 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32702 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32704 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32705 of this flag from insn conditions due to removal from r247495.
32707 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32709 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32711 (arm_early_store_addr_dep_ptr): Likewise.
32712 * config/arm/aarch-common-protos.h
32713 (arm_early_load_addr_dep_ptr): Add prototype.
32714 (arm_early_store_addr_dep_ptr): Likewise.
32715 * config/arm/cortex-a53.md: Add new bypasses.
32717 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32719 * tree.c (next_type_uid): Change type to unsigned.
32720 (type_hash_canon): Decrement back next_type_uid if
32721 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32722 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32725 2017-05-04 Martin Sebor <msebor@redhat.com>
32727 * builtins.c: Fix a trivial typo in a comment.
32729 PR middle-end/79234
32730 * builtins.c (check_sizes): Adjust to handle reading past the end.
32731 Avoid printing excessive upper bound of ranges. Use %E to print
32732 tree nodes instead of converting them to %wu.
32733 (expand_builtin_memchr): New function.
32734 (compute_dest_size): Rename...
32735 (compute_objsize): ...to this.
32736 (expand_builtin_memcpy): Adjust.
32737 (expand_builtin_mempcpy): Adjust.
32738 (expand_builtin_strcat): Adjust.
32739 (expand_builtin_strcpy): Adjust.
32740 (check_strncat_sizes): Adjust.
32741 (expand_builtin_strncat): Adjust.
32742 (expand_builtin_strncpy): Adjust and simplify.
32743 (expand_builtin_memset): Adjust.
32744 (expand_builtin_bzero): Adjust.
32745 (expand_builtin_memcmp): Adjust.
32746 (expand_builtin): Handle memcmp.
32747 (maybe_emit_chk_warning): Check strncat just once.
32749 2017-05-04 Martin Sebor <msebor@redhat.com>
32751 PR preprocessor/79214
32752 PR middle-end/79222
32753 PR middle-end/79223
32754 * builtins.c (check_sizes): Add inlining context and issue
32755 warnings even when -Wno-system-headers is set.
32756 (check_strncat_sizes): Same.
32757 (expand_builtin_strncat): Same.
32758 (expand_builtin_memmove): New function.
32759 (expand_builtin_stpncpy): Same.
32760 (expand_builtin): Handle memmove and stpncpy.
32762 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32764 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32765 which is not used any more.
32767 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32769 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32771 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32773 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32774 (cortexa53_tunings): Likewise.
32775 (cortexa57_tunings): Likewise.
32776 (cortexa72_tunings): Likewise.
32777 (cortexa73_tunings): Likewise.
32779 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32781 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32782 Set loop alignment to 8.
32784 2017-05-04 Martin Sebor <msebor@redhat.com>
32786 PR translation/80280
32787 * builtins.c (expand_builtin_object_size): Add missing quoting to
32788 %D and like directives.
32789 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32790 (hsa_type_for_tree_type): Same.
32791 (verify_function_arguments): Same.
32792 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32793 * varasm.c (get_section): Same.
32796 2017-05-04 Martin Sebor <msebor@redhat.com>
32798 PR translation/80280
32799 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32801 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32803 * config/aarch64/aarch64.c (generic_addrcost_table):
32804 Change HI/TI mode setting.
32806 2017-05-04 Martin Jambor <mjambor@suse.cz>
32808 PR tree-optimization/80622
32809 * tree-sra.c (comes_initialized_p): New function.
32810 (build_accesses_from_assign): Only set write lazily when
32811 comes_initialized_p is false.
32812 (analyze_access_subtree): Use comes_initialized_p.
32813 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32814 instead of testing for PARM_DECL.
32816 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32818 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32819 constraint on operand 0 to allow more general addressing modes.
32820 Adjust output template.
32821 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32823 * config/aarch64/aarch64-protos.h
32824 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32825 * config/aarch64/constraints.md (Dp): New address constraint.
32826 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32829 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32831 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32832 update use of estimate_ipcp_clone_size_and_time.
32833 (estimate_local_effects): Update use of
32834 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32835 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32836 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32837 Return nonspecialized time.
32839 2017-05-04 Richard Biener <rguenther@suse.de>
32841 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32842 for the last VUSE which def dominates the PHI. Directly call
32844 (get_continuation_for_phi_1): Remove.
32846 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32848 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32849 to explain the use of truncating division. Cap the number of
32850 iterations to the maximum given by nb_iterations_upper_bound,
32853 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32855 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32856 * configure: Regenerate.
32857 * config.in: Regenerate.
32858 * config/i386/driver-mingw32.c: new file.
32859 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32860 * config.host: Link driver-mingw32.o on MinGW host.
32861 * doc/install.texi: Document new --enable-mingw-wildcard configure
32864 2017-05-04 Marek Polacek <polacek@redhat.com>
32866 PR tree-optimization/80612
32867 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32869 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32870 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32872 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32873 (movt splitter): Likewise.
32874 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32875 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32876 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32877 block for Thumb-1 with MOVT.
32878 (thumb2_legitimate_address_p): Move code block ...
32879 (can_avoid_literal_pool_for_label_p): ... into this new function.
32880 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32882 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32883 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32884 "M-profile targets with the MOVT instruction".
32886 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32888 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32889 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32890 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32892 2017-05-04 Martin Liska <mliska@suse.cz>
32894 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32895 variable cond_code.
32897 2017-05-04 Richard Biener <rguenther@suse.de>
32899 * tree.c (array_at_struct_end_p): Handle arrays at struct
32900 end with flexarrays more conservatively. Refactor and treat
32901 arrays of arrays or aggregates more strict. Fix
32902 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32903 * tree.c (array_at_struct_end_p): Adjust prototype.
32904 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32905 * gimple-fold.c (get_range_strlen): Likewise.
32906 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32908 2017-05-04 Richard Biener <rguenther@suse.de>
32910 PR tree-optimization/31130
32911 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32913 (supports_overflow_infinity): Likewise.
32914 (is_negative_overflow_infinity): Likewise.
32915 (is_positive_overflow_infinity): Likewise.
32916 (is_overflow_infinity): Likewise.
32917 (stmt_overflow_infinity): Likewise.
32918 (overflow_infinity_range_p): Likewise.
32919 (usable_range_p): Remove as always returning true.
32920 (make_overflow_infinity): Remove.
32921 (negative_overflow_infinity): Likewise.
32922 (positive_overflow_infinity): Likewise.
32923 (avoid_overflow_infinity): Likewise.
32924 (set_value_range): Adjust accordingly.
32925 (set_value_range_to_nonnegative): Likewise, remove now unused
32926 overflow_infinity arg.
32927 (vrp_operand_equal_p): Adjust.
32928 (update_value_range): Likewise.
32929 (range_int_cst_singleton_p): Likewise.
32930 (operand_less_p): Likewise.
32931 (compare_values_warnv): Likewise.
32932 (extract_range_for_var_from_comparison_expr): Likewise.
32933 (vrp_int_const_binop): Likewise.
32934 (zero_nonzero_bits_from_vr): Likewise.
32935 (extract_range_from_multiplicative_op_1): Likewise.
32936 (extract_range_from_binary_expr_1): Likewise.
32937 (extract_range_from_unary_expr): Likewise.
32938 (extract_range_from_comparison): Likewise.
32939 (extract_range_basic): Likewise.
32940 (adjust_range_with_scev): Likewise.
32941 (compare_ranges): Likewise.
32942 (compare_range_with_value): Likewise.
32943 (dump_value_range): Likewise.
32944 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32946 (simplify_cond_using_ranges): Adjust.
32948 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32950 * brig-builtins.def: Added a builtin for class_f64.
32951 * builtin-types.def: Added a builtin type needed by class_f64.
32953 2017-05-03 Jason Merrill <jason@redhat.com>
32955 * timevar.def: Add TV_CONSTEXPR.
32957 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32959 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32961 2017-05-03 Martin Jambor <mjambor@suse.cz>
32963 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32964 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32965 * ipa-cp.c (ipcp_propagate_stage): Do not call
32966 ipa_update_after_lto_read.
32967 * ipa-inline.c (ipa_inline): Likewise.
32969 2017-05-03 Martin Jambor <mjambor@suse.cz>
32971 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32972 tag. Added a default constructor and a destructor.
32973 (ipa_edge_args_sum_t): New class;
32974 (ipa_edge_args_sum): Declare.
32975 (ipa_edge_args_vector): Remove declaration.
32976 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32977 (ipa_free_edge_args_substructures): Remove declaration.
32978 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32979 (ipa_edge_args_info_available_for_edge_p): Likewise.
32980 * ipa-prop.c (ipa_edge_args_vector): Removed.
32981 (edge_removal_hook_holder): Likewise.
32982 (edge_duplication_hook_holder): Likewise.
32983 (ipa_edge_args_sum): New variable.
32984 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32985 ipa_edge_args_vector.
32986 (ipa_free_edge_args_substructures): Likewise.
32987 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32988 ipa_edge_args_vector.
32989 (ipa_edge_removal_hook): Turned into method
32990 ipa_edge_args_sum_t::remove.
32991 (ipa_edge_duplication_hook): Turned into method
32992 ipa_edge_args_sum_t::duplicate.
32993 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32994 registering edge hooks.
32995 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32996 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32997 ipa_edge_args_sum instead of ipa_edge_args_vector.
32998 * ipa-profile.c (ipa_profile): Likewise.
33000 2017-05-03 Martin Jambor <mjambor@suse.cz>
33002 * symbol-summary.h (function_summary): New method exists.
33003 (function_summary::symtab_removal): Deallocate through release.
33004 (call_summary): New class.
33005 (gt_ggc_mx): New overload.
33006 (gt_pch_nx): Likewise.
33007 (gt_pch_nx): Likewise.
33009 2017-05-03 Jeff Law <law@redhat.com>
33011 PR tree-optimization/78496
33012 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
33013 from simplify_cond_using_ranges. Split off code to walk
33014 backwards through casts into ...
33015 (simplify_cond_using_ranges_2): New function.
33016 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
33017 (execute_vrp): After identifying jump threads, call
33018 simplify_cond_using_ranges_2.
33020 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
33023 * ipa-inline.h (inline_summary): Add ctor.
33024 (create_ggc): Do not use ggc_cleared_alloc.
33026 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
33027 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33029 * gcc.c (handle_braces): Support escaping in switch matching
33031 * doc/invoke.texi (Spec Files): Document it.
33032 Remove superfluous @code markup in items.
33034 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33036 * diagnostic-show-locus.c (struct column_range): New struct.
33037 (get_affected_columns): New function.
33038 (get_printed_columns): New function.
33039 (struct correction): New struct.
33040 (correction::ensure_capacity): New function.
33041 (correction::ensure_terminated): New function.
33042 (struct line_corrections): New struct.
33043 (line_corrections::~line_corrections): New dtor.
33044 (line_corrections::add_hint): New function.
33045 (layout::print_trailing_fixits): Reimplement in terms of the new
33047 (selftest::test_overlapped_fixit_printing): New function.
33048 (selftest::diagnostic_show_locus_c_tests): Call it.
33050 2017-05-03 Nathan Sidwell <nathan@acm.org>
33052 Canonicalize canonical type hashing
33053 * tree.h (type_hash_canon_hash): Declare.
33054 * tree.c (type_hash_list, attribute_hash_list): Move into
33055 type_hash_canon_hash.
33056 (build_type_attribute_qual_variant): Break out hash code calc into
33057 type_hash_canon_hash.
33058 (type_hash_canon_hash): New. Generic type hash computation.
33059 (build_range_type_1, build_array_type_1, build_function_type,
33060 build_method_type_directly, build_offset_type, build_complex_type,
33061 make_vector_type): Call it.
33063 2017-05-03 Richard Biener <rguenther@suse.de>
33065 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33066 When all DRs have unknown misaligned do not always peel
33067 when there is a store but apply the same costing model as if
33068 there were only loads.
33070 2017-05-03 Richard Biener <rguenther@suse.de>
33073 PR tree-optimization/80492
33074 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33075 compare_base_decls returning dont-know properly.
33077 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33079 * config/arm/iterators.md (CCSI): New mode iterator.
33080 (arch): New mode attribute.
33081 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33082 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33083 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33084 code iterator for success result mode.
33085 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33086 the corresponding new insn generators.
33088 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33091 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33092 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33094 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33096 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33097 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33098 (DDR_A): Wrap DDR argument in brackets.
33099 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33100 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33101 (DDR_REVERSED_P): Likewise.
33103 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33105 PR tree-optimization/79472
33106 * tree-switch-conversion.c (struct switch_conv_info): Add
33107 contiguous_range and default_case_nonstandard fields.
33108 (collect_switch_conv_info): Compute contiguous_range and
33109 default_case_nonstandard fields, don't clear final_bb if
33110 contiguous_range and only the default case doesn't have the required
33112 (check_all_empty_except_final): Set default_case_nonstandard instead
33113 of failing if contiguous_range and the default case doesn't have empty
33115 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33116 and only the default case doesn't have the required constants. Skip
33118 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33119 if default_case_nonstandard.
33120 (build_constructors): Build constant 1 just once. Assert that default
33121 values aren't inserted in between cases if contiguous_range. Skip
33123 (build_arrays): Skip virtual phis.
33124 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33125 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33126 Handle virtual phis.
33127 (gen_inbound_check): Handle default_case_nonstandard case.
33128 (process_switch): Adjust check_final_bb caller. Call
33129 gather_default_values with the first non-default case instead of
33130 default case if default_case_nonstandard.
33132 2017-05-02 Nathan Sidwell <nathan@acm.org>
33134 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33135 check. Fix formatting.
33137 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33139 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33140 errors when comparing specialized and unspecialized times.
33142 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33144 * diagnostic-show-locus.c
33145 (layout::should_print_annotation_line_p): Make private.
33146 (layout::print_annotation_line): Make private.
33147 (layout::annotation_line_showed_range_p): Make private.
33148 (layout::show_ruler): Make private.
33149 (layout::print_source_line): Make private. Pass in line and
33150 line_width, rather than calling location_get_source_line. Drop
33152 (layout::print_leading_fixits): New method.
33153 (layout::print_any_fixits): Rename to...
33154 (layout::print_trailing_fixits): ...this, and make private.
33155 Don't print newline fixits.
33156 (diagnostic_show_locus): Move logic for printing one row into...
33157 (layout::print_line): ...this new function. Move the
33158 location_get_source_line call and error-handling from
33159 print_source_line to here. Call print_leading_fixits, and rename
33160 print_any_fixits to print_trailing_fixits.
33161 (selftest::test_fixit_insert_containing_newline): Update now that
33162 newlines are partially supported.
33163 (selftest::test_fixit_insert_containing_newline_2): New test.
33164 (selftest::test_fixit_replace_containing_newline): Update comments.
33165 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33166 * edit-context.c (class added_line): New class.
33167 (class edited_line): Describe newline handling in comment.
33168 (edited_line::actually_edited_p): New method.
33169 (edited_line::print_content): Delete redundant decl.
33170 (edited_line::m_predecessors): New field.
33171 (edited_file::print_content): Call edited_line::print_content.
33172 (edited_file::print_diff): Update to support newlines.
33173 (edited_file::print_diff_hunk): Likewise.
33174 (edited_file::print_run_of_changed_lines): New function.
33175 (edited_file::print_diff_line): Convert to...
33176 (print_diff_line): ...this.
33177 (edited_file::get_effective_line_count): New function.
33178 (edited_line::edited_line): Initialize new field m_predecessors.
33179 (edited_line::~edited_line): Clean up m_predecessors.
33180 (edited_line::apply_fixit): Handle newlines.
33181 (edited_line::get_effective_line_count): New function.
33182 (edited_line::print_content): New function.
33183 (edited_line::print_diff_lines): New function.
33184 (selftest::test_applying_fixits_insert_containing_newline): New
33186 (selftest::test_applying_fixits_replace_containing_newline): New
33188 (selftest::insert_line): New function.
33189 (selftest::test_applying_fixits_multiple_lines): Add example of
33191 (selftest::edit_context_c_tests): Call the new tests.
33193 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33195 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33196 parameter cand. Update dump information.
33197 (get_computation_cost): Update uses.
33199 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33201 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33202 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33203 (get_computation_at, rewrite_use_address): Update use of
33204 get_computation_aff.
33206 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33208 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33209 (get_computation): Delete.
33210 (get_computation_cost): Implement like get_computation_cost_at.
33211 Use get_computation_at.
33212 (get_computation_cost_at): Delete.
33213 (rewrite_use_nonlinear_expr): Use get_computation_at.
33214 (rewrite_use_compare, remove_unused_ivs): Ditto.
33216 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33218 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33220 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33222 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33223 (ivopts_global_cost_for_size): Rename parameter and update uses.
33224 (iv_ca_recount_cost): Update uses.
33225 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33226 candidates seperately in n_invs and n_cands.
33227 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33229 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33231 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33232 (find_inv_vars_cb): New.
33233 (find_depends): Renamed to ...
33234 (find_inv_vars): ... this.
33235 (add_candidate_1, force_var_cost): Call find_inv_vars.
33236 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33238 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33240 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33241 inv_vars. Add inv_exprs.
33242 (struct iv_cand): Rename depends_on to inv_vars.
33243 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33244 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33245 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33246 (dump_cand): Dump inv_vars.
33247 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33248 (record_invariant, find_depends, add_candidate_1): Ditto.
33249 (set_group_iv_cost, force_var_cost): Ditto.
33250 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33251 (get_computation_cost_at, get_computation_cost): Ditto.
33252 (determine_group_iv_cost_generic): Ditto.
33253 (determine_group_iv_cost_address): Ditto.
33254 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33255 (determine_group_iv_costs): Ditto.
33256 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33257 (iv_ca_set_remove_invariants): Renamed to ...
33258 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33259 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33260 (iv_ca_set_add_invariants): Renamed to ...
33261 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33262 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33263 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33264 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33265 (create_new_ivs): Remove useless dump.
33267 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33269 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33271 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33272 (iv_ca_set_no_cp, create_new_iv): Ditto.
33274 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33276 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33278 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33280 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33281 function tree_check2.
33283 2017-05-02 Martin Liska <mliska@suse.cz>
33285 * doc/gcov.texi: Add missing preposition.
33286 * gcov.c (function_info::function_info): Properly fill up
33287 all member variables.
33289 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33291 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33293 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33295 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33297 2017-05-02 Martin Liska <mliska@suse.cz>
33300 * lto-streamer-in.c (lto_read_tree_1): Remove
33301 LTO_STREAMER_DEBUG.
33302 * lto-streamer.c (struct tree_hash_entry): Likewise.
33303 (struct tree_entry_hasher): Likewise.
33304 (tree_entry_hasher::hash): Likewise.
33305 (tree_entry_hasher::equal): Likewise.
33306 (lto_streamer_init): Likewise.
33307 (lto_orig_address_map): Likewise.
33308 (lto_orig_address_get): Likewise.
33309 (lto_orig_address_remove): Likewise.
33310 * lto-streamer.h: Likewise.
33311 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33312 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33314 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33316 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33317 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33318 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33319 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33320 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33321 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33322 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33323 (mm_maskz_sub_ss): New intrinsics.
33324 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33325 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33326 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33327 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33328 (__builtin_ia32_subss_mask_round): New builtins.
33329 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33330 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33331 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33333 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33334 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33336 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33339 2017-05-02 Martin Jambor <mjambor@suse.cz>
33341 PR tree-optimization/78687
33342 * tree-sra.c (access): New field parent.
33343 (process_subtree_disqualification): New function.
33344 (disqualify_candidate): Call it.
33345 (build_accesses_from_assign): Reset write flag if creating an
33347 (build_access_subtree): Fill in parent field and also prpagate
33348 down grp_write flag.
33349 (create_artificial_child_access): New parameter set_grp_write, set
33350 grp_write to its value.
33351 (propagate_subaccesses_across_link): Also propagate grp_write flag
33353 (propagate_all_subaccesses): Push the closest parent back to work
33354 queue if add_access_to_work_queue returned true.
33356 2017-05-02 Richard Biener <rguenther@suse.de>
33358 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33359 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33360 -fstrict-overflow documentation.
33361 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33362 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33363 flag_strict_overflow.
33364 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33365 * lto-opts.c (lto_write_options): Do not stream it.
33366 * lto-wrapper.c (merge_and_complain): Do not handle it.
33367 * opts.c (default_options_table): Do not set -fstrict-overflow.
33368 (finish_options): Likewise do not clear it when sanitizing.
33369 * simplify-rtx.c (simplify_const_relational_operation): Do not
33370 test flag_strict_overflow.
33372 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33374 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33375 using enabled attribute.
33376 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33377 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33378 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33379 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33380 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33381 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33382 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33383 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33384 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33385 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33387 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33389 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33391 2017-05-02 Richard Biener <rguenther@suse.de>
33393 PR tree-optimization/80591
33395 2017-04-10 Richard Biener <rguenther@suse.de>
33397 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33400 2017-05-02 Richard Biener <rguenther@suse.de>
33402 PR tree-optimization/80549
33403 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33404 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33405 headers if they do not have a preheader.
33407 2017-05-02 Martin Liska <mliska@suse.cz>
33410 * common.opt: Fix typo.
33411 * doc/invoke.texi: Likewise.
33413 2017-05-01 Jan Beulich <jbeulich@suse.com>
33415 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33416 swapping, add (x,x,m,x,n) alternative.
33418 2017-05-01 Nathan Sidwell <nathan@acm.org>
33420 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33421 unnecessary unadjusted_alignment check.
33423 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33426 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33428 * gimplify.c (gimplify_cilk_detach): New function.
33429 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33430 * tree-core.h: Document EXPR_CILK_SPAWN.
33431 * tree.h (EXPR_CILK_SPAWN): Define.
33433 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33435 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33436 to use new fixit_hint representation, using the "replace" logic.
33437 (get_line_span_for_fixit_hint): Likewise.
33438 (layout::print_any_fixits): Likewise.
33439 (selftest::test_one_liner_many_fixits): Rename to...
33440 (selftest::test_one_liner_many_fixits_1): ...this, and update
33441 comment and expected output to reflect that the multiple fix-it
33442 hints are now consolidated into one insertion.
33443 (selftest::test_one_liner_many_fixits_2): New test.
33444 (selftest::test_diagnostic_show_locus_one_liner): Update for
33446 (selftest::test_fixit_consolidation): Update for fix-it API
33448 * diagnostic.c (print_parseable_fixits): Likewise.
33449 * edit-context.c (edited_line::m_line_events): Convert from
33450 auto_vec <line_event *> to auto_vec <line_event>.
33451 (class line_event): Convert from abstract base class to a concrete
33452 class, taking over the role of replace_event.
33453 (class insert_event): Delete.
33454 (class replace_event): Rename to class line_event. Convert to
33456 (edit_context::add_fixits): Reimplement.
33457 (edit_context::apply_insert): Delete.
33458 (edit_context::apply_replace): Rename to...
33459 (edit_context::apply_fixit): ...this. Convert to half-open range.
33460 (edited_file::apply_insert): Delete.
33461 (edited_file::apply_replace): Rename to...
33462 (edited_file::apply_fixit): ...this.
33463 (edited_line::~edited_line): Drop deletion of events.
33464 (edited_line::apply_insert): Delete.
33465 (edited_line::apply_replace): Rename to...
33466 (edited_line::apply_fixit): ...this. Convert to half-open range.
33467 Update for change to type of m_line_events.
33468 * edit-context.h (edit_context::apply_insert): Delete.
33469 (edit_context::apply_replace): Rename to...
33470 (edit_context::apply_fixit): ...this.
33472 2017-05-01 Martin Sebor <msebor@redhat.com>
33474 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33477 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33480 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33481 __get_cpuid_max returns 0.
33482 (__get_cpuid_count): Ditto.
33484 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33486 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33487 replacement expression is another instance of one of its arguments.
33489 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33492 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33493 check for stack push/pop autoinc.
33494 * config/i386/i386.c (ix86_agi_dependent): Return false
33495 if the only reason why modified_in_p returned true is that
33496 addr is SP based and set_insn is a push or pop.
33498 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33500 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33503 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33506 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33507 (account_size_time): Use two predicates - exec_pred and
33509 (evaluate_conditions_for_known_args): Compute both clause and
33511 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33512 (inline_summary_t::duplicate): Update.
33513 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33515 (compute_inline_parameters): Likewise.
33516 (estimate_edge_size_and_time): Update caluclation of time.
33517 (estimate_node_size_and_time): Compute both time and nonspecialized
33519 (estimate_ipcp_clone_size_and_time): Update.
33520 (inline_merge_summary): Update.
33521 (do_estimate_edge_time): Update.
33522 (do_estimate_edge_size): Update.
33523 (do_estimate_edge_hints): Update.
33524 (inline_read_section, inline_write_summary): Stream both new predicates.
33525 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33527 (compute_inlined_call_time): Cleanup.
33528 (big_speedup_p): Update.
33529 (edge_badness): Update.
33530 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33531 (size_time_entry): Replace predicate by exec_predicate and
33532 nonconst_predicate.
33533 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33534 (estimate_edge_time): Return also nonspec_time.
33535 (reset_edge_growth_cache): Update.
33537 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33539 PR rtl-optimization/80491
33540 * ifcvt.c (noce_process_if_block): When looking for x setter
33541 with missing else_bb, don't check only the insn right before
33542 cond_earliest, but look for the last insn that x is modified in
33543 within the same bb.
33545 PR rtl-optimization/80491
33546 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33548 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33550 PR tree-optimization/80487
33551 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33553 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33555 PR tree-optimization/79697
33556 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33557 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33558 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33560 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33561 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33563 2017-04-28 Martin Sebor <msebor@redhat.com>
33565 PR tree-optimization/80523
33566 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33567 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33569 (maybe_warn, format_directive, parse_directive): Use new functions.
33570 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33572 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33574 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33576 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33578 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33579 target_header_dir): Set correctly.
33580 * configure: Regenerated.
33581 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33582 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33583 instead of SYSTEM_HEADER_DIR.
33585 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33587 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33588 (estimate_local_effects): Likewise.
33589 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33590 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33591 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33592 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33593 do_estimate_edge_time, estimate_edge_time): Likewise.
33594 * ipa-inline-analysis.c (estimate_node_size_and_time,
33595 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33596 (estimate_time_after_inlining): Remove.
33598 2017-04-28 Martin Liska <mliska@suse.cz>
33600 * doc/gcov.texi: Enhance documentation of gcov.
33602 2017-04-28 Martin Liska <mliska@suse.cz>
33604 * doc/gcov.texi: Sort options in alphabetic order.
33605 * doc/gcov-dump.texi: Likewise.
33606 * doc/gcov-tool.texi: Likewise.
33607 * gcov.c (print_usage): Likewise.
33608 * gcov-dump.c (print_usage): Likewise.
33609 * gcov-tool.c (print_merge_usage_message): Likewise.
33610 (print_rewrite_usage_message): Likewise.
33611 (print_overlap_usage_message): Likewise.
33613 2017-04-28 Martin Liska <mliska@suse.cz>
33615 PR gcov-profile/53915
33616 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33618 2017-04-28 Martin Liska <mliska@suse.cz>
33620 PR gcov-profile/79891
33621 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33622 is marked by compiler as living on a line.
33623 (get_cycles_count): Remove usage of the union.
33624 (output_intermediate_file): Likewise.
33625 (find_source): Fix GNU coding style.
33626 (accumulate_line_counts): Remove old non-all block mode.
33627 (output_lines): Remove usage of the union.
33628 * profile.c (output_location): Include all BBs, even if
33629 belonging to a same line (and file) as a previous BB.
33631 2017-04-28 Martin Liska <mliska@suse.cz>
33633 * gcov.c (process_args): Handle new argument 'w'.
33634 (read_graph_file): Assign ID to BBs.
33635 (output_branch_count): Display BB # if verbose flag is set.
33636 (output_lines): Likewise for arcs.
33637 (print_usage): Add '--verbose' option help.
33638 * doc/gcov.texi: Document --verbose (-w) option.
33640 2017-04-28 Martin Liska <mliska@suse.cz>
33642 * gcov.c (struct block_location_info): New struct.
33643 (process_file): Fill up the new structure.
33644 (read_graph_file): Replace usage of encoding by the newly added
33646 (add_line_counts): Likewise.
33647 (accumulate_line_counts): Remove usage of the union.
33648 (function_info::function_info): New function.
33649 (function_info::~function_info): Likewise.
33650 (process_file): Call delete instead of release_function.
33651 (release_function): Release the function.
33652 (release_structures): Call delete instead of release_function.
33653 (solve_flow_graph): Replace usage of num_blocks.
33654 (find_exception_blocks): Likewise.
33655 (output_lines): Fix GNU coding style.
33657 2017-04-28 Martin Liska <mliska@suse.cz>
33660 * coverage.c (coverage_remove_note_file): New function.
33661 * coverage.h: Declare the function.
33662 * toplev.c (finalize): Clean if an error has been seen.
33664 2017-04-28 Martin Liska <mliska@suse.cz>
33666 PR gcov-profile/80031
33667 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33668 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33669 * gcov.c (read_graph_file): Read just number of blocks.
33670 * profile.c (branch_prob): Do not stream 0 flags per a basic
33673 2017-04-28 Martin Liska <mliska@suse.cz>
33675 * gcov-dump.c (tag_*): Add new argument to declarations.
33676 (dump_gcov_file): Likewise.
33677 (tag_blocks): Add and use new argument depth.
33678 (tag_arcs): Likewise.
33679 (tag_lines): Likewise.
33680 (tag_counters): Likewise.
33681 (tag_summary): Likewise.
33682 (dump_working_sets): Use depth to do a proper indentation.
33684 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33687 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33688 * symtab.c (symtab_node::debug_symtab): Move definition here.
33690 2017-04-28 Richard Biener <rguenther@suse.de>
33692 * lto-streamer.h (LTO_major_version): Bump to 7.
33694 2017-04-28 Richard Biener <rguenther@suse.de>
33696 * tree-vrp.c (assert_info): New struct.
33697 (add_assert_info): New helper.
33698 (register_edge_assert_for_2): Refactor to add asserts to a vector
33700 (register_edge_assert_for_1): Likewise.
33701 (register_edge_assert_for): Likewise.
33702 (finish_register_edge_assert_for): New helper actually registering
33703 asserts where live on edge.
33704 (find_conditional_asserts): Adjust.
33705 (find_switch_asserts): Likewise.
33706 (evrp_dom_walker::try_find_new_range): Generalize.
33707 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33709 2017-04-27 Marek Polacek <polacek@redhat.com>
33712 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33713 arg10 and arg11 to itype.
33715 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33717 * doc/extend.texi (Object Size Checking): Improve grammar.
33719 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33722 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33723 that the logic for permitting reciprocal estimates matches that
33726 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33729 * tree.c (type_cache_hasher::equal): Only compare
33730 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33731 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33732 non-aggregate element types.
33733 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33734 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33736 2017-04-27 Richard Biener <rguenther@suse.de>
33738 PR middle-end/80533
33739 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33740 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33741 keeping a reference to a trailing array.
33743 2017-04-27 Richard Biener <rguenther@suse.de>
33745 PR middle-end/80539
33746 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33747 being in loop-closed SSA form conservatively.
33748 (chrec_fold_multiply_poly_poly): Likewise.
33750 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33752 PR middle-end/79665
33753 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33754 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33756 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33759 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33760 (aarch64_function_arg_alignment): Return unsigned int again, but still
33761 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33762 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33763 Don't emit -Wpsabi note.
33764 (aarch64_function_arg_boundary): Likewise.
33765 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33768 2017-04-26 Nathan Sidwell <nathan@acm.org>
33770 * tree.h (crc32_unsigned_n): Declare.
33771 (crc32_unsigned, crc32_unsigned): Make inline.
33772 * tree.c (crc32_unsigned_bits): Replace with ...
33773 (crc32_unsigned_n): ... this.
33774 (crc32_unsigned, crc32_byte): Remove.
33775 (crc32_string): Remove unnecessary braces.
33777 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33779 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33780 * ipa-inline-analysis.c (MAX_TIME): Remove.
33781 (account_size_time): Use sreal for time.
33782 (dump_inline_summary): Update.
33783 (estimate_function_body_sizes): Update.
33784 (estimate_edge_size_and_time): Update.
33785 (estimate_calls_size_and_time): Update.
33786 (estimate_node_size_and_time): Update.
33787 (inline_merge_summary): Update.
33788 (inline_update_overall_summary): Update.
33789 (estimate_time_after_inlining): Update.
33790 (inline_read_section): Update.
33791 (inline_write_summary): Update.
33792 * ipa-inline.c (compute_uninlined_call_time): Update.
33793 (compute_inlined_call_time): Update.
33794 (recursive_inlining): Update.
33795 (inline_small_functions): Update.
33796 (dump_overall_stats): Update.
33797 * ipa-inline.h: Include sreal.h.
33798 (size_time_entry): Turn time to sreal.
33799 (inline_summary): Turn self_time nad time to sreal.
33801 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33803 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33805 (sreal::stream_out, sreal::stream_in): New.
33806 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33808 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33810 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33813 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33816 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33818 (dimode_scalar_chain::compute_convert_gain): Ditto.
33819 (dimode_scalar_chain::make_vector_copies): Ditto.
33820 (dimode_scalar_chain::convert_reg): Ditto.
33821 (dimode_scalar_chain::convert_insn): Ditto.
33822 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33823 (VI248_AVX512BW_1): New mode iterator.
33824 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33825 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33828 2017-04-25 Martin Sebor <msebor@redhat.com>
33830 PR tree-optimization/80497
33831 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33832 constants are representable in HOST_WIDE_INT.
33833 (parse_directive): Ditto.
33835 2017-04-25 Martin Sebor <msebor@redhat.com>
33838 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33839 (new_zero_array): Adjust signature.
33840 (dom_info::dom_init): Used unsigned rather that size_t.
33841 (dom_info::dom_info): Same.
33843 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33844 Jakub Jelinek <jakub@redhat.com>
33847 * config/arm/arm.c: Include gimple.h.
33848 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33849 returns negative, increment ncrn only if it returned positive.
33850 (arm_needs_doubleword_align): Return int instead of bool,
33851 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33852 members, but if there is any such non-FIELD_DECL
33853 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33854 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33855 returns negative, increment nregs only if it returned positive.
33856 (arm_setup_incoming_varargs): Likewise.
33857 (arm_function_arg_boundary): Emit -Wpsabi note if
33858 arm_needs_doubleword_align returns negative, return
33859 DOUBLEWORD_ALIGNMENT only if it returned positive.
33861 2017-04-25 Marek Polacek <polacek@redhat.com>
33864 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33865 first argument to type.
33867 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33870 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33871 type checks to test for compatibility instead of equality.
33873 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33874 Jakub Jelinek <jakub@redhat.com>
33877 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33879 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33880 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33881 the alignment computation, but return their maximum in warn_alignment.
33882 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33883 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33885 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33886 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33889 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33891 * config/arc/simdext.md (dmpyh): Fix typo.
33893 2017-04-25 Richard Biener <rguenther@suse.de>
33895 PR tree-optimization/80492
33896 * alias.c (compare_base_decls): Handle registers with asm
33897 specification conservatively.
33898 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33899 compare_base_decls returning dont-know properly.
33901 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33903 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33904 (legitimate_offset_address_p): New function.
33905 (arc_legitimate_address_p): Use above function.
33907 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33909 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33911 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33913 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33914 ACCH registers whenever they are available.
33916 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33918 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33919 double regs fix when not used.
33921 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33923 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33925 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33926 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33928 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33930 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33931 when emitting short ADD instructions.
33933 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33935 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33937 (cmpsi_cc_c_insn): Likewise.
33938 (cbranchsi4_scratch): Compute proper instruction length using
33939 compact_hreg_operand.
33940 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33942 2017-04-25 Richard Biener <rguenther@suse.de>
33944 PR middle-end/80509
33945 * passes.c (pass_manager::pass_manager): Initialize
33946 m_name_to_pass_map.
33948 2017-04-25 Richard Biener <rguenther@suse.de>
33950 PR tree-optimization/79201
33951 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33953 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33956 * config/s390/vector.md: Split MEM->GPR vector moves for
33957 non-s_operand addresses.
33959 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33962 * config/s390/predicates.md (reload_const_wide_int_operand): New
33964 * config/s390/s390.md ("movti"): Remove d/P alternative.
33965 ("movti_bigconst"): New pattern definition.
33967 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33970 * s390-protos.h (s390_expand_cs_hqi): Removed.
33971 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33972 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33973 modes as well as CCZ1mode and CCZmode.
33974 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33975 signature of s390_emit_compare_and_swap.
33976 (s390_expand_cs_hqi): Likewise, make static.
33977 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33978 compare-and-swap, in some cases.
33979 (s390_expand_cs): Wrapper function.
33980 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33982 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33983 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33984 patterns for small and large integers. Forbid symref memory operands.
33985 Move expander to s390.c. Require cc register.
33986 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33987 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33988 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33989 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33990 symref memory operands. Remove CC mode and call s390_match_ccmode
33992 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33994 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33996 * config/s390/s390.md (define_peephole2): New peephole to help
33997 combining the load-and-test pattern with volatile memory.
33999 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34001 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
34002 with CCZmode for TARGET_Z196.
34004 2017-04-25 Jakub Jelinek <jakub@redhat.com>
34006 PR rtl-optimization/80501
34007 * combine.c (make_compound_operation_int): Set subreg_code to SET
34008 even for AND with mask of the sign bit of mode.
34010 PR rtl-optimization/80500
34011 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
34012 sum's initial value.
34014 2017-04-25 Julian Brown <julian@codesourcery.com>
34015 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34017 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
34019 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
34021 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
34023 2017-04-25 Julian Brown <julian@codesourcery.com>
34024 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34026 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
34027 (thunderx2t99_sha): New Reservation.
34029 2017-04-25 Julian Brown <julian@codesourcery.com>
34030 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34032 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
34033 type for 1-element load.
34035 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
34037 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
34039 2017-04-24 Martin Jambor <mjambor@suse.cz>
34041 PR tree-optimization/80293
34042 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
34043 char arrays not totally scalarizable if it is false.
34044 (analyze_all_variable_accesses): Pass correct value in the new
34045 parameter. Add a statistics counter.
34047 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
34049 PR middle-end/79931
34050 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
34052 2017-04-24 Richard Biener <rguenther@suse.de>
34054 PR tree-optimization/80494
34055 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
34056 out for complex types.
34058 2017-04-24 Richard Biener <rguenther@suse.de>
34060 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34061 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34062 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34063 (DFS): Adjust and never fail.
34064 (sccvn_dom_walker::fail): Remove.
34065 (sccvn_dom_walker::before_dom_children): Adjust.
34066 (run_scc_vn): Likewise and never fail.
34067 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34068 (pass_fre::execute): Likewise.
34070 2017-04-24 Richard Biener <rguenther@suse.de>
34072 PR tree-optimization/79725
34073 * tree-ssa-sink.c (statement_sink_location): Return whether
34074 failure reason was zero uses. Move that check later.
34075 (sink_code_in_bb): Deal with zero uses by removing the stmt
34078 2017-04-24 Richard Biener <rguenther@suse.de>
34081 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34082 pointer-based references.
34084 2017-04-24 Richard Biener <rguenther@suse.de>
34087 * pass_manager.h (pass_manager::operator new): Remove.
34088 (pass_manager::operator delete): Likewise.
34089 * passes.c (pass_manager::operator new): Remove.
34090 (pass_manager::operator delete): Likewise.
34091 (pass_manager::pass_manager): Zero individual pass members.
34093 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34096 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34097 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34098 Check "XEXP (src, 1)" operand here.
34099 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34100 Check "XEXP (src, 1)" operand here.
34101 (dimode_scalar_chain::make_vector_copies): Detect count register
34102 of a shift instruction. Zero extend count register from QImode
34103 to DImode to satisfy vector shift pattern count operand predicate.
34104 Substitute vector shift count operand with a DImode copy.
34105 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34108 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34110 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34111 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34112 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34113 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34114 (UNSPEC_NOREX_MEM): Remove definition.
34116 2017-04-21 Richard Biener <rguenther@suse.de>
34118 PR tree-optimization/79547
34119 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34120 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34121 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34122 without any constraints.
34124 2017-04-21 Richard Biener <rguenther@suse.de>
34126 PR tree-optimization/78847
34127 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34129 2017-04-21 Richard Biener <rguenther@suse.de>
34131 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34132 (build_distinct_type_copy): Likewise.
34133 (build_variant_type_copy): Likewise.
34134 * tree.c (build_qualified_type): Pass down mem-stat info.
34135 (build_distinct_type_copy): Likewise.
34136 (build_variant_type_copy): Likewise.
34138 2017-04-21 Richard Biener <rguenther@suse.de>
34140 PR tree-optimization/80237
34141 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34143 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34144 for a simplified result.
34146 2016-04-21 Richard Biener <rguenther@suse.de>
34148 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34149 sth as strict as a simple_iv but a chrec without symbols and an
34150 operand defined in the loop we are peeling (and not some subloop).
34151 (propagate_constants_for_unrolling): Propagate all constants.
34153 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34156 * config/i386/i386.c (print_reg): Remove assert for disalowed
34157 regno values, call output_operand_lossage instead.
34159 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34162 * config/i386/constraints.md (Yc): New register constraint.
34163 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34164 Use Yc constraint for alternative 2 of operand 0. Remove
34165 preferred_for_speed attribute.
34167 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34169 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34170 lastprivate clauses in SIMT case.
34172 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34174 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34176 2017-04-20 Richard Biener <rguenther@suse.de>
34178 PR tree-optimization/57796
34179 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34180 as N scalar stores.
34181 (vect_model_load_cost): Cost gathers as N scalar loads.
34183 2017-04-20 Richard Biener <rguenther@suse.de>
34185 * ggc-page.c (ggc_allocated_p): Rename to ...
34186 (safe_lookup_page_table_entry): ... this and return the lookup
34188 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34190 2017-04-20 Richard Biener <rguenther@suse.de>
34192 PR tree-optimization/80453
34193 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34194 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34195 from the conditions.
34196 (vn_phi_eq): Pass them down.
34197 (vn_phi_lookup): Record them.
34198 (vn_phi_insert): Likewise.
34200 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34202 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34203 uninitialized variable warning to avoid buffer overrun.
34205 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34208 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34209 is suppressed for '{ 0 }' in C.
34211 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34213 * BASE-VER: Set to 8.0.0.
34215 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34217 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34218 priority .init_array and .fini_array section with SECTION_NOTYPE
34221 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34223 PR middle-end/80423
34224 * tree.h (build_array_type): Add typeless_storage default argument.
34225 * tree.c (type_cache_hasher::equal): Also compare
34226 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34227 (build_array_type): Add typeless_storage argument, set
34228 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34230 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34231 (build_array_type): Likewise. Add typeless_storage argument.
34233 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34234 Jakub Jelinek <jakub@redhat.com>
34236 PR tree-optimization/80426
34237 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34238 operation on symbolic operands, also compute the overflow for the
34239 invariant part when the operation degenerates into a negation.
34241 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34244 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34245 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34248 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34250 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34253 * config/avr/avr.c (tree.h): Include it.
34254 (cgraph.h): Include it.
34255 (avr_encode_section_info): Don't warn for uninitialized progmem
34256 variable if it's just an alias.
34258 2017-04-19 Richard Biener <rguenther@suse.de>
34261 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34262 when needed by AutoPGO.
34264 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34267 * doc/lto.texi: Remove an extra 'that'.
34269 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34271 PR rtl-optimization/80429
34272 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34273 are only used in debug insns.
34275 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34276 Vladimir Makarov <vmakarov@redhat.com>
34278 * config/sparc/predicates.md (input_operand): Add comment. Return
34279 true for any memory operand when LRA is in progress.
34280 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34282 2017-04-18 Jeff Law <law@redhat.com>
34285 * mips.md ({return,simple_return}_internal): Do not overwrite
34288 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34290 PR tree-optimization/80443
34291 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34292 instead of adding 1, subtract -1 and similarly instead of subtracting
34295 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34297 PR rtl-optimization/80357
34298 * haifa-sched.c (tmp_bitmap): New variable.
34299 (model_recompute): Handle duplicate use records.
34300 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34301 (free_global_sched_pressure_data): Free it.
34303 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34306 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34307 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34308 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34309 instead of SYSTEM_HEADER_DIR.
34311 2017-04-18 Jeff Law <law@redhat.com>
34313 PR middle-end/80422
34314 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34315 predecessors after walking up the insn chain.
34317 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34320 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34321 sizetype type into debug info.
34323 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34326 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34327 unneeded test for TARGET_UPPER_REGS_SF.
34328 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34330 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34333 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34334 instead of gsi_after_labels.
34336 2017-04-18 Jeff Law <law@redhat.com>
34338 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34342 2017-04-13 Jeff Law <law@redhat.com>
34343 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34344 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34346 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34349 * config/avr/avr.c (intl.h): Include it.
34350 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34352 2017-04-18 Martin Liska <mliska@suse.cz>
34354 PR gcov-profile/78783
34355 * gcov-tool.c (gcov_output_files): Validate that destination
34356 file is either removed by the tool or by a user.
34358 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34359 Guy Benyei <guybe@mellanox.com>
34361 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34362 block, and do not negate it, the stored id is already negative.
34364 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34366 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34368 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34371 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34372 masks of options that should be turned off if the VSX vector
34373 options are turned off.
34374 (OTHER_P8_VECTOR_MASKS): Likewise.
34375 (OTHER_VSX_VECTOR_MASKS): Likewise.
34376 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34377 rs6000_disable_incompatible_switches to validate no type switches
34379 (rs6000_incompatible_switch): New function to disallow turning on
34380 other vector options if -mno-vsx, -mno-power8-vector, or
34381 -mno-power9-vector are specified.
34383 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34385 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34387 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34389 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34390 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34391 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34392 (ARG_POINTER_CFA_OFFSET): Likewise.
34394 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34396 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34397 conditions to take advantage of various optimizations.
34399 2017-04-13 Jeff Law <law@redhat.com>
34401 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34402 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34403 (zero_extendsidi2_dext): Likewise.
34405 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34408 * fold-const.c (fold_ternary_loc): Revert
34409 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34412 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34414 PR rtl-optimization/80343
34415 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34416 new scratch pseudo.
34418 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34421 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34422 to ubsan_encode_value.
34424 2017-04-13 Jeff Law <law@redhat.com>
34426 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34427 appearing in DEBUG_INSNs.
34429 2017-04-13 Martin Liska <mliska@suse.cz>
34431 PR gcov-profile/80413
34432 * gcov-io.c (gcov_write_string): Copy to buffer just when
34433 allocated size is greater than zero.
34435 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34438 * dwarf2out.c (decls_for_scope): Ignore declarations of
34439 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34441 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34444 * ipa-visibility.c (non_local_p): Fix typos.
34445 (localize_node): When localizing symbol in same comdat group,
34446 dissolve the group only when we know external symbols are going
34448 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34450 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34452 PR tree-optimization/79390
34453 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34454 order does not result in usable sequence, retry with reversed operand
34460 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34461 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34462 op0 instead of fold_convert_loc (loc, type, arg0).
34464 2017-04-12 Jeff Law <law@redhat.com>
34466 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34467 has a delay slot in the generated code.
34469 * config/cris/cris.md (cris_preferred_reload_class): Return
34470 GENNONACR_REGS rather than GENERAL_REGS.
34472 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34475 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34476 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34477 signedness of the result type.
34479 2017-04-12 Richard Biener <rguenther@suse.de>
34480 Jeff Law <law@redhat.com>
34482 PR tree-optimization/80359
34483 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34484 trim stores to TARGET_MEM_REFs.
34486 2017-04-12 Richard Biener <rguenther@suse.de>
34488 PR tree-optimization/79390
34489 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34490 threading case even more.
34492 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34495 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34496 for quad_address_p for TImode, instead of just not indexed_address.
34498 2017-04-12 Richard Biener <rguenther@suse.de>
34499 Bernd Edlinger <bernd.edlinger@hotmail.de>
34501 PR middle-end/79671
34502 * alias.c (component_uses_parent_alias_set_from): Handle
34503 TYPE_TYPELESS_STORAGE.
34504 (get_alias_set): Likewise.
34505 * tree-core.h (tree_type_common): Add typeless_storage flag.
34506 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34507 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34508 for types containing members with TYPE_TYPELESS_STORAGE.
34509 (place_field): Likewise.
34510 (layout_type): Likewise for ARRAY_TYPE.
34511 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34512 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34513 TYPE_TYPELESS_STORAGE.
34514 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34516 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34519 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34520 first argument to type.
34522 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34526 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34527 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34528 (rs6000_expand_binop_builtin): Likewise.
34529 (rs6000_expand_ternop_builtin): Likewise; also add missing
34530 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34531 vshasigma built-ins.
34532 * doc/extend.texi: Document that vec_xxpermdi's third argument
34533 must be a constant.
34535 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34537 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34538 Use shift_const cost parameter when calculating gain of STV shifts.
34540 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34542 PR rtl-optimization/70478
34543 * lra-constraints.c (process_alt_operands): Check memory for
34544 disfavoring memory insn operand.
34546 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34548 PR middle-end/80100
34549 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34550 left shift in unsigned HOST_WIDE_INT type.
34552 PR rtl-optimization/80385
34553 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34554 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34557 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34558 if they have any depend clauses.
34560 2017-04-11 Martin Liska <mliska@suse.cz>
34563 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34564 * ipa-split.c (split_function): Create a local comdat symbol
34565 if caller is in a comdat group.
34567 2017-04-11 Martin Liska <mliska@suse.cz>
34570 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34573 2017-04-11 Martin Sebor <msebor@redhat.com>
34575 PR middle-end/80364
34576 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34577 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34579 (directive::set_width, directive::set_precision, format_character):
34581 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34584 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34587 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34588 conflict, set target->arch_name instead of target->cpu_name.
34590 2017-04-11 Richard Biener <rguenther@suse.de>
34592 PR tree-optimization/80374
34593 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34594 build_zero_cst, remove fold_convertible_p check again.
34596 2017-04-11 Martin Liska <mliska@suse.cz>
34599 * ubsan.c (instrument_object_size): Do not instrument register
34602 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34605 * config/i386/i386-builtin-types.def
34606 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34607 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34608 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34609 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34610 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34611 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34612 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34613 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34614 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34615 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34616 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34617 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34618 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34619 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34620 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34621 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34622 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34623 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34624 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34625 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34626 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34627 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34628 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34629 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34630 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34631 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34632 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34633 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34634 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34635 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34636 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34637 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34638 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34639 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34640 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34641 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34642 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34643 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34644 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34645 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34646 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34647 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34648 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34649 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34650 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34651 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34653 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34654 flag to second_arg_count, handle 4 argument function type _COUNT
34655 aliases, handle second_arg_count on second argument rather than last.
34657 2017-04-10 Jeff Law <law@redhat.com>
34659 PR tree-optimization/80374
34660 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34661 record anything if we can not convert integer_zero_node to the
34664 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34667 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34668 Enhance special handling given to the TARGET_P9_MINMAX option in
34669 relation to certain other options.
34671 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34673 PR tree-optimization/80153
34674 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34675 remove POINTER_PLUS_EXPR's base part directly, rather than through
34678 2017-04-10 Richard Biener <rguenther@suse.de>
34679 Bin Cheng <bin.cheng@arm.com>
34681 PR tree-optimization/80153
34682 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34683 the first element of pointer type aff_tree. Build result expr in
34685 (add_elt_to_tree): Convert to type unconditionally. Remove other
34686 fold_convert calls.
34687 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34688 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34690 2017-04-10 Richard Biener <rguenther@suse.de>
34692 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34695 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34697 PR rtl-optimization/70478
34698 * lra-constraints.c (curr_small_class_check): New.
34699 (update_and_check_small_class_inputs): New.
34700 (process_alt_operands): Update curr_small_class_check. Disfavor
34701 alternative insn memory operands. Check available regs for small
34704 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34707 * config/mips/mips.opt (-mvirt): Update description.
34708 * doc/invoke.texi (-mvirt): Likewise.
34710 2017-04-10 Richard Biener <rguenther@suse.de>
34712 PR middle-end/80362
34713 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34714 looking for NEGATE_EXPR in -A / -B to A / B folding.
34716 2017-04-10 Martin Liska <mliska@suse.cz>
34718 PR gcov-profile/80224
34719 * gcov.c (print_usage): Fix usage string.
34720 (get_gcov_intermediate_filename): Remove.
34721 (output_gcov_file): Use both for normal and intermediate format.
34722 (generate_results): Do not initialize special file for
34723 intermediate format.
34725 2017-04-10 Richard Biener <rguenther@suse.de>
34727 PR tree-optimization/80304
34728 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34731 2017-04-10 Nathan Sidwell <nathan@acm.org>
34734 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34735 (rs6000_init_builtins): Use it.
34737 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34739 * config/arm/arm.md (<mrc>): Add mode to SET source.
34740 (<mrrc>): Likewise.
34742 2017-04-10 Richard Biener <rguenther@suse.de>
34744 PR middle-end/80344
34745 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34747 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34750 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34751 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34752 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34753 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34754 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34755 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34756 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34757 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34758 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34759 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34760 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34761 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34762 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34763 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34764 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34765 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34766 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34767 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34768 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34769 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34770 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34771 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34772 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34774 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34776 PR rtl-optimization/70478
34777 * lra-constraints.c: Reverse the last patch.
34779 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34781 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34782 Add comment for WCHAR_T.
34784 2017-04-08 Martin Liska <mliska@suse.cz>
34787 2017-04-07 Martin Liska <mliska@suse.cz>
34790 * ipa-split.c (split_function): Add function part to a same comdat
34793 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34796 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34798 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34800 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34801 (vec_pairing): ...to this.
34802 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34803 (rs6000_sched_init): Adjust for name change.
34804 (struct rs6000_sched_context): Likewise.
34805 (rs6000_init_sched_context): Likewise.
34806 (rs6000_set_sched_context): Likewise.
34808 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34814 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34816 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34817 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34818 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34820 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34822 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34824 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34826 PR rtl-optimization/70703
34827 * ira-color.c (update_conflict_hard_regno_costs): Use
34828 int64_t instead of HOST_WIDE_INT.
34830 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34832 PR rtl-optimization/70478
34833 * lra-constraints.c (process_alt_operands): Disfavor alternative
34834 insn memory operands.
34836 2017-04-07 Jeff Law <law@redhat.com>
34838 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34839 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34841 2017-04-07 Martin Liska <mliska@suse.cz>
34844 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34845 Show error message instead of an ICE.
34847 2017-04-07 Martin Liska <mliska@suse.cz>
34850 * ipa-split.c (split_function): Add function part to a same comdat
34853 2017-04-07 Richard Biener <rguenther@suse.de>
34855 PR middle-end/80341
34856 * tree.c (get_unwidened): Also handle ! for_type case for
34858 * convert.c (do_narrow): Split out from ...
34859 (convert_to_integer_1): ... here. Do not pass final truncation
34860 type to get_unwidened for TRUNC_DIV_EXPR.
34862 2017-04-07 Richard Biener <rguenther@suse.de>
34864 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34866 (aff_combination_const): Adjust.
34867 (aff_combination_scale): Likewise.
34868 (aff_combination_add_elt): Likewise.
34869 (aff_combination_add_cst): Likewise.
34870 (aff_combination_convert): Likewise.
34871 (add_elt_to_tree): Likewise. Remove unused argument.
34872 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34874 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34876 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34878 * config/arm/arm.c (arm_default_short_enums): Use
34879 ARM_DEFAULT_SHORT_ENUMS.
34880 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34882 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34885 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34886 members with redundant out-of-class redeclaration.
34888 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34891 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34892 * config/i386/i386.md (*zero_extendsidi2):
34893 Add (?*x,*x) and (?*v,*v) alternatives.
34895 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34898 * config/i386/i386.c (ix86_expand_builtin)
34899 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34900 mode from insn data. Convert operands to insn operand mode.
34901 Copy operands that don't satisfy insn predicate to a register.
34903 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34905 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34908 2017-04-06 Richard Biener <rguenther@suse.de>
34910 PR tree-optimization/80334
34911 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34912 preserve alignment of accesses.
34914 2017-04-06 Richard Biener <rguenther@suse.de>
34916 PR tree-optimization/80262
34917 * tree-sra.c (build_ref_for_offset): Preserve address-space
34919 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34920 Drop useless address-space information on MEM_REF offsets.
34922 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34924 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34926 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34928 PR rtl-optimization/70703
34929 * ira-color.c (update_conflict_hard_regno_costs): Use
34930 HOST_WIDE_INT instead of long.
34932 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34935 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34936 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34938 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34939 for x86_64 target. Handle -m3dnowa option.
34941 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34943 PR rtl-optimization/70703
34944 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34945 (update_conflict_hard_regno_costs): Use long instead of unsigned
34946 arithmetic for cost calculation.
34948 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34949 Bernd Edlinger <bernd.edlinger@hotmail.de>
34952 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34955 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34958 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34959 ptr_mode with Pmode throughout.
34960 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34961 into probe_stack_range and use DImode.
34963 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34966 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34967 call_eh_return is true.
34969 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34971 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34972 Initialize last_match_fntype_index.
34974 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34977 * tree-nvr.c: Include internal-fn.h.
34978 (pass_return_slot::execute): Ignore internal calls without
34981 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34982 Richard Biener <rguenther@suse.de>
34985 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34986 captures used multiple times, except for the last use.
34987 * generic-match-head.c: Include gimplify.h.
34989 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34991 PR tree-optimization/79390
34992 * target.h (struct noce_if_info): Declare.
34993 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34994 * target.def (noce_conversion_profitable_p): New target hook.
34995 * ifcvt.h (struct noce_if_info): New type, moved from ...
34996 * ifcvt.c (struct noce_if_info): ... here.
34997 (noce_conversion_profitable_p): Renamed to ...
34998 (default_noce_conversion_profitable_p): ... this. No longer
35000 (noce_try_store_flag_constants, noce_try_addcc,
35001 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
35002 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
35003 instead of noce_conversion_profitable_p.
35004 * config/i386/i386.c: Include ifcvt.h.
35005 (ix86_option_override_internal): Don't override
35006 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
35007 (ix86_noce_conversion_profitable_p): New function.
35008 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
35009 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
35010 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
35011 * doc/tm.texi: Regenerated.
35013 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35015 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
35018 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
35021 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
35022 instructions for small multiply cores.
35024 2017-04-04 Jeff Law <law@redhat.com>
35026 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
35028 (mips_expand_vec_perm_const): Initialize elements in orig_perm
35029 that are not set by the loop over the elements.
35031 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35034 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
35035 int mode, convert_modes it to mode as unsigned, otherwise use
35036 lowpart_subreg to mode rather than SImode.
35037 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
35038 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
35039 Use DImode instead of SImode for the shift count operand.
35040 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
35043 2017-04-04 Richard Biener <rguenther@suse.de>
35045 PR middle-end/80281
35046 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
35047 arithmetic done for the negate or the plus. Simplify.
35048 (A - (-B) -> A + B): Likewise.
35049 * fold-const.c (split_tree): Make sure to not negate pointers.
35051 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
35053 PR rtl-optimization/60818
35054 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
35055 a compare of comparisons with the thing compared if this results
35056 in a different machine mode.
35058 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35060 * alias.c (base_alias_check): Fix typo in comment.
35061 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35062 * cgraphunit.c (symbol_table::compile): Likewise.
35063 * collect2.c (maybe_run_lto_and_relink): Likewise.
35064 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35065 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35066 * config/avr/avr.c (avr_map_op_t): Likewise.
35067 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35068 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35069 * config/epiphany/epiphany.md (movcc): Likewise.
35070 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35071 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35073 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35074 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35075 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35076 * config/sh/sh.c (sh_rtx_costs): Likewise.
35077 * fold-const.c (fold_truth_andor): Likewise.
35078 * genautomata.c (collapse_flag): Likewise.
35079 * gengtype.h (struct type::u::s): Likewise.
35080 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35081 * input.c (FORMAT_AMOUNT): Likewise.
35082 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35083 (known_aggs_to_agg_replacement_list): Likewise.
35084 * ipa-inline-analysis.c: Likewise.
35085 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35086 * ipa-polymorphic-call.c
35087 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35088 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35089 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35091 * modulo-sched.c (apply_reg_moves): Likewise.
35092 * omp-expand.c (build_omp_regions_1): Likewise.
35093 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35094 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35095 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35096 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35097 * value-prof.c: Likewise.
35098 * var-tracking.c (val_reset): Likewise.
35100 2017-04-03 Richard Biener <rguenther@suse.de>
35102 PR tree-optimization/80275
35103 * fold-const.c (split_address_to_core_and_offset): Handle
35106 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35108 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35109 descriptors is at least equal to that of functions.
35111 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35113 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35115 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35118 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35119 (mov<IMOD4:mode>): New expander.
35120 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35122 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35124 PR rtl-optimization/79405
35125 * fwprop.c (propagations_left): New variable.
35126 (forward_propagate_into): Decrement it.
35127 (fwprop_init): Initialize it.
35128 (fw_prop): If the variable has reached zero, stop propagating.
35129 (fwprop_addr): Ditto.
35131 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35134 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35135 a FUNCTION_DECL, pass it as decl instead of origin to
35138 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35140 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35143 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35146 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35147 TARGET_VSX_SMALL_INTEGER.
35149 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35151 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35152 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35154 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35156 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35157 extraction from odd-numbered MSA register.
35159 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35161 PR middle-end/80173
35162 * expmed.c (store_bit_field_1): Don't attempt to create
35163 a word subreg out of hard registers wider than word if they
35164 have HARD_REGNO_NREGS of 1 for their mode.
35166 PR middle-end/80163
35167 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35168 conversions to integer types wider than word and pointer.
35171 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35172 (rtx_equal_for_cselib_p): Pass 0 to it.
35173 * cselib.c (cselib_hasher::equal): Likewise.
35174 (rtx_equal_for_cselib_1): Add depth argument. If depth
35175 is 128, don't look up VALUE locs and punt. Increment
35176 depth in recursive calls when walking VALUE locs.
35178 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35180 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35181 (make_gcov_file_name): Use the canonical path name for generating
35183 (read_line): Fix handling of files with ascii null bytes.
35185 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35187 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35188 to initialise a vector register instead
35189 of using a const_int.
35191 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35193 PR translation/80189
35194 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35195 diagnostic messages.
35197 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35200 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35201 (dfp_diex_<mode>): Update mode of operand 1.
35202 * doc/extend.texi (dxex, dxexq): Document change to return type.
35203 (diex, diexq): Document change to argument type.
35205 2017-03-30 Martin Jambor <mjambor@suse.cz>
35208 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35209 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35210 it reflects the signature changes performed at the callee side.
35211 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35212 to cgraph_build_function_type_skip_args.
35213 (build_function_decl_skip_args): Adjust call to the above function.
35215 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35218 * config/i386/sse.md
35219 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35220 register as dest whenever it is a MEM not rtx_equal_p to the
35221 corresponding dup operand, and when forcing into reg move the
35222 reg into the memory afterwards.
35223 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35224 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35225 for the force_reg mode.
35226 (avx512vl_vextractf128<mode>): Use register as dest either
35227 always when a MEM, or when it is a MEM not rtx_equal_p to the
35228 corresponding dup operand, or even not when it is a CONST_VECTOR
35229 depending on the mode and lo vs. hi.
35230 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35232 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35233 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35234 Likewise. Require that operands[2] is even.
35235 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35236 Remove extraneous parens. Require that operands[2] is a multiple
35238 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35239 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35240 disallow memory then.
35242 2017-03-30 Richard Biener <rguenther@suse.de>
35244 PR tree-optimization/77498
35245 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35246 to non-constants over backedges.
35248 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35250 PR rtl-optimization/80233
35251 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35252 as last_combined_insn. Do not test for BARRIER_P separately.
35254 2017-03-29 Andreas Schwab <schwab@suse.de>
35257 * calls.c (prepare_call_address): Convert funexp to Pmode before
35258 copying to temp reg.
35260 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35262 PR tree-optimization/80158
35263 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35264 Handle possible future case of more than one alternate
35266 (replace_rhs_if_not_dup): Likewise.
35267 (replace_one_candidate): Likewise.
35269 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35271 PR rtl-optimization/80193
35272 * ira.c (ira): Do not check allocation for LRA.
35274 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35276 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35277 (nvptx_output_simt_exit): Declare.
35278 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35279 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35280 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35281 (nvptx_declare_function_name): ...here. Emit declaration of local
35282 memory space buffer for omp_simt_enter insn.
35283 (nvptx_output_unisimt_switch): New.
35284 (nvptx_output_softstack_switch): New.
35285 (nvptx_output_simt_enter): New.
35286 (nvptx_output_simt_exit): New.
35287 * config/nvptx/nvptx.h (struct machine_function): New fields
35288 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35289 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35290 (UNSPECV_SIMT_EXIT): Ditto.
35291 (omp_simt_enter_insn): New insn.
35292 (omp_simt_enter): New expansion.
35293 (omp_simt_exit): New insn.
35294 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35296 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35297 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35298 (expand_GOMP_SIMT_EXIT): New.
35299 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35300 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35301 (GOMP_SIMT_EXIT): Ditto.
35302 * target-insns.def (omp_simt_enter): New insn.
35303 (omp_simt_exit): Ditto.
35304 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35306 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35307 (lower_rec_input_clauses): Likewise.
35308 (lower_lastprivate_clauses): Handle SIMT privatization.
35310 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35311 (ompdevlow_adjust_simt_enter): New.
35312 (find_simtpriv_var_op): New.
35313 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35314 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35316 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35317 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35318 (copy_decl_for_dup_finish): Ditto.
35320 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35322 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35325 * config/i386/i386.c (ix86_option_override_internal): Always
35326 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35328 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35330 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35332 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35334 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35335 mark new edge's irreducible flag accordign to it.
35336 (vect_do_peeling): Check loop preheader edge's irreducible flag
35337 and pass it to function slpeel_add_loop_guard.
35339 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35341 PR tree-optimization/80218
35342 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35343 Update block frequencies and counts.
35345 2017-03-28 Richard Biener <rguenther@suse.de>
35347 PR tree-optimization/78644
35348 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35349 of a simplification result we may not use it at all.
35351 2017-03-28 Richard Biener <rguenther@suse.de>
35354 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35355 without arguments, generate default definition of a SSA name.
35357 2017-03-28 Richard Biener <rguenther@suse.de>
35359 PR middle-end/80222
35360 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35361 TYPE_REF_CAN_ALIAS_ALL references.
35362 * fold-const.c (fold_indirect_ref_1): Likewise.
35364 2017-03-28 Martin Liska <mliska@suse.cz>
35367 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35368 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35370 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35371 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35373 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35374 (EXTRA_SPECS): Define.
35375 (SUBTARGET_EXTRA_SPECS): Likewise.
35376 (SUBTARGET_CPP_SPEC): Likewise.
35377 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35378 SUBTARGET_EXTRA_SPECS.
35379 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35381 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35383 * config/arc/simdext.md (vst64_insn): Update pattern.
35384 (vld32wh_insn): Likewise.
35385 (vld32wl_insn): Likewise.
35386 (vld64_insn): Likewise.
35387 (vld32_insn): Likewise.
35389 2017-03-28 Marek Polacek <polacek@redhat.com>
35392 * fold-const.c (fold_comparison): Use protected_set_expr_location
35393 instead of SET_EXPR_LOCATION.
35395 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35397 * tree.c (add_expr): Avoid name lookup warning.
35399 2017-03-27 Jeff Law <law@redhat.com>
35401 PR tree-optimization/80216
35402 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35403 function name. Limit recursion depth.
35404 (record_temporary_equivalences): Corresponding changes.
35406 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35408 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35411 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35414 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35416 * cfgcleanup.c (reg_note_cfa_p): New array.
35417 (insns_have_identical_cfa_notes): New function.
35418 (old_insns_match_p): Don't cross-jump in between /f
35419 and non-/f instructions. If both i1 and i2 are frame related,
35420 verify all CFA notes, their order and content.
35422 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35425 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35426 HImode and SImode with zero extend to DImode to one insn.
35427 (bswap<mode>2_extenddi): Likewise.
35428 (bswapsi2_extenddi): Likewise.
35429 (bswaphi2_extendsi): Likewise.
35430 (bswaphi2): Combine bswap HImode and SImode into one insn.
35431 Separate memory insns from swapping register.
35432 (bswapsi2): Likewise.
35433 (bswap<mode>2): Likewise.
35434 (bswaphi2_internal): Delete, no longer used.
35435 (bswapsi2_internal): Likewise.
35436 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35437 store, and gpr<-gpr swap insns.
35438 (bswap<mode>2_store): Likewise.
35439 (bswaphi2_reg): Register only splitter, combine with the splitter.
35440 (bswaphi2 splitter): Likewise.
35441 (bswapsi2_reg): Likewise.
35442 (bswapsi2 splitter): Likewise.
35443 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35444 the insns into load, store, and register/register insns.
35445 (bswapdi2_ldbrx): Likewise.
35446 (bswapdi2_load): Likewise.
35447 (bswapdi2_store): Likewise.
35448 (bswapdi2_reg): Likewise.
35450 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35452 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35453 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35455 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35458 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35460 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35461 special handling for target option conflicts between dform
35462 options (-mpower9-dform, -mpower9-dform-vector,
35463 -mpower9-dform-scalar) and -mno-direct-move.
35465 2017-03-27 Richard Biener <rguenther@suse.de>
35467 PR tree-optimization/80181
35468 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35470 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35472 * config/arc/predicates.md (move_double_src_operand): Replace the
35473 call to move_double_src_operand with a call to address_operand.
35475 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35477 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35478 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35479 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35481 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35483 * config/arc/predicates.md (long_immediate_loadstore_operand):
35484 Consider scaled addresses cases.
35486 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35488 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35489 restored when in interrupt.
35490 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35491 doesn't have delay slot.
35493 2017-03-27 Richard Biener <rguenther@suse.de>
35496 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35497 inlined thunk clones.
35499 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35502 * asan.c (instrument_derefs): Copy over last operand from
35503 original COMPONENT_REF to the new COMPONENT_REF with
35504 DECL_BIT_FIELD_REPRESENTATIVE.
35505 * ubsan.c (instrument_object_size): Likewise.
35507 2017-03-27 Richard Biener <rguenther@suse.de>
35509 PR tree-optimization/80170
35510 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35511 sure DR/SCEV didnt fold in constants we do not see when looking
35512 at the reference base alignment.
35514 2017-03-27 Richard Biener <rguenther@suse.de>
35516 PR middle-end/80171
35517 * gimple-fold.c (fold_ctor_reference): Properly guard against
35518 NULL return value from canonicalize_constructor_val.
35520 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35523 * config/i386/i386.c (ix86_expand_builtin)
35524 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35525 flags reg setting and flags reg using instructions.
35526 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35527 clobbering instructions to zero extend op2.
35529 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35531 * doc/install.texi (Configuration) <--with-aix-soname>:
35532 Update link to AIX ld.
35534 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35536 PR rtl-optimization/80160
35537 PR rtl-optimization/80159
35538 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35539 reg_alternate_class into account.
35541 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35544 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35545 to consider in curr_insn_transform.
35547 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35549 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35550 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35551 and emit_mode_inner.
35553 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35555 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35556 argument to the overloaded builtin variants. Use the new flag to
35557 deprecate certain builtin variants.
35558 * config/s390/s390-builtin-types.def: Add new builtin types.
35559 * config/s390/s390-builtins.h: Support new flags field for
35560 overloaded builtins.
35561 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35562 (s390_macro_to_expand): Enable vector float data type.
35563 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35564 builtins by incrementing the __VEC__ version number.
35565 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35567 (s390_resolve_overloaded_builtin): Emit error messages depending
35568 on the builtin flags.
35569 * config/s390/s390.c (s390_expand_builtin): Support additional
35570 flags argument. Change error message to match the messages
35571 emitted in s390-c.c.
35572 * config/s390/s390.md: New UNSPEC_* constants.
35573 (op_type): Add new instruction types.
35574 * config/s390/vecintrin.h: Add new builtins and test data class
35576 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35577 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35578 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35579 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35580 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35581 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35583 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35584 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35585 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35586 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35588 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35589 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35590 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35591 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35593 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35594 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35595 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35596 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35597 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35598 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35599 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35601 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35602 ("vec_scatter_element<V_HW_4:mode>_DI")
35603 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35604 ("vec_fpint<mode>", "vflls")
35605 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35606 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35607 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35608 ("*vec_cmphe<mode>_cc"): ... these.
35610 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35611 mode constant instead of magic value.
35613 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35615 * config/s390/s390.c (s390_expand_vec_compare): Support other
35616 vector floating point modes than just V2DF.
35617 (s390_expand_vcond): Likewise.
35618 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35619 (s390_cannot_change_mode_class): Prevent mode changes between TF
35620 and V1TF in vector registers.
35621 * config/s390/s390.md (DF, SF): New mode attributes.
35622 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35623 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35624 SFmode support for VRs.
35625 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35627 (VFT, VF_HW): New mode iterators.
35628 (vw, sdx): New mode attributes.
35629 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35630 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35631 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35632 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35633 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35634 also the new vector floating point modes. Renaming to ...
35636 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35637 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35638 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35639 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35640 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35641 ("vec_unordered<mode>"): ... these.
35643 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35644 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35645 ("*vec_extendv2df"): New insn definitions.
35647 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35649 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35650 ("mulditi3_2", "*muldi3_sign"): New patterns.
35651 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35652 rename the pattern definition.
35654 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35656 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35658 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35660 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35662 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35663 instruction if possible.
35664 * config/s390/vector.md (vec_halfnumelts): New mode
35666 ("*vec_vllezlf<mode>"): New pattern.
35668 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35670 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35671 ("popcountv4si2", "popcountv2di2"): Rename to ...
35672 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35673 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35675 ("popcount<mode>2_vxe"): New pattern.
35677 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35679 * common/config/s390/s390-common.c (processor_flags_table): Add
35681 * config.gcc: Add arch12.
35682 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35683 Default to arch12 for unknown CPU model numbers.
35684 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35685 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35686 PROCESSOR_max sanity check.
35687 * config/s390/s390-opts.h (enum processor_type): Add
35689 * config/s390/s390.c (processor_table): Add arch12.
35690 (s390_expand_builtin): Add check for B_VXE flag.
35691 (s390_issue_rate): Add PROCESSOR_ARCH12.
35692 (s390_get_sched_attrmask): Likewise.
35693 (s390_get_unit_mask): Likewise.
35694 (s390_sched_score): Enable z13 scheduling for arch12.
35695 (s390_sched_reorder): Likewise.
35696 (s390_sched_variable_issue): Likewise.
35697 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35699 (s390_tune_attr): Use z13 scheduling also for arch12.
35700 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35701 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35702 (TARGET_VXE_P): New macros.
35703 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35704 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35705 * config/s390/s390.opt: Add arch12 as processor_type.
35707 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35709 * config/s390/s390.md
35710 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35711 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35712 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35714 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35715 Rename expanders to ...
35717 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35718 ("fixuns_truncdddi2_emu"): ... these.
35720 ("fixuns_trunc<mode>si2_emu"): New expander.
35722 ("*fixuns_truncdfdi2_z13"): Rename to ...
35723 ("*fixuns_truncdfdi2_vx"): ... this.
35725 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35727 * config/s390/2964.md: Remove the single element vector compare
35728 instructions which are no longer used.
35729 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35731 (s390_canonicalize_comparison): Remove handling of DFmode
35733 (s390_expand_vec_compare_scalar): Remove function.
35734 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35735 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35737 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35739 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35741 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35743 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35744 will anyway by matched by mov<mode>_64dfp.
35746 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35748 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35749 vlef/vstef. Add missing operand to vleif.
35751 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35753 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35754 pair for all vector types with 64 bit elements.
35755 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35756 * config/s390/vector.md (V_HW_64): ... here.
35757 (V_128_NOSINGLE): New mode iterator.
35758 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35759 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35760 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35761 ("*vec_load_pairv2di"): Change to ...
35762 ("*vec_load_pair<mode>"): ... this one.
35764 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35766 * config/s390/constraints.md: Add comments.
35767 (jKK): Reject element sizes > 8 bytes.
35768 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35770 * config/s390/s390.md: Add the s_operand checks formerly in
35771 s390_split_ok_p to various splitters where they are still
35773 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35774 for 128 bit vectors. Plus two splitters.
35776 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35778 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35781 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35784 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35785 error if the boundary argument is not constant.
35787 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35789 PR rtl-optimization/80112
35790 * loop-doloop.c (doloop_condition_get): Don't check condition
35791 if cmp isn't SET with IF_THEN_ELSE src.
35793 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35795 PR tree-optimization/80158
35796 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35797 replacing a candidate statement, also replace it for the
35798 candidate's alternate interpretation.
35799 (replace_rhs_if_not_dup): Likewise.
35800 (replace_one_candidate): Likewise.
35802 2017-03-24 Richard Biener <rguenther@suse.de>
35804 PR tree-optimization/80167
35805 * graphite-isl-ast-to-gimple.c
35806 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35808 (translate_isl_ast_to_gimple::get_rename): Likewise.
35810 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35812 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35813 handling of certain combinations of target options, including the
35814 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35815 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35817 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35820 * config/arm/arm.md (*load_multiple): Add reload_completed to
35821 matching condition.
35823 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35824 Richard Biener <rguenth@suse.de>
35826 PR tree-optimization/79908
35827 PR tree-optimization/80136
35828 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35829 been cast away, gimplify_and_add suffices.
35831 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35833 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35835 2017-03-23 Richard Biener <rguenther@suse.de>
35837 PR tree-optimization/80032
35838 * gimplify.c (gimple_push_cleanup): Forced unconditional
35839 cleanups still have to go to the conditional_cleanups
35842 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35844 PR tree-optimization/80072
35845 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35847 (next_operand_entry_id): Change type to unsigned int.
35848 (sort_by_operand_rank): Make sure to return the right return value
35849 even if unsigned fields are bigger than INT_MAX.
35850 (struct oecount): Change cnt and id type to unsigned int.
35851 (oecount_hasher::equal): Formatting fix.
35852 (oecount_cmp): Make sure to return the right return value
35853 even if unsigned fields are bigger than INT_MAX.
35854 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35857 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35858 TREE_READONLY on result if writing it more than once.
35861 * doc/invoke.texi (-fsanitize=thread): Document that with
35862 -fnon-call-exceptions atomics are not able to throw
35866 * tsan.c: Include tree-eh.h.
35867 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35868 maybe_clean_or_replace_eh_stmt where needed.
35869 (instrument_memory_accesses): Add cfg_changed argument.
35870 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35871 if it returned true.
35872 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35874 PR rtl-optimization/63191
35875 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35876 wrapper function, moved the whole old content into ...
35877 (ix86_delegitimize_address_1): ... this. New inline function.
35878 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35879 true as last argument instead of ix86_delegitimize_address.
35881 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35883 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35884 cortexa57_branch_cost.
35886 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35888 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35890 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35893 * doc/md.texi (Constraints): Document wA constraint.
35894 * config/rs6000/constraints.md (wA): New.
35895 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35896 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35897 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35898 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35900 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35903 * gimplify.c (is_oacc_declared): New function.
35904 (oacc_default_clause): Use it to set default flags for acc declared
35905 variables inside parallel regions.
35906 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35907 declared variables.
35908 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35909 declare attribute to any decl as necessary.
35911 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35914 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35915 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35916 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35917 (arm_arch_lpae): This.
35918 * config/arm/arm.c (arm_arch7ve): Rename into ...
35919 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35920 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35923 2017-03-22 Martin Liska <mliska@suse.cz>
35926 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35927 error message instead of an ICE.
35929 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35931 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35933 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35935 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35937 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35940 2017-03-21 Martin Sebor <msebor@redhat.com>
35942 * doc/extend.texi: Use "cannot" instead of "can't."
35943 * doc/hostconfig.texi: Same.
35944 * doc/install.texi: Same.
35945 * doc/invoke.texi: Same.
35946 * doc/loop.texi: Same.
35947 * doc/md.texi: Same.
35948 * doc/objc.texi: Same.
35949 * doc/rtl.texi: Same.
35950 * doc/tm.texi: Same.
35951 * doc/tm.texi.in: Same.
35952 * doc/trouble.texi: Same.
35954 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35957 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35958 (collect_checksum_attributes): Set it.
35959 (die_checksum_ordered): Use it.
35961 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35963 PR tree-optimization/79908
35964 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35965 change: For a VA_ARG whose LHS has been cast away, use
35966 force_gimple_operand to construct the side effects.
35968 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35970 PR translation/80001
35971 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35972 more amenable to translation.
35973 (oacc_loop_auto_partitions): Likewise.
35975 2017-03-21 Marek Polacek <polacek@redhat.com>
35976 Martin Sebor <msebor@redhat.com>
35978 PR tree-optimization/80109
35979 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35980 on INTEGRAL_TYPE_P.
35982 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35983 Segher Boessenkool <segher@kernel.crashing.org>
35986 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35987 check reg_used_between_p between insn and one of succ or succ2
35988 depending on if succ is artificial insn not inserted into insn
35991 2017-03-21 Martin Liska <mliska@suse.cz>
35993 PR gcov-profile/80081
35994 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35995 * doc/gcc.texi: Include gcov-dump stuff.
35996 * doc/gcov-dump.texi: New file.
35998 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
36000 PR rtl-optimization/79150
36001 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
36002 conditional jump, if the jump is the last insn of the loop.
36004 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36005 Richard Biener <rguenth@suse.de>
36007 PR tree-optimization/79908
36008 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36009 been cast away, use force_gimple_operand to construct the side
36012 2017-03-21 Martin Liska <mliska@suse.cz>
36014 PR libfortran/79956
36015 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
36018 2017-03-21 Brad Spengler <spender@grsecurity.net>
36021 * plugin.c (htab_hash_plugin): New function.
36022 (add_new_plugin): Use it and adjust.
36023 (parse_plugin_arg_opt): Adjust.
36024 (init_one_plugin): Likewise.
36026 2017-03-21 Richard Biener <rguenther@suse.de>
36028 PR tree-optimization/80032
36029 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
36030 if set force the cleanup to happen unconditionally.
36031 (gimplify_target_expr): Push inserted clobbers with force_uncond
36032 to avoid them being removed by control-dependent DCE.
36034 2017-03-21 Richard Biener <rguenther@suse.de>
36036 PR tree-optimization/80122
36037 * tree-inline.c (copy_bb): Do not expans va-arg packs or
36038 va_arg_pack_len when the inlined call stmt requires pack
36040 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
36042 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36045 * tsan.c (instrument_builtin_call): If the memory model argument
36046 is not a constant, assume it is valid.
36049 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
36052 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36054 PR rtl-optimization/79910
36055 * combine.c (can_combine_p): Do not allow combining an I0 or I1
36056 if its dest is used by an insn before I2 (other than the combined
36057 insns themselves, which are properly handled already).
36059 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36062 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36064 * combine.c (record_used_regs): New static function.
36065 (try_combine): Handle situations where there is an additional
36066 instruction between I2 and I3 which needs to have a LOG_LINK
36070 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36072 * combine.c (try_combine): Delete redundant i1 test. Call
36073 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36075 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36078 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36079 alternatives 13/14.
36081 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36083 PR tree-optimization/80054
36084 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36085 the optimization if a PHI or any of its arguments is not dominated
36086 by the candidate's basis. Use gphi* rather than gimple* as
36088 (replace_profitable_candidates): Clean up a gimple* variable that
36089 should be a gphi* variable.
36091 2017-03-20 Martin Sebor <msebor@redhat.com>
36094 * doc/extend.texi (attribute constructor): Document present limitation.
36096 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36099 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36100 __POWER9_VECTOR__ #ifdef control, change template definition to
36101 use Power9-specific built-in function.
36102 (vec_any_eq): Likewise.
36103 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36104 to control outcomes from this test.
36105 (vector_ae_<mode>p): For VEC_F modes, likewise.
36107 2017-03-20 Ian Lance Taylor <iant@google.com>
36109 * config/i386/i386.c (ix86_function_regparm): Save an extra
36110 register for -fsplit-stack with DECL_STATIC_CHAIN.
36112 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36115 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36116 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36118 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36120 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36122 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36125 2017-03-20 Marek Polacek <polacek@redhat.com>
36128 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36130 2017-03-20 Richard Biener <rguenther@suse.de>
36132 PR tree-optimization/80113
36133 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36134 allocate extra SSA name for PHI def.
36135 (add_close_phis_to_outer_loops): Likewise.
36136 (add_close_phis_to_merge_points): Likewise.
36137 (copy_loop_close_phi_args): Likewise.
36138 (copy_cond_phi_nodes): Likewise.
36140 2017-03-20 Martin Liska <mliska@suse.cz>
36142 PR middle-end/79753
36143 * tree-chkp.c (chkp_build_returned_bound): Do not build
36144 returned bounds for a LHS that's not a BOUNDED_P type.
36146 2017-03-20 Martin Liska <mliska@suse.cz>
36150 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36151 COMPLEX_CST and VECTOR_CST.
36153 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36156 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36157 target operand. A new splitter adds the clobber statement in case
36158 the target operand is dead anyway.
36160 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36162 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36163 to age-old versions of binutils and glibc.
36165 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36167 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36169 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36171 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36173 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36175 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36176 requirement for binutils 2.13.
36178 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36180 * combine.c (try_combine): Delete redundant i1 test. Call
36181 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36183 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36185 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36186 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36188 <riscv64-*-elf>: Re-arrange section
36189 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36190 <riscv32-*-linux>: Likewise.
36191 <riscv64-*-elf>: Likewise
36192 <riscv64-*-linux>: Likewise.
36194 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36197 * aarch64.opt(verbose-cost-dump): Fix typo.
36199 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36202 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36203 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36205 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36207 * reload.c (find_reloads): When reloading a nonoffsettable address,
36208 use RELOAD_OTHER for it and its address reloads.
36210 PR rtl-optimization/79910
36211 * combine.c (record_used_regs): New static function.
36212 (try_combine): Handle situations where there is an additional
36213 instruction between I2 and I3 which needs to have a LOG_LINK
36216 2017-03-17 Jeff Law <law@redhat.com>
36218 PR tree-optimization/71437
36219 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36220 conditional in the hash table first.
36221 (vrp_dom_walker::before_dom_children): Extract condition from
36222 ASSERT_EXPR. Record condition, its inverion and any implied
36223 conditions as well.
36225 2017-03-17 Marek Polacek <polacek@redhat.com>
36226 Markus Trippelsdorf <markus@trippelsdorf.de>
36228 PR tree-optimization/80079
36229 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36232 2017-03-17 Richard Biener <rguenther@suse.de>
36234 PR middle-end/80075
36235 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36236 Properly verify the LHS before the RHS possibly claims to be
36238 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36241 2017-03-17 Martin Jambor <mjambor@suse.cz>
36243 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36244 (List of -O2 options): Likewise.
36245 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36248 2017-03-17 Tom de Vries <tom@codesourcery.com>
36250 * gcov-dump.c (print_usage): Print bug_report_url.
36252 2017-03-17 Richard Biener <rguenther@suse.de>
36254 PR middle-end/80050
36255 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36256 (parser::peek): Likewise.
36258 2017-03-17 Richard Biener <rguenther@suse.de>
36260 PR tree-optimization/80048
36261 * sese.c (free_sese_info): Properly release rename_map and
36262 copied_bb_map elements.
36264 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36266 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36267 Add linked-list forward and backlinks. Insert on
36268 construction, remove on destruction.
36269 (class pass_store_merging): Add m_stores_head field.
36270 (pass_store_merging::terminate_and_process_all_chains):
36271 Iterate over m_stores_head list.
36272 (pass_store_merging::terminate_all_aliasing_chains):
36274 (pass_store_merging::execute): Check for debug stmts first.
36275 Push new chains onto the m_stores_head stack.
36277 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36280 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36281 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36282 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36284 2017-03-16 Jeff Law <law@redhat.com>
36286 PR tree-optimization/71437
36287 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36288 member function. Implementation moved into after_dom_children
36289 member function and into the threader's thread_outgoing_edges
36291 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36292 some code into new thread_outgoing_edges.
36293 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36294 definition. Simplify marker handling (do it here). Assume we always
36295 have the available expression and the const/copies tables.
36296 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36298 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36299 * tree-vrp.c (equiv_stack): No longer file scoped.
36300 (vrp_dom_walker): New class.
36301 (vrp_dom_walker::before_dom_children): New member function.
36302 (vrp_dom_walker::after_dom_children): Likewise.
36303 (identify_jump_threads): Setup domwalker. Use it rather than
36304 walking edges in a random order by hand. Simplify setup/finalization.
36305 (finalize_jump_threads): Remove.
36306 (vrp_finalize): Do not call identify_jump_threads here.
36307 (execute_vrp): Do it here instead and call thread_through_all_blocks
36310 PR tree-optimization/71437
36311 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36313 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36315 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36316 (dom_opt_dom_walker::thread_across_edge): Remove
36317 handle_dominating_asserts argument. All callers changed.
36318 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36319 changes. Remove calls to lhs_of_dominating_assert. Other
36320 uses of handle_dominating_asserts turn into unconditional code
36321 (simplify_control_stmt_condition_1): Likewise.
36322 (simplify_control_stmt_condition): Likewise.
36323 (thread_through_normal_block, thread_across_edge): Likewise.
36324 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36325 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36326 object if it is not an SSA_NAME.
36327 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36328 before calling into the VRP specific simplifiers.
36329 (identify_jump_threads): Remove handle_dominating_asserts
36332 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36335 * tree-diagnostic.c (default_tree_printer): No longer static.
36336 * tree-diagnostic.h (default_tree_printer): New prototype.
36338 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36340 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36341 Change ins into fmov.
36343 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36345 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36346 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36347 Use h_con constraint for operand 1.
36348 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36349 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36351 2017-03-15 Jeff Law <law@redhat.com>
36353 PR tree-optimization/71437
36354 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36355 (record_temporary_equivalences): Use it.
36357 PR tree-optimization/71437
36358 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36359 tree-ssa-scopedtables.
36360 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36361 (record_conditions, record_cond, vuse_eq): Likewise.
36362 (record_edge_info): Adjust to API tweak of record_conditions.
36363 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36364 (record_temporary_equivalences, optimize_stmt): Likewise.
36365 (eliminate_redundant_computations): Likewise.
36366 (record_equivalences_from_stmt): Likewise.
36367 * tree-ssa-scopedtables.c: Include options.h and params.h.
36368 (vuse_eq): New function, moved from tree-ssa-dom.c
36369 (build_and_record_new_cond): Likewise.
36370 (record_conditions): Likewise. Accept vector of conditions rather
36371 than edge_equivalence structure for first argument.
36372 for the first argument.
36373 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36374 from tree-ssa-dom.c.
36375 (avail_exprs_stack::record_cond): Likewise.
36376 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36377 from tree-ssa-dom.c.
36378 (avail_exprs_stack): Add new member functions lookup_avail_expr
36380 (record_conditions): Declare.
36382 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36385 * lra-constraints.c (process_alt_operands): Increase reject for
36386 reloading an input/output operand.
36388 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36391 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36392 insns to convert from signed/unsigned char/short to IEEE 128-bit
36394 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36396 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36399 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36400 subreg of inner mode for values already in registers.
36402 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36404 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36405 iteration reg is used after the loop.
36407 2017-03-14 Martin Sebor <msebor@redhat.com>
36409 PR tree-optimization/79800
36410 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36411 precision in negative-positive range.
36412 (format_floating): Call non-const overload with adjusted precision.
36414 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36417 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36420 2017-03-14 Martin Sebor <msebor@redhat.com>
36422 PR middle-end/80020
36423 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36424 * builtins.def (aligned_alloc): Use it.
36427 * Makefile.in (GTFILES): Add calls.c.
36428 * calls.c: Include "gt-calls.h".
36430 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36432 PR rtl-optimization/79728
36433 * regs.h (struct target_regs): New field
36434 x_contains_allocatable_regs_of_mode.
36435 (contains_allocatable_regs_of_mode): New macro.
36436 * reginfo.c (init_reg_sets_1): Initialize it, and change
36437 contains_reg_of_mode so it includes global regs as well.
36438 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36439 rather than contains_regs_of_mode.
36441 2017-03-14 Martin Liska <mliska@suse.cz>
36443 * doc/invoke.texi: Document options that can't be combined with
36444 -fcheck-pointer-bounds.
36446 2017-03-14 Martin Liska <mliska@suse.cz>
36448 PR middle-end/79831
36449 * doc/invoke.texi (-Wchkp): Document the option.
36451 2017-03-14 Martin Liska <mliska@suse.cz>
36453 * Makefile.in: Install gcov-dump.
36455 2017-03-14 Martin Liska <mliska@suse.cz>
36457 * multiple_target.c (expand_target_clones): Bail out for
36458 an invalid attribute.
36460 2017-03-14 Richard Biener <rguenther@suse.de>
36462 * alias.c (struct alias_set_entry): Pack properly.
36463 * cfgloop.h (struct loop): Likewise.
36464 * cse.c (struct set): Likewise.
36465 * ipa-utils.c (struct searchc_env): Likewise.
36466 * loop-invariant.c (struct invariant): Likewise.
36467 * lra-remat.c (struct cand): Likewise.
36468 * recog.c (struct change_t): Likewise.
36469 * rtl.h (struct address_info): Likewise.
36470 * symbol-summary.h (function_summary): Likewise.
36471 * tree-loop-distribution.c (struct partition): Likewise.
36472 * tree-object-size.c (struct object_size_info): Likewise.
36473 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36474 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36475 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36476 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36477 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36478 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36479 (struct _stmt_vec_info): Likewise.
36481 2017-03-14 Martin Liska <mliska@suse.cz>
36484 * multiple_target.c (create_dispatcher_calls): Check that
36485 a target can create a function dispatcher.
36487 2017-03-14 Martin Liska <mliska@suse.cz>
36490 * multiple_target.c (expand_target_clones): Drop local.local
36491 flag for default implementation.
36493 2017-03-14 Richard Biener <rguenther@suse.de>
36495 PR tree-optimization/80030
36496 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36498 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36500 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36501 gcc_fallthrough() instead of __attribute__((fallthrough));
36503 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36505 * doc/gcc.texi: Remove "up" link to (DIR).
36506 * doc/gccint.texi: Ditto.
36508 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36510 * doc/install.texi (Specific) <avr>: Remove reference to
36513 2017-03-13 Jeff Law <law@redhat.com>
36515 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36516 attribute rather than comments.
36518 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36519 match_scratch operand is highest.
36521 2017-03-13 Martin Liska <mliska@suse.cz>
36523 PR middle-end/78339
36524 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36525 is a CHKP clone, use original declaration.
36527 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36529 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36530 (arc_conditional_register_usage): Use a different allocation order
36531 when optimizing for size.
36532 * common/config/arc/arc-common.c (arc_option_optimization_table):
36533 Section anchors default on when optimizing for size.
36535 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36537 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36539 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36541 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36542 * config/arc/arc.md (cpu_facility): Add cd variant.
36543 (*movqi_insn): Add code density variant.
36544 (*movhi_insn): Likewise.
36545 (*movqi_insn): Likewise.
36546 (*addsi3_mixed): Likewise.
36547 (subsi3_insn): Likewise.
36549 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36551 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36553 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36555 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36556 expressions with MINUS and UNARY ops.
36558 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36561 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36563 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36564 between vec_select and vector argument.
36565 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36566 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36567 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36568 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36569 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36570 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36572 2017-03-13 Richard Biener <rguenther@suse.de>
36575 * params.def (vect-max-peeling-for-alignment): Fix typo.
36577 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36579 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36580 issue that only occurred with binutils below 2.18.
36582 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36584 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36585 refer to binutils 2.11/2.12 minimum.
36587 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36589 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36590 ftp.kernel.org and simplify binutils requirement.
36592 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36594 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36596 (Optimize Options): Ditto. Also remove redundancy.
36598 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36600 PR translation/79848
36601 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36603 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36604 to G_ to avoid double translation.
36606 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36608 PR translation/79923
36609 * auto-profile.c (get_combined_location): Convert leading
36610 character of diagnostics to lower case and remove trailing period.
36611 (read_profile): Likewise for various diagnostics.
36612 * config/arm/arm.c (arm_option_override): Remove trailing period
36613 from various diagnostics.
36614 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36615 (msp430_expand_delay_cycles): Likewise.
36617 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36620 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36621 full command-line argument, rather than just "str".
36622 (aarch64_validate_march): Likewise.
36623 (aarch64_validate_mtune): Likewise.
36625 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36627 PR rtl-optimization/78911
36628 * lra-assigns.c (must_not_spill_p): New function.
36629 (spill_for): Use it.
36631 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36633 PR tree-optimization/79981
36634 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36635 ATOMIC_COMPARE_EXCHANGE ifn result.
36636 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36637 IFN_ATOMIC_COMPARE_EXCHANGE.
36639 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36642 * opts.c (parse_sanitizer_options): Add missing question mark to
36643 "did you mean" message.
36645 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36647 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36649 (VMULEUH_UNS): Likewise.
36650 (VMULOUB_UNS): Likewise.
36651 (VMULOUH_UNS): Likewise.
36652 * config/rs6000/rs6000.c (builtin_function_type): Remove
36653 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36655 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36658 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36659 x with result of extra_parsing_for_operand_code_0.
36660 (function_reader::extra_parsing_for_operand_code_0): Convert
36661 return type from void to rtx, returning x. When reading
36662 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36663 larger size containing struct block_symbol.
36665 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36667 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36668 -mfloat128-hardware without -m64.
36670 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36673 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36674 entries to the case statement that marks unsigned arguments to
36675 overloaded functions.
36677 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36679 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36680 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36682 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36685 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36686 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36688 2017-03-10 Martin Liska <mliska@suse.cz>
36692 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36693 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36696 2017-03-10 Olivier Hainque <hainque@adacore.com>
36698 * tree-switch-conversion (array_value_type): Start by resetting
36699 candidate type to it's main variant.
36701 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36703 PR rtl-optimization/79909
36704 * combine.c (try_combine): Use simplify_replace_rtx on individual
36705 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36706 of the whole CALL_INSN_FUNCTION_USAGE.
36708 PR tree-optimization/79972
36709 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36710 get_range_info on SSA_NAMEs. Formatting fixes.
36712 2017-03-10 Richard Biener <rguenther@suse.de>
36713 Jakub Jelinek <jakub@redhat.com>
36715 PR tree-optimization/77975
36716 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36717 edge to be constant.
36718 (get_val_for): For constant x return it. Formatting fix.
36719 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36720 would use the same bases as the current one.
36722 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36724 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36725 instead of vec_select for V1TImode.
36726 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36728 (VSX_LE_128): Add V1TI to this mode iterator.
36729 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36730 (*vsx_le_perm_store_<mode>): Likewise.
36731 (pre-reload splitter for VSX stores): Likewise.
36732 (post-reload splitter for VSX stores): Likewise.
36733 (*vsx_xxpermdi2_le_<mode>): Likewise.
36734 (*vsx_lxvd2x2_le_<mode>): Likewise.
36735 (*vsx_stxvd2x2_le_<mode>): Likewise.
36737 2017-03-09 Michael Eager <eager@eagercon.com>
36739 Correct failures with --enable-checking=yes,rtl.
36741 * config/microblaze/microblaze.c (microblaze_expand_shift):
36742 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36743 test for const0_rtx.
36744 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36745 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36747 2017-03-09 Richard Biener <rguenther@suse.de>
36749 PR tree-optimization/79977
36750 * graphite-scop-detection.c (scop_detection::merge_sese):
36751 Handle the case of extra exits to blocks dominating the entry.
36753 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36755 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36758 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36760 PR rtl-optimization/79949
36761 * lra-constraints.c (process_alt_operands): Check memory when
36762 trying to predict a cycle. Print about the overall increase.
36764 2017-03-09 Richard Biener <rguenther@suse.de>
36766 PR middle-end/79971
36767 * gimple-expr.c (useless_type_conversion_p): Preserve
36768 TYPE_SATURATING for fixed-point types.
36770 2017-03-09 Richard Biener <rguenther@suse.de>
36773 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36774 alignment of BLKmode params.
36776 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36779 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36780 (VALL_NO_V2Q): Likewise.
36782 * config/aarch64/aarch64-simd.md
36783 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36785 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36786 VALL_NO_V2Q mode iterator.
36787 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36789 2017-03-09 Martin Liska <mliska@suse.cz>
36791 PR tree-optimization/79631
36792 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36793 tree_int_cst_sign_bit just for INTEGER constants.
36795 2017-03-09 Martin Liska <mliska@suse.cz>
36799 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36802 2017-03-09 Marek Polacek <polacek@redhat.com>
36805 * tree.c (inchash::add_expr): Handle TREE_VEC.
36807 2017-03-09 Martin Liska <mliska@suse.cz>
36810 (chkp_narrow_size_and_offset): New function.
36811 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36812 (void chkp_parse_bit_field_ref): New function.
36813 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36814 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36816 2017-03-09 Martin Liska <mliska@suse.cz>
36819 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36820 (chkp_find_bounds_1): Remove gcc_unreachable.
36822 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36825 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36826 BUILT_IN_SYNC*, determine the access type from the size suffix and
36827 always build a MEM_REF with that type. Handle forgotten
36828 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36831 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36832 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36833 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36834 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36835 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36836 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36837 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36838 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36839 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36840 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36841 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36842 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36843 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36844 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36845 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36846 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36847 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36848 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36849 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36850 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36851 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36852 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36853 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36854 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36855 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36856 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36857 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36858 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36859 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36860 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36861 definitions outside of __OPTIMIZE__ guarded section.
36864 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36865 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36866 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36867 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36870 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36872 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36873 ("vfenez<mode>"): Add missing constraints.
36875 2017-03-08 Martin Sebor <msebor@redhat.com>
36878 * config/nds32/nds32.c (nds32_option_override):
36879 Fix misspelled diagnostic.
36881 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36884 * gimplify.c (gimplify_omp_for): Replace index var in outer
36885 taskloop statement with an artificial variable and add
36886 OMP_CLAUSE_PRIVATE clause for it.
36888 2017-03-08 Richard Biener <rguenther@suse.de>
36890 PR tree-optimization/79955
36891 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36892 for accesses that are completely outside of the variable.
36894 2017-03-08 Andrew Haley <aph@redhat.com>
36896 PR tree-optimization/79943
36897 * tree-ssa-loop-split.c (compute_new_first_bound): When
36898 calculating the new upper bound, (END-BEG) should be added, not
36901 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36903 * config/avr/avr.md (setmemhi): Make sure match_dup
36904 operand number comes before match_scratch.
36906 2017-03-08 Richard Biener <rguenther@suse.de>
36908 PR tree-optimization/79920
36909 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36910 with ncopies == 1 to ...
36911 (vect_transform_slp_perm_load): ... here. Properly compute
36912 all element loads by iterating VF times over the group. Do
36913 not handle ncopies (computed in a broken way) in
36914 vect_create_mask_and_perm.
36916 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36919 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36920 is a uniform vector, use uniform_vector_p return value instead of
36921 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36923 2017-03-07 Marek Polacek <polacek@redhat.com>
36925 PR middle-end/79809
36926 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36927 (alloca_call_type): Likewise.
36929 2017-03-07 Martin Liska <mliska@suse.cz>
36931 * gcov.c (process_args): Put comment to correct location.
36933 2017-03-07 Martin Liska <mliska@suse.cz>
36935 PR middle-end/68270
36936 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36937 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36938 (chkp_narrow_bounds_for_field): Likewise.
36939 (chkp_parse_array_and_component_ref): Pass one more argument to
36942 2017-03-07 Richard Biener <rguenther@suse.de>
36944 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36947 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36949 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36950 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36952 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36955 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36956 to end of description.
36957 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36959 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36961 PR rtl-optimization/79901
36962 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36964 (*avx512f_<code><mode>3<mask_name>): ... this.
36965 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36966 iterator instead of VI8_AVX2_AVX512BW.
36968 PR rtl-optimization/79901
36969 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36970 min/max expander, expand it using expand_vec_cond_expr.
36973 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36976 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36979 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36980 to void * for PCH reasons.
36981 * dwarf2out.c (output_loc_operands, output_die): Cast
36982 v.val_vec.array to unsigned char *.
36984 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36987 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36990 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36992 PR rtl-optimization/79571
36993 * lra-constraints.c (process_alt_operands): Calculate static
36994 reject and subtract it from overall when only addresses will be
36997 2017-03-06 Julia Koval <julia.koval@intel.com>
37000 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
37001 incoming stack boundary to 128 for 64-bit targets.
37003 2017-03-06 Richard Biener <rguenther@suse.de>
37005 PR tree-optimization/79894
37006 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
37007 to NULL after folding it.
37009 2017-03-06 Richard Biener <rguenther@suse.de>
37011 PR tree-optimization/79824
37012 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
37013 check disabling peeling for gaps.
37015 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
37017 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
37018 attributes): Document gettimeofday.
37020 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37022 * config/s390/s390.c (s390_option_override_internal): Set
37023 PARAM_MIN_VECT_LOOP_BOUND
37025 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37027 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
37028 * config/s390/s390.md: Likewise.
37030 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37033 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
37034 (<avx2_avx512>_perm<mode>): Rename to ...
37035 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
37037 (<avx512>_perm<mode>_mask): Rename to ...
37038 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
37040 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
37041 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
37042 instead of VI8F_256_512.
37043 (avx512f_perm<mode>): New define_expand.
37044 (avx512f_perm<mode>_mask): Likewise.
37045 (avx512f_perm<mode>_1<mask_name>): New define_insn.
37046 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
37048 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37050 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
37051 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
37053 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
37055 2017-03-06 Martin Liska <mliska@suse.cz>
37058 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37059 when having a SSA NAME w/o VAR_DECL assigned to it.
37061 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37063 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37064 msa_dpsub_<su>_d): Fix MODE for vec_select.
37066 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37068 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37070 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37072 2017-03-06 Richard Biener <rguenther@suse.de>
37074 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37075 * plugin.c (register_plugin_info): Likewise.
37076 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37078 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37080 * config/i386/sse.md (sse_storehps, sse_storelps,
37081 avx_<castmode><avxsizesuffix>_<castmode>,
37082 avx512f_<castmode><avxsizesuffix>_<castmode>,
37083 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37084 in condition that at least one operand is not a MEM.
37086 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37088 PR middle-end/79805
37089 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37090 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37092 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37093 gimple_call_nothrow_p flag based on whether original builtin can throw.
37094 If it can, emit following stmts on the fallthrough edge.
37095 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37096 don't create new bb if inserting just debug stmts on the edge, try to
37097 insert them on the fallthru bb or just reset debug stmts.
37099 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37102 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37103 restore recog_data (including the operand rtxes inside it) around
37104 the call to get_insn_template.
37106 2017-03-03 Martin Sebor <msebor@redhat.com>
37108 PR tree-optimization/79699
37109 * context.c (context::~context): Free MPFR caches to avoid
37110 a memory leak on program exit.
37112 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37114 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37115 Use wide_int::ulow () instead of .elt (0).
37117 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37119 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37120 (*pushxf): Limit oF constraint to 32bit targets and add oC
37121 constraint for 64bit targets.
37122 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37123 (*pushdf): Change rmF constraint to rmC.
37125 2017-03-03 Martin Liska <mliska@suse.cz>
37127 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37128 Remove unused variable.
37130 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37133 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37134 is a memory operand, increase num_memory.
37135 (ix86_expand_args_builtin): Likewise.
37137 2017-03-03 Jan Hubicka <jh@suse.cz>
37140 * ipa-devirt.c (maybe_record_node): Properly handle
37141 __cxa_pure_virtual visibility.
37143 2017-03-03 Martin Liska <mliska@suse.cz>
37145 PR tree-optimization/79803
37146 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37148 (pass_loop_prefetch::execute): Disabled optimization if an
37149 assumption about L1 cache size is not met.
37151 2017-03-03 Martin Liska <mliska@suse.cz>
37153 PR rtl-optimization/79574
37154 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37155 (hash_scan_set): Likewise.
37156 (dump_hash_table): Likewise.
37157 (hoist_code): Likewise.
37159 2017-03-03 Richard Biener <rguenther@suse.de>
37161 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37162 in place of uhigh ().
37163 (fixed_convert_from_real): Likewise.
37165 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37168 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37170 2017-03-03 Richard Biener <rguenther@suse.de>
37172 PR middle-end/79818
37173 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37174 TYPE_OVERFLOW_UNDEFINED check.
37176 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37178 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37180 (vector_ae_<mode>_p): Likewise.
37181 (vector_nez_<mode>_p): Likewise.
37182 (vector_ne_v2di_p): Likewise.
37183 (vector_ae_v2di_p): Likewise.
37184 (vector_ne_<mode>_p): Likewise.
37185 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37187 (vsx_tsqrt<mode>2_fe): Likewise.
37189 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37192 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37194 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37196 PR rtl-optimization/79780
37197 * cprop.c (one_cprop_pass): When second and further conditional trap
37198 in a single basic block is turned into an unconditional trap, turn it
37199 into a deleted note to avoid RTL verification failures.
37201 2017-03-02 Richard Biener <rguenther@suse.de>
37203 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37205 2017-03-02 Richard Biener <rguenther@suse.de>
37207 PR tree-optimization/79345
37209 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37210 param and abort the walk, returning -1 if it is hit.
37211 (walk_aliased_vdefs): Take a limit param and pass it on.
37212 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37213 defaulting to 0 and return a signed int.
37214 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37215 (check_defs): New helper.
37216 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37217 about uninitialized memory.
37218 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37219 bogus uninitialized warning.
37220 (fixed_convert_from_real): Likewise.
37222 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37224 PR tree-optimization/66768
37225 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37226 iv_use if base object can't be determined.
37228 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37230 PR tree-optimization/79345
37231 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37232 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37233 (get_pattern_stats): Initialize it.
37234 * genemit.c (gen_expand): Verify match_scratch numbers come after
37235 match_operand/match_dup numbers.
37236 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37237 match_scratch numbers.
37238 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37240 * config/s390/s390.md (trunctdsd2): Likewise.
37242 2017-03-02 Richard Biener <rguenther@suse.de>
37244 * wide-int.h (wide_int_storage::operator=): Implement in terms
37247 2017-03-02 Richard Biener <rguenther@suse.de>
37249 PR tree-optimization/79777
37250 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37251 the to insert expression to sth existing.
37253 2017-03-01 Martin Sebor <msebor@redhat.com>
37255 PR middle-end/79692
37256 * gimple-ssa-sprintf.c
37257 (directive::known_width_and_precision): New function.
37258 (format_integer): Use it.
37259 (get_mpfr_format_length): Consider the full range of precision
37260 when computing %g output with the # flag. Set the likely byte
37261 count to 3 rather than 1 when precision is indeterminate.
37262 (format_floating): Correct the lower bound of precision.
37264 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37266 * doc/invoke.texi: Document default code model for 64-bit Linux.
37268 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37271 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37272 udiv rather than div since input pattern is unsigned.
37274 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37276 * config/i386/i386.c (print_reg): Warn for values of
37277 unsupported size in integer register.
37279 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37282 * config/rs6000/predicates.md (current_file_function_operand): Do
37283 not allow self calls to be local if the function is replaceable.
37285 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37288 * config/rs6000/altivec.h (vec_ctz and others): Change the
37289 preprocessor macro that controls conditional compilation from
37290 _ARCH_PWR9 to __POWER9_VECTOR__.
37291 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37292 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37293 control (instead of _ARCH_PWR9 control) so that template
37294 definition uses power9-specific function.
37295 (vec_any_eq): Likewise.
37296 (vec_all_ne): Change macro definition to use a power9-specific
37297 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37298 _ARCH_PWR9 control).
37299 (vec_any_eq) Likewise.
37300 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37301 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37302 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37303 support for xvcmpnedp instruction.
37304 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37305 macro expansion so that Power9 implementation of vec_all_ne does
37306 not use the AltiVec predicate framework.
37307 (VCMPNEH_P): Likewise.
37308 (VCMPNEW_P): Likewise.
37309 (VCMPNED_P): Likewise.
37310 (VCMPNEFP_P): Likewise.
37311 (VCMPNEDP_P): Likewise.
37312 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37313 implementation of vec_any_eq to not use AltiVec predicate
37315 (VCMPAEH_P): Likewise.
37316 (VCMPAEW_P): Likewise.
37317 (VCMPAED_P): Likewise.
37318 (VCMPAEFP_P): Likewise.
37319 (VCMPAEDP_P): Likewise.
37320 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37321 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37322 not use the AltiVec predicate framework.
37323 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37324 of vec_any_eq to not use AltiVec predicate framework.
37325 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37326 support for predefined __POWER9_VECTOR__ macro to indicate that
37327 Power9 instruction selection is enabled.
37328 (altivec_overloaded_builtins): Remove extraneous
37329 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37330 function argument types RS6000_BTI_bool_V16QI and
37331 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37332 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37333 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37334 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37335 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37336 Power9 for implementations of vec_cmpne. Change the signature for
37337 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37338 (representing vec_all_ne) to remove the previously described first
37339 argument of type RS6000_BTI_INTSI, as this was an artifact of
37340 reliance on the AltiVec predicate framework, which is no longer
37341 used in the implementation of these functions. Add
37342 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37343 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37344 since, unlike the AltiVec predicate framework implementation, we
37345 do not share function descriptors between vec_alle and vec_anyeq.
37346 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37347 set of modes that receive special treatment even when
37348 TARGET_P9_VECTOR is true. The special treatment emits code that
37349 does not depend on Power9 instructions.
37350 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37351 define_expand to not rely on AltiVec predicate framework.
37352 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37354 (vector_ne_v2di_p): Change this define_expand to not rely on
37355 AltiVec predicate framework.
37356 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37358 (vector_ne_<mode>_p): Change this define_expand to not rely on
37359 AltiVec predicate framework.
37360 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37362 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37363 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37364 define_insn pattern.
37365 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37366 define_insn pattern because the xvcmpne<VSs>. instruction is not
37368 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37369 instruction is not supported.
37371 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37373 * config/nvptx/nvptx.c: Include intl.h.
37375 2017-03-01 Martin Jambor <mjambor@suse.cz>
37378 * ipa-prop.h (ipa_bits): Removed field known.
37379 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37380 to pointers. Adjusted their comments to warn about their sharing.
37381 (ipcp_transformation_summary): Change bits to a vector of pointers.
37382 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37383 (ipa_get_ipa_bits_for_value): Declare.
37384 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37385 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37386 (ipa_bits_hash_table): Likewise.
37387 (ipa_vr_ggc_hash_traits): Likewise.
37388 (ipa_vr_hash_table): Likewise.
37389 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37390 being pointers and vr_known being removed.
37391 (ipa_set_jf_unknown): Likewise.
37392 (ipa_get_ipa_bits_for_value): New function.
37393 (ipa_set_jfunc_bits): Likewise.
37394 (ipa_get_value_range): New overloaded functions.
37395 (ipa_set_jfunc_vr): Likewise.
37396 (ipa_compute_jump_functions_for_edge): Use the above functions to
37397 construct bits and vr parts of jump functions.
37398 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37399 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37401 (ipcp_grow_transformations_if_necessary): Also allocate
37402 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37404 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37405 them. Fix too long lines.
37406 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37407 vr_known being removed.
37408 (ipa_read_jump_function): Use new setter functions to construct bits
37409 and vr parts of jump functions or set them to NULL.
37410 (write_ipcp_transformation_info): Adjust for bits being pointers.
37411 (read_ipcp_transformation_info): Likewise.
37412 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37414 Include gt-ipa-prop.h.
37415 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37417 (ipcp_store_bits_results): Likewise.
37418 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37419 Do not write to existing jump functions but use a temporary instead.
37421 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37424 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37425 attempt to use its first operand as BIT_FIELD_REF base.
37427 2017-03-01 Richard Biener <rguenther@suse.de>
37429 PR middle-end/79721
37430 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37431 interpolating formula in wrapping arithmetic.
37432 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37434 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37436 PR tree-optimization/79734
37437 * tree-vect-generic.c (expand_vector_condition): Optimize
37438 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37439 Handle VEC_COND_EXPR where comparison has different inner width from
37440 type's inner width.
37442 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37444 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37445 markup, and similar issues. Remove @opindex entries for things
37446 that aren't options. Add missing -mmpy-option entries.
37448 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37450 PR tree-optimization/79737
37451 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37452 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37453 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37454 instead of byte_size. Formatting fix.
37455 (shift_bytes_in_array_right): Formatting fix.
37457 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37460 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37461 condition on optimize for the leaf function test.
37463 2017-02-28 Martin Liska <mliska@suse.cz>
37466 * read-rtl-function.c (function_reader::handle_unknown_directive):
37467 Bail out when one uses -flto.
37469 2017-02-28 Martin Liska <mliska@suse.cz>
37471 * common.opt: Replace space with tabular for options of <number>
37473 * config/i386/i386.opt: Show <number> value for
37474 -mlarge-data-threshold.
37475 * opts.c (print_filtered_help): Do not display number in hexadecimal
37478 2017-02-28 Martin Liska <mliska@suse.cz>
37480 * common.opt: Fix --help=option -Q for options which are of
37483 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37485 * config/i386/i386.c (print_reg): Error out for values
37486 of 8-bit size in invalid integer register.
37488 2017-02-28 Martin Sebor <msebor@redhat.com>
37490 PR tree-optimization/79691
37491 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37493 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37496 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37497 gcc_unreachable with output_operand_lossage.
37499 2017-02-28 Richard Biener <rguenther@suse.de>
37501 PR tree-optimization/79740
37502 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37504 (visit_nary_op): Insert the nary into the hashtable if we
37505 pattern-matched sth.
37506 * tree-ssa-pre.c (eliminate_insert): Robustify.
37508 2017-02-28 Richard Biener <rguenther@suse.de>
37510 PR middle-end/79731
37511 * fold-const.c (decode_field_reference): Reject out-of-bound
37514 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37516 * config/i386/i386.c: Include intl.h.
37517 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37518 instead of just cond ? "..." : "...".
37519 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37520 * coverage.c (read_counts_file): Likewise.
37521 * omp-offload.c: Include intl.h.
37522 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37523 of just cond ? "..." : "...".
37524 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37525 of just cond ? "..." : "...".
37527 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37530 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37532 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37533 'tune for' CPU name.
37534 * config/arm/arm-cpu-data.h: Regenerated.
37536 2017-02-28 Richard Biener <rguenther@suse.de>
37538 PR tree-optimization/79732
37539 * tree-inline.c (expand_call_inline): Do not shadow var.
37541 2017-02-28 Richard Biener <rguenther@suse.de>
37543 PR tree-optimization/79723
37544 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37545 address-space properly.
37547 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37549 * doc/optinfo.texi (Optimization groups): Fix option used for
37551 * doc/invoke.texi (-fopt-info): Document "omp".
37552 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37553 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37554 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37555 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37556 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37558 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37560 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37561 OPTGROUP_OMP with "omp".
37563 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37566 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37567 for arithmetic shift of unsigned V2DI.
37569 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37571 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37572 arc/linux.h headers.
37573 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37574 (LINK_SPEC): Likewise.
37575 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37576 (EXTRA_SPECS): Likewise.
37577 (STARTFILE_SPEC): Likewise.
37578 (ENDFILE_SPEC): Likewise.
37579 (LIB_SPEC): Likewise.
37580 (TARGET_SDATA_DEFAULT): Likewise.
37581 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37582 (MULTILIB_DEFAULTS): Likewise.
37583 (DWARF2_UNWIND_INFO): Likewise.
37584 * config/arc/big.h: New file.
37585 * config/arc/elf.h: Likewise.
37586 * config/arc/linux.h: Likewise.
37587 * config/arc/t-uClibc: Remove.
37589 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37591 PR tree-optimization/77536
37592 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37593 (tree_transform_and_unroll_loop): Use above function to compute the
37594 estimated niter of unrolled loop and use it when scaling profile.
37595 Also use count info rather than frequency if it's non-zero.
37596 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37597 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37598 (vect_transform_loop): Call above function.
37600 2017-02-27 Richard Biener <rguenther@suse.de>
37602 PR tree-optimization/45397
37603 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37604 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37605 (visit_nary_op): Add pattern matching for CSEing sign-changed
37606 or truncated operations with wider ones.
37608 2017-02-27 Richard Biener <rguenther@suse.de>
37610 PR tree-optimization/79690
37611 * tree-vect-stmts.c (vectorizable_store): Use vector type
37612 built from the DR with address-space.
37614 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37616 * doc/invoke.texi (Optimize Options): Refine the description
37617 of asan-use-after-return.
37619 2017-02-25 Alan Modra <amodra@gmail.com>
37621 PR rtl-optimization/79584
37622 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37623 base, not ad->base_term, the reg within base. Remove assertion
37624 that ad->base == ad->base_term. Replace gen_int_mode using
37625 bogus mode with const0_rtx.
37627 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37629 PR middle-end/79396
37630 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37631 FMA_EXPR like tcc_binary or tcc_unary.
37633 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37636 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37638 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37639 (output_loc_operands): Handle DW_OP_call_ref and
37640 DW_OP_GNU_variable_value.
37641 (struct variable_value_struct): New type.
37642 (struct variable_value_hasher): Likewise.
37643 (variable_value_hash): New variable.
37644 (string_types): Remove.
37645 (copy_loc_descr): New function.
37646 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37647 (prepend_loc_descr_to_each): New function.
37648 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37649 instead of add_loc_descr_to_each if the first argument is single
37650 location list and the second has multiple.
37651 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37652 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37653 when looking for variable value which doesn't have other location info.
37654 (loc_list_from_tree): Formatting fix.
37655 (gen_array_type_die): Simplify DW_AT_string_length handling.
37656 (adjust_string_types): Remove.
37657 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37658 string_types. Call resolve_variable_values.
37659 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37660 (resolve_addr_in_expr): Likewise. Add A argument.
37661 (copy_deref_exprloc): Remove deref argument. Adjust for the
37662 original expression being DW_OP_GNU_variable_value with optionally
37663 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37664 optionally after it.
37665 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37666 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37667 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37668 (variable_value_hasher::hash, variable_value_hasher::equal): New
37670 (resolve_variable_value_in_expr, resolve_variable_value,
37671 resolve_variable_values, note_variable_value_in_expr,
37672 note_variable_value): New functions.
37673 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37676 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37679 * opts.h (handle_generated_option): Add GENERATED_P argument.
37680 * opts-common.c (handle_option): Adjust function comment.
37681 (handle_generated_option): Add GENERATED_P argument, pass it to
37683 (control_warning_option): Pass false to handle_generated_option
37685 * opts.c (maybe_default_option): Pass true to handle_generated_option
37687 * optc-gen.awk: Likewise.
37689 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37691 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37692 a REG, look at the REG it is a SUBREG of.
37693 (splitter for cmpeqsi_t): Ditto.
37695 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37697 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37698 the special USEs with the pattern of the insn, not the insn itself.
37700 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37703 * doc/invoke.texi: Document -mload-store-pairs.
37705 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37706 Sandra Loosemore <sandra@codesourcery.com>
37708 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37709 argument isn't a CONST_INT.
37710 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37711 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37712 (nios2_validate_compare): Bypass alternate compare logic if *op2
37713 is not a CONST_INT.
37714 (ldstwm_operation_p): Return false if first_base is not a REG or
37715 if first_offset is not a CONST_INT.
37717 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37719 * config/cris/cris.md: Use correct operand in a define_peephole2.
37721 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37723 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37725 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37727 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37728 this_insn if it is an INSN or JUMP_INSN.
37729 (force_offsettable): Look at base, not at addr.
37730 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37731 on things that aren't necessarily CONST_INTs.
37733 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37735 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37736 -mfpmath=sse is the default also for x86-32 targets with SSE2
37737 instruction set when @option{-ffast-math} is enabled
37739 2017-02-24 Jeff Law <law@redhat.com>
37741 PR rtl-optimizatoin/79286
37742 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37745 2017-02-24 Richard Biener <rguenther@suse.de>
37747 PR tree-optimization/79389
37748 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37751 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37753 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37754 function comment to reflect reality.
37755 (loop_exits_before_overflow): Fix typo in function description.
37757 2017-02-24 Richard Biener <rguenther@suse.de>
37759 PR tree-optimization/79389
37760 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37761 properly that a threading opportunity exists. Detect conditional
37762 copy/constant propagation opportunities.
37764 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37766 * config/visium/visium.md (type): Add trap.
37767 (b): New mode attribute.
37768 (*btst): Rename into...
37769 (*btst<mode>): ...this and adjust.
37770 (*cbranchsi4_btst_insn): Rename into...
37771 (*cbranch<mode>4_btst_insn): ...this and adjust.
37772 (trap): New define_insn.
37774 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37776 PR tree-optimization/79389
37777 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37778 (noce_reversed_cond_code): New function.
37779 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37780 reversed_comparison_code. Formatting fix.
37781 (noce_try_store_flag): Test rev_cond != NULL in addition to
37782 reversed_comparison_code.
37783 (noce_try_store_flag_constants): Likewise.
37784 (noce_try_store_flag_mask): Likewise.
37785 (noce_try_addcc): Use rev_cond if non-NULL instead of
37786 reversed_comparison_code.
37787 (noce_try_cmove_arith): Likewise. Formatting fixes.
37788 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37789 (noce_find_if_block): Initialize rev_cond.
37790 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37791 instead of false as last argument never attempt to reverse it
37794 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37796 PR tree-optimization/79663
37797 * tree-predcom.c (combine_chains): Process refs in reverse order
37798 only for ZERO length chains, and add explaining comment.
37800 2017-02-23 Jeff Law <law@redhat.com>
37802 PR tree-optimization/79578
37803 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37804 in call to operand_equal_p.
37806 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37809 * config/i386/cpuid.h: Fix another undefined behavior.
37811 2017-02-23 Richard Biener <rguenther@suse.de>
37813 PR tree-optimization/79683
37814 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37815 vector types for data-refs.
37817 2017-02-23 Martin Liska <mliska@suse.cz>
37819 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37821 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37823 PR middle-end/79665
37824 * internal-fn.c (get_range_pos_neg): Moved to ...
37825 * tree.c (get_range_pos_neg): ... here. No longer static.
37826 * tree.h (get_range_pos_neg): New prototype.
37827 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37828 are known to be in between 0 and signed maximum inclusive, try to
37829 expand both unsigned and signed divmod and use the cheaper one from
37832 2017-02-22 Jeff Law <law@redhat.com>
37834 PR tree-optimization/79578
37835 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37836 to compare base operands.
37838 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37841 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37842 gpc_reg_operand instead of fpr_reg_operand.
37844 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37846 * config/mips/mips.c (mips_return_in_memory): Force FP
37847 vector types to be returned in memory for o32 ABI.
37849 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37851 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37852 instead of DW_TAG_member for static data member declarations and don't
37853 set no_linkage_name for static inline data members.
37854 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37857 2017-02-22 Martin Liska <mliska@suse.cz>
37859 * doc/invoke.texi: Replace inequality signs with square brackets
37862 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37865 * lra-constraints.c (simplify_operand_subreg): Handle
37866 WORD_REGISTER_OPERATIONS targets.
37868 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37871 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37872 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37873 elimination by swapping fld*.
37875 2017-02-22 Richard Biener <rguenther@suse.de>
37877 PR tree-optimization/79673
37878 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37879 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37880 irrelevant address-space qualifiers and avoiding a
37881 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37883 2017-02-22 Richard Biener <rguenther@suse.de>
37885 PR tree-optimization/79666
37886 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37887 to not symbolically negate if that may introduce undefined
37890 2017-02-22 Martin Liska <mliska@suse.cz>
37893 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37894 * data-streamer-out.c (streamer_write_gcov_count_stream):
37896 * value-prof.c (stream_out_histogram_value): Make assert more
37897 precise based on type of counter.
37899 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37902 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37903 Use nonimmediate_operand instead of memory_operand for operand 1.
37904 (float-extend standard_x87sse_constant_load splitter): Ditto.
37906 2017-02-21 Jeff Law <law@redhat.com>
37908 PR tree-optimization/79621
37909 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37910 blocks with edges to themselves.
37912 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37915 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37916 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37917 Use gimple_call_builtin_p.
37920 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37921 on temporarily removed DEBUG_INSNs.
37923 PR tree-optimization/79649
37924 * tree-loop-distribution.c (classify_partition): Give up on
37925 non-generic address space loads/stores.
37927 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37929 * doc/loop.texi (Loop manipulation): Remove nonexistent
37930 tree_ssa_loop_version from the documentation.
37931 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37933 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37936 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37937 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37938 * config/rs6000/rs6000.c: Include except.h.
37939 (rs6000_expand_split_stack_prologue): Call
37940 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37942 2017-02-21 Martin Jambor <mjambor@suse.cz>
37945 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37946 have been analyzed.
37948 2017-02-21 Martin Jambor <mjambor@suse.cz>
37950 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37951 for backward compatibility only.
37952 * doc/invoke.texi (Option Summary): Remove all references to
37953 -fipa-cp-alignment.
37955 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37959 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37961 * lra-constraints.c (curr_insn_transform): Handle
37962 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37964 2017-02-21 Martin Liska <mliska@suse.cz>
37966 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37968 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37970 PR translation/79638
37971 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37973 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37976 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37977 (arm_function_ok_for_sibcall): Return false for an indirect call by
37978 descriptor if all the argument registers are used.
37979 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37980 alignment of the function.
37982 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37984 PR tree-optimization/61441
37985 * simplify-rtx.c (simplify_const_unary_operation): For
37986 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37987 the sNaN unmodified.
37989 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37991 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37992 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37993 instead of SYSTEM_HEADER_DIR.
37995 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37996 Martin Liška <mliska@suse.cz>
37998 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37999 Fix typos and grammar, use active voice, and clarify.
38001 2017-02-20 Marek Polacek <polacek@redhat.com>
38003 PR middle-end/79537
38004 * gimplify.c (gimplify_expr): Handle unused *&&L;.
38007 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
38009 2017-02-20 Jakub Jelinek <jakub@redhat.com>
38012 * config/i386/i386.c (ix86_expand_builtin): Handle
38013 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
38014 ix86_builtins_isa[fcode].isa as a requirement of those
38015 flags and any other flag in the bitmask.
38016 (ix86_init_mmx_sse_builtins): Use 0 instead of
38017 ~OPTION_MASK_ISA_64BIT as mask.
38018 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
38019 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
38020 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
38021 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
38023 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38026 * lra-constraints.c (split_reg): Check requested split mode
38027 is supported by the register.
38029 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38031 * lra-constraints.c (simplify_operand_subreg): Remove early
38034 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38037 * lra-constraints.c (curr_insn_transform): Tighten condition
38038 for converting SUBREG reloads from OP_OUT to OP_INOUT.
38040 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38043 * lra-constraints.c (curr_insn_transform): Handle
38044 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38046 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
38049 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
38051 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
38053 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
38056 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38059 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38061 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38064 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38066 * final.c (last_columnnum, override_columnnum): New variables.
38067 (final_start_function): Set last_columnnum, pass it to begin_prologue
38068 hook and pass 0 to dwarf2out_begin_prologue.
38069 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38070 to source_line debug hook.
38071 (notice_source_line): Compute last_columnnum and for debug_column_info
38072 return true on column changes.
38073 * debug.h (struct gcc_debug_hooks): Add column argument to
38074 source_line and begin_prologue hooks.
38075 (debug_nothing_int_charstar_int_bool): Remove prototype.
38076 (debug_nothing_int_int_charstar,
38077 debug_nothing_int_int_charstar_int_bool): New prototypes.
38078 (dwarf2out_begin_prologue): Add column argument.
38079 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38080 begin_prologue hooks.
38081 (debug_nothing_int_charstar_int_bool): Remove.
38082 (debug_nothing_int_int_charstar,
38083 debug_nothing_int_int_charstar_int_bool): New functions.
38084 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38085 through to dwarf2out_source_line.
38086 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38087 (dwarf2out_source_line): Add column argument, emit it if requested.
38088 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38090 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38091 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38092 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38093 through to dwarf2out_begin_prologue.
38094 (vmsdbgout_source_line): Add column argument, pass it through to
38095 dwarf2out_source_line.
38096 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38097 dbxout_source_line caller.
38098 (dbxout_source_line): Add column argument.
38100 * common.opt (gno-column-info, gcolumn-info): New options.
38101 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38102 (check_die): Also test for multiple DW_AT_decl_column attributes.
38103 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38104 DW_AT_decl_column if requested.
38105 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38107 (gen_variable_die): Likewise.
38108 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38109 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38112 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38113 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38114 (ix86_handle_option): Handle OPT_m3dnowa.
38115 * doc/invoke.texi (-m3dnowa): Document.
38116 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38117 -m3dnowa instead of -m3dnow -march=athlon.
38120 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38121 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38123 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38126 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38127 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38128 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38129 generator for vsx_xxpermdi_<mode>_be.
38130 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38131 force big-endian semantics.
38132 (vsx_xxpermdi_<mode>_be): New define_expand with same
38133 implementation as previous version of vsx_xxpermdi_<mode>.
38135 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38137 PR tree-optimization/79327
38138 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38139 variable, its initialization and use.
38141 2017-02-17 Julia Koval <julia.koval@intel.com>
38143 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38144 (OPTION_MASK_ISA_PKU_UNSET): New.
38145 (ix86_handle_option): Handle -mrdpid.
38146 * config/i386/cpuid.h (bit_RDPID): New.
38147 * config/i386/driver-i386.c (host_detect_local_cpu):
38148 Detect RDPID feature.
38149 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38150 * config/i386/i386-c.c (ix86_target_macros_internal):
38152 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38153 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38154 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38155 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38156 * config/i386/i386.md (define_insn "rdpid"): New.
38157 * config/i386/i386.opt Add -mrdpid.
38158 * config/i386/immintrin.h (_rdpid_u32): New.
38160 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38162 PR rtl-optimization/79541
38163 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38164 instead of transforming it into USE.
38166 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38168 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38169 If HONOR_SNANS (SFmode) force the input to a register.
38170 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38171 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38172 an frsp or similar insn.
38174 2017-02-17 Martin Liska <mliska@suse.cz>
38176 PR rtl-optimization/79577
38177 * params.def (selsched-max-sched-times): Increase minimum to 1.
38179 2017-02-17 Martin Liska <mliska@suse.cz>
38181 PR rtl-optimization/79574
38182 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38184 2017-02-17 Martin Liska <mliska@suse.cz>
38186 PR tree-optimization/79529
38187 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38188 ssa_defined_default_def_p to handle cases which are implicitly
38190 * tree-ssa.c (ssa_defined_default_def_p): New function.
38191 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38192 which are implicitly defined.
38193 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38195 2017-02-17 Richard Biener <rguenther@suse.de>
38197 PR middle-end/79576
38198 * params.def (max-ssa-name-query-depth): Limit to 10.
38200 2017-02-17 Richard Biener <rguenther@suse.de>
38202 PR tree-optimization/79552
38203 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38206 2017-02-17 Richard Biener <rguenther@suse.de>
38209 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38211 2017-02-17 Marek Polacek <polacek@redhat.com>
38213 PR middle-end/79536
38214 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38215 (fold_negate_expr): New wrapper.
38217 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38219 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38220 Correct terminology and de-emphasize pre-standard behavior.
38222 2017-02-16 Alan Modra <amodra@gmail.com>
38224 PR rtl-optimization/79286
38225 * ira.c (def_dominates_uses): New function.
38226 (update_equiv_regs): Don't create an equivalence for insns that
38227 may trap where the register def does not dominate the use.
38229 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38231 PR rtl-optimization/78127
38232 * lra.c (lra): Call lra_eliminate before finish the loop after
38235 2017-02-16 Richard Biener <rguenther@suse.de>
38237 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38239 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38240 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38241 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38242 (isl_val_int_from_wi): New function.
38243 (extract_affine_gmp): Rename to ...
38244 (extract_affine_wi): ... this, take a widest_int.
38245 (extract_affine_int): Just wrap extract_affine_wi.
38246 (add_param_constraints): Use isl_val_int_from_wi.
38247 (add_loop_constraints): Likewise, and extract_affine_wi.
38249 2017-02-15 Jeff Law <law@redhat.com>
38251 PR middle-end/79521
38252 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38253 ira_init_register_move_cost_if_necessary.
38255 2017-02-15 Martin Sebor <msebor@redhat.com>
38257 PR middle-end/32003
38258 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38259 removed in a prior commit.
38261 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38263 PR tree-optimization/79347
38264 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38265 counters during peeling.
38267 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38269 * Makefile.in (site.exp): Remove "set ISLVER".
38271 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38274 * real.c (real_from_integer): Call real_convert even for decimal.
38276 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38279 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38281 2017-02-14 Andrew Pinski <apinski@cavium.com>
38283 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38284 cores and change the partno/implementer to be correct.
38285 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38286 the 'B" as the implementer.
38287 * config/aarch64/aarch64-tune.md: Regenerate.
38289 2017-02-14 Carl Love <cel@us.ibm.com>
38291 * config/rs6000/rs6000.c: Add case statement entry to make the
38292 xvcvuxdsp built-in argument unsigned.
38293 * config/rs6000/vsx.md: Fix the source and return operand types so they
38294 match the instruction definitions from the ISA document. Fix typo
38295 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38298 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38301 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38302 member early_clobber_alts.
38303 * lra-lives.c (reg_early_clobber_p): New.
38304 (process_bb_lives): Use it.
38305 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38306 (debug_operand_data): Initialize early_clobber_alts.
38307 (setup_operand_alternative): Set up early_clobber_alts.
38308 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38309 alternatives to new_insn_reg.
38310 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38312 (lra_update_insn_regno_info): Pass the new arg.
38314 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38316 PR middle-end/79505
38317 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38318 (new_oacc_loop_raw): Don't clear already cleared fields.
38321 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38322 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38323 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38325 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38328 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38330 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38333 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38334 the extra instruction to the right place to store 128-bit constant
38337 2017-02-14 Martin Sebor <msebor@redhat.com>
38339 PR middle-end/79448
38340 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38341 warning for strings of unknown length.
38343 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38345 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38347 2017-02-14 Jeff Law <law@redhat.com>
38350 * ira-costs.c (scan_one_insn): Initialize register move costs
38351 for pseudos seen in USE/CLOBBER insns.
38353 PR tree-optimization/79095
38354 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38355 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38356 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38357 if the operands are known to be not equal, then the resulting range
38359 (intersect_ranges): If the new range is ~[0,0] and the old range is
38360 wide, then prefer ~[0,0].
38361 * tree-vrp.c (overflow_comparison_p_1): New function.
38362 (overflow_comparison_p): New function.
38363 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38364 if NAME is used in an overflow test.
38365 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38366 overflow check that can be expressed as an equality test, then adjust
38367 ops to be that equality test.
38369 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38371 * config/s390/s390-builtin-types.def: Remove flags argument.
38372 * config/s390/s390.c (s390_init_builtins): Likewise.
38374 2017-02-14 Martin Liska <mliska@suse.cz>
38376 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38377 vector. Fix trailing white spaces.
38379 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38381 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38384 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38386 PR rtl-optimization/68664
38387 * config/arm/arm.c (arm_sched_can_speculate_insn):
38388 New function. Declare prototype.
38389 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38391 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38393 PR rtl-optimization/68664
38394 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38396 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38398 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38400 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38401 max skip bytes for function, loop and jump.
38403 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38405 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38406 ABS_EXPR for gimple dump.
38408 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38411 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38413 PR tree-optimization/79408
38414 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38415 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38416 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38417 also if rhs1 is INTEGER_CST.
38419 2017-02-14 Richard Biener <rguenther@suse.de>
38421 PR middle-end/79432
38422 * tree-into-ssa.c (insert_phi_nodes): When the function can
38423 have abnormal edges rewrite SSA names with broken use-def
38424 dominance out of SSA and register them for PHI insertion.
38426 2017-02-13 Martin Sebor <msebor@redhat.com>
38428 PR middle-end/79496
38429 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38430 clearing info.nowrite flag when snprintf size argument is a range.
38432 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38434 * cprop.c (cprop_jump): Add missing space in string literal.
38435 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38436 (get_constraint_for_component_ref): Likewise.
38437 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38438 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38439 * lra-constraints.c (process_alt_operands): Likewise.
38440 * ipa-inline.c (inline_small_functions): Likewise.
38441 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38442 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38443 * trans-mem.c (diagnose_tm_1_op): Likewise.
38444 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38445 (grid_parallel_clauses_gridifiable): Likewise.
38447 * config/nvptx/mkoffload.c (process): Add space in between
38450 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38451 "MOD4_SSE_REGS" and "ALL_REGS".
38453 * spellcheck.c (test_data): Add , in between "foo" and "food".
38455 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38458 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38459 boundary crossing check and subsequent code generation agree.
38461 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38463 * config/aarch64/aarch64.c (has_memory_op): Delete.
38464 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38467 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38469 PR rtl-optimization/79388
38470 PR rtl-optimization/79450
38471 * combine.c (distribute_notes): When removing TEM_INSN for which
38472 corresponding dest has last value recorded, invalidate that last
38475 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38477 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38478 of explicit '@'. Add missing assembly comment marker on branch costs
38481 2017-02-13 Nathan Sidwell <nathan@acm.org>
38483 * gengtype-lex.l (<in_struct>): Add '/'.
38485 2017-02-13 Martin Liska <mliska@suse.cz>
38488 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38490 2017-02-13 Richard Biener <rguenther@suse.de>
38492 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38494 * configure: Re-generate.
38495 * config.in: Likewise.
38496 * graphite-dependences.c: Simplify as if
38497 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38498 * graphite-isl-ast-to-gimple.c: Likewise.
38499 * graphite-optimize-isl.c: Likewise.
38500 * graphite-poly.c: Likewise.
38501 * graphite-sese-to-poly.c: Likewise.
38502 * graphite.h: Likewise.
38503 * toplev.c: Include isl/version.h and use isl_version () for
38504 printing the ISL version.
38505 * doc/install.texi: Update ISL requirement.
38507 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38509 * doc/standards.texi (Standards): Update reference to
38512 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38514 * doc/extend.texi (Named Address Spaces): sourceware.org now
38516 * doc/install.texi (Binaries): Ditto.
38519 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38521 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38522 standard terminology "stringize"/"stringizing" throughout.
38523 * doc/cppinternals.texi: Likewise.
38525 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38527 * doc/extend.texi: Fix some spelling mistakes and typos.
38528 * doc/invoke.texi: Likewise.
38530 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38533 * params.def (inline-min-speedup) Change from 10 to 8.
38535 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38537 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38540 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38543 * ipa-inline-analysis.c (get_minimal_bb): New function.
38544 (record_modified): Use it.
38545 (remap_edge_change_prob): Handle also ancestor functions.
38547 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38549 * doc/contrib.texi (Contributors): Remove broken link into
38550 the Mauve CVS repository.
38552 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38554 PR middle-end/79454
38555 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38556 result computation whenever lhs doesn't have vector mode, not
38557 just when it has BLKmode.
38559 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38561 * doc/makefile.texi (profiledbootstrap): Refer to the
38562 installation instructions only in textual form.
38564 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38567 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38569 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38571 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38572 (Specific): Update mingw-w64 reference.
38574 (Specific): Remove broken link to Renesas RX processor.
38576 2017-02-10 Richard Biener <rguenther@suse.de>
38578 * toplev.c (process_options): Do not mention obsolete graphite
38579 options when printing sorry message about missing graphite support.
38580 Mention -floop-nest-optimize.
38582 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38584 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38585 (vtst_p16): Likewise.
38586 (vtstq_p8): Likewise.
38587 (vtstq_p16): Likewise.
38589 (vtstq_p64): Likewise.
38590 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38591 (vset_lane_p64): New.
38592 (vsetq_lane_p64): New.
38594 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38596 PR tree-optimization/79411
38597 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38598 stmt operands are SSA_NAMEs used in abnormal phis.
38599 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38602 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38605 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38608 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38610 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38612 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38614 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38618 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38619 not arbitrary TREE_CONSTANT.
38622 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38623 "omp declare target link" attribute unless is_global_var.
38624 * omp-offload.c (find_link_var_op): Likewise.
38626 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38627 Chung-Lin Tang <cltang@codesourcery.com>
38629 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38631 (gimplify_adjust_omp_clauses): Don't delete TILE.
38632 (gimplify_omp_for): Deal with TILE.
38633 * internal-fn.c (expand_GOACC_TILE): New function.
38634 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38636 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38637 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38639 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38640 avoid DIV for outermost collapse var.
38641 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38642 Remove out of date comments, fix whitespace.
38643 * omp-general.c (omp_extract_for_data): Deal with tiling.
38644 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38645 adjust OLF_DIM_BASE value.
38646 (struct omp_for_data): Add tiling field.
38647 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38648 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38649 for auto loops. Remove default auto determining, moved to
38650 oacc_loop_fixed_partitions.
38651 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38652 stmts, add e_mask field.
38653 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38654 (oacc_thread_numbers): Use oacc_dim_call.
38655 (oacc_xform_tile): New.
38656 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38657 (finish_oacc_loop): Adjust for ifns vector.
38658 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38659 add case for GOACC_TILE fns.
38660 (oacc_loop_xform_loop): Delete.
38661 (oacc_loop_process): Iterate over call list directly, and add
38662 handling for GOACC_TILE fns.
38663 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38665 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38666 vector partitioning to outer loops. Assign 2 partitions to loops
38667 when available. Add TILE handling.
38668 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38669 (execite_oacc_device_lower): Process GOACC_TILE fns,
38670 ignore unknown specs.
38671 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38672 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38673 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38675 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38677 * configure.ac (ACX_BUGURL): Update.
38678 * configure: Regenerate.
38680 2017-02-09 Richard Biener <rguenther@suse.de>
38682 PR tree-optimization/69823
38683 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38684 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38686 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38688 * config/arc/arc-c.def: Add __NPS400__ definition.
38689 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38690 (TARGET_NPS400): Define.
38692 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38694 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38696 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38697 pointer, arch_info.
38698 (arc_cpu_types): Fill the arch_info field with a pointer into the
38699 arc_arch_types table.
38700 (arc_selected_cpu): Declare.
38701 * config/arc/arc.c (arc_selected_cpu): Make global.
38702 (arc_selected_arch): Delete.
38703 (arc_base_cpu): Delete.
38704 (arc_override_options): Remove references to deleted variables,
38705 update access to arch information.
38706 (ARC_OPT): Update access to arch information.
38707 (ARC_OPTX): Likewise.
38708 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38709 (TARGET_ARC600): Update access to arch information.
38710 (TARGET_ARC601): Likewise.
38711 (TARGET_ARC700): Likewise.
38712 (TARGET_EM): Likewise.
38713 (TARGET_HS): Likewise.
38714 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38717 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38720 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38721 condition/operands for integer GE/LE/GEU/LEU operations.
38723 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38725 PR translation/79397
38726 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38729 2017-02-08 Martin Jambor <mjambor@suse.cz>
38732 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38733 whether allocation happened.
38734 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38735 nothing was allocated.
38737 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38739 PR tree-optimization/79408
38740 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38741 constant, but SSA_NAME with a known integer range, use the minimum
38742 of that range instead of op1 to determine if modulo can be replaced
38743 with its first operand.
38745 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38747 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38749 2017-02-08 Richard Biener <rguenther@suse.de>
38751 PR tree-optimization/71824
38752 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38753 Check all loops contained in the merged region.
38755 2017-02-07 Andrew Pinski <apinski@cavium.com>
38757 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38759 2017-02-07 Andrew Pinski <apinski@cavium.com>
38761 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38762 (thunderxt88): Likewise.
38763 (thunderxt81): Disable LSE and change v8.1 to v8.
38764 (thunderxt83): Likewise.
38766 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38767 Richard Biener <rguenther@suse.de>
38769 PR middle-end/79399
38770 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38771 type from int to size_t.
38772 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38774 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38776 PR rtl-optimization/79386
38777 * cprop.c (bypass_conditional_jumps): Initialize
38778 bypass_last_basic_block already before splitting bbs after
38779 unconditional traps...
38780 (bypass_conditional_jumps): ... rather than here.
38783 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38784 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38785 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38786 fix -masm=intel patterns.
38788 2017-02-07 Richard Biener <rguenther@suse.de>
38790 PR tree-optimization/79256
38791 PR middle-end/79278
38792 * builtins.c (get_object_alignment_2): Use min_align_of_type
38793 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38794 and ADJUST_FIELD_ALIGN.
38796 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38798 * doc/tm.texi: Regenerate.
38799 * stor-layout.c (layout_decl): Adjust.
38800 (update_alignment_for_field): Likewise.
38801 (place_field): Likewise.
38802 (min_align_of_type): Likewise.
38803 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38804 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38805 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38806 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38807 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38808 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38809 * config/i386/i386.c (x86_field_alignment): Likewise.
38810 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38811 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38812 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38813 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38814 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38815 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38819 2017-01-30 Richard Biener <rguenther@suse.de>
38821 PR tree-optimization/79256
38822 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38823 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38826 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38828 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38829 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38830 builtins to SImode and emit a zero-extend, if necessary.
38832 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38834 * docs/invoke.texi (RISC-V Options): Alphabetize.
38836 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38838 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38841 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38843 * config/riscv/riscv.c: New file.
38844 * common/config/riscv/riscv-common.c: Likewise.
38845 * config.gcc: Likewise.
38846 * config/riscv/constraints.md: Likewise.
38847 * config/riscv/elf.h: Likewise.
38848 * config/riscv/generic.md: Likewise.
38849 * config/riscv/linux.h: Likewise.
38850 * config/riscv/multilib-generator: Likewise.
38851 * config/riscv/peephole.md: Likewise.
38852 * config/riscv/pic.md: Likewise.
38853 * config/riscv/predicates.md: Likewise.
38854 * config/riscv/riscv-builtins.c: Likewise.
38855 * config/riscv/riscv-c.c: Likewise.
38856 * config/riscv/riscv-ftypes.def: Likewise.
38857 * config/riscv/riscv-modes.def: Likewise.
38858 * config/riscv/riscv-opts.h: Likewise.
38859 * config/riscv/riscv-protos.h: Likewise.
38860 * config/riscv/riscv.h: Likewise.
38861 * config/riscv/riscv.md: Likewise.
38862 * config/riscv/riscv.opt: Likewise.
38863 * config/riscv/sync.md: Likewise.
38864 * config/riscv/t-elf-multilib: Likewise.
38865 * config/riscv/t-linux: Likewise.
38866 * config/riscv/t-linux-multilib: Likewise.
38867 * config/riscv/t-riscv: Likewise.
38868 * configure.ac: Likewise.
38869 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38870 Waterman as RISC-V maintainers.
38871 * doc/install.texi: Add RISC-V entries.
38872 * doc/invoke.texi: Add RISC-V options section.
38873 * doc/md.texi: Add RISC-V constraints section.
38874 * configure: Regenerated.
38876 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38879 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38880 false values to be constant vectors with all 0 or all 1 bits set.
38881 (vcondu<mode><mode>): Likewise.
38882 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38884 (fpmask_comparison_operator): Update comment.
38885 (vecint_comparison_operator): New predicate.
38886 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38887 vector conditionals when the true and false values are constant
38888 vectors with all 0 bits or all 1 bits set.
38890 2017-02-06 Martin Sebor <msebor@redhat.com>
38892 PR tree-optimization/79376
38893 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38895 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38897 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38898 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38899 to simplify split condition.
38901 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38903 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38904 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38907 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38909 PR rtl-optimization/68664
38910 * target.def (can_speculate_insn): New hook.
38911 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38912 * doc/tm.texi: Regenerate.
38913 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38914 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38915 (rs6000_sched_can_speculate_insn): New function.
38917 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38919 PR tree-optimization/79284
38920 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38921 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38922 vectorizable_mask_load_store, vectorizable_operation,
38923 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38924 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38925 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38926 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38927 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38928 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38929 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38930 is_gimple_assign (stmt). Replace another such test with
38931 is_gimple_assign (stmt).
38933 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38936 * config/avr/avr.c (rtl-iter.h): Include it.
38937 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38938 (avr_legitimate_combined_insn): ...and implementation.
38940 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38942 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38943 * config/s390/s390.c (s390_const_operand_ok)
38944 (s390_canonicalize_comparison, s390_extract_part)
38945 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38946 (s390_contiguous_bitmask_p, s390_rtx_costs)
38947 (legitimize_pic_address): Likewise.
38948 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38949 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38950 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38951 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38952 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38954 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38956 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38957 REGNO($0) == REGNO($1).
38959 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38961 * config/s390/linux.h(SIZE_TYPE): Add comment.
38963 2017-02-06 Julian Brown <julian@codesourcery.com>
38964 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38965 Virendra Pathak <virendra.pathak@broadcom.com>
38967 * config/aarch64/aarch64-cores.def: Change the scheduler
38969 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38970 * config/aarch64/thunderx2t99.md: New file.
38972 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38974 * doc/standards.texi (Go Language): Update link to language
38977 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38979 * tree-eh.c (lower_resx): Sanitize profile.
38980 (cleanup_empty_eh_move_lp): Likewise.
38982 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38985 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38987 * cfgloopmanip.h (loop_version): Update prototype.
38988 * modulo-sched.c (sms_schedule): Update call of loop_version.
38989 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38990 * tree-parloops.c (gen_parallel_loop): Likewise.
38991 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38992 * tree-ssa-loop-split.c (split_loop): Likewise.
38993 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38994 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38996 2017-02-05 Martin Liska <mliska@suse.cz>
38999 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
39001 (print_operand_address): Initialize a struct to zero.
39003 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39005 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
39006 garbage collector only in textual form.
39008 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39010 * doc/extend.texi (x86 specific memory model extensions for
39011 transactional memory): Simplify a phrase.
39013 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
39016 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
39017 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
39018 (atomic_storedi_1): Likewise.
39020 2017-02-04 Jakub Jelinek <jakub@redhat.com>
39022 PR tree-optimization/79338
39023 * tree-parloops.c (gather_scalar_reductions): Don't call
39024 vect_analyze_loop_form for loop->inner before destroying loop's
39027 2017-02-03 Martin Sebor <msebor@redhat.com>
39029 PR tree-optimization/79327
39030 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
39031 when precision has resulted in leading zeros.
39032 (format_integer): Adjust the likely counter to assume an unknown
39033 argument that may be zero is non-zero.
39035 2017-02-03 Jason Merrill <jason@redhat.com>
39038 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
39039 avoid copying non-taken branch.
39041 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39043 PR tree-optimization/79340
39044 * tree-vect-loop.c (vectorizable_reduction): Release
39045 vec_defs elements after safe_splicing them into other vectors.
39048 PR tree-optimization/79327
39049 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
39050 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
39052 (format_integer): Use wide_int_to_tree instead of build_int_cst
39053 + to_?hwi. If argmin is NULL, just set argmin and argmax to
39054 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
39055 of shortest and longest sequence.
39057 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39059 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39060 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39062 2017-02-03 Walter Lee <walt@tilera.com>
39065 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39066 after initial stackframe link reg save.
39067 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39069 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39072 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39073 wu for stxssp alternative.
39075 2017-02-03 Martin Sebor <msebor@redhat.com>
39077 PR tree-optimization/79352
39078 * gimple-fold.c (get_range_strlen): Add argument.
39079 (get_range_strlen): Change return type to bool.
39080 (get_maxval_strlen): Pass in a dummy argument.
39081 * gimple-fold.h (get_range_strlen): Change return type to bool.
39082 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39083 * tree.h (array_at_struct_end_p): Add argument.
39084 * tree.c (array_at_struct_end_p): Handle it.
39086 2017-02-03 Martin Liska <mliska@suse.cz>
39089 * multiple_target.c (create_dispatcher_calls): Redirect edge
39090 from a caller of a dispatcher.
39091 (expand_target_clones): Make the clones local.
39092 (ipa_target_clone): Do both target clones and resolvers.
39093 (ipa_dispatcher_calls): Remove the pass.
39094 (pass_dispatcher_calls::gate): Likewise.
39095 (make_pass_dispatcher_calls): Likewise.
39096 * passes.def (pass_target_clone): Put as very first IPA early
39099 2017-02-03 Martin Liska <mliska@suse.cz>
39101 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39102 in case of a function with ifunc attribute.
39104 2017-02-03 Martin Liska <mliska@suse.cz>
39106 * cgraph.c (cgraph_node::dump): Dump function version info.
39107 * symtab.c (symtab_node::dump_base): Add missing new line.
39109 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39111 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39112 (ifcombine_ifandif): Use it.
39114 2017-02-03 Martin Liska <mliska@suse.cz>
39116 * doc/invoke.texi: Document default value for
39117 use-after-scope-direct-emission-threshold.
39119 2017-02-03 Martin Liska <mliska@suse.cz>
39121 PR tree-optimization/79339
39122 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39123 (format_floating): Likewise.
39125 2017-02-03 Martin Liska <mliska@suse.cz>
39128 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39130 (ipa_node_params_t::remove): Likewise.
39131 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39132 initialization from removed ipa_node_params_t::insert.
39133 (ipa_node_params::~ipa_node_params): Move from removed
39134 ipa_node_params_t::release.
39135 * symbol-summary.h (symbol_summary::m_released): New member.
39136 Do not release a summary twice. Do not allow to call finalizer
39137 for types of a summary that live in GGC memory.
39139 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39141 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39144 2017-02-02 Martin Sebor <msebor@redhat.com>
39146 PR middle-end/79275
39147 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39148 (format_string): Tighten up the range of output for non-constant
39149 strings and correct the expected range for wide non-constant strings.
39151 2017-02-02 Martin Sebor <msebor@redhat.com>
39153 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39155 PR middle-end/32003
39156 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39158 (-fdump-tree-@var): Add to index and document how to come up
39159 with pass-specific option and dump file names.
39160 (-fdump-passes): Clarify where to look for output.
39162 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39164 PR middle-end/77445
39165 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39166 statistics of the analyzed path; allow threading for speed when
39167 any of BBs along the path are optimized for speed.
39169 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39171 PR middle-end/78468
39172 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39173 settings of the virtual registers.
39176 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39178 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39179 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39182 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39184 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39185 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39187 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39189 * config/s390/s390.md: Add missing comments with the expanded
39191 * config/s390/vector.md: Likewise.
39192 * config/s390/vx-builtins.md: Likewise.
39194 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39197 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39198 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39199 conditions on a single line.
39201 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39203 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39204 __S390_VX__ to __VX__.
39206 2017-02-01 Andrew Pinski <apinski@cavium.com>
39208 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39209 stmt_info to record_stmt_cost.
39210 (vect_get_known_peeling_cost): Pass stmt_info if known to
39212 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39213 cpu_vector_cost field into
39214 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39215 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39216 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39217 splitting of scalar_stmt_cost and vec_stmt_cost.
39218 (thunderx_vector_cost): Likewise.
39219 (cortexa57_vector_cost): LIkewise.
39220 (exynosm1_vector_cost): Likewise.
39221 (xgene1_vector_cost): Likewise.
39222 (thunderx2t99_vector_cost): Improve after the splitting of the two
39224 (aarch64_builtin_vectorization_cost): Update for the splitting of
39225 scalar_stmt_cost and vec_stmt_cost.
39227 2017-02-01 Torvald Riegel <triegel@redhat.com>
39228 Richard Henderson <rth@redhat.com>
39230 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39231 conditional on existance of a fast atomic load.
39232 * optabs-query.c (can_atomic_load_p): New function.
39233 * optabs-query.h (can_atomic_load_p): Declare it.
39234 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39235 no fast atomic load is available for the particular size of access.
39236 (expand_atomic_compare_and_swap): Likewise.
39237 (expand_atomic_load): Likewise.
39238 (expand_atomic_store): Likewise.
39239 (expand_atomic_fetch_op): Likewise.
39240 * testsuite/lib/target-supports.exp
39241 (check_effective_target_sync_int_128): Remove x86 because it provides
39242 no fast atomic load.
39243 (check_effective_target_sync_int_128_runtime): Likewise.
39245 2017-02-01 Richard Biener <rguenther@suse.de>
39247 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39248 (graphite_transform_loops): Provide opt-info for optimized nests.
39249 * tree-parloop.c (parallelize_loops): Provide opt-info for
39250 parallelized loops.
39252 2017-02-01 Richard Biener <rguenther@suse.de>
39254 PR middle-end/79315
39255 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39256 was not set before.
39258 2017-02-01 Richard Biener <rguenther@suse.de>
39260 PR tree-optimization/71824
39261 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39262 Verify the loops are valid in the merged SESE region.
39263 (scop_detection::can_represent_loop_1): Check analyzing the
39264 evolution of the number of iterations in the region succeeds.
39266 2017-01-31 Ian Lance Taylor <iant@golang.org>
39268 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39269 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39271 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39273 PR preprocessor/79210
39274 * input.c (get_substring_ranges_for_loc): Replace line_width
39275 assertion with error-handling.
39277 2017-01-31 Richard Biener <rguenther@suse.de>
39279 PR tree-optimization/77318
39280 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39281 (create_pw_aff_from_tree): Take loop parameter.
39282 (add_condition_to_pbb): Pass loop of the condition to
39283 create_pw_aff_from_tree.
39285 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39287 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39288 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39290 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39294 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39296 (convert_int_to_float128): Likewise.
39297 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39298 (convert_int_to_float128): Likewise.
39299 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39300 (UNSPEC_IEEE128_CONVERT): Likewise.
39301 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39302 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39303 Use local variables for IBM extended format.
39304 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39305 (fix_trunc<mode>si2_fprs): Likewise.
39306 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39307 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39308 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39309 to know that we can now have integers of all sizes in vector
39311 (fix<uns>_<mode>di2_hw): Likewise.
39312 (float<uns>_<mode>si2_hw): Likewise.
39313 (fix_<mode>si2_hw): Likewise.
39314 (fixuns_<mode>si2_hw): Likewise.
39315 (float<uns>_<mode>di2_hw): Likewise.
39316 (float_<mode>di2_hw): Likewise.
39317 (float_<mode>si2_hw): Likewise.
39318 (floatuns_<mode>di2_hw): Likewise.
39319 (floatuns_<mode>si2_hw): Likewise.
39320 (xscvqp<su>wz_<mode>): Delete, no longer used.
39321 (xscvqp<su>dz_<mode>): Likewise.
39322 (xscv<su>dqp_<mode>): Likewise.
39323 (ieee128_mfvsrd_64bit): Likewise.
39324 (ieee128_mfvsrd_32bit): Likewise.
39325 (ieee128_mfvsrwz): Likewise.
39326 (ieee128_mtvsrw): Likewise.
39327 (ieee128_mtvsrd_64bit): Likewise.
39328 (ieee128_mtvsrd_32bit): Likewise.
39330 2017-01-31 Martin Liska <mliska@suse.cz>
39333 * ipa-prop.c (ipa_free_all_node_params): Call release method
39334 instead of ~sumbol_summary to not to trigger double times
39337 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39339 PR tree-optimization/71691
39340 * bitmap.h (class auto_bitmap): New.
39341 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39342 is_maybe_undefined instead of ssa_undefined_value_p.
39344 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39346 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39347 __S390_ARCH_LEVEL__ to __ARCH__.
39349 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39351 PR tree-optimization/79267
39352 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39353 if should_remove_lhs_p is true.
39355 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39358 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39359 (add_alignment_attribute): New.
39360 (base_type_die): Add alignment attribute.
39361 (subrange_type_die): Likewise.
39362 (modified_type_die): Likewise.
39363 (gen_array_type_die): Likewise.
39364 (gen_descr_array_type_die: Likewise.
39365 (gen_enumeration_type_die): Likewise.
39366 (gen_subprogram_die): Likewise.
39367 (gen_variable_die): Likewise.
39368 (gen_field_die): Likewise.
39369 (gen_ptr_to_mbr_type_die): Likewise.
39370 (gen_struct_or_union_type_die): Likewise.
39371 (gen_subroutine_type_die): Likewise.
39372 (gen_typedef_die): Likewise.
39373 (base_type_cmp): Compare alignment attribute.
39375 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39378 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39379 (setb_unsigned) New pattern for setb with CCUNS.
39380 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39381 subfc./subfe sequence to avoid overflow problems. Generate a
39382 shorter sequence with cmpld/setb for power9.
39383 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39384 for generating subfc. instruction.
39385 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39386 now uses this instruction.
39388 2017-01-30 Ian Lance Taylor <iant@google.com>
39391 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39392 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39394 2017-01-30 Martin Sebor <msebor@redhat.com>
39396 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39397 Move constant to the right of a relational operator.
39398 (get_mpfr_format_length, format_character, format_string): Ditto.
39399 (should_warn_p, maybe_warn): Same.
39401 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39403 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39406 * asan.c (get_translation_unit_decl): Remove function.
39407 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39409 2017-01-30 Martin Liska <mliska@suse.cz>
39411 PR gcov-profile/79259
39412 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39413 -fprofile-generate.
39415 2017-01-30 Martin Liska <mliska@suse.cz>
39418 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39419 Initialize variables with NULL value.
39421 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39424 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39426 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39428 2017-01-30 Richard Biener <rguenther@suse.de>
39430 PR tree-optimization/79276
39431 * tree-vrp.c (process_assert_insertions): Properly adjust common
39432 when removing a duplicate.
39434 2017-01-30 Richard Biener <rguenther@suse.de>
39436 PR tree-optimization/79256
39437 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39438 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39440 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39442 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39445 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39446 ("*r<noxa>sbg_<mode>_sll_bitmask")
39447 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39448 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39449 Use contiguous_bitmask_nowrap_operand.
39451 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39454 * config/rs6000/altivec.h (vec_xl): Revise #define.
39455 (vec_xst): Likewise.
39457 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39459 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39461 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39463 PR rtl-optimization/79194
39464 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39465 traps before call to bypass_conditional_jumps.
39467 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39469 PR tree-optimization/71374
39470 * lra-constraints.c (check_conflict_input_operands): New.
39471 (match_reload): Use it.
39473 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39476 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39477 account to calculate conflict_set.
39479 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39481 PR rtl-optimization/78559
39482 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39483 other_insn in combine.
39485 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39487 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39488 uint16_type_node for BT_UINT16.
39490 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39492 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39493 "RTL Tests" to menu.
39494 (GIMPLE Tests): New node.
39495 (RTL Tests): New node.
39497 2017-01-27 Richard Biener <rguenther@suse.de>
39499 PR tree-optimization/79245
39500 * tree-loop-distribution.c (distribute_loop): Apply cost
39501 modeling also to detected patterns.
39503 2017-01-27 Richard Biener <rguenther@suse.de>
39505 PR tree-optimization/71433
39506 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39507 (compare_assert_loc): New function.
39508 (process_assert_insertions): Sort and optimize assert locations
39509 to remove duplicates and push down identical assertions on
39510 edges to their destination block.
39512 2017-01-27 Richard Biener <rguenther@suse.de>
39514 PR tree-optimization/79244
39515 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39516 out SSA names even if abnormal.
39518 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39520 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39521 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39522 instead of MPFR_RNDN.
39524 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39527 * arm.c (arm_option_override): Don't call build_target_option_node
39528 until after doing all option overrides.
39529 (arm_valid_target_attribute_tree): Likewise.
39531 2017-01-27 Martin Liska <mliska@suse.cz>
39533 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39534 for {cd}tors and C++ {cd}tors.
39536 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39538 * config/s390/s390.md ("*setmem_long_and")
39539 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39541 2017-01-26 Martin Sebor <msebor@redhat.com>
39543 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39546 2017-01-26 Martin Sebor <msebor@redhat.com>
39548 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39549 HAVE_DFmode before using XFmode or DFmode.
39550 (parse_directive): Avoid using the z length modifier to avoid
39551 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39553 PR middle-end/78703
39554 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39555 to accept adjustment as an array.
39556 (get_int_range): New function.
39557 (struct directive): Make width and prec arrays.
39558 (directive::set_width, directive::set_precision): Call get_int_range.
39559 (format_integer, format_floating): Handle width and precision ranges.
39560 (format_string, parse_directive): Same.
39562 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39565 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39566 comdat_type_p set, just clone them, but keep the children in the
39570 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39571 which have direct callers with -fvar-tracking-assignments enabled
39573 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39574 inside of type units.
39576 2017-01-26 Martin Sebor <msebor@redhat.com>
39578 PR middle-end/78703
39579 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39581 (struct format_result): Replace number_chars, number_chars_min,
39582 and number_chars_max with a single member of struct result_range.
39584 (format_result::operator+=): Adjust.
39585 (struct fmtresult): Remove bounded. Handle likely and unlikely
39587 (fmtresult::adjust_for_width_or_precision): New function.
39588 (fmtresult:type_max_digits): New function.
39589 (bytes_remaining): Handle likely and unlikely counters.
39590 (min_bytes_remaining): Remove.
39591 (format_percent): Simplify.
39592 (format_integer, format_floating): Set likely and unlikely counters.
39593 (get_string_length, format_character, format_string): Same.
39594 (format_plain, should_warn_p): New function.
39595 (maybe_warn): Call should_warn_p. Update diagnostic messages
39596 and handle those for all directives, including plain strings.
39597 (format_directive): Handle likely and unlikely counters.
39598 Remove unnecessary quoting from diagnostics. Add an informational
39600 (add_bytes): Remove.
39601 (pass_sprintf_length::compute_format_length): Simplify.
39602 (try_substitute_return_value): Handle likely and unlikely counters.
39604 2017-01-26 Carl Love <cel@us.ibm.com>
39606 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39607 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39609 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39612 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39613 endianess for subregs into account.
39614 * lra-constraints.c (lra_constraints): Do risky transformations
39615 always on the first iteration.
39616 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39617 last_call_used_reg_set.
39618 (process_bb_lives): Define and use last_call_used_reg_set.
39619 * lra.c (lra): Always continue after lra_constraints on the first
39622 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39624 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39626 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39628 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39630 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39631 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39632 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39633 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39634 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39635 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39636 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39637 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39638 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39640 2017-01-26 Marek Polacek <polacek@redhat.com>
39643 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39644 for the third operand.
39646 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39648 PR middle-end/79236
39649 * omp-low.c (struct omp_context): Add simt_stmt field.
39650 (scan_omp_for): Return omp_context *.
39651 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39652 context to the _simt_ SIMD stmt.
39653 (lower_omp_for): For combined SIMD with sibling _simt_
39654 SIMD, make sure to use the same decls in _looptemp_
39655 clauses as in the sibling.
39657 2017-01-26 David Sherwood <david.sherwood@arm.com>
39659 PR middle-end/79212
39660 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39663 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39666 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39667 emit fld b; fld a; if possible.
39669 * brig-builtins.def: Update copyright years.
39670 * config/arm/arm_acle_builtins.def: Update copyright years.
39672 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39675 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39676 constraint instead of o for the stxsd instruction.
39678 2017-01-25 Carl Love <cel@us.ibm.com>
39680 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39681 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39683 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39685 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39687 2017-01-25 Richard Biener <rguenther@suse.de>
39689 PR tree-optimization/69264
39690 * target.def (vector_alignment_reachable): Improve documentation.
39691 * doc/tm.texi: Regenerate.
39692 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39694 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39695 earlier changes with respect to TYPE_USER_ALIGN.
39696 (vector_alignment_reachable_p): Likewise. Improve dumping.
39698 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39701 * config/arm/arm.md (xordi3): Force constant operand into a register
39704 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39706 * doc/invoke.texi (-fstore-merging): Correct default optimization
39707 levels at which it is enabled.
39708 (-O): Move -fstore-merging from list to...
39711 2017-01-25 Richard Biener <rguenther@suse.de>
39714 * omp-expand.c: Include debug.h.
39715 (expand_omp_taskreg): Make sure to generate early debug before
39716 outlining anything from a function.
39717 (expand_omp_target): Likewise.
39718 (grid_expand_target_grid_body): Likewise.
39720 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39723 * asan.c (get_translation_unit_decl): New function.
39724 (asan_add_global): Extract modules file name from globals
39725 TRANSLATION_UNIT_DECL name.
39727 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39730 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39731 for long calls with APCS frame and VFP.
39733 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39735 * cfg.c (original_copy_tables_initialized_p): New function.
39736 * cfg.h (original_copy_tables_initialized_p): New decl.
39737 * cfgrtl.c (relink_block_chain): Guard the call to
39738 free_original_copy_tables with a call to
39739 original_copy_tables_initialized_p.
39740 * cgraph.h (symtab_node::native_rtl_p): New decl.
39741 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39742 (symtab_node::needed_p): Don't assert for early assembly output
39743 for __RTL functions.
39744 (cgraph_node::finalize_function): Set "force_output" for __RTL
39746 (cgraph_node::analyze): Bail out early for __RTL functions.
39747 (analyze_functions): Update assertion to support __RTL functions.
39748 (cgraph_node::expand): Bail out early for __RTL functions.
39749 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39751 * function.h (struct function): Update comment for field
39753 * gimple-expr.c: Include "tree-pass.h".
39754 (gimple_has_body_p): Return false for __RTL functions.
39755 * Makefile.in (OBJS): Add run-rtl-passes.o.
39756 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39758 (gcc::pass_manager::get_clean_slate): New accessor.
39759 * passes.c: Include "insn-addr.h".
39760 (should_skip_pass_p): Add logging. Update logic for running
39761 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39762 property-provider override so it is only done for gimple passes.
39764 (skip_pass): New function.
39765 (execute_one_pass): Call skip_pass when skipping passes.
39766 * read-md.c (md_reader::read_char): Support filtering
39767 the input to a subset of line numbers.
39768 (md_reader::md_reader): Initialize fields
39769 m_first_line and m_last_line.
39770 (md_reader::read_file_fragment): New function.
39771 * read-md.h (md_reader::read_file_fragment): New decl.
39772 (md_reader::m_first_line): New field.
39773 (md_reader::m_last_line): New field.
39774 * read-rtl-function.c (function_reader::create_function): Only
39775 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39776 curr_properties. Set DECL_INITIAL to a dummy block.
39777 (read_rtl_function_body_from_file_range): New function.
39778 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39780 * run-rtl-passes.c: New file.
39781 * run-rtl-passes.h: New file.
39783 2017-01-24 Jeff Law <law@redhat.com>
39785 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39788 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39790 PR tree-optimization/79159
39791 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39792 (record_nonwrapping_iv): Improve boundary using above function if no
39793 value range information.
39795 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39796 Martin Jambor <mjambor@suse.cz>
39798 * brig-builtins.def: New file.
39799 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39800 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39801 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39802 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39803 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39804 * builtin-types.def (BT_INT8): New.
39805 (BT_INT16): Likewise.
39806 (BT_UINT8): Likewise.
39807 (BT_UINT16): Likewise.
39808 (BT_FN_ULONG): Likewise.
39809 (BT_FN_UINT_INT): Likewise.
39810 (BT_FN_UINT_ULONG): Likewise.
39811 (BT_FN_UINT_LONG): Likewise.
39812 (BT_FN_UINT_PTR): Likewise.
39813 (BT_FN_ULONG_PTR): Likewise.
39814 (BT_FN_INT8_FLOAT): Likewise.
39815 (BT_FN_INT16_FLOAT): Likewise.
39816 (BT_FN_UINT32_FLOAT): Likewise.
39817 (BT_FN_UINT16_FLOAT): Likewise.
39818 (BT_FN_UINT8_FLOAT): Likewise.
39819 (BT_FN_UINT64_FLOAT): Likewise.
39820 (BT_FN_UINT16_UINT32): Likewise.
39821 (BT_FN_UINT32_UINT16): Likewise.
39822 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39823 (BT_FN_INT_PTR_INT): Likewise.
39824 (BT_FN_UINT_PTR_UINT): Likewise.
39825 (BT_FN_LONG_PTR_LONG): Likewise.
39826 (BT_FN_ULONG_PTR_ULONG): Likewise.
39827 (BT_FN_VOID_UINT64_UINT64): Likewise.
39828 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39829 (BT_FN_INT8_INT8_INT8): Likewise.
39830 (BT_FN_INT16_INT16_INT16): Likewise.
39831 (BT_FN_INT_INT_INT): Likewise.
39832 (BT_FN_UINT_FLOAT_UINT): Likewise.
39833 (BT_FN_FLOAT_UINT_UINT): Likewise.
39834 (BT_FN_ULONG_UINT_UINT): Likewise.
39835 (BT_FN_ULONG_UINT_PTR): Likewise.
39836 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39837 (BT_FN_UINT_UINT_UINT): Likewise.
39838 (BT_FN_VOID_UINT_PTR): Likewise.
39839 (BT_FN_UINT_UINT_PTR: Likewise.
39840 (BT_FN_UINT32_UINT64_PTR): Likewise.
39841 (BT_FN_INT_INT_UINT_UINT): Likewise.
39842 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39843 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39844 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39845 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39846 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39847 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39848 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39849 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39850 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39851 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39852 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39853 * doc/frontends.texi: List BRIG FE.
39854 * doc/install.texi (Testing): Add BRIG tesring requirements.
39855 * doc/invoke.texi (Overall Options): Mention BRIG.
39856 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39858 2017-01-24 Richard Biener <rguenther@suse.de>
39860 PR translation/79208
39861 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39863 2017-01-24 Martin Jambor <mjambor@suse.cz>
39866 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39867 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39868 and known_contexts.
39870 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39872 PR middle-end/79123
39873 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39874 casts from signed to unsigned really don't have a range.
39876 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39878 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39879 GMP_RNDx for compatiblity.
39881 2017-01-24 Martin Liska <mliska@suse.cz>
39884 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39885 that would prevent us to call alloca with -1 as argument.
39887 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39889 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39890 Avoid -Wformat-security warning.
39892 2017-01-23 Andrew Pinski <apinski@cavium.com>
39894 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39897 2017-01-23 Martin Sebor <msebor@redhat.com>
39899 PR middle-end/78703
39900 * gimple-ssa-sprintf.c (warn_level): New global.
39901 (format_integer): Use it here and throughout the rest of the file.
39902 Use the same switch to compute sign as base.
39903 (maybe_warn): New function.
39904 (format_directive): Factor out warnings into maybe_warn.
39905 Add debugging output. Use warn_level.
39906 (add_bytes): Use warn_level.
39907 (pass_sprintf_length::compute_format_length): Add debugging output.
39908 (try_substitute_return_value): Same.
39909 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39911 PR middle-end/78703
39912 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39913 (struct fmtresult, format_integer, format_floating): Adjust.
39914 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39915 (get_string_length, format_string,format_directive): Same.
39916 (pass_sprintf_length::compute_format_length): Same.
39917 (try_substitute_return_value): Simplify slightly.
39919 PR middle-end/78703
39920 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39921 (fmtresult::operator+=): Outlined.
39922 (struct fmtresult): Add ctors.
39923 (struct conversion_spec): Rename...
39924 (struct directive): ...to this. Add and remove data members.
39925 (directive::set_width, directive::set_precision): New functions.
39926 (format_percent): Use fmtresult ctor.
39927 (get_width_and_precision): Remove.
39928 (format_integer): Make naming changes. Avoid computing width and
39930 (format_floating): Same. Adjust indentation.
39931 (format_character, format_none): New functions.
39932 (format_string): Moved character handling to format_character.
39933 (format_directive): Remove arguments, change return type.
39934 (parse_directive): New function.
39935 (pass_sprintf_length::compute_format_length): Move directive
39936 parsing to parse_directive.
39938 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39940 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39941 (assign_assembler_name_if_needed): ... this.
39942 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39943 (assign_assembler_name_if_needed): ... this.
39944 (free_lang_data_in_cgraph): Adjust callers.
39945 * cgraphunit.c (cgraph_node::analyze): Likewise.
39946 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39949 2017-01-23 Richard Biener <rguenther@suse.de>
39951 PR tree-optimization/79088
39952 PR tree-optimization/79188
39953 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39954 resetting loop bounds after last path deletion. Reset loop
39955 bounds of the target loop, make code match the comments.
39956 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39957 Make sure loops need no fixups.
39959 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39961 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39962 exponent support with double type for first argument.
39963 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39964 type returned by __builtin_vec_extract_sig,
39965 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39966 functions from "vector int" to "vector unsigned int" or from
39967 "vector long long int" to "vector unsigned long long int".
39968 Changed type returned by __builtin_vec_extract_exp,
39969 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39970 functions from "vector int" to "vector unsigned int" or from
39971 "vector long long int" to "vector unsigned long long int".
39972 Changed return type of __builtin_vec_test_data_class,
39973 __builtin_vec_test_data_class_sp, and
39974 __builtin_vec_test_data_class_dp from "vector int" to
39975 "vector bool int" or from "vector long long int" to "vector bool
39976 long long int" and changed second argument type from "unsigned
39977 int" to "int". Added new overloaded function forms "vector float
39978 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39979 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39980 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39981 double, vector unsigned long long int)" and "vector double
39982 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39983 long int)". Changed return type of
39984 __builtin_scalar_test_data_class and
39985 __builtin_scalar_test_data_class_sp and
39986 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39987 int" and changed second argument from "unsigned int" to "int".
39988 Changed type returned by __builtin_scalar_test_neg,
39989 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39990 from "int" to "bool int". Added new overloaded function form
39991 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39992 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39993 exponent double-precision with floating point first argument.
39994 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39995 documentation of scalar_test_data_class, scalar_test_neg,
39996 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39997 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39998 vec_test_data_class built-in functions to reflect refinements in
39999 their type signatures.
40001 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
40003 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
40005 (aarch64_elf_asm_destructor): Likewise.
40007 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
40009 PR rtl-optimization/78634
40010 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
40011 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
40012 * ifcvt.c (noce_try_cmove): Add missing cost check.
40014 PR rtl-optimization/71724
40015 * combine.c (if_then_else_cond): Look for situations where it is
40016 beneficial to undo the work of one of the recursive calls.
40018 2017-01-23 Bin Cheng <bin.cheng@arm.com>
40020 PR tree-optimization/70754
40021 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
40022 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
40023 combined stmt before it if not NULL.
40024 (combine_chains): Process refs reversely and compute dominance point
40027 2017-01-23 Martin Liska <mliska@suse.cz>
40029 PR tree-optimization/79196
40030 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
40031 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
40033 (strlen_optimize_stmt): Call the renamed function.
40035 2017-01-23 Michael Matz <matz@suse.de>
40037 PR tree-optimization/78384
40038 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
40040 2017-01-23 Richard Biener <rguenther@suse.de>
40042 PR tree-optimization/79186
40043 * tree-vrp.c (register_new_assert_for): Make sure we've seen
40044 both incoming edges before moving an assert.
40046 2017-01-23 Martin Jambor <mjambor@suse.cz>
40048 * ipa-prop.c (load_from_param_1): Removed.
40049 (load_from_unmodified_param): Bits from load_from_param_1 put back
40051 (load_from_param): Removed.
40052 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
40053 with stmt. Reverted back to use of load_from_unmodified_param.
40055 2017-01-23 Martin Jambor <mjambor@suse.cz>
40058 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40059 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40060 field a pointer to garbage collected vector, mark lattices and
40061 ipcp_orig_node with GTY((skip)).
40062 (ipa_get_param_count): Adjust to descriptors being a pointer.
40063 (ipa_get_param): Likewise.
40064 (ipa_get_type): Likewise.
40065 (ipa_get_param_move_cost): Likewise.
40066 (ipa_set_param_used): Likewise.
40067 (ipa_get_controlled_uses): Likewise.
40068 (ipa_set_controlled_uses): Likewise.
40069 (ipa_is_param_used): Likewise.
40070 (ipa_node_params_t): Move into garbage collector. New methods insert
40072 (ipa_node_params_sum): Annotate wth GTY(()).
40073 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40075 (ipa_load_from_parm_agg): Adjust declaration.
40076 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40077 * ipa-profile.c (ipa_profile): Likewise.
40078 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40079 (ipa_populate_param_decls): Make descriptors parameter garbage
40081 (ipa_dump_param): Adjust to descriptors being a pointer.
40082 (ipa_alloc_node_params): Likewise.
40083 (ipa_initialize_node_params): Likewise.
40084 (load_from_param_1): Make descriptors parameter garbage collected.
40085 (load_from_unmodified_param): Likewise.
40086 (load_from_param): Likewise.
40087 (ipa_load_from_parm_agg): Likewise.
40088 (ipa_node_params::~ipa_node_params): Removed.
40089 (ipa_free_all_node_params): Remove call to delete operator.
40090 (ipa_node_params_t::insert): New.
40091 (ipa_node_params_t::remove): Likewise.
40092 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40093 copy known_csts and known_contexts vectors.
40094 (ipa_read_node_info): Adjust to descriptors being a pointer.
40095 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40097 (ipcp_transform_function): Make descriptors variable garbage
40100 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40102 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40103 * config/i386/avx512dqintrin.h: Ditto.
40104 * config/i386/avx512fintrin.h: Ditto.
40105 * config/i386/i386.c: Handle new builtins.
40106 * config/i386/i386-builtin.def: Add new builtins.
40107 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40108 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40110 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40111 Martin Liska <mliska@suse.cz>
40113 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40114 * asan.c (asan_expand_poison_ifn): Support stores and use
40115 appropriate ASAN report function.
40116 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40117 * internal-fn.def (ASAN_POISON_USE): Declare.
40118 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40119 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40120 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40121 ASAN_POISON calls w/o LHS.
40122 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40123 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40124 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40125 * gimplify.c (asan_poison_variables): Add attribute
40126 use_after_scope_memory to variables that really needs to live
40128 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40129 having the attribute.
40131 2017-01-23 Martin Liska <mliska@suse.cz>
40133 * asan.c (create_asan_shadow_var): New function.
40134 (asan_expand_poison_ifn): Likewise.
40135 * asan.h (asan_expand_poison_ifn): New declaration.
40136 * internal-fn.c (expand_ASAN_POISON): Likewise.
40137 * internal-fn.def (ASAN_POISON): New builtin.
40138 * sanopt.c (pass_sanopt::execute): Expand
40139 asan_expand_poison_ifn.
40140 * tree-inline.c (copy_decl_for_dup_finish): Make function
40142 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40143 * tree-ssa.c (is_asan_mark_p): New function.
40144 (execute_update_addresses_taken): Rewrite local variables
40145 (identified just by use-after-scope as addressable) into SSA.
40147 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40149 * doc/install.texi (Specific): opensource.apple.com uses https
40150 now. Remove trailing slash.
40152 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40154 * README.Portability: Remove note on an Irix compatibility issue.
40156 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40158 * gcov.c (INCLUDE_ALGORITHM): Define.
40159 (INCLUDE_VECTOR): Define.
40160 No longer include <vector> and <algorithm> directly.
40162 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40164 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40166 * doc/invoke.texi (Code Gen Options): Ditto.
40168 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40171 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40173 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40175 rtl-optimization/79125
40176 * cprop.c (local_cprop_pass): Handle cases where we make an
40177 unconditional trap.
40179 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40183 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40184 read from, for big endian.
40186 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40188 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40189 register pauth builtins for LP64 only.
40191 2017-01-20 Marek Polacek <polacek@redhat.com>
40194 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40197 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40199 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40201 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40202 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40203 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40205 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40208 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40209 in big-endian mode when they are not a single duplicated value.
40211 2017-01-20 Richard Biener <rguenther@suse.de>
40213 * BASE-VER: Bump to 7.0.1.
40215 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40217 * omp-low.c (omplow_simd_context): New struct. Use it...
40218 (lower_rec_simd_input_clauses): ...here and...
40219 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40220 references to idx, lane, max_vf, is_simt.
40222 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40224 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40227 2017-01-20 Martin Jambor <mjambor@suse.cz>
40229 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40230 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40232 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40233 (GTFILES): Rename hsa.c to hsa-common.c.
40234 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40235 * hsa-dump.c: Likewise.
40236 * hsa-gen.c: Likewise.
40237 * hsa-regalloc.c: Likewise.
40238 * ipa-hsa.c: Likewise.
40239 * omp-expand.c: Likewise.
40240 * omp-low.c: Likewise.
40241 * toplev.c: Likewise.
40243 2017-01-20 Marek Polacek <polacek@redhat.com>
40246 * doc/invoke.texi: Document -Wduplicated-branches.
40247 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40248 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40249 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40250 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40251 return 0 only when not OEP_LEXICOGRAPHIC.
40252 (fold_build_cleanup_point_expr): Use the expression
40253 location when building CLEANUP_POINT_EXPR.
40254 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40255 * tree.c (add_expr): Handle error_mark_node.
40257 2017-01-20 Martin Liska <mliska@suse.cz>
40260 * tree-profile.c (init_ic_make_global_vars): Do not call
40262 (gimple_init_gcov_profiler): Likewise.
40264 2017-01-20 Martin Liska <mliska@suse.cz>
40267 * cgraph.h (maybe_create_reference): Remove argument and
40269 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40271 * ipa-cp.c (create_specialized_node): Likewise.
40272 * symtab.c (symtab_node::maybe_create_reference): Handle
40273 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40275 2017-01-20 Martin Liska <mliska@suse.cz>
40277 * read-rtl-function.c (function_reader::create_function): Use
40278 build_decl instread of build_decl_stat.
40280 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40282 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40283 * config/i386/avx512dqintrin.h: Ditto.
40284 * config/i386/avx512fintrin.h: Ditto.
40285 * config/i386/i386-builtin-types.def: Add new types.
40286 * config/i386/i386.c: Handle new types.
40287 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40288 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40289 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40290 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40291 (__builtin_ia32_kshiftridi): New.
40292 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40294 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40298 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40299 define to rs6000_init_stack_protect_guard.
40300 (rs6000_init_stack_protect_guard): New function.
40302 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40303 Yunqiang Su <yunqiang.su@imgtec.com>
40305 * config.gcc (supported_defaults): Add madd4.
40306 (with_madd4): Add validation.
40307 (all_defaults): Add madd4.
40308 * config/mips/mips.opt (mmadd4): New option.
40309 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40311 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40313 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40314 (ISA_HAS_FUSED_MADD4): Likewise.
40315 * doc/invoke.texi (-mmadd4): Document the new option.
40316 * doc/install.texi (--with-madd4): Document the new option.
40318 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40320 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40321 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40322 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40323 (aarch64_init_pauth_hint_builtins): New.
40324 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40325 (aarch64_expand_builtin): Expand new builtins.
40327 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40329 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40330 * combine-stack-adj.c (no_unhandled_cfa): Handle
40331 REG_CFA_TOGGLE_RA_MANGLE.
40332 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40333 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40334 info for return address signing.
40335 (aarch64_expand_epilogue): Likewise.
40337 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40339 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40340 * config/aarch64/aarch64-protos.h
40341 (aarch64_return_address_signing_enabled): New declaration.
40342 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40344 (aarch64_expand_prologue): Sign return address before it's pushed onto
40346 (aarch64_expand_epilogue): Authenticate return address fetched from
40348 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40349 (aarch64_attributes): New function attributes for "sign-return-address".
40350 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40351 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40352 ("*do_return"): Generate combined instructions according to key index.
40353 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40354 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40356 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40357 * config/aarch64/aarch64.opt (msign-return-address=): New.
40358 * doc/extend.texi (AArch64 Function Attributes): Documents
40359 "sign-return-address=".
40360 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40362 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40364 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40365 overall option summary.
40367 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40369 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40370 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40371 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40372 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40374 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40376 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40377 -mpower9-minmax by default for -mcpu=power9.
40378 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40379 128-bit floating point.
40381 2017-01-20 Alan Modra <amodra@gmail.com>
40383 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40384 optimizing for size.
40386 2017-01-20 Alan Modra <amodra@gmail.com>
40389 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40390 for strcmp and strncmp from corresponding builtin decl.
40392 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40394 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40395 instead of i386/rtems-64.h.
40396 * config/i386/rtems-64.h: Remove.
40398 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40402 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40404 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40406 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40408 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40409 Change int to HOST_WIDE_INT.
40410 * config/aarch64/aarch64-protos.h
40411 (aarch64_simd_gen_const_vector_dup): Likewise.
40412 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40414 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40416 * langhooks-def.h (lhd_type_for_size): New decl.
40417 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40418 * langhooks.c (lhd_type_for_size): New function, taken from
40421 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40423 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40424 define_bypass for CR latency.
40425 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40426 (power9-alu2): Add define_bypass for CR latency.
40428 (power9-mul): Update insn latency.
40429 (power9-mul-compare): Update insn latency, bypass latency and remove
40432 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40434 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40436 * config/aarch64/aarch64.md
40437 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40438 aarch64_nopcrelative_literal_loads.
40439 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40441 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40443 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40444 TARGET_LOONGSON_3A.
40445 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40447 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40450 * config.gcc (supported_defaults): Add lxc1-sxc1.
40451 (with_lxc1_sxc1): Add validation.
40452 (all_defaults): Add lxc1-sxc1.
40453 * config/mips/mips.opt (mlxc1-sxc1): New option.
40454 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40456 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40457 __mips_no_lxc1_sxc1.
40458 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40459 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40460 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40462 2017-01-19 Richard Biener <rguenther@suse.de>
40464 PR tree-optimization/72488
40465 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40466 sure to restore SSA info.
40467 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40469 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40471 PR rtl-optimization/79121
40472 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40473 of the inner type when shifting an extended value.
40475 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40478 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40479 interposable aliases.
40481 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40484 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40485 Use the evmergelohi instruction.
40486 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40487 (mov_sitf_e500_subreg8_2_be): Likewise.
40488 (mov_sitf_e500_subreg12_2_le): Likewise.
40489 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40490 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40491 (mov_sitf_e500_subreg8_2_le): Likewise.
40492 (mov_sitf_e500_subreg12_2_be): Likewise.
40494 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40496 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40497 attribute from vecsimple to vecperm.
40498 (altivec_vbpermq2): Likewise.
40500 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40503 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40505 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40506 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40507 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40508 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40509 case where N arg is SIZE_MAX.
40510 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40511 (cmpstrsi): Add pattern.
40513 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40515 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40516 __builtin_vec_revb builtins.
40517 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40518 built-in functions to support generation of the ISA 3.0 XXBR<x>
40519 vector byte reverse instructions.
40520 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40521 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40522 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40523 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40524 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40525 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40526 (P9V_BUILTIN_VEC_REVB): Likewise.
40527 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40528 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40529 (p9_xxbrq_v16qi): Likewise.
40530 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40531 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40532 (p9_xxbrh_v8hi): Likewise.
40533 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40534 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40535 vec_revb built-in functions.
40537 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40539 PR rtl-optimization/78952
40540 * config/i386/i386.md (any_extract): New code iterator.
40541 (*insvqi_2): Use any_extract for source operand.
40542 (*insvqi_3): Use any_shiftrt for source operand.
40544 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40546 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40548 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40550 2017-01-18 Matthias Klose <doko@ubuntu.com>
40552 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40554 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40556 * config/rs6000/altivec.h (vec_bperm): Change #define.
40557 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40558 (altivec_vbpermq2): New define_insn.
40559 (altivec_vbpermd): Likewise.
40560 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40561 function interface.
40562 (VBPERMD): Likewise.
40563 (VBPERM): New polymorphic function interface.
40564 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40565 Add entries for P9V_BUILTIN_VEC_VBPERM.
40566 * doc/extend.texi: Add interfaces for vec_bperm.
40568 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40570 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40571 first letter of error messages.
40572 (s390_resolve_overloaded_builtin): Likewise.
40573 * config/s390/s390.c (s390_expand_builtin): Likewise.
40574 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40575 (s390_valid_target_attribute_inner_p): Likewise.
40576 * config/s390/s390.md ("tabort"): Likewise.
40578 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40580 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40581 (ISA_AVOID_DIV_HILO): New macro.
40582 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40583 (ISA_HAS_DDIV): Likewise.
40585 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40587 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40589 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40591 * doc/invoke.texi (fabi-version): Spelling fix.
40593 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40596 * doc/invoke.texi (fabi-version): Mention mangling fix for
40599 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40602 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40604 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40607 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40608 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40610 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40612 (stack_protect_test): Ditto.
40613 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40614 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40616 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40617 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40618 -mstack-protector-guard-offset=.
40619 (RS/6000 and PowerPC Options): Ditto.
40621 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40623 * config/i386/i386.h (MASK_CLASS_P): New define.
40624 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40625 there are no registers from different register sets also when
40626 mask registers are used. Update function comment.
40627 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40628 to (*k/*r) and (*k/*km) alternatives.
40630 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40632 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40633 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40634 (EH_RETURN_HANDLER_RTX): New define.
40635 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40636 Force frame pointer in EH return functions.
40637 (aarch64_expand_epilogue): Add barrier for eh_return.
40638 (aarch64_final_eh_return_addr): Remove.
40639 (aarch64_eh_return_handler_rtx): New function.
40640 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40642 (aarch64_eh_return_handler_rtx): New prototype.
40644 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40646 * config/rs6000/altivec.h (vec_rlmi): New #define.
40647 (vec_vrlnm): Likewise.
40648 (vec_rlnm): Likewise.
40649 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40650 (UNSPEC_VRLNM): Likewise.
40651 (VIlong): New mode iterator.
40652 (altivec_vrl<VI_char>mi): New define_insn.
40653 (altivec_vrl<VI_char>nm): Likewise.
40654 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40656 (VRLDNM): Likewise.
40657 (RLNM): New polymorphic function entry.
40658 (VRLWMI): New monomorphic function entry.
40659 (VRLDMI): Likewise.
40660 (RLMI): New polymorphic function entry.
40661 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40662 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40663 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40666 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40669 * dwarf2out.c (field_byte_offset): Restore the
40670 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40671 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40672 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40675 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40678 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40680 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40683 * dwarf2out.c (add_data_member_location_attribute): For constant
40684 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40685 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40686 DW_AT_byte_size attributes.
40688 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40690 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40691 after forcing to constant memory when the code model is medium.
40693 2017-01-17 Julia Koval <julia.koval@intel.com>
40696 * config/i386/avx512fintrin.h
40697 (_mm512_i32gather_ps): Change __addr type to void const*.
40698 (_mm512_mask_i32gather_ps): Ditto.
40699 (_mm512_i32gather_pd): Ditto.
40700 (_mm512_mask_i32gather_pd): Ditto.
40701 (_mm512_i64gather_ps): Ditto.
40702 (_mm512_mask_i64gather_ps): Ditto.
40703 (_mm512_i64gather_pd): Ditto.
40704 (_mm512_mask_i64gather_pd): Ditto.
40705 (_mm512_i32gather_epi32): Ditto.
40706 (_mm512_mask_i32gather_epi32): Ditto.
40707 (_mm512_i32gather_epi64): Ditto.
40708 (_mm512_mask_i32gather_epi64): Ditto.
40709 (_mm512_i64gather_epi32): Ditto.
40710 (_mm512_mask_i64gather_epi32): Ditto.
40711 (_mm512_i64gather_epi64): Ditto.
40712 (_mm512_mask_i64gather_epi64): Ditto.
40713 (_mm512_i32scatter_ps): Change __addr type to void*.
40714 (_mm512_mask_i32scatter_ps): Ditto.
40715 (_mm512_i32scatter_pd): Ditto.
40716 (_mm512_mask_i32scatter_pd): Ditto.
40717 (_mm512_i64scatter_ps): Ditto.
40718 (_mm512_mask_i64scatter_ps): Ditto.
40719 (_mm512_i64scatter_pd): Ditto.
40720 (_mm512_mask_i64scatter_pd): Ditto.
40721 (_mm512_i32scatter_epi32): Ditto.
40722 (_mm512_mask_i32scatter_epi32): Ditto.
40723 (_mm512_i32scatter_epi64): Ditto.
40724 (_mm512_mask_i32scatter_epi64): Ditto.
40725 (_mm512_i64scatter_epi32): Ditto.
40726 (_mm512_mask_i64scatter_epi32): Ditto.
40727 (_mm512_i64scatter_epi64): Ditto.
40728 (_mm512_mask_i64scatter_epi64): Ditto.
40729 * config/i386/avx512pfintrin.h
40730 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40731 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40732 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40733 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40734 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40735 (_mm512_prefetch_i32scatter_ps): Ditto.
40736 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40737 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40738 (_mm512_prefetch_i64scatter_pd): Ditto.
40739 (_mm512_prefetch_i64scatter_ps): Ditto.
40740 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40741 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40742 * config/i386/avx512vlintrin.h
40743 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40744 (_mm_mmask_i32gather_ps): Ditto.
40745 (_mm256_mmask_i32gather_pd): Ditto.
40746 (_mm_mmask_i32gather_pd): Ditto.
40747 (_mm256_mmask_i64gather_ps): Ditto.
40748 (_mm_mmask_i64gather_ps): Ditto.
40749 (_mm256_mmask_i64gather_pd): Ditto.
40750 (_mm_mmask_i64gather_pd): Ditto.
40751 (_mm256_mmask_i32gather_epi32): Ditto.
40752 (_mm_mmask_i32gather_epi32): Ditto.
40753 (_mm256_mmask_i32gather_epi64): Ditto.
40754 (_mm_mmask_i32gather_epi64): Ditto.
40755 (_mm256_mmask_i64gather_epi32): Ditto.
40756 (_mm_mmask_i64gather_epi32): Ditto.
40757 (_mm256_mmask_i64gather_epi64): Ditto.
40758 (_mm_mmask_i64gather_epi64): Ditto.
40759 (_mm256_i32scatter_ps): Change __addr type to void*.
40760 (_mm256_mask_i32scatter_ps): Ditto.
40761 (_mm_i32scatter_ps): Ditto.
40762 (_mm_mask_i32scatter_ps): Ditto.
40763 (_mm256_i32scatter_pd): Ditto.
40764 (_mm256_mask_i32scatter_pd): Ditto.
40765 (_mm_i32scatter_pd): Ditto.
40766 (_mm_mask_i32scatter_pd): Ditto.
40767 (_mm256_i64scatter_ps): Ditto.
40768 (_mm256_mask_i64scatter_ps): Ditto.
40769 (_mm_i64scatter_ps): Ditto.
40770 (_mm_mask_i64scatter_ps): Ditto.
40771 (_mm256_i64scatter_pd): Ditto.
40772 (_mm256_mask_i64scatter_pd): Ditto.
40773 (_mm_i64scatter_pd): Ditto.
40774 (_mm_mask_i64scatter_pd): Ditto.
40775 (_mm256_i32scatter_epi32): Ditto.
40776 (_mm256_mask_i32scatter_epi32): Ditto.
40777 (_mm_i32scatter_epi32): Ditto.
40778 (_mm_mask_i32scatter_epi32): Ditto.
40779 (_mm256_i32scatter_epi64): Ditto.
40780 (_mm256_mask_i32scatter_epi64): Ditto.
40781 (_mm_i32scatter_epi64): Ditto.
40782 (_mm_mask_i32scatter_epi64): Ditto.
40783 (_mm256_i64scatter_epi32): Ditto.
40784 (_mm256_mask_i64scatter_epi32): Ditto.
40785 (_mm_i64scatter_epi32): Ditto.
40786 (_mm_mask_i64scatter_epi32): Ditto.
40787 (_mm256_i64scatter_epi64): Ditto.
40788 (_mm256_mask_i64scatter_epi64): Ditto.
40789 (_mm_i64scatter_epi64): Ditto.
40790 (_mm_mask_i64scatter_epi64): Ditto.
40791 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40792 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40793 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40794 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40795 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40796 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40797 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40798 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40799 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40800 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40801 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40802 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40803 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40804 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40805 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40806 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40807 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40808 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40809 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40810 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40811 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40812 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40813 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40814 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40815 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40816 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40817 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40818 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40819 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40820 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40821 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40822 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40823 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40824 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40825 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40826 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40827 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40828 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40829 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40830 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40831 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40832 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40833 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40834 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40835 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40836 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40837 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40838 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40839 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40840 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40841 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40842 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40843 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40844 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40845 definitions accordingly.
40847 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40848 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40851 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40854 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40857 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40858 hard regs for inner regno.
40860 2017-01-17 Martin Liska <mliska@suse.cz>
40863 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40864 assumption and add comment.
40866 2017-01-17 Nathan Sidwell <nathan@acm.org>
40868 * ipa-visibility.c (localize_node): New function, broken out of ...
40869 (function_and_variable_visibility): ... here. Call it.
40871 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40873 PR middle-end/77445
40874 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40875 correctly set frequency of oudgoing edge.
40876 (duplicate_thread_path): Fix profile updating.
40878 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40881 * configure.ac: Add GCC_BASE_VER.
40882 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40883 version from BASE-VER file.
40884 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40885 (gcc.o): Depend on $(BASEVER).
40886 * common.opt (dumpfullversion): New option.
40887 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40888 * doc/invoke.texi: Document -dumpfullversion.
40889 * doc/install.texi: Document --with-gcc-major-version-only.
40890 * configure: Regenerated.
40892 2017-01-17 Richard Biener <rguenther@suse.de>
40894 PR tree-optimization/71433
40895 * tree-vrp.c (register_new_assert_for): Merge same asserts
40896 on all incoming edges.
40897 (process_assert_insertions_for): Handle insertions at the
40900 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40902 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40903 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40905 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40908 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40911 2017-01-17 Alan Modra <amodra@gmail.com>
40914 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40915 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40916 symbolic stack limit when pic.
40918 2017-01-16 Martin Sebor <msebor@redhat.com>
40920 PR tree-optimization/78608
40921 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40923 2017-01-16 Jeff Law <law@redhat.com>
40926 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40927 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40928 for several include directories that may be relative to sysroot.
40929 * config/i386/x-mingw32 (gplus_includedir): Define.
40930 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40931 (native_system_includedir): Likewise.
40932 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40933 override if TARGET_SYSTEM_ROOT is defined.
40934 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40936 PR tree-optimization/79090
40937 PR tree-optimization/33562
40938 PR tree-optimization/61912
40939 PR tree-optimization/77485
40940 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40941 and computed trims into the dump file.
40943 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40945 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40947 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40950 * gimplify.c (gimplify_init_constructor): If want_value and
40951 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40955 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40956 sequence. Formatting fixes.
40957 (doloop_optimize): Formatting fixes.
40960 * gcc.c (debug_level_greater_than_spec_func): New function.
40961 (static_spec_functions): Add debug-level-gt spec function.
40962 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40964 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40965 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40966 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40967 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40968 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40969 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40971 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40973 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40974 QImode fixups to general and mask registers only.
40976 2017-01-16 Carl Love <cel@us.ibm.com>
40978 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40979 for built-in functions
40980 vector signed char vec_nabs (vector signed char)
40981 vector signed short vec_nabs (vector signed short)
40982 vector signed int vec_nabs (vector signed int)
40983 vector signed long long vec_nabs (vector signed long long)
40984 vector float vec_nabs (vector float)
40985 vector double vec_nabs (vector double)
40986 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40988 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40989 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40990 * doc/extend.texi: Update the documentation file for the new built-in
40993 2017-01-16 Martin Sebor <msebor@redhat.com>
40995 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40998 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41000 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
41001 UNSPEC_VSX__XXSPLTD to require special splat handling.
41003 2017-01-16 David Malcolm <dmalcolm@redhat.com>
41006 * system.h: Poison strndup.
41008 2017-01-16 Alan Modra <amodra@gmail.com>
41011 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
41014 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
41016 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
41018 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
41020 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
41021 call recog here. Assert that INSN_CODE (insn) is non-negative.
41023 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41026 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
41028 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
41029 in the currently scheduled RTL fragment.
41031 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41033 PR rtl-optimization/78751
41034 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
41037 2017-01-14 Jeff Law <law@redhat.com>
41039 PR tree-optimization/79090
41040 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
41041 variable length stores.
41042 (compute_trims): Delete dead assignment to *trim_tail.
41043 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
41046 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
41048 PR rtl-optimization/78626
41049 PR rtl-optimization/78727
41050 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
41051 of a block, and split such blocks after everything else is finished.
41053 2017-01-14 Alan Modra <amodra@gmail.com>
41056 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
41057 target legitimate_combined_insn.
41058 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41059 (rs6000_legitimate_combined_insn): New function.
41060 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41062 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41063 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41064 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41066 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41068 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41070 2017-01-13 Jeff Law <law@redhat.com>
41072 PR tree-optimization/33562
41073 PR tree-optimization/61912
41074 PR tree-optimization/77485
41075 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41077 (delete_dead_assignment): Likewise.
41078 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41079 statement to delete_dead_call and delete_dead_assignment.
41081 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41084 * substring-locations.c (format_warning_va): Strengthen case 1 so
41085 that both endpoints of the substring must be within the format
41086 range for just the substring to be printed.
41088 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41090 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41091 * config/i386/i386.c (ix86_target_string): Add missing options
41092 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41093 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41094 flags_other and ix86_target_other to flags2_other. Display unknown
41096 (ix86_valid_target_attribute_inner_p): Add missing options and
41097 reorder options by implied ISAs, as in ix86_target_string.
41099 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41101 * hash-table.h (hash_table::too_empty_p): New function.
41102 (hash_table::expand): Use it.
41103 (hash_table::traverse): Likewise.
41104 (hash_table::empty_slot): Use sizeof (value_type) instead of
41105 sizeof (PTR) to convert bytes to elements. Shrink the table
41106 if the current size is excessive for the current number of
41109 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41111 * ira-costs.c (record_reg_classes): Break from the inner loop
41112 early once alt_fail is known to be true. Update outer loop
41113 handling accordingly.
41115 2017-01-13 Jeff Law <law@redhat.com>
41117 * tree-ssa-dse.c (decrement_count): New function.
41118 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41119 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41120 when we know the partially dead statement is a mem* function.
41122 PR tree-optimization/61912
41123 PR tree-optimization/77485
41124 * tree-ssa-dse.c: Include expr.h.
41125 (maybe_trim_constructor_store): New function.
41126 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41128 PR tree-optimization/33562
41129 PR tree-optimization/61912
41130 PR tree-optimization/77485
41131 * doc/invoke.texi: Document new dse-max-object-size param.
41132 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41133 * tree-ssa-dse.c: Include params.h.
41134 (dse_store_status): New enum.
41135 (initialize_ao_ref_for_dse): New, partially extracted from
41137 (valid_ao_ref_for_dse, normalize_ref): New.
41138 (setup_live_bytes_from_ref, compute_trims): Likewise.
41139 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41140 (maybe_trim_partially_dead_store): Likewise.
41141 (maybe_trim_complex_store): Likewise.
41142 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41143 Track what bytes live from the original store. Return tri-state
41144 for dead, partially dead or live.
41145 (dse_dom_walker): Add constructor, destructor and new private members.
41146 (delete_dead_call, delete_dead_assignment): New extracted from
41148 (dse_optimize_stmt): Make a member of dse_dom_walker.
41149 Use initialize_ao_ref_for_dse.
41151 PR tree-optimization/33562
41152 PR tree-optimization/61912
41153 PR tree-optimization/77485
41154 * sbitmap.h (bitmap_count_bits): Prototype.
41155 (bitmap_clear_range, bitmap_set_range): Likewise.
41156 * sbitmap.c (bitmap_clear_range): New function.
41157 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41159 2017-01-13 Martin Liska <mliska@suse.cz>
41162 * function.c (set_cfun): Add new argument force.
41163 * function.h (set_cfun): Likewise.
41164 * ipa-inline-transform.c (inline_call): Use the function when
41165 strict alising from is dropped for function we inline to.
41167 2017-01-13 Richard Biener <rguenther@suse.de>
41169 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41170 for dumping GIMPLE INTEGER_CSTs.
41172 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41174 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41175 to 201112L since C++17.
41177 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41180 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41181 if -fsanitize=kernel-address is present.
41183 2017-01-13 Richard Biener <rguenther@suse.de>
41185 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41186 as _Literal ( type ) number in case usual suffixes do not
41187 preserve all information.
41189 2017-01-13 Richard Biener <rguenther@suse.de>
41191 PR tree-optimization/77283
41192 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41193 and ssa-iterators.h.
41194 (is_feasible_trace): Implement a cost model based on joiner
41197 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41200 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41201 char or short to __float128/_Float128 directly.
41203 2017-01-12 Martin Sebor <msebor@redhat.com>
41205 to -Wformat-overflow.
41206 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41207 (min_bytes_remaining): Same.
41208 (get_string_length): Same.
41209 (format_string): Same.
41210 (format_directive): Same.
41212 (pass_sprintf_length::handle_gimple_call): Same.
41214 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41216 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41217 info.nowrite calls with no lhs that can't throw. Return bool
41218 whether gsi_remove has been called or not.
41219 (pass_sprintf_length::handle_gimple_call): Return bool whether
41220 try_substitute_return_value called gsi_remove. Formatting fix.
41221 (pass_sprintf_length::execute): Don't use gsi_remove if
41222 handle_gimple_call returned true.
41225 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41226 be removed due to side-effects, don't remove following barrier nor
41227 turn the successor edge into fallthru edge.
41229 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41232 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41233 element-reversing loads and stores as not swappable.
41235 2017-01-12 Nathan Sidwell <nathan@acm.org>
41236 Nicolai Stange <nicstange@gmail.com>
41238 * combine.c (try_combine): Don't ignore result of overlap checking
41239 loop. Combine overlap & asm check into single loop.
41241 2017-01-12 Richard Biener <rguenther@suse.de>
41243 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41244 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41246 2017-01-12 Richard Biener <rguenther@suse.de>
41248 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41249 and TS_TARGET_OPTION directly derive from TS_BASE.
41250 * tree-core.h (tree_optimization_option): Derive from tree_base.
41251 (tree_target_option): Likewise.
41253 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41255 * config/i386/i386.c (memory_address_length): Increase len
41256 only when rip_relative_addr_p returns false.
41258 2017-01-11 Julia Koval <julia.koval@intel.com>
41260 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41261 (OPTION_MASK_ISA_SGX_SET): New.
41262 (ix86_handle_option): Handle OPT_msgx.
41263 * config.gcc: Added sgxintrin.h.
41264 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41265 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41266 * config/i386/i386.c (ix86_target_string): Add -msgx.
41268 (ix86_option_override_internal): Handle new options.
41269 (ix86_valid_target_attribute_inner_p): Add sgx.
41270 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41271 * config/i386/i386.opt: Add msgx.
41272 * config/i386/sgxintrin.h: New file.
41273 * config/i386/x86intrin.h: Add sgxintrin.h.
41275 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41278 * fold-const.c (maybe_nonzero_address): Return 1 for function
41280 (tree_single_nonzero_warnv_p): Don't handle function local objects
41284 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41287 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41290 * opts.c: Include "spellcheck.h"
41291 (struct string_fragment): New struct.
41292 (struct edit_distance_traits<const string_fragment &>): New
41294 (get_closest_sanitizer_option): New function.
41295 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41297 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41299 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41301 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41302 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41303 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41304 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41305 for initial die_offset if dwarf_split_debug_info.
41306 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41307 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41309 (output_skeleton_debug_sections): Formatting fix. Use
41310 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41311 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41313 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41315 * config/arm/cortex-a53.md: Add bypasses for
41316 cortex_a53_r2f_cvt.
41317 (cortex_a53_r2f): Only use for transfers.
41318 (cortex_a53_f2r): Likewise.
41319 (cortex_a53_r2f_cvt): Add reservation for conversions.
41320 (cortex_a53_f2r_cvt): Likewise.
41322 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41324 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41325 to all inlined functions, change static to extern.
41327 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41330 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41332 (arm_assemble_integer): Likewise.
41334 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41336 * config.gcc: Use new awk script to check CPU, FPU and architecture
41337 parameters for --with-... options.
41338 * config/arm/parsecpu.awk: New file
41339 * config/arm/arm-cpus.in: New file.
41340 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41342 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41344 * config/arm/t-arm: Update dependency rules.
41345 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41346 of processing .def files.
41347 * config/arm/genopt.sh: Deleted.
41348 * config/arm/gentune.sh: Deleted.
41349 * config/arm/arm-cores.def: Deleted.
41350 * config/arm/arm-arches.def: Deleted.
41351 * config/arm/arm-fpus.def: Deleted.
41352 * config/arm/arm-tune.md: Regenerated.
41353 * config/arm/arm-tables.opt: Regenerated.
41354 * config/arm/arm-cpu.h: New generated file.
41355 * config/arm/arm-cpu-data.h: New generated file.
41356 * config/arm/arm-cpu-cdata.h: New generated file.
41358 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41361 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41363 (input_varpool_node): Unpack dynamically_initialized bit.
41365 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41367 PR rtl-optimization/79032
41368 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41369 the alignment of the adjusted memory reference against that of MODE,
41370 instead of the alignment of the original memory reference.
41372 2017-01-11 Martin Jambor <mjambor@suse.cz>
41374 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41376 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41377 decorated functions.
41379 2017-01-11 Richard Biener <rguenther@suse.de>
41381 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41382 set range/nonnull info for PHI results. Do not set it on
41383 stmts marked for removal.
41385 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41387 * expr.c (store_field): In the bitfield case, fetch the return value
41388 from the registers before applying a single big-endian adjustment.
41389 Always do a final load for a BLKmode value not larger than a word.
41391 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41394 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41395 that we correctly handle column numbers greater than
41396 LINE_MAP_MAX_COLUMN_NUMBER.
41398 2017-01-10 Martin Sebor <msebor@redhat.com>
41400 PR middle-end/78245
41401 * gimple-ssa-sprintf.c (get_destination_size): Call
41402 {init,fini}object_sizes.
41403 * tree-object-size.c (addr_object_size): Adjust.
41404 (pass_through_call): Adjust.
41405 (pass_object_sizes::execute): Adjust.
41406 * tree-object-size.h (fini_object_sizes): Declare.
41408 2017-01-10 Martin Sebor <msebor@redhat.com>
41410 PR tree-optimization/78775
41411 * builtins.c (get_size_range): Move...
41412 * calls.c: ...to here.
41413 (alloc_max_size): Accept zero argument.
41414 (operand_signed_p): Remove.
41415 (maybe_warn_alloc_args_overflow): Call get_size_range.
41416 * calls.h (get_size_range): Declare.
41418 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41420 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41421 from TI's devices.csv file as of September 2016.
41422 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41424 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41426 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41427 * doc/invoke.texi: Likewise.
41428 * doc/md.texi: Likewise.
41429 * doc/objc.texi: Likewise.
41431 2017-01-10 Joshua Conner <joshconner@google.com>
41433 * config/arm/fuchsia-elf.h: New file.
41434 * config/fuchsia.h: New file.
41435 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41436 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41438 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41440 2016-01-10 Richard Biener <rguenther@suse.de>
41442 PR tree-optimization/79034
41443 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41444 Propagate out degenerate PHIs in the joiner.
41446 2017-01-10 Martin Liska <mliska@suse.cz>
41448 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41449 (sort_congruence_classes_by_decl_uid): Likewise.
41450 (sort_congruence_class_groups_by_decl_uid): Likewise.
41451 (sem_item_optimizer::merge_classes): Sort class, groups in these
41452 classes and members in the groups by DECL_UID of declarations.
41453 This would make merge operations stable.
41455 2017-01-10 Martin Liska <mliska@suse.cz>
41457 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41458 usage of m_classes_vec.
41459 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41460 (sem_item_optimizer::get_group_by_hash): Likewise.
41461 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41462 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41463 (sem_item_optimizer::verify_classes): Likewise.
41464 (sem_item_optimizer::process_cong_reduction): Likewise.
41465 (sem_item_optimizer::dump_cong_classes): Likewise.
41466 (sem_item_optimizer::merge_classes): Likewise.
41467 * ipa-icf.h (congruence_class_hash): Rename from
41468 congruence_class_group_hash. Remove declaration of m_classes_vec.
41470 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41472 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41473 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41474 * config.gcc: Add avx512vpopcntdqintrin.h.
41475 * config/i386/avx512vpopcntdqintrin.h: New.
41476 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41477 * config/i386/i386-builtin-types.def: Add new types.
41478 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41479 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41480 __builtin_ia32_vpopcountq_v8di_mask): New.
41481 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41482 __AVX512VPOPCNTDQ__.
41483 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41484 (PTA_AVX512VPOPCNTDQ): Define.
41485 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41486 TARGET_AVX512VPOPCNTDQ_P): Define.
41487 * config/i386/i386.opt: Add mavx512vpopcntdq.
41488 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41489 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41491 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41493 PR middle-end/77484
41494 * predict.def (PRED_CALL): Set to 67.
41496 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41498 * expr.c (store_field): In the bitfield case, if the value comes from
41499 a function call and is of an aggregate type returned in registers, do
41500 not modify the field mode; extract the value in all cases if the mode
41501 is BLKmode and the size is not larger than a word.
41503 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41506 * config/i386/cpuid.h: Fix undefined behavior.
41508 2017-01-04 Jeff Law <law@redhat.com>
41510 PR tree-optimization/79007
41511 PR tree-optimization/67955
41512 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41513 conservative for pt.null when flag_non_call_exceptions is on.
41515 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41517 PR translation/79019
41518 PR translation/79020
41519 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41520 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41521 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41523 * config/avr/avr.opt (maccumulate-args): Likewise.
41524 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41525 * common.opt (freport-bug): Likewise.
41526 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41527 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41528 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41529 translatable string.
41530 * config/i386/i386.c (function_value_32): Likewise.
41531 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41532 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41534 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41535 * common/config/msp430/msp430-common.c (msp430_handle_option):
41537 * symtab.c (symtab_node::verify_base): Likewise.
41538 * opts.c (set_debug_level): Likewise.
41539 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41540 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41541 missing whitespace to translatable strings.
41542 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41543 * config/sh/superh.h: Likewise.
41544 * config/i386/xopintrin.h: Likewise.
41545 * config/i386/znver1.md: Likewise.
41546 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41547 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41548 * double-int.h (struct double_int): Likewise.
41549 * double-int.c (div_and_round_double): Likewise.
41550 * wide-int.cc: Likewise.
41551 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41552 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41553 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41554 (crossjumps_occurred): ... this.
41555 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41558 PR tree-optimization/78899
41559 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41560 returning bool return struct loop *, NULL for failure and the new
41562 (versionable_outer_loop_p): Don't version outer loop if it has
41563 dont_vectorized bit set.
41564 (tree_if_conversion): When versioning outer loop, ensure
41565 tree_if_conversion is performed also on the inner loop of the
41566 non-vectorizable outer loop copy.
41567 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41568 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41569 prevent vectorization of it.
41570 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41571 the outer loop vectorization of the non-scalar version is attempted
41572 before vectorization of the inner loop in scalar version. If
41573 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41574 vectorization of its inner loop.
41575 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41576 has 2 inner loops, rename also on edges from bb whose single pred
41577 is outer_loop->header. Fix typo in function comment.
41579 2017-01-09 Martin Sebor <msebor@redhat.com>
41582 * asan.c (asan_emit_stack_protection): Increase local buffer size
41583 to avoid snprintf truncation warning.
41585 2017-01-09 Andrew Pinski <apinski@cavium.com>
41587 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41588 to reference thunderx2t99 for the tuning structure
41589 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41591 (thunderx2t99_extra_costs): This.
41592 * config/aarch64/aarch64-tune.md: Regenerate.
41593 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41594 (vulcan_addrcost_table): This.
41595 (vulcan_regmove_cost): Rename to ...
41596 (thunderx2t99_regmove_cost): This.
41597 (vulcan_vector_cost): Rename to ...
41598 (thunderx2t99_vector_cost): this.
41599 (vulcan_branch_cost): Rename to ...
41600 (thunderx2t99_branch_cost): This.
41601 (vulcan_tunings): Rename to ...
41602 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41603 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41605 2017-01-09 Martin Jambor <mjambor@suse.cz>
41609 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41610 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41611 (propagate_vr_accross_jump_function): Use the above function for all
41612 value range computations for pass-through jump functions and type
41613 converasion from explicit value range values.
41614 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41615 parameters from TYPE_ARG_TYPES.
41616 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41617 (ipa_write_node_info): Stream type of the actual argument.
41618 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41620 2017-01-09 Martin Liska <mliska@suse.cz>
41623 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41624 (lookup_compiler): Do not show error message with have_E.
41626 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41628 PR tree-optimization/78938
41629 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41630 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41631 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41632 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41635 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41637 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41640 2017-01-09 Richard Biener <rguenther@suse.de>
41642 PR tree-optimization/78997
41643 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41644 name condition properly.
41646 2017-01-09 Richard Biener <rguenther@suse.de>
41649 * dwarf2out.c (is_cxx): New overload with context.
41650 (is_naming_typedef_decl): Use it.
41652 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41654 * invoke.texi (Option Summary): Correct spacing in option lists
41655 and add line breaks to fix over-long lines.
41657 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41659 PR middle-end/17660
41661 * extend.texi (Common Variable Attributes): Add xref to GCC
41662 Internals manual to explain mode attribute keywords.
41664 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41667 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41668 and Preprocessor Options.
41669 (Options for Linking): Document -pthread here....
41670 (RS/6000 and PowerPC Options): ...not here.
41671 (Solaris 2 Options): ...or here.
41672 * doc/cppopts.texi: Document -pthread.
41674 2017-01-08 Martin Sebor <msebor@redhat.com>
41676 PR middle-end/77708
41677 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41678 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41679 New member functions.
41680 (format_directive): Used them.
41682 (pass_sprintf_length::handle_gimple_call): Same.
41683 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41684 to avoid truncation for any argument.
41685 (extract_affine_mul): Same.
41686 * tree.c (get_file_function_name): Same.
41688 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41690 PR middle-end/77484
41691 * predict.def (PRED_INDIR_CALL): Set to 86.
41693 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41695 PR preprocessor/54124
41696 * doc/cppopts.texi: Reformat -d subtable to list the full name
41697 of the options. Add cross-reference to the docs for the general
41698 compiler -d options.
41699 * doc/invoke.texi (Developer Options): Add cross-reference to the
41700 preprocessor-specific -d option documentation.
41702 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41704 PR preprocessor/13498
41705 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41706 redudant material, and reflect new command-line options.
41707 (System Headers): Likewise.
41709 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41711 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41712 -isystem, and -idirafter. Copy-edit.
41713 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41714 default for -ftrack-macro-expansion. Delete obsolete and
41715 badly-formatted implementation details about -fdebug-cpp output.
41716 * doc/cppwarnopts.texi: Copy-edit.
41718 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41721 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41722 that the transition from a max line width >= 1<<10 to narrower
41723 lines works correctly.
41725 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41727 * doc/options.texi (PerFunction): New.
41728 * opt-functions.awk (switch_flags): Map both Optimization and
41729 PerFunction to CL_OPTIMIZATION.
41730 * opth-gen.awk: Test for PerFunction flag along with
41732 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41733 it only when the latter is present. Skip those that don't in
41734 the hash function generator.
41735 * common.opt (fvar-tracking): Mark as PerFunction instead of
41737 (fvar-tracking-assignments): Likewise.
41738 (fvar-tracking-assignments-toggle): Likewise.
41739 (fvar-tracking-uninit): Likewise.
41741 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41743 PR translation/79018
41744 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41747 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41750 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41751 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41752 TARGET_LONG_JUMP_TABLE_OFFSETS.
41753 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41754 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41755 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41756 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41757 * config/m68k/m68k.md (tablejump expander): Likewise.
41758 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41759 TARGET_LONG_JUMP_TABLE_OFFSETS.
41760 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41761 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41763 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41764 David Holsgrove <david.holsgrove@xilinx.com>
41766 * common/config/microblaze/microblaze-common.c
41767 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41768 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41770 * config/microblaze/microblaze.c (microblaze_must_save_register)
41771 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41773 (microblaze_eh_return): New function.
41774 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41775 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41776 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41777 * config/microblaze/microblaze.md (eh_return): New pattern.
41779 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41781 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41782 GCC_DIAGNOSTIC_STRINGIFY): Define.
41784 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41786 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41788 * config/arm/arm.md (<mcrr>): New.
41790 * config/arm/arm.c (arm_arch5te): New.
41791 (arm_option_override): Set arm_arch5te.
41792 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41794 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41795 (arm_mcrr_qualifiers): ... this. New.
41796 (MRRC_QUALIFIERS): Define to...
41797 (arm_mrrc_qualifiers): ... this. New.
41798 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41800 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41801 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41802 (MRRCI, mrrc, MRRC): New.
41803 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41804 VUNSPEC_MRRC2): New.
41806 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41808 * config/arm/arm.md (<mcr>): New.
41810 * config/arm/arm.c (arm_coproc_builtin_available): Add
41811 support for mcr, mrc, mcr2 and mrc2.
41812 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41813 (arm_mcr_qualifiers): ... this. New.
41814 (MRC_QUALIFIERS): Define to ...
41815 (arm_mrc_qualifiers): ... this. New.
41816 (MCR_QUALIFIERS): Define to ...
41817 (arm_mcr_qualifiers): ... this. New.
41818 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41820 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41821 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41822 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41823 VUNSPEC_MRC2): New.
41825 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41827 * config/arm/arm.md (*ldc): New.
41831 * config/arm/arm.c (arm_coproc_builtin_available): Add
41832 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41833 (arm_coproc_ldc_stc_legitimate_address): New.
41834 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41835 'qualifier_const_pointer'.
41836 (LDC_QUALIFIERS): Define to...
41837 (arm_ldc_qualifiers): ... this. New.
41838 (STC_QUALIFIERS): Define to...
41839 (arm_stc_qualifiers): ... this. New.
41840 * config/arm/arm-protos.h
41841 (arm_coproc_ldc_stc_legitimate_address): New.
41842 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41843 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41844 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41845 stc2, stcl, stc2l): New.
41846 * config/arm/constraints.md (Uz): New.
41847 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41848 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41849 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41850 VUNSPEC_STC2L): New.
41852 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41854 * config/arm/arm.md (<cdp>): New.
41855 * config/arm/arm.c (neon_const_bounds): Rename this ...
41856 (arm_const_bounds): ... this.
41857 (arm_coproc_builtin_available): New.
41858 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41859 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41860 (CDP_QUALIFIERS): Define to...
41861 (arm_cdp_qualifiers): ... this. New.
41863 (arm_expand_builtin_args): Add case for 6 arguments.
41864 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41865 (arm_const_bounds): ... this.
41866 (arm_coproc_builtin_available): New.
41867 * config/arm/arm_acle.h (__arm_cdp): New.
41869 * config/arm/arm_acle_builtins.def (cdp): New.
41871 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41872 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41873 'arm_const_bounds'.
41874 * config/arm/types.md (coproc): New.
41875 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41876 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41877 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41878 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41880 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41882 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41883 (UBINOP_QUALIFIERS): New.
41885 (acle_builtin_data): New. Change comment.
41886 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41887 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41888 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41889 arm_acle_builtins.def.
41890 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41891 (arm_init_acle_builtins): New.
41892 (CRC32_BUILTIN): Remove.
41893 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41894 crc32cb, crc32ch and crc32cw.
41895 (arm_init_crc32_builtins): Remove.
41896 (arm_init_builtins): Use arm_init_acle_builtins rather
41897 than arm_init_crc32_builtins.
41898 (arm_expand_acle_builtin): New.
41899 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41900 * config/arm/arm_acle_builtins.def: New.
41902 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41904 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41905 (arm_builtin_datum): ... this.
41906 (arm_init_neon_builtin): Rename to ...
41907 (arm_init_builtin): ... this. Add a new parameters PREFIX
41908 and USE_SIG_IN_NAME.
41909 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41910 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41911 'arm_builtin_datum'.
41912 (arm_init_vfp_builtins): Likewise.
41913 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41914 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41915 (arm_expand_neon_args): Rename to ...
41916 (arm_expand_builtin_args): ... this. Rename builtin_arg
41917 enum values and differentiate between ARG_BUILTIN_MEMORY
41918 and ARG_BUILTIN_NEON_MEMORY.
41919 (arm_expand_neon_builtin_1): Rename to ...
41920 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41921 values, arm_expand_builtin_args and add bool parameter NEON.
41922 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41923 (arm_expand_vfp_builtin): Likewise.
41924 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41926 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41928 PR middle-end/77484
41929 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41930 * predict.c (tree_estimate_probability_bb): Reverse direction of
41931 polymorphic call predictor.
41933 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41935 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41936 (determine_pass_name_match): ...this new function and...
41937 (should_skip_pass_p): ...this new function.
41939 2017-01-06 Nathan Sidwell <nathan@acm.org>
41941 * ipa-visibility.c (function_and_variable_visibility): Reformat
41942 comments and long lines. Remove extrneous if.
41943 * symtab.c (symtab_node::make_decl_local): Fix code format.
41944 (symtab_node::set_section_for_node): Fix comment typo.
41946 2017-01-06 Martin Liska <mliska@suse.cz>
41949 * lra-constraints.c: Rename invariant to lra_invariant.
41950 * predict.c (set_even_probabilities): Initialize e to NULL.
41952 2017-01-05 Martin Sebor <msebor@redhat.com>
41954 PR tree-optimization/78910
41955 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41956 (format_integer): Correct off-by-one error in the handling
41957 of precision with negative numbers in signed conversions..
41959 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41961 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41963 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41965 PR tree-optimization/71016
41966 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41967 factor_out_conditional_conversion. Formatting fix.
41968 (factor_out_conditional_conversion): Add cond_stmt argument.
41969 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41970 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41973 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41975 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41976 read-rtl-function.o, and selftest-rtl.o.
41977 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41978 (selftest::aarch64_test_loading_full_dump): New function.
41979 (selftest::aarch64_run_selftests): New function.
41980 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41981 selftest::aarch64_run_selftests.
41982 * config/i386/i386.c
41983 (selftest::ix86_test_loading_dump_fragment_1): New function.
41984 (selftest::ix86_test_loading_call_insn): New function.
41985 (selftest::ix86_test_loading_full_dump): New function.
41986 (selftest::ix86_test_loading_unspec): New function.
41987 (selftest::ix86_run_selftests): Call the new functions.
41988 * emit-rtl.c (maybe_set_max_label_num): New function.
41989 * emit-rtl.h (maybe_set_max_label_num): New decl.
41990 * function.c (instantiate_decls): Guard call to
41991 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41992 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41994 * gensupport.c (gen_reader::gen_reader): Pass "false"
41995 for new "compact" param of rtx_reader.
41996 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41997 rather than an empty string for NULL strings.
41998 * read-md.c: Potentially include config.h rather than bconfig.h.
41999 Wrap include of errors.h with #ifdef GENERATOR_FILE.
42000 (have_error): New global, copied from errors.c.
42001 (md_reader::read_name): Rename to...
42002 (md_reader::read_name_1): ...this, adding "out_loc" param,
42003 and converting "missing name or number" to returning false, rather
42005 (md_reader::read_name): Reimplement in terms of read_name_1.
42006 (md_reader::read_name_or_nil): New function.
42007 (md_reader::read_string): Handle "(nil)" by returning NULL.
42008 (md_reader::md_reader): Add new param "compact".
42009 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
42010 (md_reader::read_file): New method.
42011 * read-md.h (md_reader::md_reader): Add new param "compact".
42012 (md_reader::read_file): New method.
42013 (md_reader::is_compact): New accessor.
42014 (md_reader::read_name): Convert return type from void to file_location.
42015 (md_reader::read_name_or_nil): New decl.
42016 (md_reader::read_name_1): New decl.
42017 (md_reader::m_compact): New field.
42018 (noop_reader::noop_reader): Pass "false" for new "compact" param
42020 (rtx_reader::rtx_reader): Add new "compact" param.
42021 (rtx_reader::read_rtx_operand): Make virtual and convert return
42022 type from void to rtx.
42023 (rtx_reader::read_until): New decl.
42024 (rtx_reader::handle_any_trailing_information): New virtual function.
42025 (rtx_reader::postprocess): New virtual function.
42026 (rtx_reader::finalize_string): New virtual function.
42027 (rtx_reader::m_in_call_function_usage): New field.
42028 (rtx_reader::m_reuse_rtx_by_id): New field.
42029 * read-rtl-function.c: New file.
42030 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
42031 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
42032 (selftest::verify_three_block_rtl_cfg): New decl.
42033 * read-rtl-function.h: New file.
42034 * read-rtl.c: Potentially include config.h rather than bconfig.h.
42035 For host, include function.h, memmodel.h, and emit-rtl.h.
42036 (one_time_initialization): New function.
42037 (struct compact_insn_name): New struct.
42038 (compact_insn_names): New array.
42039 (find_code): Handle insn codes in compact dumps.
42040 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
42041 (bind_subst_iter_and_attr): Likewise.
42042 (add_condition_to_string): Likewise.
42043 (add_condition_to_rtx): Likewise.
42044 (apply_attribute_uses): Likewise.
42045 (add_current_iterators): Likewise.
42046 (apply_iterators): Likewise.
42047 (initialize_iterators): Guard usage of apply_subst_iterator with
42048 #ifdef GENERATOR_FILE.
42049 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
42050 (md_reader::read_mapping): Likewise.
42051 (add_define_attr_for_define_subst): Likewise.
42052 (add_define_subst_attr): Likewise.
42053 (read_subst_mapping): Likewise.
42054 (check_code_iterator): Likewise.
42055 (rtx_reader::read_rtx): Likewise. Move one-time initialization
42057 (one_time_initialization): New function.
42058 (rtx_reader::read_until): New method.
42059 (read_flags): New function.
42060 (parse_reg_note_name): New function.
42061 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42062 Handle reuse_rtx ids.
42063 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42064 Add parsing support for RTL dumps, mirroring the special-cases in
42065 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42066 values, and calling handle_any_trailing_information.
42067 (rtx_reader::read_rtx_operand): Convert return type from void
42068 to rtx, returning return_rtx. Handle case 'e'. Call
42069 finalize_string on XSTR and XTMPL fields.
42070 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42071 "(nil)" values were omitted. Call the postprocess vfunc on the
42073 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42074 class ctor. Initialize m_in_call_function_usage. Call
42075 one_time_initialization.
42076 * rtl-tests.c (selftest::test_uncond_jump): Call
42077 set_new_first_and_last_insn.
42078 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42079 * selftest-rtl.c: New file.
42080 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42081 (selftest::get_insn_by_uid): New decl.
42082 * selftest-run-tests.c (selftest::run_tests): Call
42083 read_rtl_function_c_tests.
42084 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42085 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42088 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42090 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42091 operands in a special way. Assert that pos+len <= mode precision.
42093 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42095 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42096 3 argument Alias with unlimited for the negative form.
42097 (fno-vect-cost-model): Removed.
42099 2017-01-05 Martin Liska <mliska@suse.cz>
42101 * hsa-gen.c (gen_hsa_divmod): New function.
42102 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42104 2017-01-05 Martin Liska <mliska@suse.cz>
42107 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42110 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42112 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42113 small constant length operands.
42115 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42117 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42118 between loop iterations.
42120 2017-01-05 Martin Liska <mliska@suse.cz>
42123 * gimplify.c (gimplify_decl_expr): Compare to
42124 asan_poisoned_variables instread of checking flags.
42125 (gimplify_target_expr): Likewise.
42126 (gimplify_expr): Likewise.
42127 (gimplify_function_tree): Conditionally initialize
42128 asan_poisoned_variables.
42130 2017-01-04 Jeff Law <law@redhat.com>
42132 PR tree-optimizatin/78812
42133 * rtl.h (contains_mem_rtx_p): Prototype.
42134 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42135 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42136 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42137 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42138 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42140 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42142 * input.c (assert_char_at_range): Default-initialize actual_range.
42144 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42146 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42147 to match the caller.
42149 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42151 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42152 insns after final jump in test to emit dummy move.
42154 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42156 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42157 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42159 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42161 * multiple_target.c (create_dispatcher_calls): Init e_next.
42162 * tree-ssa-loop-split.c (split_loop): Init border.
42163 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42166 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42171 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42172 (altivec_register_operand): Do not return true if the operand
42173 contains a SUBREG mixing SImode and SFmode.
42174 (vsx_register_operand): Likewise.
42175 (vsx_reg_sfsubreg_ok): New predicate.
42176 (vfloat_operand): Do not return true if the operand contains a
42177 SUBREG mixing SImode and SFmode.
42178 (vint_operand): Likewise.
42179 (vlogical_operand): Likewise.
42180 (gpc_reg_operand): Likewise.
42181 (int_reg_operand): Likewise.
42182 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42183 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42184 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42186 (rs6000_emit_move_si_sf_subreg): New helper function.
42187 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42188 fixup SUBREGs involving SImode and SFmode.
42189 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42190 numbers for the new peephole2 optimization.
42191 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42192 the GLIBC math library that do AND/IOR/XOR operations on single
42193 precision floating point.
42194 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42195 target macros to say whether we need to avoid SUBREGs mixing
42197 (TARGET_ALLOW_SF_SUBREG): Likewise.
42198 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42199 (UNSPEC_SI_FROM_SF): Likewise.
42200 (iorxor): Change spacing.
42201 (and_ior_xor): New iterator for AND, IOR, and XOR.
42202 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42203 (movdi_from_sf_zero_ext): Likewise.
42204 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42205 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42206 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42207 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42208 (fms<mode>4): Likewise.
42209 (fnma<mode>4): Likewise.
42210 (fnms<mode>4): Likewise.
42211 (nfma<mode>4): Likewise.
42212 (nfms<mode>4): Likewise.
42214 2017-01-04 Marek Polacek <polacek@redhat.com>
42217 * doc/invoke.texi: Document -Wpointer-compare.
42219 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42221 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42224 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42225 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42228 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42231 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42232 documentation of the powerpc_popcntb_ok attribute.
42233 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42234 code to issue warning messages if a requested CPU configuration is
42235 not supported by the binary (assembler and loader) toolchain.
42236 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42237 made to define a built-in function that has been disabled.
42238 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42239 made to define a built-in function that has been disabled.
42240 (altivec_init_builtins): Add comment explaining why definition
42241 of the DST built-in functions is not preceded by an assertion
42242 check. Add assertions to prevent ICE if attempts are made to
42243 define an altivec predicate or an abs* built-in function that has
42245 (htm_init_builtins): Add comment explaining why definition of the
42246 htm built-in functions is not preceded by an assertion check.
42248 2017-01-04 Jeff Law <law@redhat.com>
42250 PR tree-optimizatin/67955
42251 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42252 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42253 the points-to solution does not include pt_null. Use DECL_PT_UID
42256 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42258 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42259 Use gen_int_mode instead of gen_lopwart for const_int operands.
42261 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42263 PR tree-optimization/71563
42264 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42265 out of range value - has low bits known to be zero.
42267 2017-01-04 Alan Modra <amodra@gmail.com>
42269 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42270 * configure: Regenerate.
42271 * config.in: Regenerate.
42273 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42276 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42277 a substring of the message, but strcmp with the whole message. Ifdef
42278 ENABLE_NLS, translate the message first using dgettext.
42280 2017-01-03 Jeff Law <law@redhat.com>
42282 PR tree-optimizatin/78856
42283 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42284 (mark_threaded_blocks): Remove code to truncate thread paths that
42285 cross multiple loop headers. Instead invalidate the cached loop
42286 iteration information and handle case of a thread path walking
42287 into an irreducible region.
42289 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42292 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42293 assertions. Add support for doing the signbit if the IEEE 128-bit
42294 floating point value is in a GPR.
42295 * config/rs6000/rs6000.md (Fsignbit): Delete.
42296 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42297 Update the length attribute if the value is in a GPR.
42298 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42299 the sign or zero extension instruction, since the value is always 0/1.
42300 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42303 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42304 extracting SImode to a GPR register so that we can generate a
42305 store, limit the vector to be in a traditional Altivec register
42306 for the vextuwrx instruction.
42308 2017-01-03 Ian Lance Taylor <iant@google.com>
42310 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42312 2017-01-03 Martin Sebor <msebor@redhat.com>
42314 PR tree-optimization/78696
42315 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42316 precision. Use MPFR for %f for greater fidelity. Correct handling
42318 (pass_sprintf_length::compute_format_length): Set width and precision
42319 specified by asrerisk to void_node for vararg functions.
42320 (try_substitute_return_value): Adjust dump output.
42322 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42324 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42326 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42328 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42329 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42330 -mlra/-mno-lra was passed to the compiler.
42332 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42334 PR rtl-optimization/65618
42335 * emit-rtl.c (try_split): Move initialization of "before" and
42336 "after" to just before the call to emit_insn_after_setloc.
42338 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42340 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42342 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42344 * dwarf2out.c (gen_enumeration_type_die): When
42345 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42347 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42349 PR tree-optimization/78965
42350 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42351 Change first argument from const call_info & to call_info &. For %n
42352 set info.nowrite to false.
42354 PR middle-end/78901
42355 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42356 possibly throwing calls.
42358 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42359 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42360 and fns handling, rather than in a separate case SSA_NAME.
42362 2017-01-02 Jeff Law <law@redhat.com>
42364 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42365 fixes for first_period and second_period variables.
42367 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42370 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42371 (*insvqi_1): New insn pattern.
42372 (*insvqi_1_mem_rex64): Ditto.
42373 (*insvqi_2): Ditto.
42374 (*insvqi_3): Rename from *insvqi.
42376 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42378 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42380 * doc/cfg.texi (Edges): Remove reference to Java.
42381 (Maintaining the CFG): Ditto.
42383 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42385 PR middle-end/77674
42386 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42387 transparent aliases.
42389 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42391 PR middle-end/77484
42392 * predict.def (PRED_CALL): Update hitrate.
42393 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42394 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42395 into direct/indirect/polymorphic variants.
42397 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42399 Update copyright years.
42401 * gcc.c (process_command): Update copyright notice dates.
42402 * gcov-dump.c (print_version): Ditto.
42403 * gcov.c (print_version): Ditto.
42404 * gcov-tool.c (print_version): Ditto.
42405 * gengtype.c (create_file): Ditto.
42406 * doc/cpp.texi: Bump @copying's copyright year.
42407 * doc/cppinternals.texi: Ditto.
42408 * doc/gcc.texi: Ditto.
42409 * doc/gccint.texi: Ditto.
42410 * doc/gcov.texi: Ditto.
42411 * doc/install.texi: Ditto.
42412 * doc/invoke.texi: Ditto.
42414 Copyright (C) 2017 Free Software Foundation, Inc.
42416 Copying and distribution of this file, with or without modification,
42417 are permitted in any medium without royalty provided the copyright
42418 notice and this notice are preserved.